University of Southern California Dissertations and Theses

Using network perspective to examine the organization of community -based elder care systems across four communities

(USC Thesis Other)

Using network perspective to examine the organization of community -based elder care systems across four communities

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. USING NETWORK PERSPECTIVE TO EXAMINE THE ORGANIZATION OF COMMUNITY-BASED ELDER CARE SYSTEMS ACROSS FOUR COMMUNITIES by Judy Yun Yip A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Gerontology and Public Policy) May 2000 Copyright 2000 Judy Yun Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3018046 _ _ ® UMI UMI Microform 3018046 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90007 This dissertation, written by .................................................. under the direction of h.er. Dissertation Committee, and approved by all its mem bers, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean of Graduate Studies Date DISSERTATION COMMITTEE Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Judy Yun Yip Kathleen H. Wilber Using a Network Perspective to Examine the Organization of Community-Based Elder Care Systems Across Four Communities Despite various efforts over the past three decades, how to organize effective eider care systems remains an enigma. This dissertation describes how elderly care systems are organized and examines interorganizational factors associated with the organization. Using a network perspective, elder care systems are perceived as networks of various resources exchanged among community service providers. Four aspects of elder care systems are described and compared across four communities in California: network development, formalization, structure underlying these systems, and their community-specificity. Multiple regression quadratic assignment procedure (MRQAP) is used to examine the interorganizational factors associated with the organization of elder care systems. Results show that across all communities, more service providers participated in client and information exchanges more frequently than money and staff resources. Service providers that send and receive resources simultaneously are more likely to be found in the client and information exchanges than in the money and staff exchanges. The majority of resources exchanges are supported by informal understandings Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rather than by formal verbal or contractual arrangement. Cohesive groups of service providers, found primarily in client and information exchanges, are generally small in size and number. While structurally equivalent subgroups are identified in each community, fine divisions of labor in each “system” are not apparent. MRQAP results show that significant differences are found across communities in factors associated with their elder care systems and the resource dependence perspective seems to be the most significant framework. These findings suggest that elder care systems are complex, with exchange dynamics differing on resources and specific to community. As socially constructed entities, elder care systems are reflective of how local service providers interact with one another and with the state and federal government along the continuum between vertical control and individual flexibility. Their complexity suggest that careful planning ought to be an essential component in system development, yet the emphasis on informal understandings among service providers may create dilemmas for policy makers in “master-planning" the entire system. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS As a Christian, I want to thank God for providing me an opportunity to explore my area of interest in gerontology. Providing me all the means to survive in the United States with grace, He gave me many chances to educate myself and experience Him through my education. Coming to the States, as I learned later, is more than studying and getting an advanced degree. Throughout the years, I know more about myself and about Him. This dissertation would not be completed without the guidance, support, and assistance of my dissertation committee members. As my committee chair, mentor, and friend for three years and, hopefully more years to come, Dr. Kathleen Wilber has been extremely supportive in coaching me to explore the various issues related to the topics of this dissertation research. I am most thankful for her thoughtful guidance in how to go about doing research beyond writing this dissertation. Her support as a valuable friend also sustains me tremendously in time of stress, distress, and desperation. I also want to thank Dr. Robert Myrtle and Dr. David Grazman for their continued guidance in doing high-quality research and their support as academic mentors. There are times when I am confused about my own role as a scholar and their achievements and roles as members of my committee have been a source of encouragement and inspiration to me. My thanks also goes to my family in Hong Kong. I want to thank my mom and my sisters for continued support in prayers and letter writing, especially during the time when my letters to home were almost come to ii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extinction. My thanks also goes to my friends Patrick Burns, Melissa Tabarrah, Susan Stewart, Sascha Enyeart, Hiroshi Ueda, Jeff Hyde, and Chris Kelly. I also want to thanks my brothers (especially c-lie Kelvin Lam) and sisters (especially Felice Ip and 9-ci Emily Wong) in church who have been supporting me through prayers, encouragement, meals, and weekend accommodations. All of you are truly my friends regardless of who I was and whom I become as a result of writing this dissertation. I want to give special thanks to individuals who have always been behind the “screens,” whether it is my all-time favorite computer lab manager Greg Dolniak, or my Chinese mom May Ng in the United States, or my “e-pal” Michael Ebanks in Illinois. You are all significant in the various stages of my dissertation writing. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Acknowledgements.............................................................................................ii List of T a b le s ................................................................................................ vi List of F ig u re s............................................................................................... ix Chapter I: INTRODUCTION Study Purposes and Hypotheses.......................................................... 1 Contribution............................................................................................. 6 Organization of the Dissertation............................................................ 7 Chapter II: BACKGROUND The Challenge of Designing Effective and Efficient Delivery System.................................................................................... 10 Public Efforts in Responding to the Challenge............................... 12 The Concept of Care Systems Re-examined................................. 27 Recent Efforts in Conceptualizing Community-based Care Systems................................................................................ 33 Research Questions and Network Perspective............................. 40 Summary...............................................................................................45 Chapter III: CONCEPTUALIZATION Describing the Organization of Community-based Care Systems..................................................................................... 47 Factors Underlying Network Relationships........................................ 63 Summary........................................................................................... 77 Chapter IV: METHODS Data and Methods................................................................................. 79 Describing the Organization of Care Systems.................................. 87 Examining the Factors Associating with Resource Exchange Patterns...................................................................................... 97 Summary.............................................................................................100 Chapter V: RESULTS Description of the Organization of Community-based Care 102 Factors Associated with Resource Exchange P a ttern s............... 157 Chapter VI: DISCUSSION AND CONCLUSION Discussion on Hypotheses.................................................................182 General Conclusion on the Organization of C are............................200 Integration in the Organization of Care.............................................205 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Future Research.................................................................................210 Limitations............................................................................................212 REFERENCE................................................................................................... 215 APPENDICES: Appendix A: SEED Questionnaire.....................................................229 Appendix B: Correlation T ables........................................................ 234 Appendix C: Network G raphs............................................................243 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES 1. State Approaches to Reform Long-Term Care Systems........................... 23 2. Studies Using Network Analyses to Examine Service System s.............. 49 3. Exchange Items in the SEED Evaluation................................................... 83 4. Service Provider Types and Their Abbreviation for Network Graphs......................................................................................................85 5. Model Perspectives and Their Variables................................................... 98 6. Density and Standard Deviation of Various Resource Networks Across the Selected Communities................................................. 104 7. Network Centralization Index of Three Resource Networks Across the Selected Communities................................................ 105 8. Extent of Formalization.............................................................................. 108 9. QAP Regression Results Between Mode of Formalization and Resource Networks............................................................................. 110 10. Subgroup Cohesion................................................................................ 119 11. CONCOR R esults..................................................................................... 121 12. Structurally Equivalent Groups in San Mateo with Their Relational Patterns in Blockmodels....................................................122 13. Structurally Equivalent Groups in Long Beach with Their Relational Patterns in Blockm odels.................................................. 130 14. Structurally Equivalent Groups in Tulare with Their Relational Patterns in Blockm odels.................................................. 134 15. Structurally Equivalent Groups in San Francisco with Their Relational Patterns in Blockmodels ...................................... 140 16. Which Program Types Are More Likely to be Found Across All the Selected Communities?.............................................. 145 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17. Perception of Inter-organizational Need................................................147 18. Multiplexity of T ie s............................................................................... 148 19. Prominent Brokers (High Indegrees and Outdegrees) by Resources and Communities............................................................150 20. The Importance of Social Services (DPIHAPS) and Senior Centers (SENIORCTR) in the Selected Communities................. 152 21. Prominent Senders (High Outdegrees) by Resources and Communities.................................................................................... 153 22. Prominent Receivers (High Indegrees) by Resources and Communities................................................................................... 154 23. Centrality Correlation Results (Four Communities)............................. 156 24. Bivariate Correlation, San Mateo: Independent Matrices with Dependent M atrices..................................................................... 158 25. Correlation Results Between Dependent Matrices, San M ateo...................................................................................... 159 26. Bivariate Correlation, Long Beach: Independent Matrices with Dependent Matrices...........................................................................160 27. Correlation Results Between Dependent Matrices, Long Beach . . .161 28. Bivariate Correlation, Tulare: Independent Matrices with Dependent M atrices..........................................................................162 29. Correlation Results Between Dependent Matrices, T ula re................ 163 30. Bivariate Correlation, San Francisco: Independent Matrices with Dependent Matrices.....................................................................165 31. Correlation Results Between Dependent Matrices, San Francisco................................................................................. 166 32. QAP Regression Results for San Mateo Care System................... 168 33. QAP Regression Results for Long Beach Care System................. 171 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34. QAP Regression Results for Tulare Care System............................... 174 35. QAP Regression Results for San Francisco Care S ystem .................177 36. Variance Explained Across Communities............................................. 179 37. Factor Comparison Across Communities............................................. 181 38. Summary of Hypotheses, Tests, and Conclusions for Hypotheses...................................................................................... 183 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES 1. Structurally Equivalent Subgroups Illustrated............................................. 58 2. Multiple-factor Model to Examine Resource Exchange P a ttern s 75 3. Illustration on the Relationships Between Density Table, Image Matrix, and Image Graph........................................................................93 4. Cliques...........................................................................................................113 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER I: INTRODUCTION An increasing older population and rapidly rising care costs have made the concern of building and improving coordinated care systems more urgent than ever before. Despite various attempts over the past three decades, the organization of care systems for older adults still is described as fragmented. Using the network perspective, this dissertation has two objectives: first, to describe the organization of community-based care systems for older adults; and second, to examine factors associated with this organization. This dissertation starts by questioning the current conceptualizations of the “care systems,” as these conceptualizations have contributed to abundant discussions and debates on how the care system is fragmented and how integration or coordination could be a potential solution. Through arguing that these conceptualizations are ill-equipped in providing sufficient understanding on current care organization and fragmentation, this dissertation proposes a network perspective to supplement understanding of care systems. Study Purposes and Hypotheses Drawing from service delivery and systems literature, this dissertation’s objective to describe elder care systems is accomplished by examining the patterns of resource exchanges among service providers. Four resources 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (client, information, money, and staff) are examined to understand the complexity of care systems because these resources are likely to perform different functions in the system. This dissertation develops hypotheses addressing the dynamics underlying the exchange patterns across these resources. Hypothesis 1: Patterns of exchanges will differ across the four resources (client, information, money, and staff). Hypothesis 2a: Patterns o f client and information exchanges are more likely to be similar to each other than those o f money and staff exchanges. Hypothesis 2b: Patterns o f money and staff exchanges are more likely to be similar to each other than those of client and information exchanges. Many studies argue that service providers emphasize the interpersonal aspect of long-term care. With the increased division of labor in caring for a population with multiple chronic conditions, the tension service providers face is between maintaining efficiency through formal contracts and retaining the interpersonal aspect of the relationship. What are the roles of formal and informal relationships in such service delivery patterns? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hypothesis 3a: Money and staff exchanges are more likely to be associated with formalized exchanges than are client and information exchanges. Hypothesis 3b: Client and information exchanges are more likely to be associated with informal exchanges than are money and staff exchanges. How elder care systems are organized is influenced by unique community circumstances. In states such as California where efforts in developing system development are diffused, communities have tremendous flexibility in shaping their own care systems for older adults. Given the fact that long-term care delivery is inherently a local matter and resource exchanges between service providers offer important clues as to how care is organized in a community, this dissertation also focuses on four communities and compare their long-term care delivery systems. Hypothesis 4: Organization o f elder care systems will differ across communities. This dissertation develops a model to examine factors associated with these resource exchange patterns. Facing the choice between providing clients with continuity of care and losing organizational autonomy and resources, what drives these service providers to develop and maintain these Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. resource exchanges? At this very local level, is government mandates an effective tool for governing resource exchange (Weiss, 1987; Woodard, 1994)? Borrowing findings from research examining dyadic inter organizational relationship, this study suggests a framework of complementing perspectives that may be at work at the network level in a community. Several perspectives of inter-organizational relations are presented in a model developed by Oliver in an earlier study (Oliver, 1990). This dissertation develops two hypotheses with regard to factors associated with resource exchange patterns. Hypothesis 5: Resource exchange relations are associated with multiple factors rather than dominated by a single factor. Hypothesis 6: Different resource networks are likely to be associated with different sets of factors. Data from four out of thirteen communities are drawn from a larger study on evaluating system development in California. Because few studies examine the organization of community-based care for older adults, this dissertation is descriptive and exploratory in nature. The purpose of describing these patterns helps reveal structures of community-based service delivery. Although the existence of a complex pattern of resource exchange among service providers may not initially suggest the presence of a “system,” the structure of a system can still be uncovered. 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Results from these analyses are helpful in two ways: first, they may suggest a way to examine the process of how these resource exchange patterns occur, which may be entirely dependent on the characteristics of the community. Understanding more about how and why service providers choose to develop a resource exchange pattern with each other provides insight into how possible integration or partnership may be pursued. Second, examining the factors associated with resource exchange patterns helps contribute to the development of theoretical models for evolving community- based systems of care. Network Perspective in Brief A network perspective is useful in both theoretical and practical ways. Theoretically, the network perspective provides an appropriate framework for examining the totality of community-based care systems by focusing on several levels of analysis (providers, groups of providers, or the entire network) and the interactions among these levels. Moreover, using network analysis to understand community-based care systems may enhance our understanding of how organizations respond to system changes and simultaneously shape the structure of the system as these organizations go through their own life cycle. Furthermore, how a particular configuration of community-based care network relations is developed can also be described and examined by this perspective. The examination of these configurations may provide bases for developing efficient and effective systems that reflect Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. social and cultural values in providing health and long-term care in the United States. Practically, a network perspective can help policy makers by suggesting where resources can be allocated according to current network patterns. It can also help individual organizations in positioning themselves in order to gain advantages for their markets. Furthermore, the network perspective helps researchers understand the milieu of aging at the individual level in the landscape of community care services. Contribution The trends towards managed care in both health care and long term care may even make understanding care systems in a systematic way more appealing. If managed care aims at improving service delivery for a population with multiple chronic conditions by functional, clinical and administrative integration (Shortell, Gillies, Anderson, Erickson, & Mitchell, 1996), the transition of care between settings, indicated as likely due to an improved relationship among service providers, should also be improved (Ory, Cooper, & Siu, 1998). By using a network perspective, one may be able to examine the impact of various transitions on care for the long-term care population and identify ways for better transitions in the delivery system. A description of what current service delivery is like provides not only a reference point for design innovation, but also a basis for which effective and strategic intervention can be implemented. Furthermore, this description also informs policy makers and service providers of the degree of discontinuity of 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. care and of the platform on which advocates can generate conviction and compassion to improve long-term care. By examining the service delivery patterns of long-term care, this study may provide a realistic picture of the extent to which the experts’ notion of “system” exists and also may suggest future research to link outcomes to structure so as to improve services for the long-term care population. In a broader perspective, this study argues that policymakers, providers, and the long-term care population can be benefited from the examination of service delivery patterns and structure. Organization of the Dissertation The following five chapters describe the process of researching community-based care systems using a network perspective. The background chapter provides an overall review on efforts of improving care delivery and conceptualizing care systems. A network perspective will be introduced also in the background chapter. Building on the network perspective, the conceptual chapter focuses on laying down a framework to describe and examine the organization of care. Hypotheses are developed for empirical assessment on the nature of care systems using the four communities. The method chapter describes sample and data used for this dissertation. Various methodologies used for network analyses on describing and examining care systems are described. The result chapter presents findings of this dissertation according to the two objectives. Finally, the discussion and conclusion chapter interprets the findings of this dissertation in 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. light of the current development of community-based care and suggests areas for future research. 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER II: BACKGROUND As the older population of the United States grows, policy makers are challenged to enact policies that ensure accessibility, affordability, and quality of care for older adults. Developing comprehensive, effective, and efficient care systems has been a goal for policy makers, service providers, and researchers to rectify the current fragmented service delivery and to respond to the increasing demand of persons with multiple chronic care needs. This chapter points out how important an understanding of “care systems” is in developing relevant care policies for older adults. Reviewing previous efforts of “care systems” development reveals little consensus among policy makers, researchers, and service providers, thus rendering these efforts as chasing the wind. In order to delineate the importance of understanding what “care systems” means, this chapter is divided into four major sections. First, it discusses briefly circumstances leading to the challenge of “care systems” development. Second, it outlines recent federal and state efforts in responding to this challenge. Through reviewing the evaluations of these efforts, it also examines the idea of “care systems” underlying these efforts. Third, it reviews recent conceptualizations of building community-based care systems and examines how these ideas facilitate the 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. understanding of “care systems.” Fourth, the chapter suggests using a new perspective to enhance knowledge about “care systems.” It introduces the network perspective as a supplementary framework to unveil the mask of “care systems.” In addition, it also introduces the two research questions of this dissertation: “How is the elder care system organized in a community?” and “What factors are associated with this organization?” The Challenge of Designing Effective and Efficient Delivery Systems The projected increase in the older population with multiple chronic care needs renders designing effective care systems an urgent and important mission for both the public and the private sectors. Although research on disability shows that the prevalence of older Americans with a chronic disability condition is declining (Manton, Corder, & Stallard, 1993), two issues are likely to create significant concerns for health policy makers and professionals. First, the aging of the population in the United States in general will increase the demand for chronic care (General Accounting Office [GAO], 1990). The elderly in general have a higher rate of chronic conditions. About 88% of community-dwelling elders had at least one chronic condition in 1987 and of these people, 69% have multiple chronic conditions to manage (Hoffman, Rice, & Sung, 1996). In 1994, nearly 40% of the community- dwelling elderly, amounting to 12 million people, were limited by chronic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. conditions. Of these, three million (10% of all elderly) were unable to perform activities associated with independent living (such as bathing, shopping, dressing, or eating). It is estimated that in 2020 with the growth of the elderly population to about 20% of the total United States population, there will be 10- 14 millions elderly needing long-term care (GAO, 1994). Moreover, the oldest old (persons age 85 or over) is now the fastest growing segment of the older population. Given that in 1995, three in five persons aged 85 or older had a chronic disability (Sofaer, 1998), the future demand for chronic care services should be of critical concern for health policy makers and health care professionals. A second issue relates to the need to manage costs of providing care to the elderly with chronic conditions. Because they are more likely to have greater acute care needs than those without chronic conditions, people with chronic conditions usually incur significantly higher medical costs. Almost 40% of total direct health care costs for this population were spent on hospital care and 25% on physician care in 1990 (Hoffman et al., 1996). In addition, people with chronic conditions also accounted for two-thirds of physician visits and over half of emergency department visits. Given the projected increase in the number of older persons with chronic care needs, finding ways to manage the high care costs for this population becomes more urgent. Currently, relatives of older adults provide a majority of chronic care assistance (Hooyman, 1990; Toseland, Smith, & McCallion, 1995). The 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. proportion of family caregiving has not changed significantly in the last several decades. In 1995, about 71% of the elderly with long-term care needs lived in community and received a significant amount of help from families, friends and volunteers (Friedland & Feder, 1998), especially wives and adult daughters (Hooyman, 1990). While family support will remain an important source of caregiving, the effect of reduced fertility over the last several decades could strain the capacity of families to care for a population with increasingly more chronic care needs. For those older persons who do not have adequate family support, care systems developed from the public sector becomes extremely crucial to their quality of life. Public Efforts in Responding to the Challenge In this section, I discuss how policy makers react to these two growing concerns in regard to providing care to older adults. I begin by describing how existing programs and services provided for older adults are relatively fragmented and biased toward institutional care. Then I review both federal and state efforts in rectifying the current conditions of care organization, especially their attempts in building care systems. A Fragmented and Institutionally-Biased Care System Since the mid-1960s, many programs and services for older adults have been developed, administered, and funded by federal and state 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. governments. The provision of affordable and accessible acute care services through Medicare since 1965 has generated a fee-for-service financing and delivery system for health care for older adults. Moreover, the Older Americans Act has established an aging network to coordinate social services and in-home care services for older adults (Alter, 1988). These varied programs and services increased the pressure to develop a more comprehensive service system. One of the major pressures lies in the current fragmented and institutionally biased service system. Over the past several decades, the delivery of services funded under the Older Americans Act, Medicare, and Medicaid has not only been fragmented as a result of their different funding mechanisms and oversight agencies, but also been overwhelmingly focused on institutional care (Kane, Kane, Ladd, & Veazie, 1998). For programs that support social services specifically for older adults, funding under the Older Americans Act has traditionally been relatively minimal compared to the Medicare and Medicaid programs. Though it is a major public program for long-term care services, Medicaid mandates states to provide nursing home care with some limited home health care. In 1991, over 85% of the national long-term care expenses were accounted for by nursing home care (American Association of Retired Person — Public Policy Institute [AARP-PPI], 1994). Although most states have applied for section 2176 Medicaid waivers to rectify the institutional bias in their existing care systems (Kane et a!., 1998; Coleman, Kassner, & Pack, 1996), few have been 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. successfully in restoring the balance between institutional and home-and- community-based care. As a result, coupled with the existing fragmented and over- institutionalized organization of care provision, the growth of the proportion of older adults with chronic care needs has bought increased attention to care system building and development. Over the years, numerous efforts have been tried to develop effective and efficient community-based care systems to reduce fragmentation and institutional bias. Many of these efforts have multiple objectives but their major goal is to develop community-based service systems. Federal Efforts This section describes major federal efforts in developing care systems. Literature shows that the federal government has made efforts to build acute care and long-term care systems and to integrate acute and long term care through innovations such as combining various funding streams or providing case management or coordination for older adults. Building Acute Care Systems Serious federal efforts in integrating acute care might be traced back to the Medicare Risk Contract program, enacted in the Tax Equity Fiscal Responsibility Act (TEFRA) of 1982. Its financing mechanism of prepaying 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95% of the Annual Adjusted Per Capita Cost (AAPCC) to health maintenance organizations began to transform the acute care provision landscape from fee- for-service to managed care (Zoroboros, 1997). Responding to the prepaid financing, managed care organizations used various means such as utilization review, hospital preauthorization, physician gatekeepers, and capitation contracting with physicians, hospitals, and specialists to restructure the fragmented care delivery under fee-for-service into a coordinated care system. Some recent systemic innovations for managed care plans include acute care units for elders (ACE), home hospitalization, subacute care units, and interdisciplinary home care (Boult, Boult, & Pacaia, 1998). Building Long-term Care Systems Several major federal efforts have been implemented to better coordinate the long-term care “system.” Earlier federal efforts focused more on reducing the growth of nursing home costs through the search for alternative home care. Since the late 1960s, nursing home alternatives have been explored in a number of states (Hennessy & Hennessy, 1990; Weissert, Cready, & Pawelak, 1988). In 1980 home care alternatives were evaluated at the national level through the National Long Term Care Channeling Demonstration (Weissert et al., 1988). This demonstration aimed at testing the substitutability of community-based services for nursing home care for older adults who are at risk of being institutionalized. Two models, both 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. providing comprehensive case management, were evaluated in ten sites. The basic case management model tested whether the difficulty for older adults in obtaining long-term care resulted from the lack of adequate information about the ability to obtain and manage services under the existing system. Therefore this model only provided case management to older adults. The financial control model argued that the current nursing home use pattern was attributed to the lack of funding for community services and hence this model created a fund pool to expand service coverage at the site and to grant case managers service authorization power (Carcagno & Kemper, 1988). Although the National Long Term Care Channeling Demonstration did not directly deal with building an effective care system, its attempt to reduce nursing home costs was related to some important system development issues. By providing case management and expanded service availability, the demonstration project proposed to change the composition and the interaction within the existing service system. Integrating Acute and Long-Term Care Given that frail older adults are more likely to suffer from both acute and chronic diseases, integrating acute and long-term care seems to be an appropriate policy goal for care system development. Linking acute care with long-term care may involve combining funding pools of Medicare and Medicaid, developing case management to coordinate services for older 16 permission of the copyright owner. Further reproduction prohibited without permission. adults, or consolidating agencies responsible for service provisions (Wiener & Skaggs, 1995; Kane, Starr, & Baker, 1996). Two major federal efforts of integrating acute and long-term care systems are the Social Health Maintenance Organizations (SHMO) and the Program of All-inclusive Care for the Elderly demonstrations (PACE). The original idea for the Social/Health Maintenance Organization Demonstration was to provide extended care for older adults enrolling in Medicare Health Maintenance Organizations (Leutz, Greenlick, & Capitman, 1994; Wiener & Skaggs, 1995). By enrolling a cross section of Medicare beneficiaries with all levels of care needs, the sponsor organization assumed full financial risk in providing all Medicare A and B benefits as well as some extended care benefits. Implemented in 1985 with four sites in the demonstration, SHMO now launched its second generation (SHMO II) which emphasized a more geriatric approach to care integration and the application of the SHMO model in rural areas (Kane, Kane, & Finch, 1995). Several organizational and financing features characterized the SHMO Demonstration (Leutz et al., 1994). First, it provided a comprehensive package of acute and long-term care services under full financial risk and a single organizational structure. Second, the provision of long-term care services, authorized by coordinated care management, was to those frail elderly who were deemed nursing home certifiable and were under a fixed but fairly modest cap (up to $1,000 a month). Third, the SHMO enrolled a cross- 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. section of the elderly population representing a similar proportion of the community so as to maintain the risk pooling principle. Fourth, the SHMO was financed by prepaid capitation of pooled funds from Medicare and Medicaid, or member premiums, and copayments. SHMOs were able to offer enriched long-term care benefits because they received 100% of the AAPCC rate. SHMO I operated on a complicated stratified enrollment (to have a group that had the same proportion of elderly persons as the general community) to balance their actuarial risk. SHMO II tested whether a more geriatric-focused strategy would yield better results. It did not use a stratified enrollment structure but would accept all eligible enrollees. Geriatric and case management efforts would be directed at all high-risk clients, not just those who were nursing home eligible. Another model integrating acute and long-term care was the Program of All Inclusive Care for the Elderly (PACE), which started as a federal demonstration project enacted under the Omnibus Budget Reconciliation Act (OBRA) of 1986. This demonstration expanded from its original 10 sites in 1986 to 15 sites in 1990 (Branch, Coulam, & Zimmerman, 1995). Since the Balanced Budget Act of 1997, forty sites have implemented the PACE programs, with 20 growing each year (Hansen, 1999). PACE provided an extensive set of acute and long-term care services on a capitated basis for older people with nursing home care needs. Hence, PACE represented an example of the specializing approach to care system development (Starr et 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. al., 1996; Wiener & Skaggs, 1995; Leutz et a!., 1994). Unlike SHMO’s mainstream approach of enrolling both healthy and frail elderly, PACE limited its enrollment to persons with severe impairments who met nursing home admission criteria. Providing a comprehensive array of acute and long-term care services either directly or contractually to its enrollees, PACE was financed through capitated Medicaid and Medicare payments, mostly coming from Medicaid. One distinctive feature of PACE was its emphasis of service delivery through adult day health centers. Based on a geriatrics-oriented, staff-model HMOs, with primary care physicians as employees of the organizations, PACE also relied on multi-disciplinary care management teams to provide community-based long-term care to its enrollees mostly in the adult day health centers (Eleazer, 1996). State-Initiated Efforts The literature also suggested that a number of states have made efforts to develop care systems for older adults. With an increasing number of older adults enrolling in the federal Medicare Risk Contract program, states were interested in reforming their Medicaid program for two main reasons. First, they want to gain similar benefits from managed care as M edicare does, and second, to avoid excessive cost shifting between the acute care provision of Medicare and the long-term care provision of Medicaid, especially among their Medi-Medi population. Moreover, since a significant portion of long-term 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. care cost was state-based (Kane et al., 1998), states have also been interested in how to reform their long-term care “systems” (Coleman et al., 1996). Building Acute Care Systems Because most acute care for older adults was provided through Medicare, states were not at stake in integrating acute care. However, with the financial reform in Medicare, states have been increasingly aware of the cost-shifting potential for their Medicaid beneficiaries. Currently, only Arizona, through the Arizona Health Care Cost Containment System, has integrated its acute care services in its Medicaid program. Contracting Medicaid providers by capitating acute care services to its Medicaid beneficiaries, Arizona was able to control the cost growth of Medicaid expenses (Kane et al., 1996). Building Long-Term Care Systems States have used several strategies to integrate long-term care. States have had an important stake in managing long-term care costs through developing better community-based care systems and some have already started to experiment with innovative strategies. Table 1 listed several approaches states have used to reform their long-term care systems. Despite many ways of improving existing community-based care systems, few have been truly vigorous. Kane et al. (1998) listed three major barriers for states to 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. commit seriously to developing community-based care systems: (1) the political pressure from existing providers, (2) the anticipated “woodwork” effect of making community-based care available, and (3) the inability to appreciate the value of these services beyond being substitutes for expensive care. Unlike federal demonstrations, state-level efforts have varied from innovations focused at one level of the existing system to more systemic innovations. These state approaches revealed reforms that take place on several levels: service/program, administration, client, and system (Kagan and Neville, 1993). Program-level approaches included limiting nursing home growth by restricting access to nursing homes, prohibiting construction, and allowing the use of swing beds as well as expanding home and community- based services increase the number of community-based providers and hinder the growth of nursing home providers. Consolidating state agencies and placing a cap on care plan cost can be considered reform strategies at the administrative level. Service researchers also noted the multi-faceted nature of these strategies in existing care systems. For instances, Zawadski (1984) summarized states’ efforts in improving long-term care service delivery in two major strategies: (1) those that develop single services to increase service comprehensiveness of a system, and (2) those that improve system integration either through merging different services and funding (the consolidated direct service model) or introducing integration mechanism in the 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. system (the brokerage model). Justice (1988) in her study of six community care systems found that, at the administrative level, states make use of different governance structures (umbrella, cabinet, and consolidated models) to manage a diverse range of long-term care services. Among these various models, consolidation at the state level seems to benefit coordination among local agencies in service delivery. Some states have also attempted to monitor how older adults enter the system and become eligible for services. What Are the Results of These Efforts? Both the federal and state governments have devoted a significant amount of time, funds, and other resources to discussing, debating, and developing care systems. How successful are these efforts in reducing fragmentation and institutional bias in the existing care systems? Since some of these federal and state efforts are system-related intervention and others are domain-specific, the following section discusses results from evaluations of these efforts by classifying them into domain-specific and systemic intervention. Domain-Specific Intervention Various strategies have been tried to fund more home and community-based services to reduce institutional bias, but the current care systems are far from balanced (Polivka, 1998; Kane etal., 1998). Evaluations of federal attempted 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1 State Approaches to Reform Long-Term Care Systems Approaches Strategies Limit the growth of nursing homes (Coleman, 1996) Expand home and community-based services (Coleman, 1996; Wiener & Stevenson, 1998) Increase residential options (Coleman, 1996) Consolidate state agencies (Coleman, 1996) Develop a single point of entry system (Coleman, 1996; Justice, 1990) Use Managed Care to integrate care in various programs/ settings (Riley, 1998) Place a cap on care plan cost (Justice, 1990) Limit the growth of nursing home beds through the use of Certificate of Need, restricting nursing home construction, converting nursing home beds for other uses, and relocating nursing home residents back to community • Restrict access to nursing homes through tightening income or functional eligibility requirements • Re-setting reimbursement rate for nursing home services • Make more flexible use of waivers, state Medicaid plan optional personal care, and state funds Offer incentives to develop supportive housing through housing loans, grants, and subsidies Impose standards through licensing and regulation Transferring long-term care programs from several state agencies to a newly created state agency or an interagency committee Designate a single agency to serve as focal point for clients to enter the system, usually a case management agency Integrate capitated Medicaid with Community-based long term care services, Medicare is FFS (e.g. ALTCS). Capitated Medicaid primary and acute care, long-term care services is FFS (e.g. Oregon Health Plan). Integrate Medicare and Medicaid for dually eligible elders (e.g. Minnesota’s Senior Health Options) Integrate Medicare and Medicaid in its Integrated Care and Financing Project in a single point of entry system (e.g. Colorado) Limit the amount of expenses on care plan 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to reduce the delivery problems in long-term care varied in their results. First, several reviews on home and community care demonstration evaluations found that home and community-based long-term care did not reduce overall health costs (Weissert, 1988; Kane, 1988; Hennessy & Hennessy, 1990). Second, the general impacts of such provision to health status (in terms of mortality, and physical and mental functioning) were limited. Weissert et al. (1988) surveyed dose to 30 demonstrations for their effectiveness in affecting service utilization, expenses, and outcomes. Mixed results were found in using community-based services to save hospital and nursing home costs, partly because of the research design in evaluating these interventions. Weissert et al. (1988) concluded that whatever savings occurred by community care programs were not maintained for a long period. In their evaluation of how these interventions affected patients’ health outcomes, they found that patients who benefited most were not those at risk for nursing home institutionalization. Overall health impacts were limited to patient and caregiver satisfaction as well as reducing unmet needs. Weissert and Hedrick (1999) updated their review in the late 1990s and came up with a similar conclusion. Failure to reduce cost was also echoed by Wiener (1996) in evaluating the effectiveness of expanding home and community-based services. 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. On the other hand, home and community-based care was found to improve client’s life satisfaction. In some cases (e.g. Channeling Demonstration), the intervention had a positive effect on the caregiver’s quality of life and morale, suggesting that expanding and improving home and community-based services had the potential to supplement the caregiving provided by informal support networks. However, these reviews cautioned that policy makers and researchers should modify their conception of using home and community care as a substitute for nursing home care (Weissert, 1988; Weissert et al., 1988). The extensive studies evaluating the National Long Term Care Channeling Demonstration and other home and community-based care as nursing home alternatives suggested that the results were selective and limited. As for the other state efforts, limited evidence of effectiveness has been found, partly because evaluations of the effectiveness of state-initiated interventions were more difficult to obtain (Kane et al., 1996; Coleman et al., 1996; Weissert & Hedrick, 1998). While many studies documented various state approaches to reform long-term care systems (Justice, 1988; Riley & Mollica, 1996), systematic state-wide evaluations have been rare. Large scale evaluations such as those assessing the Arizona Long Term Care System (ALTCS) found that any cost savings achieved were almost entirely from limiting services to a more narrow group of beneficiaries (those at high risk of institutionalization) (Wiener, 1996). 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Systemic Intervention Evaluations of the PACE and SHMO demonstrations have been more systematic, yet they were not without dispute. In general, evaluations of both demonstrations have shown that utilization of acute care services was reduced due to the capitated funding arrangement. Both operations have encouraged Congress to make the PACE program permanent under the Balanced Budget Act of 1997 (Gage, 1998) and to extend the SHMO to a second generation of demonstrations with a focus on the geriatric approach and on integrating care in rural areas (Online available - http://www.hcfa.gov/ HCFA website, 1999; Kaneetal., 1996). Nevertheless, evaluations of these two demonstrations have both come under scrutiny. For the PACE demonstration, the issue of the extent to which cost savings were results of “favorable selection” has been raised, suggesting that the outcomes could be related to targeting a group of older adults that did not resemble those living in institutions (Branch et al., 1995). Controversies on the evaluations of the SHMO have focused on the original intent of the SHMOs in system development. Harrington, Lynch, and Newcomer (1993) argued that the implementation of the demonstration did not incorporate an adequate degree of integration to achieve its goal of integrating acute, social, and long-term care services to reduce service fragmentation. In a letter responding to Harrington et al.’s (1993) article, Leutz, Greenlicck, Ripley, Ervin, and Feldman (1995) pointed that Harrington et al.’s (1993) argument 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. was based on a different conception of the demonstration than its original goal. According to Leutz et al. (1995), the SHMO was not a clinical trial of a medical intervention but “a systems intervention that was designed to change the context of Medicare financing, benefits, service delivery, and marketing (p.6)”. Leutz et al. (1995) argued that SHMO tested for expanding long-term care benefits to existing HMO setting; hence, this was not a geriatric restructuring service delivery model. Harrington et al. (1995) argued that Leutz et al.’s letter attributed the failure of SHMO to the methodological problems in its evaluation rather than to the weakness of the intervention. After reviewing approaches of integrating acute and long-term care, Wiener and Stevenson (1998) pointed out that the limited progress found in managed care and acute and long-term care integration have slowed many states in adopting the approaches to respond to cost pressure. At the same time, Wiener and Skaggs (1995) have questioned whether integration added value to capitation in controlling costs. Wilber and Myrtle (1998) argued that linking better integration to improved outcomes was an untested assumption of many service providers and policy makers. The Concept of Care Systems Re-examined These results raised an important question: How can care systems be improved? To address this question, understanding what the current care 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. system looks like becomes essential. This section discusses care systems as conceived from federal and state efforts of the past several decades. Common sense tells us that a gap exists between merely offering an array of services for older adults and providing a functional system of care. Yet, given the current understanding of the care provided to older adults, the appropriateness of using the term “care systems” is questionable when it is used to describe the current “organization of care” for older adults. Does our current service system possess a satisfactory level of “systemness” so that it is “qualified” to be considered care systems? As this section attempts to argue, the lack of consensus in defining “care systems” renders its usage short of conceptual clarity. Despite many efforts described in the previous section, the understanding of how to develop better care systems for older adults is still far from adequate. While financial integration and case management are essential features in developing such systems, a framework is lacking to comprehend such development. Although the various federal and state efforts have granted numerous opportunities to seek deeper understanding of care systems, they are inadequate in providing a common understanding of the various components in a community-based care system and the implications of their interrelationships with each other. These efforts do not even provide what the current care systems look like. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In fact, the concept of a care system has rarely been articulated thoroughly among policy makers and health care professionals. Based on the various innovations and demonstrations, care systems can be examined based on two criteria: unit of analysis and phases of system development. Care systems can be provider-based (as some would characterize adult day health care or nursing homes as care systems), network-based (alliances), regional (such as state-based case management system), or national. In some other cases, care systems are described by the nature of the care, such as acute care, long-term care, or comprehensive care. As a result, the term “care systems” is used just like an amoeba, changing its shape to see as circumstances change. Some researchers would depict current long-term care as a “non system,” comprised of a variety of programs or services (Hennessy & Hennessy, 1990; GAO, 1993); other researchers would argue that a system does exist, but that it is fragmented. They suggest a “need to move away from a fragmented approach to health service delivery and toward more comprehensive, continuous service strategies" (Counte, 1998, 404). They seem to be comfortable using the phrase “long-term care system,” yet they do not provide an explicit description of what the term denotes or which aspect of the system is fragmented. Some researchers use the term to describe a list of services (Wacker, Roberto, & Piper, 1998). Others use it synonymously with 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. “continuum of care.” All these suggest a lack of consensus of what “care systems” means. In addition to the relative ambiguity in defining “care systems,” components and levels within the care systems have rarely been explicated in previous proposals. Innovations identified as having potential improvements for the care system for older adults are proposed but without descriptions or explanations of the level and the aspect of the entire system in which they are expected to make an impact. For example, Boult et al. (1998) described many innovative programs to care for older adults with varying levels of health risk. Programs that may potentially benefit older adults with acute conditions include Acute Care for Elderly (ACE) units, home hospitalization, and subacute care units. Depending on the unit and level of analysis, these innovative programs can be conceptualized as (1) care systems themselves, (2) parts of a changing acute care system, or (3) strategies aimed at restoring the balance of various components in an acute care system. Furthermore, federal and state efforts aforementioned did not provide information on what constitutes a good conception of a system. Some efforts proposed to suggest the benefits of an essential system feature to the existing organization of care, but do not delineate how this feature fits into the “system" and interacts with other system components. This lack of consensus and incomplete conception of “care systems" may be related to two reasons. First, efforts to develop “care systems” tend to 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. be on an ad hoc basis, and second, the unit of analysis and implementation of where the care system should be developed or reformed is not clear to policy makers from different government levels. A review of the various demonstrations shows that these strategies attempt to improve system functioning at various levels on a scattered and ad hoc basis. These demonstration efforts include a variety of strategies, ranging from complete development of a new system to patient-level innovations. They are fragmented “patchwork” solutions, each suggesting a different view of “care systems” results when successfully implemented. For those demonstrations and evaluations that emphasize one level of intervention, mixed results could be related to the failure of taking into account the macro/meso system environment or the interrelationships among multiple levels within the system (Hennessy & Hennessy, 1990). For demonstrations that focus on building new systems, the experiments are relatively small and are extremely sensitive to political feasibility, making such experiment and its subsequent network development relatively unstable. The incomplete conception of “care system” is also manifested in the ambiguity found in defining the level and unit of analysis and implementation for such a system. Determining the appropriate level of analysis and implementation is essential in understanding care systems (Luke & Wholey, 1999). Unfortunately, some confusion is raised in terms of where care systems should be described and examined and at what government level 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. long-term care policy should, if needed, be enacted. While the local nature of care provision to older adults is recognized, pressure to reduce cost growth in long term care at the federal and state level has taken precedence. Realizing the need to find cost-saving solutions at the national and state level while preserving a sensitivity to local needs for flexibility would only increase the tension (Justice, 1988). For example, many system-level reforms that consolidate state agencies and develop case management system tend to overlook the current care system organization at the community level where older adults reside and have direct interaction with their service providers. Oftentimes, the various demonstrations may have overlooked the unique context of the local community-based care systems. Each demonstration project includes not only the model of care for older adults but also the community-specific inter-organizational arrangements that prove its success. Consequently, when a proposed idea is considered as a potential demonstration project, it is not just the model of care but also the inter- organizational arrangement that is being tested. As a result, the logic of expanding the On-Lok model into the PACE program therefore not only examines whether such a delivery model works in other communities but also whether unique inter-organizational arrangements can be developed. However, the success of these demonstrations may depend on how closely the inter-organizational arrangements in the demonstration sites are to those of the original community. If little similarity is found, the project may actually 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. impose an “unwelcome” inter-organizational relationship on the demonstration sites in providing integrated care. The demonstration may underestimate how the intervention might modify the dynamics of service delivery at the community level. Recent Efforts in Conceptualizing Community-Based Care Systems Recently, some experts began to characterize the system in a more concrete way. For example, Estes and Close (1998) described the system as having fluid and permeable boundaries among health and social services, public and private spheres, and formal and informal service providers. As discussed below, other health care researchers have made efforts to conceptualize an ideal community-based care system. Acute Care Systems Based on their research on nine health systems, Shorten et al. (1996) formalized a model for organized delivery systems and suggested three essential processes to build such systems: functional integration, physician- system integration, and clinical integration. Functional integration represents how the key support functions (e.g. financial management, human resources, information systems, quality management) were coordinated across different operating units. Physician-system integration referred to the degree of physician involvement in the system through the use of the system’s facilities 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and services, as well as in its planning, management, and governance. Clinical integration denoted the extent to which patient care is coordinated. Shortell et al. (1996) argued that creating an organized delivery system was a dynamic building process, shaped by various market and external forces such as the degree of overall managed care penetration in the marketplace, the activities of major employers and business coalitions, and state health reform initiatives. Scott (1993) attempted to develop an integrated theoretical model to understand the changes occurred in the health care organizations for the past several decades. He argued that certain organizational perspectives were useful in explaining the specific level of the entire organization of health care. For example, the institutional perspective was more applicable in describing how the norms and values of health care changed at the macro societal level, whereas the transaction cost perspective presented a more relevant description for health care organizational changes. Consequently, Scott (1993) pointed out that one should seek to clarify the levels each of these perspectives best explained and to combine them in a complementary fashion. He suggested that understanding the medical care system started best at the macro institutional and technical environment, then proceeding to the organizational field level, and finally to the organizational set level. Scott’s conceptualization was not to show how an acute care system should work but 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to focus on how to understand the various aspects of a medical care system using different organizational perspectives. Comprehensive Care Systems First defining the continuum of care as “a comprehensive, coordinated system of care including extended care, acute care, ambulatory care, home care, outreach, wellness/health promotion, and housing services,” Evashwick (1996) identified several integrating mechanisms essential for developing such a system. These mechanisms were: inter-entity structure of planning and management, care coordination, information systems, and financing. Similar to Shortell, et al.’s 1996 conceptualization but extending to providers other than those of acute care, Evashwick’s notion of comprehensive care systems suggested that these systems were more than a collection of fragmented services. Moreover, joint planning and management, information sharing, and capitated adequate financing were supportive to efficient and comprehensive care provision to older adults. Callahan (1981) presented the organization of long-term care from a system perspective, thus beginning to delineate the various system components within a long-term care system. Callahan (1981) argued that the input of the care system consisted mainly of patient’s characteristics, and the system is likely to have desired outcomes as the output. System operation occurred at multiple levels: systems-management level, operational- 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. management level, and patient-management level. Each of these levels maintains several areas of specific functions. For instances, planning, system development, system control, and evaluation are four main functions observed at the systems-management level. On the other hand, at the patient- management level, focuses should be on outreach, entry, assessment, eligibility certification, case management, service provision, patient information, and quality control. The way how the entire long-term care system is organized, as argued by Callahan, is governed by how it is financed. Callahan’s conceptualization of care systems has been the most comprehensive one to date because not only does it outline the various components within the system but also points out the goals of the care system and the relevant factors affecting the system. Unfortunately, since few studies have focused on examining the various components in Callahan’s conceptualization systematically, the functioning and configuration of the “long-term care system” remains an enigma. The Concept of Integration In addition to these conceptualizations, the term “integration” is also popular among policy makers, researchers, and service providers. Integration is a buzzword for policy makers and professionals to describe the process leading to an effective and efficient care system (Stone & Katz, 1996), 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. although the definition of integration has rarely been specified (Wiener & Skaggs, 1995). Instead, integration is often interchangeably used with cooperation, coordination, and collaboration (GAO, 1992). Among social service researchers, service integration possesses multiple meanings. Some refer to it as “coordinating programs and organizations;” others use the term to mean “physical conglomeration of networked services” (Adams & Nelson, 1997). Still others interpret it as “the fundamental restructuring of the organizations of human services to achieve collective goals” (Austin, 1997). Some define the term as “systemic efforts to solve problems of service fragmentation and matching clients with services” (Waldfogel, 1997). Others distinguish it from “coordination,” which involves “joint activity without much intervening too much of each organizations’ own sets of goals and expectations” (Hassett & Austin, 1997). As a result, researchers conceptualize integration as having multiple dimensions. For example, Redburn (1977) distinguished the administrative/structural aspect of integration, by which he means the extent of centralization of authority, from the service delivery aspect. Bolland and Wilson (1994) noted the three faces of integrative coordination: service delivery, administration, and planning. Agranoff and Pattakos (1979) divided service integration into four dimensions: service approaches, program linkages, policy management, and organizational structures. More recently, Agranoff (1991) distinguished current integration efforts from the earlier efforts 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and identified activities at three interdependent levels that characterize successful integration: the policy or strategy level, the operational level, and the delivery level. Kagan and Neville (1993) proposed that, instead of using terms like “services” and “systems” to add to the confusion, “client-centered integration,” “program-centered integration,” “policy-centered integration,” and “organization-centered integration” should be used to describe the various levels of integrative efforts. Hassett and Austin (1997) recited Bruner's (1991) three levels of interagency collaboration (used interchangeably with integration): (1) establishing interagency groups such as task forces, commissions, committees, or councils; (2) building multi-site demonstration projects; and (3) designing comphensive, statewide collaborative approaches. Waldforgel (1997) characterized the broad range of service integration activities in three dimensions: type (administrative, governance, financing, and casework), level (state, county, level), and locus (program, worker, client). The diverse conceptualizations of integration suggest that integration is a complicated process requiring careful planning and implementation among organizations at multiple levels of the entire system. Evaluations (Reflections) on These Conceptualizations These conceptual models provide some tools with which to think about community-based care systems, but few studies have been able to make use of these models to understand the existing care systems. More importantly, 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. these conceptual models, to a large extent, are developed in a vacuum, assuming that a care system can be built from scratch or that one organization can take the lead building a care system. For example, Shorten et al. (1996) conceptualized the development of an organized delivery system from a management perspective, starting from one health care provider or insurer to develop the care system. They identified three processes: functional integration, physician-system integration, and clinical integration, with the former two supportive to and preceding the development of the latter one. Similarly, Evashwick's conceptualization bore similar assumptions that service providers had consensus about system goals and what was needed to be done and in what order. These conceptualizations may overlook the absence of a formal structure governing various organizations (Chisholm, 1989) and how different providers’ cultures and positions work within the community-based care systems and how the overall existing network of relationships in a care system may facilitate (or hinder) further system development. Because long-term care was provided by a diverse set of providers, initiatives to integrate or coordinate various services often ignored important organizational issues. As a result, one can argue that these conceptualizations are biased toward a rational model of system development from which an ideal system acts like a purposive entity with consensus of goals among entities within the system (Wilber & Myrtle, 1998). 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Literature on integration pointed to its multifaceted nature in building service systems. Depending on the aspect or level within the system, integration is the mechanism holding the various components of a system together. While discussions on integration abound, few systematic studies provide guidance as to what kind of integration is beneficial, or to how to understand the care system from the “integration” perspective. The lack of data to support the claimed benefits of service integration models is troubling because considerable efforts by both federal and state governments have been implemented on this basis for the past several decades (Lago & Zarit, 1992). Moreover, the assumption that integration leads to better system outcomes and improved client well-being is rarely questioned (Wilber & Myrtle, 1998). Research Questions and Network Perspective Previous sections indicate that despite the existence of many conceptualizations of “care systems," how the current care systems are organized remains relatively unknown. To be able to take up the challenge of building comprehensive, efficient, and effective care systems, understanding this organization provides the foundational basis for effecting solutions and facilitating evaluations. Therefore, it is the purpose of this dissertation to describe and to examine the current care systems by using the network perspective, which emphasizes the community nature of care systems and 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. their multifaceted complexity. More specifically, this dissertation asks two research questions: (1) “How is the elder care system organized at the community level?" and (2) “What factors are associated with this organization?” In this dissertation, elder care system is conceptualized as networks of service providers exchanging multiple resources with each other. Network Perspective Emphasized by structure and network, a network perspective provides a port of entry for studying care systems in a systemic way (Berkowitz, 1982 & 1988). Because networks are structures of interdependence involving multiple organizations (O’Toole, 1997), a network perspective would lead to understanding how service providers interact with each other and therefore may provide some preliminary information to evaluate how a network functions as a system. The inter-organizational linkages and network configuration among service providers hence can foreshadow how community-based long-term care is organized. Closely related to structural analysis, a network perspective argues that decision making and behavior of an organization can be influenced by the environment in which it operates (Perrow, 1987). More specifically, this perspective argues that the “environment” here refers to the linkages and networking patterns the organization initiates and maintains. Networks highlight how interdependence is structured among organizations through 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. various types of glues that congeals their interdependence (O’Toole, 1997). These glues may include exchange relations, interagency entity, institutional thought structure, shared norms, or mandates (O’Toole, 1997; Wilber & Myrtle, 1998; Woodward, 1994). At the organizational level, changes in inter- organizational relationships may influence an organization’s behavior and perception of its relationships with others (Wellman, 1988). The behavior of these organizations will also influence the functioning of their partners, which in turn reflect structural and contextual differences in the entire network of organizations (Perrow, 1972). On the other hand, at the network level, the characteristics of a network are contingent upon behaviors and interactions among individual organizations within the network boundary. In the organizational literature, the use of the network perspective has mainly focused on understanding the role of individuals within an organization and the development of inter-organizational networks such as strategic alliances and joint ventures (Gulati, 1998). For the former, the relationships among individuals working within an organization and with the centrality of power are the major emphases. For the latter, discussion has been related to the value of developing networks as an alternative form of organization between market and hierarchy. The evolution, development, and the demise of a network, for example, are interests of an increasing number of organizational researchers (O'Toole, 1997; Ring and Van de Ven, 1994). 42 permission of the copyright owner. Further reproduction prohibited without permission. Mental health service researchers also use the network perspective to understand mental health systems. For example, Morrissey et al. (1985) used various network concepts to describe the fragmentation in the mental health systems of two communities and to identify mental health providers that perform similarly in service provision. In other articles, Johnson, Morrissey, and Calloway (1996) used a network perspective to evaluate the effectiveness of federal and state innovations on local mental health systems. Similar attempts are found in the service delivery of aging programs to community-dwelling older adults. For example, Alter (1988) assessed the effects of changes in Medicaid and the Older Americans Act on the structure of elderly service systems in two counties. She used network concepts such as centrality and density to describe the structural differences in two counties that differ in their compliance to the consolidation of various programs. Aiming at understanding the different aspects of coordination, Bolland and Wilson (1994) used the network perspective to examine the level of integrative coordination in elderly service delivery in four communities. These studies represent scattered interests in using network perspective and offer techniques to further understand the current status of community-based care systems. This dissertation suggests that the network perspective may enhance current understanding of community-based care systems in two ways. First, it emphasizes that the interdependencies exist among service providers in a 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. community serving older adults. While not assuming a network would function as a “system”, the network perspective provides a framework for describing how care is organized at the community level. Second, it brings our inquiry of the care systems to the community level. From this perspective, community- based care systems can be understood through networks of relationships among service providers. A network of service providers indicates the presence of relationships among three or more service providers. Given the many health, long term care, and social services available in the community, the network perspective defines the care system in terms of relationships among diverse service providers who identify themselves as serving older adults in their community. Community-based care systems should be described, examined, and understood at the community level. Wiener (1993) argues that long-term care is a local business. Without an understanding of what the care organization look like at the community level, solutions would be limited in combating problems associated with such an organization and in facing challenges in building care systems for that community. The current status of the community-based care systems is still largely an enigma. In evaluating various interventions, there has been little focus on how the current care system operates. While care provision for older adults is mostly characterized by its institutional bias, the extent to which fragmentation exists in the system 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and the extent of the gaps and duplication of services are in service delivery are largely unknown. It is difficult to propose innovations to integrated care without relating them to network relationships among service providers, since care for older adults with chronic care needs necessitates close relationships between acute and long-term care providers, funding agencies, and integrative mechanisms. For community-based care to operate efficiently, extensive cooperation and sharing among providers, agencies, and payers are needed (Lago and Zarit, 1992). Summary This chapter focuses on two major concerns resulted from the growing population with chronic care needs. With the increasing demand and anticipated costs for chronic care, designing effective and efficient care systems is an urgent policy task. Despite several decades of federal and state efforts in developing care systems, the current organization of care for older adults has remained fragmented. The progress toward building a care system with community-based alternatives has been slow. Given this sluggish improvement, “care systems,” as conceived by various interventions, provides no concise definition with clear description of components. Efforts to improve “care systems” have been implemented on an ad hoc, scattered basis, resulting in relatively diffused efforts in building strong systems. This chapter also reviewed several recent conceptualizations 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of care systems. Unfortunately, gaps have been found in how these conceptualizations could be realized within the current organization of care. This chapter introduces the network perspective as a framework to understand the current community-based care systems. / 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER III: CONCEPTUALIZATION This dissertation aims to use network perspective to answer two questions: first, “How is the elder care system organized at the community level?,” and second, “What factors are associated with this organization?” This chapter presents the conceptual framework of the dissertation. More specifically, it delineates what in the elder care system is to be studied from the network perspective. This chapter will be divided into two sections. The first section focuses on what to describe in the elder care systems. The second section discusses perspectives used to explain inter-organizational relations and illustrates how these perspectives are used to examine how the elder care systems are organized. Describing the Organization of Community-Based Care Systems The following section addresses several aspects of how the elder care systems are organized from the network perspective. These aspects include (1) the various resources involved in the description of the care organization, (2) the degree of formalization in such an organization, (3) the patterns of resource exchanges and the “system,” and (4) the uniqueness of the care organization embedded in its community context. 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Resource Exchanges in the Organization of Care Although aging researchers and policy analysts have suggested innovations in improving service delivery for older adults that may involve interventions at multiple levels as described in the previous chapter, few empirical studies have examined the multiple levels existed in organizing community-based care. As shown in the background chapter, both the intervention efforts and the conceptualization of care systems and integration suggest that care systems are not organized along a single dimension, but involve intertwining dependence between different levels (service, administrative, organizational). Different resources are predominant in each of these levels. The network perspective suggests that care systems can be described in the patterns of how service providers exchange their resources in a network. While no consensus is found in determining what resources to be examined, most researchers have recognized clients, funding, and information as the three basic resources (see Table 2). In addition to these three, Van de Ven and Ferry (1980) suggested that organizations also exchange staff, equipment, and office space. More specifically, studies examining service systems for other chronically ill populations indicate that client, information, and money are probable resources for studying networks of service organizations. Table 2 reviews various studies of service systems and what resources the authors 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2 Studies Using Network Analyses to Examine Service Systems Authors and Years Population/ Network Resources or Dimensions Examined Wright and Shuff (1995) HIV/AIDS Information, assistance, client Provan and Milward (1994) Severely Mentally III referrals received, sent, case coordination, joint programs, and service contracts Heflinger (1996) Mental health for children and youth Awareness of agencies, staff interaction, client referrals, formal agreements, information exchange, activity coordination, satisfaction Morrissey, Calloway, Johnsen, and Ullman (1997) Homeless clients, information, funds Randolph, Blasinsky, Leginski, Parker, Goldman (1997) Mentally III Homeless clients, information, funds Rosenheck et al. (1998) Severe Mentally III Homeless Information, clients, and funds Tausig (1987) Mental health clients, funds, information, presence of a legal mandate, and need to settle disputes Fried, Johnsen, Calloway, and Morrissey (1998) Severe Mentally III individuals Information, clients, funds, grants, contract, office space, staff McAuley and Safewright (1991) Area Agencies on Aging interagency agreements, types of Interaction actually occur Alter (1988) Elder service systems Client referrals, impersonal program, personal feedback, and group methods Hall, Clark, Giordano, Johnson, and Van Roekel (1977) Problem youth Various dimensions of contacts, coordination, conflict, quality of conflict resolution, and quality of communication Gamm and Benson (1998) Community health Case study approach Bassoli et al. (1997) Health & human service Collaboration/partnership relationships j*. C D have identified as appropriate to portray the systems of their interest. Examining exchange patterns of different resources among service organizations therefore may offer a good starting point to examine the current organization of community-based care: how do service providers organize or interact among themselves in terms of these resources in a community? Following what other service research has done, this study uses client, information, funding, and staff as essential resources to describe and examine the organization of community-based care among service providers. The importance of client referral to different service providers within the network is obvious: given the chronic and acute medical needs of older adults, client referral is an essential component of system functioning. Examining the pattern of client referral also recognizes the core as well as the peripheral service providers in the current organization, as well as whether certain agencies designated as service hubs have performed their functions effectively. Accompanying the client referral is the information exchanges between service providers. Much of this information is related to clients, although some may be of an administrative nature. Exchanges of money serve another support function to client referral (Shortell et al., 1996). However, since most service programs in the United States are currently funded by federal and state governments, the interorganizational exchanges of money will be less likely to be as prevalent as those of client and information. Finally, the increased popularity of using interdisciplinary care 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. teams, establishing alliance, and planning joint projects has encouraged the use of staff interchangeably across disciplines and professional circles. As a result, patterns of staff exchange may add another dimension that is increasingly evident in current organization of community-based care. Network Dynamics in Resource Exchanges Few service system studies describe in detail the patterns of various resource exchanges. While some studies portrayed the complexity of their respective service systems (Morrissey, Calloway, Johnsen, & Ullman, 1997; Randolph, Blasinsky, Leginski, Parker, & Goldman, 1997; Johnsen, Morrissey, & Calloway, 1996; Tausig, 1987), only a couple pointed out how and why these patterns differed from one another (Bolland and Wilson, 1994; Fried, Johnsen, Calloway, 8 c Morrissey, 1998). Kagan and Neville (1993) argued that different levels within the system might require various approaches to achieve integration. Similarly, differences in resource exchange patterns might reflect various levels of system functions. Without knowing how they functioned in a particular community, designing effective strategies would be difficult. This section argues that resources in the current organization of care display different dynamics of exchange patterns among service providers. At least two reasons explained different exchange patterns across resources. First, differences in resource exchanges may reflect the extent to which service organizations trust each other through exchanges of resources 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with various levels of importance to their organizational survival. As service organizations may develop their inter-organizational linkages at different stages, their patterns of resource exchange may reflect such variation (Ring & Van de Ven, 1994). A second reason for the different dynamics in various resource networks may relate to the notion of comparative advantages among service organizations in the exchange relationships. According to the resource dependence theory of inter-organizational linkages, organizations may establish linkages with each other for resources they depend on1 (Aldrich, 1976). While such resource interdependence may not directly suggest all linkages be asymmetrical in terms of resource being exchanged, the theory infers that resource patterns among service organizations are different at the network level. Hypothesis 1: Patterns o f exchanges will differ across the four resources (client, information, money, and staff). It is reasonable to speculate that client and information exchanges would be more like each other than to money or to staff. The predominant 1 A more extended description of this theory is found in a later section of this chapter. 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reason for differentiating the two lies in their basic system functions. Boiland and Wilson (1994) discussed how Reid (1965) and Gans and Horton (1975) characterized money and staff as “administrative resources from client" and information as “resources serving the delivery function of the system.” Money and staff resources differ from client and information in that organizing alliances occurs at the administrative level, but not at the client level. Joint use of staff and exchange of experts across different services is certainly at a different level of intervention than employing service coordinators or case managers to coordinate services for older adults. A second reason for the different exchange patterns of client and information from those of money and staff lies in the current financing mechanism and program implementation in the aging network and community-based care systems. Most social and health programs are financed currently at the federal and state levels. Rather than giving older adults cash or vouchers to purchase needed services, funding is funneled through programs to service providers. Consequently, it is expected that there will be much less money exchange activity among service providers in the community because most of the exchange is vertical, from either federal or state government. 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hypothesis 2a: Patterns o f client an d information exchanges are more likely to be similar to each other than those o f m oney and staff exchanges. Hypothesis 2b: Patterns o f m oney an d staff exchanges are m ore likely to be similar to each other than those o f client and information exchanges. Degree of Formalization in Resource Exchanges Different resources also display different degrees of formalization (Aldrich & Whetten, 1981). For example, a telephone call asking for a client’s previous information or gathering other providers’ information to help in a client care plan is very common for case managers or service coordinators. On the other hand, money or staff exchange between organizations are more likely to undergo a formal process of negotiation, discussion, and approval, therefore, their relationships are more formal than are information and client exchanges (Ring & Ven de Van, 1994). Although client and information exchanges can be guided by formal contracts, informal relations are less likely to be replaced by rules and guidebooks. Therefore, if the relations of client and information exchanges are institutionalized, they are more likely to be characterized by both formal and informal mechanisms. 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hypothesis 3a: M oney and staff exchanges are more likely to be associated with formalized exchanges than are client and information exchanges. Hypothesis 3b: Client and information exchanges are more likely to be associated with informal exchanges than are m oney and staff exchanges. Patterns of Resource Exchanges Relatively few studies in service system research focus on how patterns of resource exchanges are structured. Common descriptions of the structural characteristics of these systems include indices such as density, centralization, formalization, or span of control (Nelson, 1986). As a result, rich information from examining the structure of networks of relationships in these systems has not been fully explored. Direct study of the structure of these relational networks may enhance our understanding of interorganizational dependence and the extent of system integration or fragmentation (Galaskiewicz & Krohn, 1984; Johnsen, Morrissey, & Calloway, 1996). Network researchers suggest two dominant approaches to understand the structure of relationships: relational and positional approaches (Fraust & Wasserman, 1997; Shah, 1998; Johnsen etal., 1996; Brieger, Boorman, & White, 1976; Burt, 1980). This section introduces these two approaches to study the patterns of resource exchanges. 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Relational Approach The central idea of the relational approach is expressed through the concept of cohesion. Burt (1980) describes the various types of model under the relational approach, and indicates that clique is the most dominant model of cohesion. Cliques are subgroups in which organizations maintain very close ties with each other and relatively loose ties with those outside the cliques (Marsden, 1990; Burt, 1980). They are complete maximal subgraphs, with subgroups of actors formed within a graph, based on characteristics such as mutuality of ties (actors must choose each other), closeness of subgroup members, frequency of ties among members, and the relative frequency of ties among subgroup members compared to non-members (Wasserman & Faust, 1994). Since cliques denote the existence of cohesion among organizations, more cliques within a system means higher connectivity. Therefore, the more cliques identified in a network, the more integrated the system is thought to be. In general, density and size of a system may affect the number of cliques. One can also examine the extent of clique overlap within a network (Provan & Sebastian, 1998). The degree of clique overlap represents the extent to which isolated cliques exists in a network. High clique overlap suggests that some organizations within a network have multiple affiliations with several cliques. These organizations could serve as “brokers” or “bridges” to the entire network relationship. From the cohesion perspective, 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. inter-organizational networks can be perceived as the organization of cliques formed by service providers with high level of resource exchange linkages. Cliques can come in different sizes, depending on the number of organizations involved and the frequency of interaction among organizations in the clique. Given that no research existed to serve as a guide to determine the normative size of cliques, this study will examine the data and identify the largest cliques found from the data. Positional Approach The structure of care systems can also be examined in terms of the presence of and the interaction between structurally equivalent subgroups through the positional approach. These structurally equivalent subgroups are sets of organizations that maintain approximately the same resource exchange patterns with other organizations within the network. Figure 1 illustrates the concept of structural equivalence. In Figure 1, only organizations A and B are structurally equivalent to each other because they have exactly the same exchange patterns with other organizations in the network: they both send resources to organization D and maintain reciprocal exchange relations with organization E. 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ID ▲ C J C D " O a > c a \_ • 4 — * c o _ 3 c o Q 3 O i— a .a 3 C O c C D c a > '3 a LD > " c a 3 O 3 03 £ 3 0 5 L l . 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In organizational research, the concept of structural equivalence can be applied relatively easily to formal hierarchy. For example, in a hospital setting where two nurses have to report to a physician about the care for 10 patients, these two nurses are structurally equivalent in terms of their relations within the network of physician, patients, and themselves. Unlike the cohesion approach, relations among organizations within a structurally equivalent subgroup are less emphasized under the positional approach than are the relationships between different structurally equivalent subgroups. It is the inter-group relationship that characterizes the role of each structurally equivalent subgroup in the overall network. As discussed in the background chapter, although long-term care service researchers and policy makers generally have conceptions about how a care system should look, few studies have empirically examined the structure of care systems (Bolland & Wilson, 1994; Alter, 1988). It is suggested that the structure is imposed or assumed by the researcher rather than empirically derived. Using the positional approach, this dissertation explores the structure of care systems through patterns of resource exchanges by empirically derived structurally equivalent subgroups. Given the vast variety of organizations serving to meet older adults’ needs in the community, how do they organize themselves through exchanges of client referral, money transaction, information communication, and common use of staff resources? 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The positional approach provides a more direct way to examine the entire structure of care systems. Rather than identifying cohesive relations, these methods treat the network of relationships as a whole, taking every organization of the care network into consideration. Hence, the outcome of the analysis is the partition of organizations into different subgroups. Unlike the cohesion subgroups in which only organizations with close ties are identified, the structural equivalent method identifies organizations affiliated with subgroups based on various exchange patterns as well as those that cannot be affiliated because of their distinct exchange pattern in the network. Organizations of the latter are usually considered as residuals (Nelson, 1986, 1988; White, Boorman, & Breiger, 1976; Burt, 1980). A structurally equivalent approach can also detect structure of network relationship not based on direct relations (Gerlach, 1992). Another advantage of using the positional approach lies in its ability to examine division of labor in the care system. Rather than following researchers’ conceptualization of what the structure of a care system should look like, this dissertation examines the data to study the presence of division of labor as an important feature of a care system. The division of labor as it exists in current care organization not only tells the degree of integration in the network but also how it functions in a particular community. Without understanding the functioning of a system, intervention superimposed would likely be inefficient and encounter tremendous resistance during 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. implementation- The structure of care systems, as revealed through the division of labor, may also relate to discussion on the extent to which care delivery is redundant, how the community-based network of organizations collectively meets the needs of elderly people, and how to relate to external demands (Galaskiewicz & Krohn, 1984). The structure of care systems conceptualized by researchers is mostly a priori client-based. For example, Alter (1988) describes the structure of elderly service delivery systems in terms of the client’s care management process (care planning, coordination, referral, assessment). It is uncertain if the structure of various providers revealed as roles of different processes is a result of analyses by Alter (1988) or an a priori conceptualization. To what extent the structure accurately describes the network of service organizations is an empirical question. Instead of following previous network research on community-based care, this study will examine data from the communities of San Mateo, Long Beach, Tulare, and San Francisco and explore the structures of these care systems. Given that the client is only one of the many resources in a service system, trying to examine the structure from the client exchange pattern is not enough. There is no preconceived notion of what the system would look like when multiple resources are considered. The focus is on the role of each service provider. Hence, two working research questions regarding the structure of elder care systems are developed. As far as “clique” is concerned, does the literature suggest any 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. close relationship among particular service providers? The clique approach can be used to examine the extent to which fragmentation prevails in a care system. As for the structurally equivalent subgroup, one could examine how service providers divide among themselves in term of service functions and scopes. Care Systems in Community Context The network perspective posits that the organization of care systems for older adults is influenced by characteristics of individual service providers and how the providers relate to each other in a community. It is also shaped by local community happenings. Consequently, the organization of care systems is community-specific, tailored towards community needs and conditioned by the important actors and resources of the local community. The extent of community uniqueness in care systems is also intimately related to the amount of vertical control exerted by the federal and state governments (Alter, 1988). Tighter vertical control is more likely to lessen the local idiosyncrasy embedded in the care organization. The history of care system development in California is critical in understanding how care is organized in individual communities. Coberly and Wilber (1991) described the history of California state government in developing care systems. The state has not been successful in developing a state-wide model for communities to 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. develop its long-term care system. Consequently, individual communities possess a high level of flexibility in designing their own “systems.” Hypothesis 4: Organization o f community-based care system s will differ across communities. Factors Underlying Network Relationships Organizations are usually reluctant to make ties with others because by doing so, much of their autonomy may be lost. Also, it is costly for organizations to coordinate with other organizations whose missions, culture, and organizational procedures differ from their own (Van de Ven & Ferry, 1980). Yet, increasingly an inter-organizational or network form of organizational arrangement is evident in the human service sector (Alter & Hage, 1993). What motivates organizations to form ties is both of theoretical interest and of practice concern. Since network relationships are constituted from ties between two or more organizations, examining what motivates organizations to engage in inter-organizational activities provides important insights into network relations. In service delivery for the frail elderly, understanding what motivates service providers to establishing ties with each other to achieve broader goals enhances the efficacy of intervention in better coordination at 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the system level (Morrissey, et al. 1982; Schmidt & Kochan, 1977). This section describes major perspectives of interorganizational relations. Several review articles have been instrumental in understanding why organizations develop relationships with one another. For example, Halpert (1982) and Alexander (1995) have provided a list of facilitating and inhibiting factors to inter-organizational relationships. These are conditions that may affect the likelihood of inter-organizational relationships actually being developed as well as the form of these relationships. Alter and Hage (1993) also provided a framework for understanding the formation of inter- organizational networks. More recently, Grandori and Soda (1995) reviewed literature on inter-firm networks and summarized factors affecting networks through different approaches. Most of the factors, however, are related to their influences on particular network forms, but may not be critical in motivating network formation. Most factors are contextual or perceptual, affecting whether a particular inter-organizational relational form would be adopted by organizations motivated to develop relationships. While it is difficult to disentangle the intertwining nature of organization-environment antecedents to network formation, Ebers (1997) conceptualizes factors affecting inter-organizational networks into motives and contingencies. This dissertation aims to seek a preliminary understanding of why network relationships in general emerge among service providers in the realm of long term care service delivery. 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Perspectives on Inter-organizational Relations This section offers a more in-depth discussion of several major organizational perspectives that may apply to how organizations establish relationships with one another. These perspectives include resource dependence framework, power perspective, institutional perspective, transactional cost perspective, and governance perspective. Resource Dependence Framework Several organizational perspectives have been useful in understanding what prompts organizations to develop relations with one another. The resource dependence/exchange perspective presents the foremost dominant view of why inter-organizational linkages occur (Grandori & Soda, 1995). According to this perspective, organizations become more interdependent to each other as a result of increased specialization and differentiation in their field. Developing inter-organizational relationships to ensure smooth and predictable resource flow, especially in a resource-scare environment, is considered an important strategy for their own survival. For the past several decades, the resource dependence perspective has been the predominant framework used to study inter-organizational relationships, especially in the social service sector. As argued by Aldrich (1976), organizations establish inter-organizational relationships because the 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. differentiation and specialization within each organizational field requires organizations to depend upon each other for survival. This resource dependence is especially acute when resources are scarce and the knowledge about the external environment is limited (Alexander, 1995; Mizruchi & Galaskiewicz, 1993). Inter-organizational relationships emerge as adaptive responses to the recognition of resource dependence and the perceived environmental uncertainty of an organization. Organizations establish ties with each other to ensure a steady flow of critical resources important to their survival. Resource dependence theorists argue that in an uncertain environment, organizations are more likely to develop inter- organizational relationships if a certain level of stability can be achieved in those relationships (Oliver, 1990; Reitan, 1998). As a result, increased awareness of resource dependence among organizations, especially if they are in the same network, is related to the emergence of inter-organizational relationships (Morrissey et al., 1982; Van de Ven & Ferry, 1980). In this study, the perception of the potential organizations being in the same network (SAMENET) will be one of the variables used in representing this perspective. The use of SAMENET suggests heightened awareness of the resource dependence among organizations that would like to establish inter- organizational linkages. Organizations also form links with one another because there are opportunities to pursue mutually beneficial goals, especially in time of 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. resource scarcity (Cook, 1982; Oliver, 1990). The relationship formed under this situation tends to be cooperative. During periods of resource scarcity, the specialization situation may even prompt service organizations to develop linkages with each other, even when they are the only available organizations in the community. In order to secure resource flow, organizations may be left with few options but to develop relations with the only available providers. As a result, the presence of only available partners (ONLYAVAIL) can also be another variable used to examine the resource dependence perspective. Within the resource dependence framework, centrally located services or co-located services provide high level of convenience for organizations to communicate with each other and for older people to access to services. Consequently, proximity would increase the likelihood of establishing inter- organizational relationships, particularly for organizations that seek for client linkages. In this study, the existence of proximate service organizations (PROXIMATE) is considered as a variable to examine this aspect of the resource dependence perspective. Power Perspective Contrary to the reciprocity and exchange perspective is the power approach to inter-organizational relationships. According to this approach, organizations make ties in order to exert control over scarce resources (Mizruchi & Galaskiewicz, 1993; Oliver, 1990; Schmidt & Kochan, 1977). The 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. desire to exert control and influence over resources has prompted organizations to make ties in an asymmetrical situation. This perspective argues that instead of encouraging cooperation, resource scarcity prompts competition and conflict between organizations. Schmidt and Kochan (1977) posited that organizations in the ‘weaker* end of an asymmetrical inter- organizational relationship continue to maintain these linkages because they consider such relations may enhance their probability to pursue their own interests, despite the benefits embedded in these relations are low. Consequently, inter-organizational relationships are the result of an organization's interest in exerting control and influence over other organizations. In the area of service delivery for frail older people, hospitals, skilled nursing facilities, or even managed care organizations have been interested in contracting with social service providers to provide care for older adults. However, the concern of over-medicalization of long-term care may create barriers for some traditional providers for social services to develop an inter- organizational relationship with acute care providers or managed care organizations. This situation can be understood as an example of looking at inter-organizational relationships through the power perspective. In this study, how organizations perceive the level of influence on each other (INFLUENCE) is used as a variable for this perspective. Power perspective would argue that organizations in an inter-organizational 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relationship are likely to have wide discrepancies on how they perceive the influence of other partners. Institutional Perspective Strangely enough, that organizations develop inter-organizational relationships may be a strategic for them to gain acceptance to their own fields. According to institutional perspective, organizations try to mimic innovators to increase their status and legitimacy (DiMaggio and Powell, 1983). This approach, which helps explain the increased prevalence of inter- organizational networks in recent decades, has become increasingly popular in studies of inter-organizational relations (O’Toole, 1997). Given the prevailing evidence of network form of organizational structure in both private and public sectors, the institutional approach, which emphasizes the socialization of organizational culture and norms, provides a relevant framework to examine networks. It suggests that organizations maintain ties because the culture of the organizational field sees it as a legitimate activity to pursue (Reitan, 1998). Applying the concept of isomorphism to the prevalence of inter-organizational relationships (DiMaggio & Powell, 1983), increased uniformity of institutional environments results when organizations mimic others and therefore establish inter-organizational relations. By doing so, organizations can increase their legitimacy within their field. 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In long-term care service delivery, the common push toward coordination, cooperation, and alliance formation have encouraged service providers to “pair up” with one another to enhance their reputation in the field of service delivery. Hence, “partnering” with other organizations to provide care for frail older adults may become a status symbol for organizations. With the increased penetration of managed care form of service delivery, organizations are even more likely to be pressured to establish ties with others. Operationalizing mimic response has been difficult for researchers of institutional approach. Given the unavailability of data appropriate for this perspective, no variable is used under this perspective for this study. Transactional Cost Perspective Organizations also may see the benefits of making inter-organizational relationships a means to reduce their transaction costs compared to developing internal units for accommodating their clients’ need. This perspective assumes that organizations make decisions based on efficiency and cost-related criteria and therefore focuses on the organizational decision of whether a particular need should be satisfied by internal development versus external contracting, of which the latter can be considered as a type of inter-organizational relationships. 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reitan (1998) argued that it is relatively easy to apply the transactional cost approach to human service delivery. In service delivery for frail older adults, an example of this perspective is the diversification of nursing homes into various functions. Instead of contracting out to services that provide subacute care to elderly population who require short-term rehabilitative care, nursing homes develop subacute care units to draw in more revenues. Similar reasons are also given for the development of in-house Alzheimer’s care units within nursing homes. Unstable political and economic environment may affect service providers in making decision as to contracting out services or developing internal functions to accommodate their clients’ needs. Whether contracting out or developing internal functions, organizations want to keep their transaction costs to a minimum, relationships established under this mentality tend to be cost effective from the participating organizations’ point of view. While contracting to low cost providers may be a common logical choice for organizations in general, the presence of such providers may be particular critical for organizations who take a transaction cost approach in deciding its goals. As a result, this study uses the presence of low cost providers (LOWCOST) as a variable representative of the transactional cost perspective. Under the transactional cost perspective, organizations belonged to the same larger corporate structure may be more likely to establish inter- 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. organizational relationships with each other. The comparison on transactional cost in this situation would be to those potential partners not under the same corporate structure. Transactional cost for organizations under the same corporate structure may be less because both organizations may have similar operational culture and goals. Procedural costs to translate language and documents to establish relationship would be less. Moreover, the uncertainty embedded in a principal-agent situation would be less because both organizations are subject to the same hierarchy of the corporate structure. In this study, organizations under the same larger corporate (SAME ORGANIZATION) is used to represent this notion under the transactional cost perspective. Governance Perspective Finally, organizations may exchange resources with each other in order to comply with a mandate. To what extent participating organizations are aware of their resource dependence and whether such a relationship is cost effective seems to be less relevant when explaining the existence of the relationship, but the awareness is important when the quality of the relationship is concerned. Involuntary (Hall et al., 1977), mandated relationships are quite common in the provision and delivery of human services (Reitan, 1998). For instances, in order to obtain funding under the Older Americans Act, service providers are to establish relations with Area 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Agencies on Aging, of which act as a coordinator for local service delivery. This study uses the presence of mandate (MANDATE) to represent the governance perspective. Domain Similarity— An Inter-organizational Relationship Facilitator Many studies have found that organizations of similar goals, client pools, scopes of services are more likely than those who do not to form inter- organizational relationships (Van de Ven & Walker, 1984). Domain similarity is a facilitator for network development. This study uses three variables ‘same goals,’ ‘same clients,’ and ‘same funding’ and their composite variable (DOMAIN SIMILARITY) to examine the relative importance of this concept to network development of aging service organizations. Single Perspective or Multiple Complementary Perspectives? These perspectives offer a variety of possible bases of inter- organizational relationships. However, few studies have empirically tested the extent to which these perspectives explain the formation of inter- organizational relationships. One possible reasons for few studies of this sort may be related to how types of inter-organizational relationships are dichotomized into mandated versus voluntary (Schopler, 1987). Hall et al. (1977) and Schmidt and Kochan (1977) integrate several perspectives to examine the complexity underlying the formation of inter-organizational 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relationships. Since organizations are generally reluctant to engage in inter- organizational activity, it may take several motivators to convince them of the benefits of that activity. Moreover, organizations prompted by a single motive to maintain inter-organizational relation are likely to withdraw from the inter- organizational relationship when environments change. Therefore, it has been hypothesized that organizations have multiple reasons to build relationships with each other (Oliver, 1990). Figure 2 presents a model for testing the relative influences of factors from different perspectives on the patterns of resource flows in community- based care systems. One of the advantages of entering all related reasons into the model is that the strength of each factor, operationalized to represent a perspective, can be evaluated independent of others. Unlike studies that examine only one or two perspectives, this study evaluates multiple perspectives in one model so that the complex reality of why inter- organizational relationships emerge can be examined more fully. Hypothesis 5: Resource exchange relations are associated with multiple factors rather than dominated by a single factor (see Figure 2). 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Resource ~ Dependence Power Perspective. Transactional Cost Perspective __ Governance Perspective Domain Sim ilarity- Same Network Proximity Mutual Needs Only Available Provider Mutual Influences Same Organization Low Cost Provider — Mandate Same Goal Same Client Same Fund Client Exchanges Information Exchanges Money Exchanges Staff Exchanges Figure 2. Multiple-factor Model to Examine Resource Exchange Patterns. -J Ol Different Sets of Factors Affecting Exchange Patterns Across Resources While these factors are hypothesized to be associated with resource exchange patterns of community-based systems of care, how they are related to the resources being exchanged are unknown. Previous studies examining these perspectives either do not specify or have not tested specific resource exchanges. This study argues that resource exchange relations are differentially associated with the various factors listed in Figure 2. Hypothesis 6: Different resource networks are likely to be associated with different sets o f factors. Because different resources have dissimilar functions to the overall system operations, it is likely that service providers would be driven by different sets of factors as they establishing linkages of various resources. For example, client exchanges would be more likely to be affected by perception of service providers as to each other in serving their clients. As a result, the pattern of client exchanges is more likely to be associated with resource dependence, power perspective, and domain similarity. Moreover, given that client exchanges would be affected by how funds are distributed from the government to individual service providers, the latter are prone to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. seek low cost providers as their partners, making the transactional cost perspective another possible framework for describing client exchanges. On the other hand, the pattern of money exchanges may be more likely associated with the governance and transactional cost perspective. Since many programs are funded categorically from the government to individual service providers, it is likely that money exchanges among service providers would be related to certain forms of mandates. Similarly staff exchanges seem to be associated more with mandates because the latter two tend to be more driven by need than by advanced strategic planning. Because few studies have focused on factors associated with specific resource exchanges, this study suggests the model as an exploratory one to examine the relative importance of these perspectives to resource exchanges. Summary This chapter lays out the theoretical and conceptual underpinnings of how the organization of care is examined in this dissertation. The various sections are discussed so as to address the two research objectives in this study: (1) to describe the organization of elder care systems and (2) to examine the factors associated with this organization. In the first objective, resources considered to be appropriate and important in a system of care are discussed and explanations presented for how their exchange patterns are likely to be different. The organization of care systems can also be described 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in terms of how formalized are the resource exchange patterns. The chapter also introduces two approaches to examine the structure underlying these resource exchange patterns. Finally, given the lack of state-wide models for systems of care, this chapter speculates that organization of care across communities differs significantly from each other. In the second objective, this chapter argues that the network relationships developed by service providers are likely to be influenced by multiple factors rather than a single one. Moreover, since resources vary in their function in the organization of care, different sets of factors will be at work for different resource networks. 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER IV: METHODS This chapter has three purposes. First, it describes the rationale for choosing the four communities for the study and provides a brief description of the chosen communities in their development in long-term care service delivery. Second, it introduces the major network packages used in the analyses. Third, it discusses the methodologies employed to answer hypotheses laid out in the conceptualization chapter. Data and Methods This section describes how data are collected and prepared for analyses in this dissertation. Issues including sampling, data collection, unit and level of analysis, data aggregation and sociomatrices. Software used will also be discussed. Sampling The data used for this study are from a larger system evaluation study conducted in 14 Californian communities1 (the SEED evaluation) between 1 The 14 communities are San Mateo, Long Beach, Alameda, Humboldt, East Los Angeles, Monterey, Stanislaus, Santa Cruz, San 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1988 and 1991. To examine the resource exchange patterns of community- based care delivery as well as differences in service complexity, this dissertation focuses on four communities (San Mateo, Long Beach, Tulare, and San Francisco) representing different rural and urban settings. Given the fact that California did not have a statewide community system protocol for local communities serving older adults, service organizations had high levels of flexibility to initiate various “bottom up” approaches to create care systems for their older adults. As a result, wide community differences in network of relationship could be anticipated. In terms of service resource availability, there were two types of communities: service-rich and service-poor communities. In service-poor communities, crucial programs such as Linkages or Multipurpose Senior Service Programs (MSSP) were not available (original SEED evaluation). Linkages was a state program intended to prevent inappropriate nursing home institutionalization by providing case management and information and referral services to community-dwelling frail elderly. MSSP was a Medicaid waiver program in which case management was provided to Medi-Cal eligible frail elderly to arrange for and monitor community service utilization so as to delay institutionalization. Since both programs coordinated their services, the absence of these programs might affect how service providers in a community Francisco, Riverside, South Central Los Angeles, Glendale, San Benito, and Tulare. 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. coordinate among themselves in order to serve their elders. Of the four communities selected, San Mateo, San Francisco and Long Beach were considered service-rich communities, whereas Tulare was service-poor. Data Collection In the SEED evaluation, questionnaires addressing the previous perception of system development and effectiveness of various integrating mechanisms in their communities were distributed to program managers of different services. Similar to other network studies (Marsden, 1990), the problem of identifying network boundaries existed in the SEED evaluation. The SEED evaluation used a key informant strategy to identify as many community organizations serving older adults as possible and distributed mailed questionnaires to program managers. The boundary created by a network of service providers might consist of organizations serving the general population. This method allowed service providers to identify their own networks rather than a network being defined by the researcher. Consequently, such a method of boundary “drawing” allowed communities to diversify their network composition of providers serving the elderly. In addition to questions concerning system development perception and effectiveness, the SEED questionnaire also contained questions pertaining to how the respondent service providers were related to those who were important to them. A copy of this latter section of the questionnaire was 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. found in Appendix A. This information was used in the current study. Respondents were first asked to identify organizations that were important to them in achieving their organizational goals and missions. They were then asked to identify the reasons for establishing an interaction with their partners. They were also asked to specify what kinds of interaction they had with their partners and what types of mechanisms were being used to maintain interaction. Furthermore, the reasons for establishing an interaction were asked in the questionnaire. Table 3 showed items asked respondents in the SEED questionnaire. Unit and Level of Analysis This dissertation used community as the main unit of analysis. Within each community, this dissertation aimed to understand the organization of community-based care systems through multiple levels of analysis. It focused on the network, sub-network, and to a lesser extent, provider levels of analysis. The relationship among those providers forming exchange networks was examined. Moreover, in order to examine components within networks of exchanges, sub-networks such as cliques and structurally equivalent subgroups as well as their relationships with other subnetworks within the community were discussed. Related to the networks of exchanges the provider level was described, but this level was not the primary level of analysis in this dissertation. 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3 Exchange Items in the SEED Evaluation Categories Specific Items Response Codes for each identified partners Reasons for Interaction • Physically close to organization • A hassle to switch • Only provider available • High quality provider • Low cost provider • Provides needed service • Part of the network • ability to service clients with multiple needs Yes/No Resources Received • clients • money • staff • space • equipment Yes/No Resources Given • information • technical assistance • clients • money • staff • space • equipment • information • technical assistance Yes/No Mechanisms to Maintain Interaction • required by government or funding regulations • contractual agreement • memorandum of understanding • formal verbal agreement • informal, interpersonal communications • advisory board membership • joint programs • part of same organization Yes/No 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Data Aggregation and Sociomatrices For each item, the respondent's information was aggregated to service- provider types. Instead of examining individual service providers in each community, this study classified service providers according to their genericservice-provider types. Table 4 showed the various service-provider types for all 14 SEED communities. Such an aggregation was justified by two reasons. First, it facilitated comparison across communities. Idiosyncrasy of individual agencies was reduced and the role of organization-type in each community could be more easily understood. Second, the aggregation allowed respondent’s information to be transformed into matrix formats for analyses. Unlike ordinary data structure, these formats, called sociomatrices, had both rows and columns as observations. Each sociomatrix represented a type of relationship (variables) for observations. Since not all service-provider types maintained mutual resource exchanges, rows and columns of each sociomatrix represented the sending service-provider types and the receiving service-provider types. In general, a binary value (0 or 1) shown in a cell of the sociomatrix indicated the presence of a relationship between a particular sending service organization and a particular receiving service organization. For example, a “1” found in the row representing hospitals and in the column representing home health care in a client referral matrix would indicate hospitals have referred clients to home health care. A valued sociomatrix 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. showed the intensity or frequency of exchanges between service-provider types. Table 4 Service-Provider Types and Their Abbreviation for Network Graphs Service-provider types____________________Abbreviation Linkages/Multipurpose Senior Service LINKMSSP Program AAA Area Agencies on Aging HSG Housing SNF Skilled Nursing Facilities HHC Home Health Care EMERG Emergency Services/Fire/Police SENIORCTR Senior Centers LEGAID Legal Aid MOW Nutrition Services l&R Information and Referral LEGPOL Legislator/Politician DPIHAPS Social Services SOCSEC Social Security MH Mental Health HOSPITAL Hospital PUBGUR Public Guardian ADHC Adult Day Health Care TRANSPRT Transportation HEACLIN Health Clinic OMBUDS Ombudsman CASEMGT Case Management MONEYMGT Money Management COUNTYHD County Health Department CHARITY Private Charity HOMELESS Homeless Services CONSORT Consortium RSVP Retired Senior Volunteer Program REHAB Rehabilitation REGCTR Regional Center ALZ Alzheimer Diseases Services Other OTHER This dissertation used binary matrices to examine the resource exchange patterns of care networks. Although the original matrices had Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. values indicating the number of connections made between program types, this dissertation converted the matrices into adjacency matrices for two reasons. First, since the network matrices were composed of aggregated program types, not individual programs, the strength of ties between program types might be misleading as representing the presence of ties. For example, because of the constraints in using measures other than nodal degree to measure centrality, number of program types in which a particular program type sent resources to or received from were more important than the strength of the sending and receiving ties. Secondly, the use of structural equivalence concepts and techniques in examining the structure of a system was commonly used in adjacent but not valued matrices. It would be more difficult to partition program types into positions if each had different values in connecting with one another. Consequently, this dissertation answered the two research questions using matrices with binary values, with “0” indicating an absence of a tie and “1” its presence. The values in the original matrices, therefore, were first dichotomized into binary values (0 and 1) for subsequent analyses. Software Two network packages were used in this study, UCINET and KrackPlot. UCINET was a network analytical package network researchers frequently used to calculate descriptive properties of networks, identify 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. network subgroups, and examine inter-relationships between networks. KrackPlot was a graphing package used to transform sociomatrices into visual graphs. Some basic network properties also could be calculated from KrackPlot. This study used UCINET for its analytical and statistical analyses and KrackPlot to generate visual images for resource exchange patterns. Describing the Organization of Care Systems Several network analyses were used to answer the first research question “how is an elder care system organized at the community level?” and to test hypotheses 1,2, and 3 as stated in the conceptual chapter. Four aspects of the organization of care systems were examined: (1) extent of network development, (2) extent of the formalization of network relationships, (3) structure of the care systems, and (4) community specificity in the organization of care systems. This section illustrates how these four aspects were examined. Extent of Network Development Density and centralization were used to describe network development and to test hypotheses 1, 2a and 2b in four communities. A more detailed discussion of the use of these measures can be found in Marsden (1990). For binary sociomatrix, network density was the proportion of connections actually present in the network relative to all possible connections. For valued 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sociomatrix, it was the mean strength of connections among service-provider types. Since it indicated the level of interorganizational activity in the network, density could be used as a rough measure of network integration. However, Marsden (1990) cautioned that density could be a problematic index of structural cohesion if a network had subgroups: within these subgroups individual entities might find high connections with each other but there might be only a few linkages between these subgroups, making the overall network less well-connected. Centralization measure was a network-level indicator based on individual degree centrality in the network. Degree centrality was a measure of an actor’s (a provider’s) position in a network and was calculated by the number of incoming and outgoing connections the provider develops with other network actors. Indegree centrality referred to the number of incoming connections of an actor whereas outdegree centrality represented the number of its outgoing links. Instead of measuring the degree of integration at the network level, degree centrality denoted the extent to which a particular actor/organization is connected to other actors in the network (Burt, 1980). The corresponding network-level measure for centrality was the network centralization index. Measuring the extent to which resource exchanges were centralized into a few actors, the index was the aggregation of all degree centrality of all actors in the network (Wasserman & Faust, 1994). This dissertation used binary sociomatrices to calculate density, degree centrality 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for individual actors, and network centralization index for different resource networks. Extent of Formalization of Network Relationships To test for hypotheses 3a (“Money and staff exchanges are more likely to be associated with formalized exchanges than client and information exchanges.”) and 3b (“Client and information exchanges are more likely to be associated with informal exchanges than money and staff exchanges.”), the distribution of exchanges in various degrees of formalization was described for four communities using chi-square tests. The various degrees of formalization were operationalized into three discrete categories: “contractual agreements” denoting the most formalized arrangement between service providers in their exchange relations, “formal verbal agreements” and “informal understanding” from which the degree of formalization was minimal. Moreover, these discrete categories were used as variables in associating with the four resource exchange patterns in four communities. Using the multiple regression quadratic assignment procedure (MRQAP), variables that were significantly associated with any of the four resource exchange patterns could be identified. Description and the explanation of using MRQAP instead of ordinary multiple regression are detailed in a later section in this chapter. 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Structure of the Organization of Care Systems As described in the conceptual chapter, two approaches were used to examine the structure of care systems. Burt (1980) discussed the relational and positional approaches to analyze the patterns of relations. Relational approach described the network structure in terms of connectivity and focused on the presence of cohesive groups in the network. These cohesive groups were called network cliques whose members were connected to one another by strong relations (Wasserman & Faust, 1997). Usually, it required at least three actors having relationships strong enough to form a clique. A network could have many large and overlapping cliques. This study used UCINET to perform clique analysis to identify network cliques in all the resource networks across the four communities. Following Provan and Sebastian’s (1998) procedure of clique identification, this study adopted the strictest clique definition whereby organizations maintaining two reciprocal ties with each other were considered in the clique. This study used mutuality of ties as the criterion for defining a clique. A clique consisted of three or more actors who maintain at least two reciprocated ties with the other two members of the cliques. This criterion was stringent because it required a higher frequency of ties among members. Before performing the clique analysis, valued matrices of the four resources were dichotomized by using a cut-off value of “2.” Any cell entry of less than 2 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. was recoded as “0.” Then these dichotomized matrices were symmetrized so that reciprocated pairs of actors could be identified in the clique analysis. The “minimum” option assigned “0" to those entries whose values were not equal, indicating that a reciprocal tie does not exist. Clique analysis was then performed. When a resource network did not have any cliques under this definition, the criterion was relaxed to contain only one exchange among at least three network clique members. In order to examine the extent of cohesion of these cliques, subgroup cohesion ratio was calculated. The ratio was the average strength of ties within the clique versus that from clique members to non-clique outsiders. Although this dissertation used binary matrices to identify cliques, valued matrices were used to calculate the subgroup cohesion ratio. For a value relation, the numerator is the average strength of intra-clique ties and the denominator is the average strength of ties between clique members and nonmembers. If the ratio equaled or was greater than 1, the strength of ties did not differ within the subgroup as compared to outside the subgroup. Positional approach, on the other hand, described network structure through status or role sets (Burt, 1980; White et al., 1976). Individual entities of the same role set jointly occupied the same position in the network. This position denoted similarity of relations between these entities and the others in the system, although entities within the same position might not have strong relations with each other. Entities having similar relations with others in the 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. system were considered structurally equivalent to each other (Faust & Wasserman, 1994). Two procedures were essential in examining the network structure of multiple relations. First, based on the sociomatrices, network actors were partitioned into positions of structural equivalence through the use of a clustering algorithm. Second, relations within and between these positions were described in terms of density tables, image matrices, and image graphs. The relations between density tables, image matrices, and image graphs are illustrated in Figure 3. Density tables consist of densities among all the positions identified from the clustering algorithm. With exactly the same number of rows and columns, image matrices are simplified version of density tables based on certain criteria to generate zero-blocks and one-blocks. General criteria include perfect-fit, fat-fit, lean-fit, and a density criteria (Faust & Wasserman, 1994). Network researchers used one of these criteria to compare intra-position and inter-position densities in the density table and generate image matrices. Based on the image matrices, they produced image graphs. Perfect-fit criterion assigned “1” only to those cells in an image matrix with “0” and “1” to those that were not ones. Fat-fit criterion assigned “1" to cells that had a density value greater than “0” and lean-fit criterion assigned “0" to cells whose density value was lower than “1.” The a density criterion assigned “1” to cells that had a density value greater than the grand density of the original matrix. 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Image Matrix Using a density criterion (0.4) Density Table 1 2 3 4 1 0.3 0.5 0.2 0.4 2 o o ' 1.0 0.7 C O o ' 3 1.0 C M o ' 0.5 o o ' 4 0.7 G O o ' 0.3 C D o ' 1 2 3 4 1 0 1 0 1 2 0 1 1 1 3 1 0 1 0 4 1 1 0 1 Image Graph Figure 3. Illustration on the Relationships Between Density Table. Image Matrix, and Image Graph. These representations were considered as part of the processes to construct blockmodels. Consisted of zero-blocks and one-blocks, blockmodels were statements in matrix form describing relational patterns among positions. They can be considered more than a mere description of structural relations between two positions (Panning, 1982; White et al., 1976). In some cases, hypotheses can be developed into blockmodels from which empirical data can be tested. Using the positional approach to examine network structure, this study stacked all the resource sociomatrices into a supermatrix. The valued matrices of each community were first converted into binary matrices because this study was interested mainly in a group of service providers that maintained structurally equivalent resource exchange linkages in the network, not in the amount or intensity of exchanges. Binary matrices were then merged into a supermatrix for Convergence of Iterated Correlations (CONCOR) analyses. The idea was to examine the role of service providers in a care system, with their exchanged resources considered. Given the diverging exchange patterns in the four networks, were there any structurally equivalent groups? Did these groups reveal a certain level of “systemness” in the community-based care system? The CONCOR technique was developed by Breiger, Boorman, and Arabie (1975). It started with correlating rows and columns of a matrix to generate a correlation matrix first. Then the CONCOR technique repeated 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the procedure iteratively until the convergence of correlation coefficients of each row to either “1” or “- 1 I t was from this converged correlation matrix that a profile similarity matrix was generated and the partition of the original matrix into positions was achieved. This dissertation used two criteria to decide the appropriate level of the CONCOR results: the number of structurally equivalent groups and the degree of R-squared for that particular split level. Since CONCOR started with treating all service providers in one group (i.e. they are all structurally equivalent to one another), increasing the partition level would increase the number of structurally equivalent groups. The R-square values represented the correlation between the original matrix and the correlation matrix on which the partition was based. Therefore, R-square for lower partition level appeared to be lower. Theoretically the number of groups could be equal to the number of service providers in the network. Intra-position and inter-position densities were calculated to form density tables for each community. The a density criterion was used to determine whether a block was a zero-block or one-block. Blockmodels were used to describe the relational pattern among positions, and image graphs were produced. The relational and positional approaches provided complementary ways to examine the structure of community-based systems of care. The relational approach helped identify cohesive groups in the system and to 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. evaluate the extent of fragmentation beyond simple descriptive network measures. It provided information on which service providers formed a core in the network. Instead of identifying “islands" of connected actors, the positional approach aimed at examining the global pattern of relations among service providers in the system. Moreover, this approach was capable of examining multiple relations in a system simultaneously, making analyses of complex systems possible (Panning, 1982). Community Specificity in the Organization of Care Systems While not explicitly stated in the form of a hypothesis, this study was also interested in whether communities displayed different patterns in their organization of care for older adults. In this study, service provider involvement was compared across communities. The number of various service providers participating in the networks, as well as their perception of interorganizational need was described. For the latter, chi-square was performed to examine if significant community differences exist. Moreover, this study also examined the intensity of service provider involvement in terms of their multiplexity of exchanges. According to Morrissey (1985), intensity represented the number of different resources being exchanged between service providers. The more resource types were exchanged among providers, the more integrated and complex the system was. Intensity also indicated the extent of service provider involvement in interorganizational 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. activities. Four categories of multiplexity were developed: “single-tie exchange,” “two-tie exchange,” “three-tie exchange,” and “four-tie exchange.” Chi-square was performed to examine significant community differences. The organization of care systems in each community also could be compared through the differences in their prominent service-provider types. In order to identify prominent service-provider types for community-based care systems, this study used the one standard deviation above the mean degree centrality of the corresponding resource network as the cut-off value for prominent service providers. Prominent receivers in the community-based care network should have an indegree centrality value equal to or above the cut-off indegree value. Similarly, prominent senders should have an outdegree centrality value equal to or above the cut-off outdegree value. For prominent brokers, their indegree and outdegree centrality values should both exceed the respective cut-off values of their resource networks. Examining the Factors Associating with Resource Exchange Patterns A second research question of this study asks what factors are associated with the organization of care systems in a community. Items from the SEED evaluation questionnaire were selected for various dimensions associated with resource exchange patterns across four communities (Table 5). In order to develop parsimony for the model, items selected were ombined into a scale if they achieved high face validity and statistical correlation. 97 permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 5 Model Perspectives and Their Variables Perspectives Variables Selected Items/Questions Original Coding Combined Scales Governance MANDATE Relationship maintained as required by government or funding regulations Yes/No Power MY INFLUENCE THEIR INFLUENCE To what extent does your program influence this program? To what extent do they influence your program? 4-point Likert Scale 4-point Likert Scale IN FLU E N C E -absolute difference between MY INFLUENCE and THEIR INFLUENCE Transactional Cost LOW COST SAME ORGAN IZATION Relationship existed because the partner is a low cost provider Relationship maintained because of being part of same organization Yes/No Yes/No Resource Dependence ONLY AVAILABLE PROXIM ATE SAME NETW ORK MY NEED THEIR NEED Relationship existed because partner is the only provider available Relationship existed because partner is physically close to organization Relationship existed because of being part of the network How much do you need this program to accomplish your program's goals? How much does this program need you to accomplish its goals? Yes/No Yes/No Yes/No 4-point Likert Scale 4-point Likert Scale INTER-DEPENDENCE - absolute difference between MY NEED and TINDER Domain Similarity SAME GOAL SAME FUND SAME CLIENT To what extent do you share the same goals? To what extent do you share the same funding source? To what extent do you share the same clients? 4-point Likert Scale 4-point Likert Scale 4-point Likert Scale DOMAIN S IM IL A R IT Y -su m of SAME GOAL, SAME FUND, and SAME CLIENT C D 0 0 Quadratic assignment procedure multiple regression (MRQAP) was used to examine the association between hypothesized factors and the resource exchange patterns. MRQAP was a network regression technique that was specifically designed to handle the problem of possible interrelations among observations (Krackhardt, 1988). Since the unit of analysis under this research question was the resource exchange relations between two service providers, service providers’ relations with one another were likely to relate to other relations with their partners in the network. MRQAP would reduce the bias resulted from the interdependence of observations if the ordinary least square regression was performed. Two steps were involved in MRQAP. First, an ordinary multiple regression was performed between corresponding cells of the dependent and independent matrices. This was exactly like the ordinary least square multiple regression commonly used in social sciences research. Second, MRQAP randomly rearranged rows and columns of the dependent matrix (called permutation) and performed the regression again, while keeping the resultant values of r-square and all coefficients for each rearrangement in store. For each coefficient, MRQAP counts the percentage of random permutations that generated a coefficient as large as the one calculated from the first step. The second step serves as a goodness-of-fit test for MRQAP. With the exception of DOMAIN SIMILARITY, INFLUENCE, and INTER DEPENDENCE, valued matrices were converted to binary matrices first and 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. their bivariate relationships with the four resource networks were examined through QAP correlation. Then all matrices were normalized before entering the model for MRQAP. Normalizing matrices generated standardized coefficients for the MRQAP results to compare the relative magnitude among statistically significant variables to the dependent variables. Summary This chapter described the nature of the data needed and various network analyses used to answer the two research questions and to test for the hypotheses stated in the conceptual chapter. The data set used in this study was collected in 1991 from a larger system development study in California. In order to modify the data to be usable for this study, information from individual service provider respondents from that study was transformed into sociomatrices. In order to test hypotheses 1, 2a, 2b, 3a, 3b and 4, four aspects of the organization of community-based care were examined. These included the extent of network development, the extent of formalization, the structure of the care system and the community specificity in care systems. Network measures (such as density and centralization) and service-provider measures (such as indegree and outdegree centrality) were used. Community-based measures such as proportion of multiplexity of exchanges, degree of service provider involvement were also examined and tested by using chi-square tests. Moreover, techniques used to identify cliques and 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. structurally equivalent groups were employed to explore the structure underlying the resource exchange patterns of care. In order to test hypotheses 5 and 6, MRQAP was used to test the multiple-factor models. The next chapter presented the results. 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER V: RESULTS This chapter presents results for the two research questions stated in the conceptual chapter. More specifically, it structures the presentation of the results into two sections. The first section displays results related to the first research question of this dissertation (how is the elder care system organized in the community level?). The second section describes results from the MRQAP models to examine the second research question which concerns factors associated with the organization of care systems. Description of Organization of Community-Based Care Systems In this section, four aspects of the organization of community-based care systems are described and discussed. First, different dynamics of resource exchanges patterns is illustrated through various network measures (such as density and network centralization). Second, findings on the extent of formalization in resource exchange patterns are presented. Third, results from using relational and positional approaches on exploring the structure underlying resource exchanges patterns are discussed. Fourth, community differences in their care system organization are presented through the level of involvement of service providers and the prominence of various service providers in the community. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Extent of Network Development This section presents network-level measures that described the extent of involvement of service providers in terms of their resource exchange relations in community-based care systems. Figures C1 through C16 (in Appendix C) depicted patterns of resource exchanges of the four communities. A casual glimpse of these graphs shows a large amount of exchange relationships is established in client and information resources, with less observed in money and staff. Network density, an important network-level index, measures the sparseness of the overall network. As introduced in the Method chapter, though a rough number, density describes structural integration among service providers in a community-based care system. Table 6 presents density values of the resource networks across the four communities and shows that more service providers are involved in client and information exchanges than in money and staff exchanges. For instances, service providers in San Mateo make 18 percent of all the possible linkages in client and information networks, but only 4 to 6 percent of money and staff linkages. The relatively higher density in the client and information networks is another way to characterize the overall care system dominated by client and information exchanges. Moreover, the density values of client and information networks are also identical, suggesting the prevalence of client- information exchanges in all the communities. 103 permission of the copyright owner. Further reproduction prohibited without permission. Table 6 Density and Standard Deviation of Various Resource Networks Across the Selected Communities Number of Service- Provider Types Client Network Information Network Money Network Staff Network San Mateo 24 0.18 0.18 0.04 0.06 (0.39) (0.39) (0.20) (0.24) Long Beach 18 0.22 0.21 0.05 0.05 (0.41) (0.4) (0.22) (0.22) Tulare 20 0.29 0.3 0.03 0.08 (0.45) (0.46) (0.18) (0.27) San Francisco 29 0.23 0.25 0.05 0.05 (0.42) (0.43) (0.22) (0.22) Table 6 also shows the different densities among the four communities. Variations seem to be greater in the client and information networks than in the money and staff networks. For instances, given more service-provider types (24) in the network, San Mateo actually has fewer client and information exchanges among their service-provider types than Long Beach (18) and Tulare (20). Among all the communities, Tulare has the densest client, information, and staff networks. Table 7 reports the network centralization measures in the four communities. The larger the centralization value, the more centralized the network; this is because high centralization suggests most of the resource exchanges are controlled by a few service-provider types. Table 7 shows that resources are in general not distributed uniformly across various 104 permission of the copyright owner. Further reproduction prohibited without permission. service-provider types. For instances, the relatively higher centralization values of client and information networks (compared to money and staff networks) suggests that there are a few service-provider types controlling client and information exchanges, a phenomenon less apparently found in the money and staff networks. Table 7 Network Centralization Index of Resource Networks Across the Selected Communities Client Information Money Staff San Mateo Sending 32.4% 27.7% 9.9% 21.7% Receiving 27.7% 32.4% 19.4% 26.5% Long Beach Sending 28.3% 36.4% 27.6% 27.6% Receiving 28.3% 36.4% 7.7% 14.3% Tulare Sending 26.3% 30.9% 13.7% 14.6% Receiving 43.9% 30.9% 7.9% 14.6% San Francisco Sending 32.9% 35.1% 21.4% 21.2% Receiving 32.9% 42.7% 21.4% 25.0% The uneven distribution of resource exchanges is also evident in the differences found in the centralization values of sending and receiving exchanges. While most of the client and information resource flows tends to be relatively balanced in sending out resources and receiving them, client referral in Tulare seems directed towards a smaller number of service- provider types: its indegree centralization (43.9 percent) is much higher than its outdegree value (26.3 percent). Similarly, compared to the out- 105 permission of the copyright owner. Further reproduction prohibited without permission. flowing centralization, the high indegree in the San Francisco information network suggests that the concentration of information flow towards a limited number of service-provider types. Despite these differences, some communities show identical outdegree and indegree centralization values. For instances, the client referral and information networks in Long Beach, the information network in Tulare, and the client and money networks in San Francisco all display identical outdegree and indegree values. This finding suggests that within the resource exchange patterns, there is no particular service-provider types dominating/controlling the resource exchange in only one direction. If these networks are centralized, they are equally centralized in terms of their incoming and outgoing exchanges. Differences are also found between resource networks, reflecting various exchange dynamics underlying each resource. In general, the distribution of the client and information exchanges tends to be more uneven and targeted towards particular service-provider types than the money and staff exchanges. Given what is learned from Table 6 on the similarities between client referral and information exchange networks, however, little similarity is found in the outdegree and indegree centralization measures between these networks. For instances, while the outdegree and indegree centralization are identical for client and information network in Long Beach (28.3% and 36.4%), the information network is actually more centralized than the client network. 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lastly, communities report different centralization measures and display different dynamics in how their resources flowed through their care networks. For examples, San Francisco tends to be relatively “balanced” in its resource exchanges. Apart from the high in degree centralization in the information network, most resources are exchanged rather evenly in their networks. Moreover, compared to the other communities, the money and staff networks in San Francisco not only have a more even distribution between their incoming and outgoing exchanges, but also are more centralized. The outgoing exchanges of money and staff in Long Beach tend to be more centralized than their incoming exchanges. Extent of Formalization of Network Relationships A second way to describe the organization of care systems is to examine how formalized these relationships are. Table 8 indicates that a significant majority of relations among service providers is informal in all the four communities. Informal relationships, which account for 48% to 65% of all relations across four communities, are the backbone of how service providers are related to one another in care networks. Besides informal agreement, combinations of formal and contractual arrangement with informal understanding are also important for service providers to maintain their inter-organizational relations. Together, they accounted for 10% to 29% of all service-provider relations, depending on the community. 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 8 Extent of Formalization San Mateo Long Beach Tulare San Francisco Contractual Arrangement Only 12% 14.5% 4.3% 12% Formal Verbal Arrangement Only 9% 1.8% 2.6% 2.4% Informal Agreement Only 61% 54.5% 65.5% 48.8% Contractual and Formal Arrangement 3% 0% 0% 1.2% Formal and Informal Arrangement 1% 5.5% 11.2% 15.1% Contractual and Informal Arrangement 9% 23.6% 12.9% 4.2% Contractual, Formal, and Informal Arrangement 5% 0% 3.4% 16.3% Chi-square p value = 0.000 There are significant community differences in the composition of formal and informal relationships among service providers. For instances, compared to the other communities, Tulare has a significantly lower percent of relations solely for contractual or formal agreement (4.3% and 2.6% respectively), yet has a much higher proportion of relations as informal understanding among service providers. San Francisco has a significantly higher percent of relations characterized as a combined “contractual, formal, and informal” type (16.3%), compared to 5% in San Mateo and 3.4% in Tulare. 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Assessing the extent to which the resource exchange patterns are associated with various relationships, multiple regression quadratic assignment procedure (MRQAP) was performed and results are shown in Table 9. Table 9 indicates that informal arrangement dominates the client resource exchanges. In all communities, informal arrangement is the only significant variable related to the client resource network at the 0.001 significance level. Variation is found in the information network where the role of formal-verbal arrangement becomes more apparent in some communities (such as San Mateo and San Francisco). As expected, contractual arrangement characterizes most of the money networks, although the unique role of informal relationship remains in associating with the money network in San Francisco. More variation is found in characterizing staff network. Contractual, formal-verbal, and informal arrangements bear their unique contribution to the staff network in different communities. For example, staff exchanges in San Francisco and Tulare are mainly supported by formal-verbal arrangement, whereas in San Mateo and Long Beach, informal arrangement seems to have a larger role in associating with staff exchanges. Structure Underlying the Organization of Care Systems This section describes and discusses findings from relational and positional analyses on the examination of the structure underlying the organization of community-based care systems. It first presents general 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9 QAP Regression Results Between Mode of Formalization and Resource Networks Client Information Money______ Staff San Mateo Intercept Contractual Formal-Verbal Informal 0 0.1859* -0.0028 0.5104** 39.6%** 0 0.0746 0.1272* 0.5910** 48.3%** 0 0.3369*' 0.0558 -0.1724 9.1%** -0.0252** 0.04768 -0.1246 0.2137** 4.8%** Long Beach Intercept 0 0 0 -0.0456 Contractual 0.1374 -0.0012 0.2941** 0.0957* Formal-Verbal 0.0831 0.085 -0.0216 0.0378 Informal 0.5937** 0.7248** 0.1815* 0.337** R2 45.9%** 53.7%** 17.1%** 35.6%** Tulare Intercept Contractual Formal-Verbal Informal R2 0 0.0047 0.0892 0.5931* 40.2%*' 0 0.1053 0.0457 0.6076** 44.1%** 0 0.3241** 0.014 -0.0195 10.3%** -0.0164 0.1641* 0.1832** 0.0657 10.5%** San Francisco Intercept 0 0 0 -0.0381 Contractual 0.1582 0.1629** 0.3805** 0.0386 Formal-Verbal 0.0482 0.0816* -0.0201 0.2062 Informal 0.6382** 0.5198** 0.1841** 0.1789 R2 57.4%** 44.3%** 23.7%** 18%** *p<0.05; ** p< 0.01 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. observation of cliques across all communities. Then it presents results from the blockmodel analysis. For each of these subsections, the findings are organized according to individual communities. Cliques General Observations Across All Communities. Figure 4 shows that not all the resource exchanges have cliques. In all four communities, service-provider types engaging in money and staff exchanges are not cohesive enough to establish cliques. It also shows that almost all the communities had cliques with the least number of service providers as members. The absence of larger cliques suggests that despite the fact that service-provider types have extensive exchanges with one another, cohesive groups are rare in these communities. A relatively high degree of clique overlap can be found in Figure 4 as well. Clique overlap occurs when service-providers belong to at least two cliques simultaneously, whether these cliques are of the same resource type. The overlap of cliques of the same resource might imply that the service provider that belong to at least two cliques act as broker for these cliques. For example, Social Services (DPIHAPS) is a significant member in all three information-cliques and two client-cliques in Tulare. The absence of Social Services (DPIHAPS) would disintegrate the structure of all the cliques in Tulare. 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Adult Day Health Care (ADHC) and Health Clinics (HEACLIN) belonged to the same client-clique and information-clique. Their “dual” membership in these cliques characterizes the interaction of the two cliques and promoted further functional integration of the care networks. Compared the multiple affiliation of Adult Day Health Care (ADHC) and Health Clinics (HEACLIN) to the cliques in San Francisco with that of the Linkage/Multi-Serivice Senior Program (LINKMSSP), Adult Day Health Care (ADHC), Home Health Care (HHC), and Hospitals (HOSPITAL) in San Mateo, the “functional integration” is considered greater in the latter because all the client-clique members are also members of the same information cliques. No service-providers are excluded from exchanging both information and clients with one another in these cliques. Extensive overlap is also evident between two client-cliques and information-cliques found in Tulare. The high degree of overlap suggests that service providers referring clients to one another are also likely to exchange information, providing further evidence in the closeness between client and information. The presence of non-overlapping cliques could suggest that service providers of each clique might not have common clients or goal with members in the other cliques. This situation was found in the two information-cliques in San Francisco. One information-clique is composed of Senior Centers (SENIORCTR), Information and Referral (l&R), and Transportation (TRANSPRT), and the other of Adult Day Health Care 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. A.San Mateo Cliques LINKMS HHC HOSPITAN DH C.Tulare Cliques MH l^QPIHAPS’ \ PUBGUR GOUNTYHD \ CJHARipl' Figure 4. Cliques. B.Lona Beach Cliques LINKMSSP DPIHAPS Client Information DPIHAP ADHC HEACLIN \ pASEMGT/ D.San Francisco Cliques /SENIORCTRv l&R \ TRANSPRi \ / (ADHC), Health Clinics (HEACLIN), and Case Management (CASEMGT). It appears that one clique focuses on social activities for older adults whereas the other concentrated on the health services. Specific Observation For Each Community. Few cliques are apparent in San Mateo (Figure 4A); there are two cliques in the client network and only one information-clique. The first client-clique is composed of the Linkages/Multi-Service Senior Program (LINKMSSP), Hospitals (HOSPITAL), and Adult Day Health Care (ADHC) and the second one is composed of Linkage/Multi-Service Senior Program (LINKMSSP), Home Health Care (HHC), and Adult Day Health Care (ADHC). Both cliques contain two members in common and differ only by a single member. However, these four service providers do not form a bigger clique, because no strong exchange relations are developed between Hospitals (HOSPITAL) and Home Health Care (HHC). Nevertheless, they do form a larger clique in the information exchange network (Linkages/MSSP Program, Hospital, Home Health Care, and Adult Day Health Care). No clique is found among organizations that maintained money or staff exchanges. On the other hand, it appears that there are close relationships between client-cliques and information-clique. The smaller cliques in the client network are “nested" within the larger clique in the information network (Figure 4A). 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The structure of cliques in Long Beach is unique. The Linkages/Multi-Service Senior Program (LINKMSSP), Home Health Care (HHC), and Social Services (DPIHAPS) form the only clique in the client network, referring and receiving at least two ties to one another (Figure 4B). No clique is found in the other three networks. If the cut-off value of the dichotomized matrices was “one” instead of “two,” meaning that service providers exchanged information at least once from one another to be considered as members in the clique, one clique was found (Senior Centers, Social Services, Hospital, and Charity). However, the information clique bear little resemblance to the client clique in Long Beach. In Tulare, clique analysis shows two cliques in the client network and three in the information network. Again, no clique is found in the money or staff network (Figure 4C). The two client network cliques are (1) Area Agencies on Aging (AAA), Senior Meal Program (MOW), Social Services (DPIHAPS) and (2) Social Services (DPIHAPS), Mental Health (MH), Public Guardian (PUBGUR), with Social Services (DPIHAPS) as the “common” clique member. The three information network cliques are (1) Area Agencies on Aging (AAA), Senior Meal Programs (MOW), Social Services (DPIHAPS); (2) Social Services (DPIHAPS), Mental Health (MH), Public Guardian (PUBGUR); and (3) Social Services (DPIHAPS), County Health Department (COUNTYHD), Charity (CHARITY). Compared to the other communities, Tulare has the highest number of cliques involved with seven service-providers, suggesting that it might be relatively more structurally 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. integrated. On the other hand, its clique formation could be related to the fact that fewer providers existed in the community, which might encourage tighter connections. Compared to the pattern in San Mateo, Tulare does not have a high degree of overlapping cliques within each resource exchange pattern, although two cliques are identical in both client and information cliques. This again might help confirm the close relation between client and information linkages. Despite its higher density and larger number of service provider types in the network, San Francisco has only one clique in its client network and two in the information network (Figure 4D). The one client clique consists of Social Services (DPIHAPS), Adult Day Health Care (ADHC), and Health Clinics (HEACLIN). The two information cliques are (1) Senior Centers (SENIORCTR), Information and Referral (l&R), Transportation (TRANSPRT); and (2) Adult Day Health Care (ADHC), Health Clinics (HEACLIN), Case Management (CASEMGMT). There is no significant clique overlap found in these networks, although Adult Day Health Care (ADHC) and Health Clinics (HEACLIN) are the two service providers treating each other as members in client and information networks. No clique is found in money network when the “at least two tie”’ criterion is considered. One clique (Senior Center, Adult Day Health Care, Transportation) was found if this criterion was loosened to include one tie. Two members in this clique are also found to be members in one of the information cliques. Again, some level of nesting in terms of the frequency 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of ties can also be found. The higher level of tie interaction in information clique might encourage developing money clique. While information-cliques tended to be more volatile in nature across the four communities, client-cliques showed some recognizable patterns. Almost all the client-cliques are composed of either health or social service- providers with a publicly-administered program such as the Linkages/Multi- Service Senior Program (LINKMSSP) or Social Services (DPIHAPS). Network Subgroup Cohesion. Table 10 presents the subgroup cohesion measures of various cliques in each community. Because the tie criterion for cliques is “two,” all the subgroup cohesion measures are above one and ranged from 3.57 to 6.74. The high ratio suggests that clique members maintain a very close resource exchange relationship with each other compared to those linkages they make with non-clique members. For example, in San Mateo, clique members on average maintain at least four client ties with one another but less than one tie with those outside the clique. To maintain high/intensive ties with clique members required high coordination cost such that most service-providers might not be able to afford to include more partners in this closely-knitted group, hence restricting the size of these cliques. For cliques that have the same service providers as members, their tie strength among members differs across networks. This finding might suggest the dynamics among clique members in different types of resource exchange. For example, in Tulare, Area Agencies on Aging-Senior Meal 117 permission of the copyright owner. Further reproduction prohibited without permission. Program-Social Services (AAA-MOW-DPIHAPS) is a clique in both client and information networks. Yet, their subgroup cohesion measures differ by one whole tie (5.04 for client clique and 3.97 for information clique). This is mainly due to a stronger tie among the clique members in their client referrals (2.65 versus 2.5 for information clique) and a stronger tie between the same group of clique members with nonmembers in exchanging information (0.63 versus 0.53 for client clique). Compared to the other clique found in the client and information networks (for example, the clique formed by Social Services (DPIHAPS), Mental Health (MH) and Public Guardian (PUBGUR), the variation of AAA-MOW-DPIHAPS in how members related to one another and to nonmembers was higher. Table 10 indicates that great variance is found in intra-clique density across resource network and communities. Intra-clique density varies from 2 in San Francisco’s client and information cliques to 4.5 in San Mateo’s client-clique. Comparatively, the variance of how clique members make ties with non-clique members is less, ranging from 0.44 to 0.68. The below- one level of the tie strength between clique members and nonmembers suggests that there are many nonmembers in the overall care network. Table 10 also shows that with higher subgroup cohesion ratios, providers in San Mateo and Long Beach are more integrated than those in Tulare and San Francisco. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10 Subgroup Cohesion Resource Clique Members Intra- clique Density Ties strength with non- members Sub group cohesion ratio San Mateo Client Linkage/MSSP Program (LINKMSSP) Hospital (HOSPITAL) Adult Day Health Care (ADHC) 4.17 0.68 6.13 Linkage/MSSP Program (LINKMSSP) Home Health Care (HHC) Adult Day Health Care (ADHC) 4.5 0.59 7.63 Information Linkage/MSSP Program (LINKMSSP) Home Health Care (HHC) Hospital (HOSPITAL) Adult Day Health Care (ADHC) 2.94 0.45 6.53 Client Long Beach Linkage/MSSP Program (LINKMSSP) Home Health Care (HHC) Social Services (DPIHAPS) 2.83 0.42 6.74 Tulare Client Area Agencies on Aging (AAA) Senior Meal Programs (MOW) Social Services (DPIHAPS) 2.67 0.53 5.04 Social Services (DPIHAPS) Mental Health (MH) Public Guardian (PUBGUR) 3 0.55 5.45 Information Area Agencies on Aging (AAA) Senior Meal Programs (MOW) Social Services (DPIHAPS) Social Services (DPIHAPS) Mental Health (MH) Public Guardian (PUBGUR) 2.5 3.33 0.63 0.59 3.97 5.64 Social Services (DPIHAPS) County Health Department (COUNTYHD) Charity (CHARITY) 3.5 0.65 5.38 San Francisco Client Social Services (DPIHAPS) Adult Day Health Care (ADHC) Health Clinic (HEACLIN) 2 0.56 3.57 Information Senior Centers (SENIORCTR) Information & Referral (l&R) Transportation (TRANSPRT) 3 0.46 6.52 Adult Day Health Care (ADHC) Health Clinic (HEACLIN) Case Management (CASEMGT) 2 0.44 4.55 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Structurally Equivalent Groups In addition to the relational analysis, the structure of a community- based care system also is examined by using positional analysis. Positional analysis generated structurally equivalent groups in which service providers bore approximately similar relational pattern with other service providers. In the following section, structurally equivalent groups are identified from the resource networks in the four communities. The convergence of iterative correlation (CONCOR) algorithm was used as the major technique to identify these structurally equivalent groups. Resource exchange patterns of these structurally equivalent groups were represented in the form of blockmodeis. The partition results of each community are presented first, followed by a discussion of how each structurally equivalent group was related to one another in the form of blockmodeis. CONCOR Results. Table 11 shows CONCOR partition results for the four communities. In general, the resultant partitions yield higher R-square values in the client and information partitioned sociomat rices than those of the money and staff. Moreover, the third-level partition seems to be capable of reflecting more of the care systems in all the other communities than in San Francisco: the R-square values in all the resource networks in San Francisco are much lower than the other communities. Some residuals are found in some communities. Although these residuals might maintain 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extensive exchange relations with other structurally equivalent groups in the network, their unique relations with other providers, and hence inability to be grouped with other providers should render careful examination of their roles in the care system. Table 11 CONCOR Results Community Partition Level Number of Groups Number of Residuals R-square (Correlation Coefficients) Client Information Money Staff San Mateo 3 7 1 0.36 0.33 0.29 0.20 Long Beach 3 6 1 0.41 0.34 0.17 0.19 Tulare 3 7 1 0.39 0.36 0.17 0.24 San Francisco 3 7 0 0.25 0.27 0.10 0.15 San Mateo Structure Table 12 displays the partition results for San Mateo and the relational patterns among the structurally equivalent groups in blockmodel format. CONCOR analysis yields seven structurally equivalent groups and one residual service-provider in San Mateo. Blockmodel relations across resources are presented in Figures C17 through C20 (in Appendix C). 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 12 Structurally Equivalent Groups in San Mateo with Their Relational Patterns in Blockmodeis 1: Linkages/Multi-Service Senior Programs, Home Health Care, Hospital, Ombudsman, Retired Senior Volunteer 2: Housing, County Health Department, Regional Center 3: Skilled Nursing Facilities, Social Security Agencies, Public Guardian 4: Emergency Services, Social Sen/ices, Adult Day Health Care 5: Senior Centers, Information and Referral, Charity, Other 6: Legal Aid, Transportation 7: Senior Meals Program, Mental Health, Money Management 8: Rehabilitation Client Information Money Staff Multiplex Ties 123 4 56 7 8 1 234 567 8 123456 78 1 234 567 8 1 2 3 4 5 6 7 8 1 10 1 10011 2 0 0 0 0 00 0 1 3 00000 001 4 110 1110 0 5 000 1 001 1 6 0 0 0 0 0 0 0 0 7 00001 00 1 8 1110 10 10 10 0 11001 0 0 0 10 001 0 0 0 0 0 0 0 1 100 1 11 0 0 000 10010 0 0 0 1 00 0 0 1000 1 00 1 1100 1000 10 0 1100 1 000 0000 1 0 0 0 0 00 0 0 0000 1 1 00 000 1 00 1 0 0 0 0 1 00 0 0 1000 100 1 010 0 00 1 0 10 0 10 000 0 0 0 110 11 0 00 0 0 0 0 0 110 11000 0 10 10000 00 0 1 0 0 0 0 1100 1000 0 10 0 0 0 10 4 0 1 4 2 0 1 3 0002 1 01 4 0 0 0 0 0 0 0 2 3 2 0 3 4 2 0 0 0 1040 03 1 0 0 0 3 0 0 0 0 3 1 0 0 4 0 0 3 2 4 1 0 2 0 3 0 Group Description— Identifying various partitioned groups is only the first step toward an understanding of the structure of the care system. How these groups are related to each other in their exchange linkages reveals this structure. For instance, Group 1 encompasses almost all of the medical related providers (Linkage/MSSP Program, Home Health Care, Hospital, Ombudsman, and Retired Senior Volunteers) and maintains exchange relations with five other structurally equivalent groups (Groups 3, 4, 5, 7, and 8). It is one of the three most active structurally equivalent groups in the network. Most of their relations with these partners are 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reciprocal in nature, especially in the client and information networks, where half of its exchange relations are reciprocal. On the other hand, money and staff exchanges do not possess this characteristic of reciprocal as frequently as client and information exchanges. Most relations between Group 1 and its partners tended to be multiplex too. Regardless of the direction of exchanges, four out of its five partners have multiple resource exchanges with Group 1, except Group 3 (Skilled Nursing Facilities, Social Security, and Public Guardian). This tight exchange pattern suggests a close relationship between Group 1 and its partners. If reciprocity is also considered under the context of muitiplexity, Group 1 has close relations with Group 4 (Emergency Services, Social Services, and Adult Day Health Care) and Group 8 (Rehabilitation), in that it exchanges all resources with Group 4. Service providers of Group 2 (Housing, County Health Department, Regional Center) also have a comparable number of structurally equivalent groups to exchange resources with. Group 2 maintains mostly reciprocal exchange relations with Group 4 (Emergency Services, Social Services, and Adult Day Health Care), 5 (Senior Centers, Informational & Referral, Charity, and Others), 7 (Senior Meal Program, Mental Health, Money Management), 8 (Rehabilitation). Nevertheless, the amount of exchanges is much less than that of Group 1, suggesting a low degree of muitiplexity between Group 2 and its partners and that Group 2 is not a very active group in the network. With the exception of exchanging all four resources 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with Group 8 (Rehabilitation), most of the exchanges contain one of the following: client, information, and staff resources. Among all the relations Group 2 establishes with its partners, staff exchanges are the most apparent activities. One very distinctive feature of this structure is the isolated role of Group 3 in the network. Group 3 (Skilled Nursing Facilities, Social Security Agencies, and Public Guardian) is one of the least integrated groups in the network, having only two partners with mostly client referral as the exchange resource. It has only one reciprocal partner (Group 8 — Rehabilitation). Group 3 receives clients from Group 1 (Linkage/MSSP Program, Home Health Care, Hospital, Ombudsman, and Retired Senior Volunteers) and has a reciprocal client linkages with Group 8 (Rehabilitation). Besides client exchanges, it also sends information to Group 8 (Rehabilitation). Group 4 (Emergency Services, Social Services, and Adult Day Health Care) is an active group in the network, with four partners (Groups 1, 2, 5, and 6). It has multiple, reciprocal exchange relations with all its partners, but exchanges the fewest amount of resource comparatively with Group 2 (Housing, County Health Department, and Regional Center). Across all its partners, Group 4’s relation with Group 1 (Linkage/MSSP Program, Home Health Care, Hospital, Ombudsman, and Retired Senior Volunteers), Group 5 (Senior Centers, Information and Referral, Charity, and Other) and 6 (Legal Aid and Transportation) remains relatively more 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stable across resources. A careful examination of the exchange pattern of Group 4 shows that its client, information, and staff exchanges are almost identical, suggesting a high degree of muitiplexity between Group 4 and its partners. Group 5 (Senior Centers, Information and Referral, Charity, and Other) maintains reciprocal exchanges with Groups 1, 2, 4, 7, and 8. Group 4 (Emergency Services, Social Services, and Adult Day Health Care) and 7 (Senior Meals Program, Mental Health, Money Management) are the major partners exchanging multiple resources with Group 5. Group 5’s exchange patterns in the client, information, and staff networks are very similar, leaving its money exchanges distinctive compared to the other resources. In general, Group 5 is an active group in the network. Given its sole relation with Group 4 (Emergency Services, Social Services, and Adult Day Health Care), Group 6 (Legal Aid and Transportation) is the most isolated structurally equivalent group in the network. Nevertheless, its relationship with Group 4 (Emergency Services, Social Services, and Adult Day Health Care) is multiple and reciprocal: the two groups exchange all resources but staff. Group 7 (Senior Meals Program, Mental Health, and Money Management) receive and send out resources to the same set of structurally equivalent groups in the network (Groups 1, 2, 5, and 8). Among these partners, Group 5 (Senior Centers, Information and Referral, Charity, and Other) and 8 (Rehabilitation) are its most integrated partners in 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. terms of number of resources exchanged. Its linkages with its partners are also reciprocal. Group 8 (Rehabilitation) maintains ties with many groups (five in total, Groups 1, 2, 3, 5, and 7) in the network, although these ties are not very expansive in terms of the number of resources involved. Group 2 (Housing, County Health Department, Regional Center) and 7 (Senior Meals Program, Mental Health, and Money Management) are its most integrated partners, with at least three resources to be exchanged. To a large extent, residuals such as Group 8 could not be considered as important as other structurally equivalent groups, because its relations with other groups are usually too idiosyncratic. Network Description — The overall structural differentiation of the care system does not exactly follow the functional differentiation as normally anticipated in community-based long-term care, with distinction between acute and long-term care providers or social versus medical providers. Instead it is more complex. Some groups are more catered to particular segments of the older population. For example, Group 5 (Senior Centers, Information & Referral, Charity, and Other) seems to be catered to healthier older adults and to be related to social services. Group 7 (Senior Meals Program, Mental Health, and Money Management) consists of service providers who serve frail older adults who are aging in their homes rather than in institutions. Group 6 (Legal Aid and Transportation) is 126 permission of the copyright owner. Further reproduction prohibited without permission. composed of providers serving the genera! population in the community but supportive to the elder care system. Other groups are mixed in terms of the types of care their service providers provided. For example, Group 4 (Emergency Services, Social Services, and Adult Day Health Care) has providers of social and health services. Group 1 (Linkages/MSSP, Home Health Care, Hospitals, Ombudsman, and Retired Senior Volunteers) has providers serving persons with different kinds of needs, ranging from acute care need (Hospitals), post-acute care (Home Health Care) to skilled nursing need (Ombudsman). These groups seem to be unable to display a clear distinction between different functions. Of all the groups considered in the network, most groups have reciprocal exchanges with one another (Figures 9 through 12). No prominent “sender" or “receiver" is found. There is high reciprocity in relationships in general, and for most of the resource exchanges, the highest reciprocity is found in San Mateo. Two structurally equivalent groups (Groups 3 and 6) are relatively isolated from the network exchanges. Groups 1 (Linkages/MSSP, Home Health Care, Hospitals, Ombudsman, Retired Senior Volunteers), 4 (Emergency Services, Social Services, and Adult Day Health Care), and 5 (Senior Centers, Information & Referral, Charity, and Other) constitute the core. The core groups also maintain close exchange relations among themselves, of which two of them (Groups 1 and 4) have direct 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relationships within their own group members (see recursive arrows in Figures 9 through 12). Through relations with the core groups, other “peripheral” groups are able to connect with one another in the network. For example, the relatively isolated group 6 (Legal Aid and Transportation) and 2 (Housing, County Health Department, and Regional Center) are mainly connected to the care system through their relationships with Group 4 (Emergency Services, Social Services, and Adult Day Health Care). By exchanging various resources with Group 7 (Senior Meals Programs, Mental Health, and Money Management) and occasionally with Group 3 (Skilled Nursing Facilities, Social Security, and Public Guardian), Group 1 (Linkages/MSSP, Home Health Care, Hospitals, Ombudsman, and Retired Senior Volunteers) is able to accommodate others needs acquired by its clients. San Mateo does not demonstrate a fine division of labor in its care system. First, “repetitive” ties are found among service providers: one can access groups that are connected to the network through one of two core groups, hence making the other one of these core groups “redundant”. For example, Group 7 (Senior Meals Program, Mental Health, and Money Management) is accessible to both Groups 1 (Linkages/MSSP, Home Health Care, Hospitals, Ombudsman, and Retired Senior Volunteers) and 5 (Senior Centers, Information & Referral, Charity, and Other). Group 5 also is connected to Groups 1 and 4 (Emergency Services, Social Services, and Adult Day Health Care). Second, few differences are found in the relations 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. among structurally equivalent groups across the four resources. High reciprocity of exchange relations in these resources leads to the inability to recognize a clear pattern of division of labor based on the directions of exchanges. Compared to the other communities, the majority of “connected” structurally equivalent groups in San Mateo are highly connected: 28 out of 56 possible intergroup exchanges (50%) are found. In terms of the intensity of these relations, the network is fairly integrated among various groups on multiple resources: 46% (13 out of 28) of its intergroup ties are exchanges involving more than two types of resources. Long Beach Structure CONCOR analysis partitioned six structurally equivalent groups and one residual program in Long Beach care system. Relations among structurally equivalent groups across the four resources are found in Figures C21 through 24 (in Appendix C). Table 13 displays the members in the various groups and their relations with one another in blockmodeis. Group Description — Group 1 (Linkages/MSSP Program and Home Health Care) develops multiple and reciprocal relations with two of its three partners (Groups 2, 5, and 7) (Table 13). its relationships with Group 2 (Housing and Retired Senior Volunteers) and 5 (Senior Meals Program, Social Services, Adult Day Health Care, Health Clinics, and Charity) are Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 13 Structurally Equivalent Groups in Long Beach with Their Relational Patterns in Blockmodeis 1: Linkages/Multi-Service Senior Program, Home Health Care 2: Housing, Retired Senior Volunteers 3: Skilled Nursing Facilities, Other 4: Senior Centers, Hospitals, Alzheimer Association 5: Senior Meals Program, Social Services, Adult Day Health Care, Health Clinics, and Charity 6: Information and Referral, Social Security Agencies, Mental Health 7: Transportation____________________________________________ Client_________ Information Money______ Staff_________ Multiplex Ties 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 1 1 0 0 1 0 0 1 1 0 0 1 0 0 1 1 0 0 1 0 1 1 0 0 0 0 0 0 4 3 0 0 3 0 1 2 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 3 0 0 0 1 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 0 0 1 0 1 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 1 1 4 2 0 5 1 0 0 1 0 1 0 1 0 0 1 0 1 0 1 1 0 1 1 0 0 1 0 0 1 1 1 0 4 1 0 4 2 3 0 6 0 0 0 1 1 1 0 0 0 0 1 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 3 2 2 0 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 reciprocal in client and information networks. In general, Group 1 sends money to its partners, but receives staff from them. Group 2 (Housing and Retired Senior Volunteers) is relatively isolated in the network. It has two partners (Group 1 and 5), among which Group 1 (Linkages/MSSP Program and Home Health Care) is its stronger partner, exchanging client and information resources with each other. Given its only relation with Group 4 (Senior Centers, Hospitals, and Alzheimer’s Association), Group 3 (Skilled Nursing Facilities and Other) is one of the two isolated groups in the network. Other than receiving client 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. referral from Group 4, there is no other link between Group 3 and the network. Group 4 (Senior Centers, Hospitals, and Alzheimer's Association) has three partners (Groups 3, 5, and 6), among them Group 5 (Senior Meals Program, Social Services, Adult Day Health Care, Health Clinics, and Charity) and 6 (Information and Referral, Social Security Agencies, and Mental Health) are its strong partners. Its relationships with Group 5 are reciprocal across all four resources. On the other hand, its linkages with Group 6 are not reciprocal. Besides the reciprocal money exchanges, Group 6 sends client and information to Group 4 in receipt of staff resources. Group 5 (Senior Meals Program, Social Services, Adult Day Health Care, Health Clinics, and Charity) is the most active group in the network, having four partners (Groups 1, 2, 4, 5, and 6) that exchanges multiple resources with it. Groups 1, 4, and 6 are its strong partners, especially with its identical exchange patterns in the client and information networks. Although Group 6 (Information and Referral, Social Security Agencies, and Mental Health) has only two partners (Groups 4 and 5), both partners develop reciprocal and multiple exchanges. This close relation constitutes a subnetwork within the overall care network in Long Beach, acting as glue to other loosely coupled structurally equivalent groups. Group 7 (Transportation) is a residual from the CONCOR partition. Its uniqueness lies in its isolation from the overall care network, maintaining 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. only money tie with Group 1 (Linkages/MSSP Program and Home Health Care). Netw ork Description - Little differentiation along the social-health dimension can be found among these structurally equivalent groups in the Long Beach care system. Most groups contain service providers from either social or health arena. For example, Group 4 is composed of service providers that serve healthy (for example, Senior Centers) and extremely frail older adults (for example, Alzheimer’s Services). Similarly, providers of Group 5 (Senior Meal Program, Social Services, Adult Day Health Care, Health Clinics, and Charity) serve different segments within the older population. One provider (Information and Referral) in Group 6 serve to a more general population whereas the other (Mental Health) has its own client niche. The basic structure of the Long Beach care system evolves around five structurally equivalent groups, with two (Group 3 — Skilled Nursing Facilities and Other; Group 7 — Transportation) barely integrated in the system. Figures C21 through 24 show that Groups 4, 5, and 6 are the core in Long Beach care system. The three groups form a clique in exchanging three resources (client, information, and staff). It seems that this core can accommodate older adults with different gradations and types of needs. Moreover, Group 5 can be considered the glue of all resource exchange patterns congealing the care system: removing Group 5 might disintegrate 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the entire Long Beach system. The relational patterns of these five groups indicate that the seemingly loosely coupled care system in Long Beach was cemented by Group 5 (Senior Meals Program, Social Services, Adult Day Health, Health Clinics, and Charity) and Group 1 (Linkages/MSSP Program and Home Health Care) in some resource networks. Group 5 connects Groups 4 and 6 to Group 1, which maintains exchange relations with Groups 2 and 7. A great deal of similarity in exchange patterns is found in client, information, and money, suggesting the prominence of muitiplexity in the Long Beach network. Of those 14 inter-group exchange relations, eight are exchanges with more than two resources (57%). This finding also implies that there is low role differentiation in terms of structurally equivalent groups dominating specific resource network. A high level of reciprocity (> 50%) is also found in the network. Long Beach seems to have a fairly sparsely integrated network. Only 14 out of 42 possible ties are across groups (33%). Only three groups have intra-group relations among their own service providers. Structurally equivalent groups in Long Beach in general do not have many partners and two groups have very minimal exchange relations with the rest of the network, making the network a very loosely and sparsely coupled one. 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Tulare Structure Seven structurally equivalent groups and one residual resulted from the CONCOR analysis. Figures C25 through C28 (in Appendix C) display the relational patterns of resource exchanges and Table 14 shows the various programs in each group. Table 14 Structurally Equivalent Groups in Tulare with Their Relational Patterns in Blockmodeis 1: Area Agencies on Aging, Retired Senior Volunteer 2: Housing 3: Skilled Nursing Facilities, Home Health Care, Mental Health 4: Emergency Services, Hospitals, County Health Department 5: Senior Centers, Legislator/Politicians, Charity 6: Legal Aid, Senior Meals Program, Information and Referral 7: Social Services, Social Security Agencies, Public Guardian 8: Adult Day Health Care, Ombudsman__________________ Client__________Information Money_______ Staff_________Multiplex Ties 12345678 12345678 1 234 5678 12345678 12 3 4 5 6 7 8 1 0 10 0 1 1 0 0 0 10 0 1 100 0 0 0 0 0 1 0 0 0 0 0 0 1 10 0 0 2 0 0 3 4 0 0 2 10 0 1 1 1 1 1 1 0 0 1 10 10 0 0 0 0 0 0 0 0 0 0 0 1 100 0 2 0 0 3 3 12 1 3 0 0 1 1 1 1 1 0 0 0 1 1 1 110 0 0 1 10 0 1 0 0 0 1 10 0 1 0 0 0 4 4 2 2 4 0 4 0 1 1 1 1 1 10 0 1 1 1 1 100 0 0 0 10 0 0 0 0 0 0 1100 0 02 2 4 32 1 0 5 1 1 1 10 0 0 0 1 1 1 10 00 0 1 00 10 1 1 0 1 1 1 10 10 0 4 3 3 4 0 2 1 0 6 1 10 1 0 1 0 0 1 1 1 10 10 0 0 00 0 0 0 0 0 1 0 1 1 110 0 3 2 2 3 1 3 0 0 7 0 1 1 0 0 0 1 0 0 1 1 0 0 0 10 0 0 10 0 0 0 0 0 0 0 0 0 0 1 0 0 2 3 0 0 0 3 0 8 0 0 0 0 0 0 0 0 0 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 0 00 00 0 0 0 0 0 0 0 0 0 Group Description — Group 1 (Area Agencies on Aging and Retired Senior Volunteers) maintains reciprocal exchange relations with Groups 2, 5, and 6 all with different magnitudes across resource types (Table 14). Its strong relationships are with Groups 5 (Senior Centers, 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Legislator/Politicians, and Charity) and 6 (Legal Aid, Senior Meals Programs, and Information and Referral), where exchanges are dominantly client, information, and staff. Being a residual from the CONCOR partition, Group 2 (Housing) maintains unique exchanges with all groups except Group 3 (Skilled Nursing Facilities, Home Health Care, and Mental Health) in the network. Service providers of Group 3 (SNF, HHC, MH) maintain relations with four other structurally equivalent groups (Groups 4, 5, 6, and 7); all of them exchange primarily client and information. Groups 4 (Emergency Services, Hospitals, and County Health Department) and 7 (Social Services, Social Security Agencies, and Public Guardian) receive money and staff from Group 3, which receive staff from Groups 5 (Senior Centers, Legislator/Politicians, and Charity) and 6 (Legal Aid, Senior Meals Program, and Information and Referral). Group 4 (Emergency Services, Hospitals, and County Health Department) also have a high degree of exchange relations with five other structurally equivalent groups (Groups 2, 3, 5, 6, and 7). What is interesting for Group 4 is that its reciprocal relations with partners are not balanced when comparing its outgoing involvement to its incoming ones. Focusing on its outgoing ties, Group 4 has a strong multiplex relation with Group 5 (Senior Centers, Legislator/Politicians, and Charity). However, when its incoming relations are considered, its strong partners include Groups 2 (Housing), 3 (Skilled Nursing Facilities, Home Health Care, and Mental 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Health), 5 (Senior Centers, Legislator/Politicians, and Charity), and 6 (Legal Aid, Senior Meals Program, and Information and Referral). Most of the relations between Group 4 and its partners are concentrated in client and information exchanges. Comparatively, most money and staff relations happen to be resources sent to Group 4. Group 5 (Senior Centers, Legislator/Politicians, and Charity) maintains exchange relations with six partners (Groups 1, 2, 3, 4, 6, and 7). Although it has the same number of partners, Group 5’s relational pattern differs from that of Group 2 (Housing): it maintains ciient and information exchanges with Group 2 and 3. Group 5 has reciprocal relations with Group 1 (Area Agencies on Aging and Retired Senior Volunteers) and 4 (Emergency Services, Hospitals, and County Health Department) at least in client, information, and staff networks. Group 6 (Legal Aid, Senior Meals Program, and Information and Referral) is another highly integrated structurally equivalent group in the network. Its partners include Groups 1 (Area Agencies on Aging and Retired Senior Volunteers), 2 (Housing), 3 (Skilled Nursing Facilities, Home Health Care, and Mental Health), 4 (Emergency Services, Hospital, and County Health Department), and 5 (Senior Centers, Legislator/Politicians, and Charity). Group 6 exchanges client, information, and staff with its partners; money exchange is relatively infrequent. Compared to other groups, Group 7 (Social Services, Social Security Agencies, and Public Guardian) does not have extensive multiple exchange 136 permission of the copyright owner. Further reproduction prohibited without permission. patterns with its partners (Groups 2, 3, 4, and 5). Outgoing exchanges are more common than incoming ones (four versus two). Other than Group 3 (Skilled Nursing Facilities, Home Health Care, and Mental Health) with which it exchanges all four resources, most of its exchanges are sparse in terms of number of resources involved. Group 8 (Adult Day Health Care1 and Ombudsman) is the least integrated group in the network. Its service providers do not have direct relations with each other and it has minimal exchange with the rest of the network: it only receives client from Group 2 (Housing). Network Description - Structurally equivalent groups in Tulare seem to be differentiated along the social-health dimension. For example, Group 3 (Skilled Nursing Facilities, Home Health Care, and Mental Health) consists of programs catered to the long-term care population. Group 5 (Senior Centers, Legislator/Politician, and Charity) is composed of providers of social nature; Group 4 (Emergency Services, Hospitals, and County Health Department) seems to focus on the acute health side of services. 1 According to some field notes, Tulare did not have adult day health care program. The fact that this service-provider type existed could be a data entry problem. Since questionnaires for Tulare were lost, there was no way to verify the information. 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Providers of Group 8 (Adult Day Health Care and Ombudsman) are mainly for a frail population. Exchange relations among the structurally equivalent groups in Tulare are fairly connected in both client and staff: it is easy to find at least three connections for each group to exchange these resources. When all resource exchanges are considered, Groups 4 (Emergency Services, Hospitals, and County Health Department) and 5 (Senior Centers, Legislator/Politician, and Charity) remain the core in this care system. The Tulare care system portrays a different structure than Long Beach and San Mateo. First, almost all the exchanges in client and information are reciprocal among structurally equivalent groups, while only a few are found in the staff network and none in the money network. Moreover, given the fact that on average, each group has at least three ties (except Group 8, Adult Day Health Care and Ombudsman), the resource allocation (degree centrality) is relatively even. Second, it appeared that minimal division of labor in terms of various resources exists. Given the high level of multiplexity among the most integrated groups (Groups 2, 3, 4, 5, and 6), functional differentiation is difficult to detect. Yet, Groups 2, 3, 4, 5, and 6 were active groups in client and information networks and Groups 3, 4, 5, and 6 are also actively involved in staff exchange (little structural differentiation). Group 2 (Housing), while active in the client and information networks, is an isolate in the money network. 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Similar to San Mateo, most exchanges are symmetric and reciprocal, with the absence of major senders or receivers in the network. More groups having intra-group relations are found in Tulare (four) than in San Mateo. Only one group is considered significantly isolated from the rest of the network. The integration level in terms of number of exchange relations (density in binary matrix) is higher than that of San Mateo (55%, 31 out of 56 possible ties). However, a lower proportion of these exchange relations is found having more than two resources involved (42%, 13 out of 31). This would suggest that although in general service providers are more integrated with each other, the magnitude/intensity of such integration is still relatively small. San Francisco Structure CONCOR results found seven structurally equivalent groups in San Francisco care system (Table 15). Their relational patterns across resources are shown in Figures C29 through C32 (in Appendix C). Group Description — Group 1 (Linkages/MSSP Program, Area Agencies on Aging, Information and Referral, Charity, and Consortium) maintains reciprocal relations with three groups (Groups 2, 6, and 7) (Table 15). Service providers have money exchanged among themselves but not other types of resources. Its stronger partners include Groups 2 (Housing, Home Health Care, Senior Centers, Legal Aid, Senior Meals Program, Social 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Services, Mental Health, and Adult Day Health Care) and 6 (Transportation, Money Management, Rehabilitation). Table 15 Structurally Equivalent Groups in San Francisco with Their Relational Patterns in Blockmodels 1: Linkages/MSSP Program, Area Agencies on Aging, Information and Referral, Charity, Consortium 2: Housing, Home Health Care, Senior Centers, Legal Aid, Senior Meals Program, Social Services, Mental Health, Adult Day Health Care 3: Skilled Nursing Facilities, Public Guardian, Retired Senior Volunteers 4: Emergency Services, Hospitals, Health Clinics, Case Management, County Health Department, Homeless Services 5: Legislator/Politicians, Social Security Agencies 6: Transportation, Money Management, Rehabilitation 7: Om budsm an, Other Client Information Money Staff Multiplex Ties 1 2 3 4 5 6 7 12 34 5 67 1 2 3 4 56 7 1234 5 67 1 2 3 4 56 7 1 0 1 0 0 0 1 0 2 1 1 0 10 10 3 0 0 0 0 0 0 0 4 0 1 0 1 0 0 0 5 0 0 0 0 0 11 6 1 0 0 0 0 0 0 7 1 0 0 0 1 0 1 0 1 0001 1 110 1000 00 000 00 0 10 1000 0 0 0 0 111 10 00 1 00 1000101 110 0 0 11 110 0 0 10 0 0 0 0 00 0 0 0 10 0 11 0 0 0 0 0 11 0 1 0 0 00 0 0 0 0 0 0 1 0 01 00 0 00 110 1000 000 0000 0 11 10 0 0 00 00 0 10 110 0 0 0 1 0000001 1 4 0 0 0 3 2 44 0 3 020 0 0 0 0 0 0 0 032 3011 0 0 0 0 143 3200 101 2 0 0 0 2 1 3 Service providers of Group 2 (Housing, Home Health Care, Senior Centers, Legal Aid, Senior Meals Program, Social Services, Mental Health, and Adult Day Health Care) have relations with three partners (Groups 1, 4, and 6), with only one group (Group 6) in common with Group 1. Groups 1 (Linkages/MSSP Program, Area Agencies on Aging, Information and Referral, Charity, and Consortium) and 4 (Emergency Services, Hospitals, Health Clinics, Case Management, County Health Department, and 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Homeless) maintain at least three types of resources (client, information, and staff) exchanged reciprocally with Group 2, and Group 1 ’s relations with Group 6 are mainly in money and staff exchanges. Group 3 (Skilled Nursing Facilities, Public Guardian, and Retired Senior Volunteers) is less integrated to other groups in the network. The only partner Group 3 maintains in the network is Group 4 (Emergency Services, Hospitals, Health Clinics, Case Management, County Health Department, and Homeless Services) through receiving money and staff. Among its four partners (Groups 2, 3, 6, and 7), Group 4 (Emergency Services, Hospitals, Health Clinics, Case Management, County Health Department, and Homeless Services) maintains reciprocal relations only with Group 2 (Housing, Home Health Care, Senior Centers, Legal Aid, Senior Meals Program, Social Services, Mental Health, and Adult Day Health Care) across client, information, and staff resource exchanges. Group 4’s relations with Groups 3, 6, and 7 are mainly found in money exchanges and to a lesser extent staff exchanges. Group 5 (Legislator/Politicians and Social Security Agencies) is a fairly isolated group in the network, with only two partners out of the possible six (Groups 6 and 7). Nevertheless, the intensity of its relations in terms of multiplexity is high and focuses on the outgoing client, information, and money ties to these groups. Group 6 (Transportation, Money Management, and Rehabilitation) has four partners (Groups 1, 2, 5, and 7), among which Group 1 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Linkages/MSSP Program, Area Agencies on Aging, Information and Referral, Charity, and Consortium) maintains reciprocal client and information exchanges. Its relation with Group 5 (Legislator/Politicians and Social Security Agencies) is relatively uneven, in the sense that Group 6 receives more resources (all four) from Group 5 than it sends to Group 5 (information ties). Service providers of Group 7 (Ombudsman and Other) exchange multiple resources with Groups 1, 4, 5, and 6. Of these partners, its relations with Groups 1 (Linkages/MSSP Program, Area Agencies on Aging, Information and Referral, Charity, and Consortium) and 5 (Legislator/Politicians and Social Security Agencies) are the strongest. Netw ork Description - Structurally equivalent groups in San Francisco can be identified relatively easily because most of them are partitioned according to the types of care. For example, Group 4 (Emergency Services, Hospitals, Health Clinics, Case Management, County Health Department, and Homeless Services) apparently is oriented toward health services. Programs of Group 3 (Skilled Nursing Facilities, Public Guardian, and Retired Senior Volunteers) serve a frail older population. Service providers in Group 2 (Housing, Home Health Care, Senior Centers, Legal Aid, Senior Meals Program, Social Services, Mental Health, and Adult Day Health Care) seem to serve community-dwelling older adults with a gradation of social and long-term care needs. 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figures C29 through C32 show that Groups 1, 2, and 4 form the basic axle of serving older adults in San Francisco. Moreover, programs of Groups 2 and 4 also have direct relationships with each other, suggesting that some relatively integrated “subsystems” might have already existed within each group. The structure of care systems in San Francisco seems quite differentiated. Compared to the other communities, San Francisco has more structurally equivalent groups whose members maintained direct relations with one another. Moreover, although structurally equivalent groups in San Francisco also maintained reciprocal ties with their partners, the number of partners they have is much fewer than San Mateo and Tulare. Only 18 out of 42 possible ties exist (42%) in the network. Five structurally equivalent groups have intra-group relations among their service providers, although not all of them have relations with many resource exchanges. About 44% of the existing exchanges involve more than two resources (8 out of 18). Community Specificity in the Organization of Care Systems Variation in the Extent of Program Involvement How extensively service providers are involved in building community-based networks becomes an important issue when the organization of care systems is examined. The extent of provider involvement through program provision is described in several manners. 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. First is the number of service-provider types in the community. More service-provider types identified in the organization of care systems suggest that more resources are available for older adults in that community to access for their care. Among the four communities, San Francisco, San Mateo, Long Beach are considered resource-rich communities whereas Tulare is resource-poor community. Table 16 indicates, based on the information collected from the SEED questionnaire, the service-provider types in the four communities. From the table, San Francisco (29) and San Mateo (24) have more service-provider types identified as aging service providers in their respective communities than Long Beach (18) and Tulare (20). Some service-provider types are more prevalent in these communities than are others. Twelve of the thirty-two service-provider types are found in all the communities and can be considered as core community-based service programs in the care system. These service- provider types are Housing, Skilled Nursing Facilities, Home Health Care, Senior Centers, Senior Meals Program, Information and Referral, Social Services, Social Security Agencies, Mental Health, Hospitals, Adult Day Health Care, and Retired Senior Volunteers. These core service-provider types are more likely to be providers that serve predominantly older adults. A second aspect of service provider involvement in community- based care system development is the extent service providers perceived that they are interdependent with one another. Table 17 reports providers’ 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 16 Which Program Types Are More Likely to be Found Across All the Selected Communities? Program Types San Long Tulare San Core _______________________Mateo Beach_________ Francisco Programs Linkages/MSSP Program X X X Area Agencies on Aging X X Housing X X X X X Skilled Nursing Facilities X X X X X Home Health Care X X X X X Emergency Services X X X Senior Centers X X X X X Legal Aid X X X Senior Meals Program X X X X X Information and Referral X X X X X Legislator/Politicians X X Social Services X X X X X Social Security Agencies X X X X X Mental Health X X X X X Hospitals X X X X X Public Guardian X X X Adult Day Health Care X X X X X Transportation X X X Health Clinics X X Ombudsman X X X Case Management X Money Management X X County Health Department X X X Charity X X X Homeless Services X Consortium X Retired Senior Volunteers X X X X X Other X X X Rehabilitation X X Regional Center X Alzheimer’s Association X 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. response to the questions related to their perception of the degree of dependence towards their partners and from them. In general, there is a discrepancy in terms of how much service providers perceived they need their partners versus how much their partners need them to accomplish respective goals. Comparatively smaller proportion (ranging from 13 percent in San Francisco to 30 percent in San Mateo) of service providers in all the four communities express needing their partners not at all or only somewhat to accomplish their goals. Responding to the question “How much does this program need you to accomplish its goals?” at least 30 percent of the service providers in all the communities perceive that their partners either do not need them or only need them somewhat to accomplish their goals. Significant community differences are found in these perceptions. For instance, significantly more service providers in San Mateo (60.7%), Long Beach (60%) and San Francisco (62.1%) recognized needing their partners or needing them a great deal to help them accomplish their goals, whereas only 45.1% of service providers in Tulare do so. The involvement of different programs in community-based system is examined through the distribution of various types of linkages. Table 18 shows the different types of resource exchanges. Only 6.4 percent of all exchanges made by programs are single ties, meaning that a majority of exchanges are multiplex in nature. Among these single ties, “client only” 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 17 Perception of Inter-organizational Need How much do you need this program to accomplish your program’s goals? Chi-square p-level = 0.004_____________________________________ San Mateo (n=112) Long Beach (n=65) Tulare (n=122) San Francisco (n=190) not at all .9% 3.1% 5.7% 1.6% some what 22.3% 15.4% 19.7% 11.6% Needed 16.1% 21.5% 29.5% 24.7% Needed great deal 60.7% 60.0% 45.1% ' 62.1% How much does this program need you to accomplish its goals? Chi-square p-level = 0.008 San Mateo (n=112) Long Beach (n=65) Tulare (n=122) San Francisco (n=189) not at all 11.6% 9.2% 13.9% 11.6% some what 39.3% 21.5% 30.3% 22.2% Needed 19.6% 40.0% 31.1% 32.3% Needed great deal 29.5% 29.2% 24.6% 33.9% * p <0.01; **p< 0.05 ties are the most common exchanges, accounting for at least 3.5 percent of all the exchanges in each community. The majority of service providers making two-tie exchanges is found in client-information exchanges. Client- information-staff exchanges are the most prevalent form of three-tie exchanges, which characterize over 24 percent of all the exchanges in these communities. Despite the fact that the majority of exchanges are in multiplex form, only 11.6% of exchanges among service providers in these communities are composed of all four resources (client, information, money, and staff). 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moreover, client and information exchanges dominate most of the relations. It appears that client is the primary resource driving inter-program exchanges, given that there is a significant higher percent of single exchange that is client-only rather than information-only. Table 18 Multplexitv of Ties San Mateo Long Beach Tulare San Francisco Tota Single-Tie Exchange Client only 3.6% 4 (3.6%) 9.2% 6 (9.2%) 6.5% 7 (5.7%) 7% 6.4% 7 (3 .8 % ) 24(5% ) Money only 1 (0.5%) 1 (0.2%) Information only 1 (0.8%) 5 (2.7%) 6 (1.2%) Client and money 2(1.8% ) 4 (6.2%) 13 (7.1%) 19 (3.9%) Two-Tie Exchange Client and information 44.7% 48 (42.9%) 55.4% 55.2% 29 65 (52.8%) (44.6%) 60.3% 93 (51.1%) 54.7% 235 (48.8%) Client and staff 2 (1.1%) 2 (0.4%) Money and information 2 (1.6%) 2 (0.4%) Money and staff 1 (0.5%) 1 (0.2%) Information and staff 3 (4.6%) 1 (0.8%) 1 (0.5%) 5 (1%) Three-Tie Exchange Client, money, and information 42% 20 (17.9%) 21.5% 3 (4.6%) 22.4% 7 (5.7%) 22% 14 (7.7%) 27.2% 44 (9.1%) Client, money, and staff 5 (4.5%) 2 (1.1%) 7 (1.5%) Money, information, and staff 3 (4.6%) 1 (0.8%) 4 (2.2%) 8 (1.7%) Client, information, and staff 22 (19.6%) 8 (12.3%) 22 (17.9%) 20 (11%) 72 (14.9%) Four-Tie Exchange Client, money, information and staff 9.8% 11 (9.8%) 13.8% 13.8% 10.4% 9(13.8% ) 17(13.8% ) 19(10.4% ) 11.6% 56 (11.6%) Total 112 100.0% 65 100.0% 123 100.0% 182 100.0% 482 100.0% Chi-square test p-level = 0.000 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Variation in the Prominent Service-Provider Types The differences in network centralization suggest that service- provider types in various community-based care systems have different roles in sending out and receiving resources. Service-provider types with high indegrees and outdegrees are considered prominent brokers in the care systems. These brokers are, in general, neither origins nor destinations of any resource flow in the entire community-based care system. Their positions in care networks have significant implications for any innovation to be implemented. Table 19 shows that more prominent brokers are found in the client and information networks than the money and staff networks. This finding may be related to the relatively denser connection between service-provider types in the client and information networks. Comparatively, given its relative sparseness, there are fewer brokers in the money and staff networks. Similarities exist in terms of which service-provider types are brokers between the client and information networks. Some service-provider types that are brokers in the client network are also in the information network in each community. On the other hand, there is really very little similarities between the money and staff network. Whereas client and information are resources in the same “category/function,” the extent of similarity between the money and staff resources is less certain. 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 19 Prominent Brokers (High Indegrees and Outdegrees) by Resources and Communities San Mateo Long Beach Tulare San Francisco Client Linkages/MSSP Program Home Health Care Adult Day Health Care Social Services Rehabilitation Social Services Charity Home Health Care Senior Centers Hospitals Social Services Home Health Care Social Services Senior Centers Housing Adult Day Health Care Information Linkages/MSSP Program Adult Day Health Care Social Services Senior Centers Social Services Charity Senior Centers Social Services Home Health Care Mental Health Social Services Senior Centers Information and Referral Adult Day Health Care Hospitals Health Clinics Money Rehabilitation Charity Adult Day Health Care Senior Centers Transportation Staff County Health Department Charity Senior Centers Charity Senior Centers Senior Meals Program Table 19 indicates the importance of Social Services as a broker in both client and information networks for all communities. Social Services certainly controls significant resource flows across the community based long-term care system: at least seven other service-provider types are 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. dependent on the client referral and information exchanges of Social Services in either incoming or outgoing connections. The high indegree centrality also suggests that the program type is a commonly recognized “hub” where other service-provider types send their resources. Besides Social Services, Senior Centers appear to be prominent brokers particularly in San Francisco and Long Beach. Senior Centers are especially prominent in the information network. Senior Centers are important in all the resource networks in San Francisco, suggesting that this service- provider type not only is an “integrator” for other service-provider types in each respective resource network, but also across different resources. To demonstrate the importance of brokers, densities were recalculated and graphs redrawn for these communities when Social Services and Senior Centers were not considered in their care systems. Table 20 presents densities for each resource network across four communities when Social Services and Senior Centers were dropped. Comparing Table 20 to Table 6, dropping the two prominent brokers reduces four to five percent of linkages in the client and information networks and about one percent in the money and staff networks. Service-provider types with high outdegrees are prominent senders in the community-based care systems, indicating that these service- provider types maintain more outgoing linkages than incoming linkages with other service providers in the care system. Table 19 shows that, in general, more senders are found in the client, money, and staff networks than in the 151 permission of the copyright owner. Further reproduction prohibited without permission. information network. Hospitals, Home Health Care, and Housing are the service-provider types recognized as prominent senders in the information network in San Mateo, Long Beach, and San Francisco respectively. Table 20 Importance of Social Services (DPIHAPS) and Senior Centers (SENIORCTR) in the Selected Communities San Mateo Long Beach Tulare San Francisco Client Network Raw# 100 67 110 186 With 0.18 0.22 0.29 0.23 Without 0.14 0.17 0.24 0.19 Information Network Raw# 100 63 114 199 With 0.18 0.21 0.03 0.25 Without 0.14 0.15 0.25 0.2 Money Network Raw# 22 15 13 41 With 0.04 0.05 0.03 0.05 Without 0.04 0.04 0.04 0.04 Staff Network Raw# 34 15 30 43 With 0.06 0.05 0.08 0.05 Without 0.04 0.03 0.07 0.04 Compared to the Table 19, Table 21 indicates less similarity in which service-provide types are prominent senders among resource networks in each community. Hospitals and Charity appear to be senders that are found in some resource networks in San Mateo, Tulare, and San Francisco. 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 21 Prominent Senders (High OutDearees) bv Resources and Communities San Mateo Long Beach Tulare San Francisco Client HOPSITAL SENIORCTR LINKMSSP HOSPITAL HSG MOW HOSPITAL HEACLIN Information HOSPITAL HHC HSG Money LINKMSSP DPIHAPS HOSPITAL CHARITY MONEYMGT LINKMSSP DPIHAPS ADHC CHARITY PUBGUR MH HOSPITAL LINKMSSP CHARITY Staff HOSPITAL DPIHAPS MH REHAB SENIORCTR RSVP AAA MH l&R MOW CHARITY MH TRANSPRT Service-providers with high indegree centrality are considered prominent receivers in this study. These receivers are major destination of resource flows in community-based care systems. In contrast to Tables 19 and 21, Table 22 shows much fewer prominent receivers in the client and information networks than in the money and staff networks. No prominent receivers are found in the information network in both Tulare and San Francisco, and only one receiver is recognized in San Mateo and Long Beach. Only one receiver is found in the client network in each community. Comparatively, significant number of receivers is found in the money and staff networks. 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 22 Prominent Receivers (High Indearees) bv Resources and Communities San Mateo Long Beach T ulare San Francisco Client MH l&R MH l&R Information HHC HOSPITAL Money ADHC HHC SENIORCTR HOSPITAL RSVP SENIORCTR HHC HHC HOSPITAL HHC MOW Staff LINKMSSP ADHC SENIORCTR CHARITY LINKMSSP l&R DPIHAPS HOSPITAL COUNTYHD ADHC HSG Similar to Table 21, Table 22 shows that it is very difficult to find major receivers or senders for all resources in a community. Adult Day Health Care (ADHC) in San Mateo and Hospitals (HOSPITAL) in Tulare is the only prominent receiver found in both money and staff networks. Moreover, prominent service-provider types that receive staff from other service providers tend to be very diverse across communities. Home Health Care (HHC) appears to be a prominent funding receiver for all communities. Given the discrepancies found in the number of prominent brokers, senders, and receivers in different resource networks, there may be associations between various types of prominent service providers among resource networks. To further understand this association, Table 23 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. presents the correlation results for all service providers and correlation results for each community that are found in Tables B1 through B4 in Appendix B. These correlation tables show that, in general, service- provider types with high indegree centrality in either client or information networks are also likely to have high outdegree centrality in either network. If a service-provider type is prominent in referring clients, it is also more likely to be prominent in receiving clients, and hence becoming a prominent broker in the client network. Moreover, given the high correlation (>0.8) in degree centrality between the client and information networks, this service- provider type is also more likely to be a prominent broker in the information network. The high correlation between client and information networks is consistent across the four communities. Such correlation patterns, however, can not be found in the money or staff networks. As a result, there are much fewer brokers in these networks. More modest (< 0.5) correlations are found between indegree and outdegree centrality values in money and staff networks. Moderate correlations (> 0.6) are found in Table 23 for the indegree centrality in the staff network with the outdegree in the client network, and both indegree and outdegree centrality in the information network. Service-provider types prominent in sending money or staff resources are more likely to become prominent receivers in either client or information networks; this is also true when the service-provider types are prominent receivers in the money or staff networks. 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 23 Centrality Correlation Results (Four Communities) In Out In Out In Out In Out- degrees degree degree degree degree degree degrees degree (Client) (Client) (Information) (Information) (Money) (Money) (Staff) (Staff) Indegrees (Client) 1.000 .88 .935 .922 .445 .375 .515 .518 Outdegree (Client) .880 1.000 .948 .946 .384 .450 .676 .477 Indegrees .935 .948 1.000 .971 .459 .421 .631 .517 (Information) Outdegrees .922 .946 .971 1.000 .388 .437 .628 .507 (Information) Indegrees (Money) .445 .384 .459 .388 1.000 .314 .470 .431 Outdegrees (Money) .375 .450 .421 .437 .314 1.000 .446 .344 Indegrees (Staff) .515 .676 .631 .628 .470 .446 1.000 .489 Outdegrees (Staff) .518 .477 .517 .507 .431 .344 .489 1.00 cn 0) Factors Associated with the Resource Exchange Patterns This section is concerned with factors underlying the resource exchanges patterns among service providers in the four communities. It begins with presenting non-normalized results on the bivariate relations among these factors as well as their relations to the exchange patterns. Normalized results from the Multiple Regression Quadratic Assignment Procedure (MRQAP) also are presented. Non-Normalized Results San Mateo A number of independent matrices are found to be highly correlated (>0.8) with each other (Table B5). SAME NETWORK is highly correlated with SAME GOAL, SAME CLIENT, SAME FUND, MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED. LOW COST (low cost provider) is also highly correlated with SAME GOAL, SAME CLIENT, MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED. Table B5 in Appendix B shows that SAME GOAL, SAME CLIENT, and SAME FUND perform similarly in terms of how they are related to other independent variables, suggesting that they can be combined as a scale. The similar correlation performance of MY INFLUENCE and THEIR INFLUENCE, as well as MY NEED and THEIR NEED also suggests that construction of a scale is appropriate. 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 24 presents bivariate correlation between each independent matrix and the four dependent matrices in San Mateo (client, information, money, and staff). In general, SAME NETWORK has a higher correlation with client and information networks, whereas SAME ORGANIZATION is more highly correlated with money and staff networks than the other two resource networks. MANDATE (mandate) maintains a low correlation with all resource networks. It is significantly associated only with client and information networks. Table 24 Bivariate Correlation. San Mateo: Independent Matrices with Dependent Matrices Matrices___________Client_______Information Money______ Staff SAME NETWORK 0.613* 0.613* 0.144* 0.260* SAME 0.412* 0.362* 0.159* 0.394* ORGANIZATION PROXIMATE 0.041 0.063 -0.047 0.009 LOW COST 0.547* 0.598* 0.137* 0.246* ONLY AVAILABLE 0.485* 0.431* 0.193* 0.219* MANDATE 0.253* 0.283* 0.026 0.007 SAME GOAL 0.641* 0.655* 0.201* 0.287* SAME CLIENT 0.655* 0.655* 0.201* 0.287* SAME FUND 0.633* 0.633* 0.203* 0.291* MY INFLUENCE 0.641* 0.641* 0.201* 0.287* THEIR 0.655* 0.655* 0.201* 0.287* INFLUENCE MY NEED 0.655* 0.655* 0.201* 0.287* THEIR NEED 0.647* 0.647* 0.203* 0.291* *P<0.05 Table 25 shows that a strong correlation (0.86) is found between the client and information networks in San Mateo. Only moderate correlations 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. between the client and information networks and the money and staff networks are found. Moreover, the money network is only modestly correlated with the staff network. Table 25 Correlation Results Between Dependent Matrices. San Mateo Client Information Money Staff Client 1.000 Information 0.859** 1.000 Money 0.525** 0.446** 1.000 Staff 0.448** 0.459** 0.347* 1.000 * p<0.05; ** p<0.01 Long Beach Tables B6 (in Appendix B) and 26 show the correlation results. Most of the independent matrices are highly correlated with each other. This is especially true among SAME GOAL, SAME CLIENT, SAME FUND, MY INFLUENCE, THEIR INFLUENCE, MY NEED, THEIR NEED, and LOS COST because they all have correlation above 0.8. These sets of highly correlated variables are also associated significantly with SAME NETWORK and PROXIMATE. This high correlation in Long Beach suggests that service providers identifying themselves as part of a “network” are more likely to have the following characteristics: (1) sharing same goals, clients, and funding sources with each other; (2) located close 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to each other; (3) seeing each other as low cost service providers; and (4) having perceived needs and influences of one another. Moderate correlations are found between ONLY AVAILABLE and other independent matrices, and between perceived need and influence. Also of concern is that MANDATE (service programs recognized their relations as a results of mandate) does not correlate highly with any other independent matrices, especially those with same goals, clients, and funding sources. Similar to SAME ORGANIZATION, in which service programs maintain relations because of corporate structure, programs related to each other as a result of mandates are less likely to maintain same goals and share similar client pool. Table 26 Bivariate Correlation. Long Beach: Independent Matrices with Dependent Matrices Matrices_____________ Client____ Information Money Staff SAME NETWORK 0.592* 0.522* 0.316* 0.361* SAME 0.235* 0.198* 0.05 0.408* ORGANIZATION PROXIMATE 0.529* 0.5* 0.311* 0.458* LOW COST 0.592* 0.546* 0.271* 0.406* ONLY AVAILABLE 0.373* 0.352* 0.083 0.287* MANDATE 0.309* 0.322* 0.058* 0.437* SAME GOAL 0.618* 0.623* 0.286* 0.54* SAME CLIENT 0.628* 0.633* 0.29* 0.547* SAME FUND 0.567* 0.568* 0.282* 0.511* MY INFLUENCE 0.65* 0.631* 0.3* 0.518* THEIR INFLUENCE 0.58* 0.605* 0.276* 0.549* MY NEED 0.618* 0.623* 0.286* 0.54* THEIR NEED 0.639* 0.644* 0.295* 0.554* * P<0.05 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 26 shows correlations between the independent and dependent matrices (client, information, money, and staff). Correlations with client and information networks are relatively similar to each other. The money network has fewer significant correlations than the other three networks, among these are SAME ORGANIZATION, ONLY AVAILABLE, and MANDATE. Correlations within the money network are the lowest among all four networks, and do not bear any similarity to the staff network. For example, SAME ORGANIZATION and MANDATE are moderately correlated with the staff network (0.408 and 0.437 respectively) but are not significant in the money network. Their correlations with MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED are also higher than with money exchanges. Table 27 Correlation Results Between Dependent Matrices. Long Beach Client Information Money Staff Client 1.000 Information 0.823** 1.000 Money 0.459** 0.338** 1.000 Staff 0.292** 0.516** 0.098 1.000 * p<0.05; ** p<0.01 Bivariate correlations among resource networks in Long Beach are shown in Table 27. Similar to Table 24 in San Mateo, client and information 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. networks in Long Beach are highly correlated with each other On the other hand, the money and staff networks are not correlated. The staff network correlates moderately with the information network (0.52). Tulare A similar correlation procedure is performed for Tulare. Table B7 (In Appendix B) shows that as with the other communities, SAME NETWORK, ONLY AVAILABLE, SAME GOAL, SAME CLIENT, SAME FUND, MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED are highly correlated to each other (>0.8). Moreover, there is perfect correlation (1.0) between SAME GOAL, SAME CLIENT, MY NEED, and THEIR NEED. Table 28 Bivariate Correlation. Tulare: Independent Matrices and Dependent Matrices Matrices Client Information Money Staff SAME NETWORK 0.56* 0.591* 0.159* 0.279* SAME ORGANIZATION 0.276* 0.299* 0.04 0.357* PROXIMATE 0.464* 0.49* 0.082 0.228* LOW COST 0.51* 0.479* 0.152* 0.28* ONLY AVAILABLE 0.515* 0.5* 0.096 0.228* MANDATE 0.328* 0.319* 0.081 0.1 SAME GOAL 0.61* 0.608* 0.177* 0.271* SAME CLIENT 0.61* 0.608* 0.177* 0.271* SAME FUND 0.595* 0.593* 0.177* 0.271* MY INFLUENCE 0.603* 0.601* 0.179* 0.274* THEIR INFLUENCE 0.603* 0.616* 0.179* 0.274* MY NEED 0.61* 0.608* 0.177* 0.271* THEIR NEED 0.61* 0.608* 0.177* 0.271* * P<0.05 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Among all the independent variables, SAME ORGANIZATION and MANDATE are not highly correlated with the rest. Both of them have a correlation lower than 0.6, with the exception that SAME ORGANIZATION and PROXIMATE are highly correlated (0.78), meaning that programs affiliating themselves to the same corporate structure (SAME ORGANIZATION) are also proximate to each other physically (probably in the same building). Table 28 indicates a similar correlation pattern between these variables and the resource exchange patterns as in other communities. Associations between the independent variables and the client and information networks are generally stronger than with money and staff networks. Moreover, SAME ORGANIZATION and MANDATE have the weakest association with all the resource networks. Table 29 Correlation Results Between Dependent Matrices. Tulare Client Information Money Staff Client Information Money Staff 1.000 0.914** 0.28** 0.325** 1.000 0.316** 0.358** 1.000 0.553** 1.000 * p<0.05; ** p<0.01 A very strong correlation between the client and information networks (0.92) is found in Tulare (Table 29). There is a moderate 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. correlation (0.55) between the money and staff networks. Other correlations, for example, those between money or staff and client or information, are relatively modest. San Francisco Analytical procedures were performed again for San Francisco. Table B8 (in Appendix B) shows the bivariate correlation results. Strong correlations (0.8) are found among these independent matrices. Of particular importance is the strong correlation among SAME GOAL, SAME CLIENT, SAME FUND, MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED. Furthermore, SAME NETWORK is also strongly related to all these independent matrices. Given the high correlation between SAME GOAL, SAME CLIENT, and SAME FUND, a combined construct has been developed to measure domain similarity. For MY INFLUENCE and THEIR INFLUENCE, as well as MY NEED and THEIR NEED, two new variables have been developed to measure the dissimilarity of perceived need and influence between pairs of service organizations. Table 30 shows that correlations of all the independent matrices to client and information networks are generally higher than to money and staff networks. None of the correlation with money and staff networks are higher than 0.4, suggesting that these factors may be more strongly related to the client and information networks than to the money and staff networks. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 30 Bivariate Correlation. San Francisco: Independent Matrices with Dependent Matrices Matrices Client Information Money Staff SAME NETWORK 0.625* 0.599* 0.256* 0.31* SAME ORGANIZATION 0.324* 0.325* 0.208* 0.263* PROXIMATE 0.454* 0.434* 0.119* 0.29* LOW COST 0.652* 0.597* 0.209* 0.28* ONLY AVAILABLE 0.531* 0.467* 0.185* 0.197* MANDATE 0.385* 0.381* 0.264* 0.101* SAME GOAL 0.725* 0.661* 0.342* 0.33* SAME CLIENT 0.705* 0.641* 0.371* 0.298* SAME FUND 0.702* 0.638* 0.342* 0.345* MY INFLUENCE 0.705* 0.641* 0.348* 0.335* THEIR INFLUENCE 0.717* 0.653* 0.377* 0.333* MY NEED 0.689* 0.626* 0.371* 0.283* THEIR NEED 0.709* 0.645* 0.369* 0.297* * P<0.05 Certain independent matrices are more highly correlated with the client and information networks than the other independent matrices. Both of the dependent matrices (client and information) are moderately related (>0.6) to SAME NEP/VORK, LOW COST, SAME GOAL, SAME CLIENT, SAME FUND, MY INFLUENCE, THEIR INFLUENCE, MY NEED, and THEIR NEED. Given the high correlation among these independent matrices in Table B8, it comes as no surprise that they are also highly related to the dependent matrices. While the client and information networks are closely associated to SAME NETWORK, its correlation to SAME ORGANIZATION is relatively low and similar to that of money and staff networks. This finding suggests that client and information exchanges 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are less likely to be characterized by relationship among programs that recognize themselves as being part of the same organization. Bivariate correlation among resource networks in San Francisco shows similar patterns to those in other communities (Table 31). Client and informatin networks are highly correlated with each other, whereas money and staff networks are only moderately correlated. Nevertheless, the information network also seems to be moderately related to both the money and staff networks. Table 31 Correlation Results Between Dependent Matrices. San Francisco Client Information Money Staff Client 1.000 Information 0.823** 1.000 Money 0.379** 0.457** 1.000 Staff 0.371** 0.467** 0.525** 1.000 * p<0.05; ** p<0.01 Given the high correlation among SAME GOAL, SAME CLIENT, and SAME FUND, a new variable called DOMAIN SIMILARITY has been created to measure the presence of domain similarity between pairs of service organizations. Discrepancy between how much influence each organization exerted on one another is measured in the factor INFLUENCE. 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INTER-DEPENDENCE is created to measure the dissimilarity in the perception of need between pairs of programs. Normalized Results In order to compare the relative magnitude of various factors associating with the configuration of the resource exchange networks in the four communities, all sociomatrices were normalized into z-scores, hence standardized coefficients could be calculated from the MRQAP regression. Normalized regression results are presented for each community, followed by comparison across communities. San Mateo Across Networks. The availability of potential partners in the community (ONLY AVAILABLE) seems to be a prevalent factor associated with development of inter-organizational resource networks (Table 32). Nevertheless, its relative magnitude is not as large as other significant variables that possess unique contributions to individual resource networks. Organizational affiliation (SAME ORGANIZATION) is another variable significantly associated with multiple resource networks. Among the three resources (client, information, and staff) with significant relationships, programs that belong to the same organizational structure seem more likely to develop staff exchange than other exchanges. 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 32 QAP Regression Results for San Mateo Care System Client Information Money Staff SAME NETWORK 0.286* 0.2747* -0.1211 0.0346 SAME ORGANIZATION 0.1258* 0.0997* 0.0127 0.2249* PROXIMATE -0.0564 -0.036 -0.0566 -0.0478 LOW COST 0.0769** 0.2449* -0.00886 -0.0344 ONLY AVAILABLE 0.1003** 0.1103* 0.1069* 0.0894** MANDATE -0.0842 -0.0273 -0.1143 -0.0959 DOMAIN SIMILARITY 0.1035 0.0611 0.2685* 0.1648** INFLUENCE 0.028 -0.0235 -0.0317 -0.0914 INTERDEPENDENCE 0.0024 -0.0646 0.1071 0.0032 R2 0.32* 0.38* 0.05* 0.13* Intercept 0 0 0 0 * p<0.05; **p<0.1 Network affiliation (SAME NETWORK) and domain similarity (DOMAIN SIMILARITY) are variables that made unique contributions to different resource networks. Whereas network affiliation is more likely to be associated with the client and information networks, domain similarity contributed significantly to the money and staff networks. What is interesting is that as shown in the correlation pattern discussed earlier, network affiliation and domain similarity were highly correlated. Client Network. Program identified as in the same network (SAME NETWORK), the same organizations (SAME ORGANIZATION), low cost providers (LOW COST), or the only available providers in the community (ONLY AVAILABLE) are more likely to make client exchanges. Among these variables, network affiliation (SAME NETWORK) stands out as the most 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significant variable influencing the client exchange patterns. In addition, SAME ORGANIZATION also contributes to the client exchange relationship. Low cost providers (LOW COST) and being the only available organization in the community (ONLY AVAILABLE) also are associated with client exchanges. The overall model explains 32% of the variance in the client network. However, even a program that possesses all the significant characteristics does not guarantee client exchange. Information Network. In the information network, the same set of variables as the client network are significant in the regression model. The overall model is able to explain more of the variance in the information network (38%). Being in the same network (SAME NETWORK) contributes the most in influencing the development of information exchanges, followed by potential partners that are low cost providers (LOW COST). Money Network. Only 5% of the variance is explained in the money network by the overall model. Two variables are significantly associated with the money network. The greater the similarity identified in goals, clients, and funding sources (DOMAIN SIMILARITY), the more likely there is a money exchange between programs. Moreover, the potential partners are the only one available in the community (ONLY AVAILABLE) is also associated with the likelihood of developing money exchanges. 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Staff Network. Significant variables in the money network also prove to be significant in the staff network. Like money exchange, staff exchange is positively associated with domain similarity (DOMAIN SIMILARITY) and that potential partners are the only available providers in the community (ONLY AVAILABLE). However, the most significant variable in the staff network is SAME ORGANIZATION. The entire model explains 16% of the variance in the staff network. Long Beach Across Networks. While being in the same organization (SAME ORGANIZATION) and perceived interdependence (INTERDEPENDENCE) seem not to have any significant impact on any of these resource networks, several variables are found to be “common” significant factors across resource networks (Table 33). Domain similarity (DOMAIN SIMILARITY) has a significant effect on the development of all four resource exchanges. The presence of an only available partner (ONLY AVAILABLE) is significantly related to client, information, and staff networks. Similarly, physical proximity (PROXIMATE) has a significant relationship with exchanges in three resources: information, money, and staff. 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 33 QAP Regression Results for Long Beach Care System Client Information Money Staff SAME NETWORK 0.2687* -0.0003 0.2074* -0.4215 SAME ORGANIZATION -0.141 -0.1629 -0.1708 -0.06 PROXIMATE 0.0662 0.1131** 0.2251* 0.2499* LOW COST 0.1945** -0.0343 -0.1156 -0.5443 ONLY AVAILABLE 0.1204** 0.2529* -0.1232 0.1063** MANDATE 0.061 0.0226 -0.0226 0.1555** DOMAIN SIMILARITY 0.2846** 0.5682* 0.3374* 1.0271* INFLUENCE -0.0519 0.0511 -0.1192 0.1142** INTERDEPENDENCE -0.1733 -0.1711 -0.1118 0.0226 Rz 0.49* 0.45* 0.16* 0.52* Intercept 0 0 0 0 * p<0.05; **p<0.1 Some variables contribute significantly to one or two resource network models. For examples, being in the same network (SAME NETWORK) is associated with client and money networks but not information and staff networks. Mandates (MANDATE) and perceived influence (INFLUENCE) have significantly associations with staff network only. Having a low cost provider (LOW COST) contributes significantly to the client network only. Client Network. Four variables are found to be significant in the client linkages and three of them represented the resource dependence perspective. Being in the same network (SAME NETWORK), having a low cost (LOW COST) or the only available (ONLY AVAILABLE) provider in the community, and being similar in domain with potential partners (DOMAIN SIMILARITY) are 171 permission of the copyright owner. Further reproduction prohibited without permission. associated with whether two programs establish client exchange relations. Among these four variables, domain similarity (DOMAIN SIMILARITY) has the strongest association with client network (0.2846), followed by perception of being in the same network (0.2687). The overall model explains 49% of the variance in client network. Information Network. Only three variables are significantly associated with information network. They are physical proximity (PROXIMATE), the presence of only one partner in the community (ONLY AVAILABLE), and domain similarity (DOMAIN SIMILARITY); among these, domain similarity is strongly associated with the information network (0.5682). The overall model explains 45% of the variance in information network. Money Network. Similar to the information network, three variables are also significant in money network. They are perception of being in the same network with the potential partner (SAME NETWORK), physical proximity (PROXIMATE), and domain similarity (DOMAIN SIMILARITY). All of them positively relate to the development of money exchange between programs. Of these three, domain similarity has the strongest relationship (0.3374), followed by having a physically proximate partner (0.2251). Sixteen percent of the variance in the money network is explained by the overall model. 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Staff Network. Compared to the other resource networks, the development of staff network seems to be influenced by significantly more variables. Significant factors include physical proximity (PROXIMATE), having the only available partner in the community (ONLY AVAILABLE), mandate (MANDATE), domain similarity (DOMAIN SIMILARITY), and perceived influence (INFLUENCE). Of all the six significant variables, domain similarity (DOMAIN SIMILARITY) is the strongest factor associated with staff network (1.0271). The overall model explains more than half (52%) of the variance in the staff network. Tulare Across Networks. Unlike the other communities, factors affecting Tulare's resource networks are more diverse, although there are fewer significant variables in each resource network (Table 34). One variable (LOW COST) is significant among three resource networks (client, money, and staff). Two variables are significant across two resource networks. They are mandate (MANDATE) associated with client and information networks, and domain similarity (DOMAIN SIMILARITY) associated with money and staff networks. 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 34 QAP Regression Results for Tulare Care System Client Information Money Staff SAME NETWORK 0.1075 0.3148* -0.1092 -0.2798 SAME ORGANIZATION -0.0093 0.0093 0.0077 0.2401* PROXIMATE 0.0954 0.0975** -0.2661 -0.1265 LOW COST 0.1152** 0.0706 0.1105** 0.0947* ONLY AVAILABLE 0.1067 -0.0229 -0.1656 0.001 MANDATE 0.2088* 0.2179* -0.0789 -0.1406 DOMAIN SIMILARITY 0.0979 0.0709 0.5965* 0.7592* INFLUENCE 0.101 -0.0464 -0.0941 -0.2248 INTERDEPENDENCE -0.0234 -0.0165 0.0327 -0.1558 R^ 0.35* 0.37* 0.08* 0.20* Intercept 0 0 0 0 * p<0.05; **p<0.1 Client Network. Two variables are significant in the client exchange network, having a low cost partner (LOW COST) and mandate (MANDATE). The strongest variable is mandates (MANDATE) with a result of 0.2088. The overall model explains 35% of the total variance in client network. Information Network. Three variables are significant by associating with the information network. These variables include perception of being in the same network with the potential partner (SAME NETWORK), physical proximity (PROXIMATE), and the presence of mandate (MANDATE). The strongest factor among the three is the perception of being in the same network with the potential partner (0.3148). The overall model explained 37% of the variance in the information network. 174 permission of the copyright owner. Further reproduction prohibited without permission. Money Network. Domain similarity (DOMAIN SIMILARITY) and the presence of low cost providers (LOW COST) are the two significant variables found in money network. Both of them are positively associated with the development of money linkages among service providers in Tulare. Domain similarity is more strongly associated (0.5965) with the money network than the presence of low cost providers. Only eight percent of the variance in the money network, however, is explained by the overall model. Staff Network. Three variables are significant in staff exchanges among programs. Being in the same organization (SAME ORGANIZATION), the presence of low cost providers (LOW COST), and domain similarity (DOMAIN SIMILARITY) all positively relate to staff network, of which domain similarity has the strongest influence (0.7592). The overall model explains 20% of the total variance in staff network. San Francisco Across Networks. Compared to other communities, relatively more variables are significant across all the resource networks in San Francisco (Table 35). Significant variables across all four networks include being in the same organization (SAME ORGANIZATION) and domain similarity (DOMAIN SIMILARITY). Three other variables are also significant in three resource 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. networks. Physical proximity (PROXIMATE) is significant in client, information, and staff networks, whereas the presence of low cost providers (LOW COST) and mandate (MANDATE) are significant in client, information, and money networks. The high frequency of significant factors suggests that a similar set of factors is instrumental in associating with network relationships in San Francisco. Client Network. Six significant variables shape the development of the client network. These variables include being in the same organization (SAME ORGANIZATION), physical proximity (PROXIMATE), the presence of low cost provider (LOW COST), of the only available partner in the community (ONLY AVAILABLE), of mandate (MANDATE), and domain similarity (DOMAIN SIMILARITY). Of the six significant variables, the presence of low cost providers (LOW COST) remains the strongest factor in associating with the client network (0.3365). The overall model, presented by these six variables, explains more than half (52%) of the total variance in client network. Information Network. Five variables significant in the client network are also significant in the information network. The presence of the only available partner in the community (ONLY AVAILABLE), a significant factor in client network, is replaced by the perception of being in the same network with the potential 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 35 QAP Regression Results for San Francisco Care System Client Information Money Staff SAME NETWORK 0.0143 0.1474* -0.0842 -0.0063 SAME ORGANIZATION 0.0841** 0.0791** 0.0557** 0.1576* PROXIMATE 0.0947* 0.1819* -0.1633 0.0737** LOW COST 0.3365* 0.166* 0.1336* 0.0031 ONLY AVAILABLE 0.0875* 0.0338 -0.1555 0.0155 MANDATE 0.0639** 0.0569** 0.1015* -0.0562 DOMAIN SIMILARITY 0.1692* 0.1544* 0.4506* 0.3059* INFLUENCE 0.0306 0.0277 -0.0409 -0.0723 INTERDEPENDENCE 0.0398 -0.0132 0.0061 -0.0316 R* 0.52* 0.44* 0.17* 0.15* Intercept 0 0 0 0 * p<0.05; **p<0.1 partner (SAME NETWORK). The strongest factor in the information network is domain similarity (DOMAIN SIMILARITY) (0.1544). The overall model explains 44% of the total variance in the information network. Money Network. Four variables are significant in associating with the money exchange relations, namely being in the same organization (SAME ORGANIZATION), the presence of low cost providers (LOW COST), the presence of mandate (MANDATE), and domain similarity (DOMAIN SIMILARITY). The strongest factor in money network is domain similarity (DOMAIN SIMILARITY) (0.4506). With the exception of the staff network, the proportion of variance explained by the overall model is relatively small in money network (17 percent), compared to the other networks. 177 permission of the copyright owner. Further reproduction prohibited without permission. Staff Network. Being in the same organization (SAME ORGANIZATION), physical proximity (PROXIMATE), and domain similarity (DOMAIN SIMILARITY) are the three significant variables affecting the establishment of staff exchange relations among programs. Among the three significant factors, domain similarity has the strongest relation with staff network (0.3059). The overall model explains 15% of the total variance in the staff network. Comparison Across Communities The set of factors as presented in the overall model seems to be more capable of describing care systems in some communities rather than the other. Comparatively, they are less likely to explain network development in Tulare but were considered favorable factors in the San Francisco networks. Each variable is more likely to have a statistically significantly relationship with more resource networks in San Francisco than in Tulare. If the proportion of variance explained is an indicator of the effectiveness/relevance of the model, Table 36 shows that similarities exist between San Mateo and Tulare, as well as Long Beach and San Francisco. The overall model explains about one-third of variance in the client network and 37 to 38% in information network in San Mateo and Tulare, whereas a higher percent of variance explained (44% to 52%) in these networks is attained in Long Beach and San Francisco. 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 36 Variance Explained Across Communities Client Information Money Staff San Mateo 32% 38% 5% 13% Long Beach 49% 45% 16% 52% Tulare 35% 37% 8% 20% San 52% 44% 17% 15% Francisco The overall model also demonstrates its applicability in explaining client and information networks more effectively than money and staff networks. With the exception of the Long Beach staff network, less than one-fourth of the variance is explained in money and staff networks for all the communities. As found in almost every community in this study, some statistically significant factors are common among various resource networks. Table 37 shows which resources are more likely to be influenced by a particular variable. Table 37 shows that more variables have statistically significant associations with the client network less so with the money network. These variables are more likely to be associated with the client and information exchanges rather than the money and staff exchanges. Nevertheless, few variables have a statistically significant impact on a particularly resource 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. network across all communities. For example, the presence of low cost providers (LOW COST) is important to whether programs established client exchange relations in all four communities. Domain similarity (DOMAIN SIMILARITY) is pivotal to all resource networks, but it is particularly critical for money and staff networks (Table 36). 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Table 37 Factor Comparison Across Communities Client Information Money Staff SM LB TL S F S M LB TL S F SM LB TL S F S M LB TL SF RESOURCE DEPENDENCE Same Network * A A A A A A Proximity A A A A A A Mutual Needs Only Available A A A A A A A A A Provider POWER A Mutual Influence TRANSACTIONAL COST Same Organization A A A A A A A A Low Cost Provider A A A A A A A A GOVERNANCE Mandate A A A A A A DOMAIN SIMILARITY Domain Similarity A A A A A A A A A A A A SM: San Mateo LB: Long Beach TL: Tulare SF: San Francisco 00 CHAPTER VI: DISCUSSION AND CONCLUSION Findings from the result chapter have shown that the organization of care for older adults is very complex. Using the network perspective, how service providers organize their care in four communities is examined in terms of the exchange patterns of four resources (client, information, money, and staff) among service providers. This chapter is organized into three sections. First, findings from the result chapter are grouped into major categories as a description of community-based care systems. Hypotheses developed in the conceptual chapter is reviewed and presented throughout the description of these categories. Second, based on general findings, this chapter discusses the extent of integration versus fragmentation of community-based care systems and the extent of community differences in their care systems. Third, the chapter offers conclusions to the study and suggests areas for future research. Discussion on Hypotheses In this section, findings presented in the results chapter are discussed according to the five hypotheses listed in the conceptual chapter. A general conclusion on the description of the organization of care follows discussion on each of these hypotheses. Table 38 summaries the findings of each hypothesis in this study. 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 38 Summary of Hypotheses. Tests, and Conclusion for Hypotheses Hypotheses Tests/Means for Examination Support 1. Patterns of exchanges will differ across the four resources (client, information, money, and staff). ^ Different network density and centralization values ^ Varying types of prominent service providers (senders, brokers, and receivers) Cliques present in client and information but not money and staff Support 2a, Patterns of client and information exchanges are more likely to be similar to each other than those of money and staff exchanges, * Moderate density values * Similarity in the prevalence of brokers ^ High QAP correlations ^ Presence of cliques * Moderately connected blockmodel relations Support 2b. Patterns of money and staff exchanges are more likely to be similar to each other than those of client and information exchanges, ^ Low density values ^ Prominent providers as senders or receivers * Absence of cliques * Sparsely connected structural relationships X staff network more likely on informal understanding than contracts, unlike money network X Only moderate QAP between the two networks Am-bivalent 3a, Money and staff exchanges are more likely to be associated with formalized exchanges than are client and information exchanges, 3b, Client and information exchanges are more likely to be associated with informal exchanges than are money and staff exchanges * MRQAP regression Support 4. Organization of elder care systems will differ across communities, • Density, Multiplexity of ties (Chi-square), Formalization of ties (Chi-square, MRQAP regression), Service-provider centrality, Cliques, Structurally equivalent subgroups • Factors associated with resource exchange patterns (MRQAP regression) Am bivalent, but mainly significant differences are found 5. Resource exchange relations are associated with multiple factors rather than dominated by a single factor, * MRQAP regression Support 6, Different resource networks are likely to be associated with different sets of factors. *' MRQAP regression Support 00 C O Different Resources Display Different Exchange Patterns in Care Systems (Hypothesis 1) The complexity of care organization can be indicated through the different exchange patterns of the four resources. In the conceptual chapter, the first hypothesis states that patterns of exchanges will differ across the four resources (client, information, money and staff). Findings from the result chapter show support for the first hypothesis in three ways. First, the four resources examined display different levels of network density and centralization in different communities. Among the four resources, client and information exchanges are the most frequent in the network and demonstrate the densest network connection. There are also more service providers participating in making client and information exchanges. Comparatively, money and staff exchanges are less frequent and therefore suggest relatively disconnected networks. The higher network centralization measures found in the client and information networks suggest that these resources tend to flow more unevenly among their service providers than money and staff. Second, differences in exchange patterns among resources can also be illustrated by the types of prominent service providers in each resource network. Prominent brokers, rather than senders or receivers, are more likely to be found in the client and information networks. This difference in the types of prominent providers can be related to the density of network connections among providers. Lower network density with an equivalent 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. number of providers may suggest that few providers send or receive resource exchanges simultaneously. Comparatively, prominent senders and receivers are more prevalent in money and staff networks. Unless the relations are mandated by the government, it seems logically to suggest that service providers are less likely to be both senders and receivers of these resources because it is inefficient to establish unnecessary ties. Imagine a broker organization assisting money exchanges between a funding organization and a service provider. In this situation, since money requires establishment of contracts, two contracts are needed to justify one organization to be a broker. The funding organization may think that the broker is unnecessary if the aim is to grant money to serve older adults. It may as well establish contracts with the service provider to save administrative costs. It is also difficult to justify how the broker organization is able to identify itself serving older adults. Therefore, prominent brokers are relatively rare in money and staff networks. The correlation of centrality measures across the four resources (Tables B1 through B4 in Appendix B) provides indirect support on the different exchange patterns across resources. In general, providers that are prominent in client exchange are also prominent in exchanging information but do not necessarily maintain a central role in money or staff networks. 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Third, the first hypothesis of differential exchange patterns across resources is also supported by findings of cliques. Cliques are cohesive subgroups formed by service-provider types that develop close direct relations with one another. The formation of cliques could be different, depending on the types of resources exchanged. Findings in the results chapter show that cliques are found in the client and information networks only, but not the money and staff networks. Cliques may be a more relevant form of inter-program entity found in the client and information networks than in the money and staff networks. That cliques are not found in the latter two networks can be attributed to two reasons. First, relationships between service providers are not deep enough to establish money and staff resources. Second, perhaps it is not necessary to form clique-like relations with service providers in the money network. This second reason resembles the reasoning of the relative scarcity of prominent brokers in these networks. High Similarity Between Client and Information, but not Between Money and Staff (Hypothesis 2) Findings Suggest High Similarity Between Client and Information Exchanges Similarity between client and information, and between money and staff networks can be examined from five aspects: (1) density, (2) brokers, (3) QAP correlations, (4) cliques, and (5) blockmodels. Findings from these aspects support hypothesis 2a (the pattern of client and information 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. exchanges are more likely to be similar to each other than that of the exchanges of money and staff). In contrast, they do not support 2b (the patterns of money and staff exchanges are more likely to be similar to each other than that of the exchanges of client and information). Findings in this study supports hypothesis 2a. First, the densities of client and information networks are similar. Both networks have differences from each other of no more than 2 percent in all four communities (Table 6), suggesting that service providers establish relatively similar amounts of client and information exchanges with one another. Second, service providers that are prominent brokers in the client network are also likely to be brokers in exchanging information. From Table 19, three out of five brokers in San Mateo and Long Beach’s client networks are also brokers in their information networks. Similarity in the types of service providers as brokers suggests that the ‘landscapes’ of resource exchanges are relatively the same between the client and information networks. Third, the high QAP correlation (> 0.8) between the networks themselves across four communities in Tables 25, 26, 29, and 31, as well as correlation between the provider centrality values of the two networks (Tables B1 through B4 in Appendix B) provide further evidence to support to the hypothesis. The high correlation indicates that service providers establish similar types and number of partners in both of these resource networks. 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fourth, with the exception of Long Beach which has no two-tie clique in the information network, moderate to significant overlap in terms of clique members is found in the client and information networks in the other three communities. Service providers that become client-clique members in exchanging clients with each other are also more likely to be information- clique members. Fifth, the biockmodel diagrams displaying structural relations between structurally equivalent groups in the four communities also indicates that client and information exchanges are closely linked. In addition to the five major findings listed above, both the client and information networks are also more likely to be characterized by informal relationship and formal verbal agreements rather than contracts (Table 9). Such high level of similarity in the exchange patterns of the client and information resources could be related to the fact that they are associated with a similar set of factors shaping the development of exchange patterns. Findings from the result chapter (Table 37) show that although there is some variation across communities, both the client and information networks are associated with approximately the same set of factors. The resemblance is especially close in the San Mateo and San Francisco care systems. For San Mateo, factors significantly associated with client exchange pattern are identical to those with information exchanges. For the San Francisco system, five out of seven significant factors commonly associated with its client and information exchange patterns. 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Findings Are Ambivalent to Support the Similarity Between Money and Staff Exchanges On the other hand, study findings are ambiguous in evaluating hypothesis 2b. The money exchanges are hypothesized to be more likely to be similar to staff exchanges and vice versa rather than with the client and information exchanges. While the exchange patterns of money and staff are distinctively different from those of the client and information networks, they are not as similar as client with information exchanges. Findings supportive of their similarity include (1) their seemingly low density values, with the exception of Tulare; (2) their tendency to have more prominent providers as senders or receivers but not as brokers; (3) the absence of cliques; and (4) their sparsely connected structural relationships among structurally equivalent groups. Despite these similarities, there is evidence showing that the two resources are not distinct. For instance, Table 9 shows that providers exchanging staff are more likely to rely on informal understanding to maintain exchange relations instead of contracts as in of the money network. The moderate QAP correlation (< 0.6) between the money and staff networks also suggests that the two resources quite dissimilar. Given that factors associated with the two networks do not resemble one another as strongly as those with client and information networks, there must be greater differences between money and staff exchange patterns. 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Formalization of Resource Exchange (Hypothesis 3) This study found that service providers rely on different modes to exchange their resources. In general, service providers are more likely to exchange clients through informal understanding or formal verbal agreements. Quite infrequently are formal contracts the dominant mode through which clients are exchanged. The nature of client care may require efficient and responsive management at the front workers’ level so that the informal exchange of a telephone call is sufficient to establish a relationship concerning the clients’ care. Similarly, requesting simple client or service provider information over the phone is also considered acceptable under informal understanding or formal verbal agreement. On the other hand, the issue of accountability increases when service providers file claims for reimbursement or details for illustrating task responsibility for their staff. As a result, formal contracts are more commonly used to characterize money and staff exchanges. This difference in exchanging resources in terms of their modes may also vary when community-based care systems become more rationalized. One can conjecture the increased influence of formal contracts as a mode in shaping exchanges of various resources including client and information exchange when the community-based care systems develop into an organized delivery system (Shortell et al., 1996). In the organized delivery system, negotiated contracts among service providers may dominate the landscape of service delivery so as to enhance system efficiency and 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. effectiveness. Functional integration and financial integration will be developed in order to support the development of clinical integration. Although Shortell et al.’s (1996) conceptualization is confined to the acute care arena, a similar notion of an organized system of long-term care or an integrated acute and long-term care system can also be applied in this context. The development of such an organized delivery system, however, does not nullify the role of informal understanding among service providers in serving older adults. Nevertheless, the role of formal contracting is likely to increase when, for instance, service providers of the aging network attempt to cooperate with managed care organizations to provide comprehensive coordinated care for older adults. Extent of Community Differences in Care System/Exchange Patterns (Hypothesis 4) This section discusses how unique community-based organization of care is across communities in providing services for their community- dwelling older adults. Given that no statewide system has been developed to provide care for older adults, communities in California have tremendous flexibility in organizing how care can be delivered to their older residents. As found in this study, there are significant differences as well as similarities across communities in how service providers organize their services for their older clients, depending on which aspects of network linkages are concerned. The extent of community differences in care organization will 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. be discussed in seven areas: (1) density, (2) multiplexity of ties, (3) formalization of ties, (4) service-provider centrality, (5) cliques, (6) structurally equivalent subgroups, and (7) factors associated with resource exchange patterns. Some of the findings in the result chapter are purely descriptive whereas other can be verified through statistical tests. Communities do not differ greatiy from each other in terms of the proportion of exchanges their service providers make within the community. In general, service providers establish less than 30 percent of all possible client and information exchanges, less than 6 percent of all possible money exchanges, and less than 9 percent of all possible staff exchanges in their respective communities. The differences in the proportion of client and information exchanges made across communities are larger than those in money and staff exchanges. For example, variation in the proportion of client exchanges can be as large as 11% points between San Mateo (18%) and Tulare (29%) while variation in money or staff networks across communities is relatively small (around 2 to 3 percent points). Significant differences have been found across communities in terms of the variety of exchanges. It appears that San Mateo stands out from the other three communities in that a small proportion of exchanges among service providers relies on one resource while its exchange relations with three resources (42%) are close to double those of the other three communities (around 22%). Over 50 percent of the exchanges established 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. by service providers in San Mateo consist of at least three resources when compared to the next closest community is Tulare (40%). Given the variation in the total number of exchanges in each community, San Mateo and Tulare lie in the middle of the four communities, with 112 and 123 exchanges respectively. The findings suggest that the number of exchanges could be related to the proportion of multiplex ties in the community. While most of the exchanges are characterized by informal agreement, chi-square tests show that there are substantial differences across communities. Among all the communities, over 80% of the relations established in San Mateo are composed by single type of arrangements, compared to 71 % in Long Beach and Tulare as well as 62% in San Francisco. In addition to informal agreements, formal verbal agreements (9%) seem to be an important characteristic for San Mateo relations. However, combined agreements are not very popular in San Mateo, as revealed in its relative low percentage in Table 8. When the relations between the formalization of ties and the various resource networks are examined (Table 9), factors associated with various resource networks are similar across communities. For example, informal agreement is considered a very influential factor in the client and information networks across all the communities, whereas in all communities, contractual agreement is the dominant factor for money exchanges. 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The dominance of Social Services as a broker in client and information exchanges across four communities (Table 19) suggests that some basic programs have been established in communities in California in general. Perhaps this arrangement is possible because Social Services is offered under the Department of Public and Social Services through programs such as the in-home supportive services. Other client brokers popular among communities include Home Health Care, Senior Centers, and Adult Day Health Care. These are small private providers that more likely provide community-based supportive services for older adults predominantly. Community differences in terms of prominent receivers and senders are widened because there are fewer similarities across communities. For example, hospitals are the only service-provider type that can be considered as prominent senders for three communities (San Mateo, Tulare, and San Francisco). As for prominent receivers, Home Health Care is the only service-provider type receiving money across four communities. A major similarity found across communities in terms of clique is that all the cliques identified in these communities tend to be really small in size and in number. No client clique consists of more than three service- provider types. No more than three cliques have been found in these communities. Moreover, no cliques are found in the money and staff networks across all communities. Although members within these cliques are not all the same across communities, they tend to show that cliques are 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. composed mainly of health-related as well as social-related service providers. For example, in San Mateo, there are actually two health-related cliques (Linkages/MSSP Program, Adult Day Health Care, and Hospitals, as well as Linkages/MSSP Program, Home Health Care, and Adult Day Health Care). In Long Beach, it is a clique formed by both health and social service providers (Linkages/MSSP Program, Home Health Care, and Social Services). Cliques of combined nature can also be found in Tulare and San Francisco. Yet in Tulare, another clique clearly consists of social service providers such as Area Agencies on Aging, Senior Meals Program, and Social Services. In San Francisco the clique consists of Social Services and health providers such as Adult Day Health Care and Health Clinics. Similar information cliques can also be found. Consequently, while different dynamics have been displayed in these four communities, there are certain similarities found in how cliques are composed. Although a similar number of structurally equivalent groups is found among the four communities, the structural relations among these groups vary, and this can be found among a similar set of service-provider types. First, the grouping of service-provider types into structurally equivalent groups suggests that the same service-provider type in different communities would display different relations with other service-provider types and there is no standard or mandate of how relations should be arranged or developed. The diverse grouping therefore highlights the importance of community context in affecting the dynamic exchange 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. patterns for each community. Second, different patterns of interaction are found among structurally equivalent groups in the four communities. While some communities such as San Francisco and Long Beach indicate a relatively differentiated division of labor, the organizations of care in other communities are not very structurally differentiated. Interactions between groups in Tulare and San Mateo are not very well defined. A lot of the community differences in the organizations of care for older adults could be related to the analysis of factors associating with the exchange patterns. Table 37 indicates the variation in factors in associating with the four networks in each community. Among all communities, San Francisco has been associated to the largest number of factors (ranged from three in staff exchanges to six in client and information exchanges) listed in the model while Tulare is associated with at most three factors (in information and staff exchanges). Different factors appear to show varying association with different resource networks across communities. For example, Physical Proximity (PHYSICAL PROXIMATE) is an important factor in Long Beach and San Francisco organizations of care, especially in their information and staff exchanges while being in the same organization (SAME ORGANIZATION) seems significantly associating with client and information exchanges in San Mateo only. Having only one service provider of that type is also important in the organization of all the exchanges in San Mateo but this is not important at all in Tulare. Similarly, while mandate seems to be 196 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. important in client and information exchanges in both Tulare and San Francisco, it is not a major factor in affecting any exchange in San Mateo at all. Factors that seem to have a general impact on most of the exchanges in all four communities are LOW COST and DOMAIN SIMILARITY. To this end, it seems that service providers across communities are not entirely distinct from each other as to how they are drawn to develop exchanges with each other. Service providers are still more likely to develop exchanges if their potential partners possess similar goals and serve similar clients. Moreover, given the fact that service providers are not obligated to coordinate services and cooperate with each other, any attempt to do so would be considered costly to them in terms of time and resources. Searching for low cost providers consequently may be an overarching strategy to achieve the goal of providing coordinated services in a fragmented organization of care. Multiple Versus Singular Perspective in Describing Resource Exchange Patterns (Hypotheses 5 and 6) Findings from the result chapter provide sufficient evidence to support hypothesis 4 that resource exchange patterns are associated with multiple factors rather than being dominated by a single factor. Based on the MRQAP results (Table 37), exchange patterns of the four resources are influenced significantly by at least two factors, with the most of six factors on the client network in San Francisco. This finding of multiple factors in 197 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. resource exchange patterns suggests that one should not limit the explanation or interpretation of resource network to one perspective. Instead, multiple perspectives complementary to one another are likely to enhance the understanding of the complex resource exchange patterns. Findings show that differences are found in terms of factors shaping the four resource networks, though such difference is not very distinctive (Table 37). For example, though it is not for all communities, being in the same organization (SAME ORGANIZATION) significantly associates with the patterns of client and staff exchanges. Moreover, it seems that whether SAME ORGANIZATION remains an important factor depends on the individual community. Furthermore, PROXIMATE, a variable that may enhance client and staff networks, is found to be important in the information network too. MANDATE, a variable thought to have a significant role in establishing money and staff exchanges, actually contributes more to the development of client and information networks. What is even more surprising is the relative insignificance of both INFLUENCE and INTER-DEPENDENCE in shaping these resource networks. With the exception of the staff network in Long Beach being associated with INFLUENCE, none of the resource networks in all the communities are significantly associated with the two variables. The lack of significant contributions of these two variables may be related to their high correlation with other factors such as DOMAIN SIMILARITY and SAME NETWORK. Given that the latter two variables contribute significantly to 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. many resource networks in various communities, the relative magnitude of INFLUENCE and INTER-DEPENDENCE may have been reduced. Hypothesis 6, which states that different resource networks are likely to be associated with different sets of factors, could be supported if we only count those factors that are commonly found significant in at least three communities. Under this criterion, client networks are associated with LOW COST and ONLY AVAILABLE, both of which represent the resource dependence perspective. Information networks are associated with PROXIMATE and SAME NETWORK. The money networks are the least explained pattern: the multiple-factor model only explains five percent of resource exchange patterns in San Mateo and at most 17 percent in San Francisco. The money networks are associated with DOMAIN SIMILARITY only. The staff networks are associated with SAME ORGANIZATION and DOMAIN SIMILARITY. It is possible to conclude that the factors underlying the resource exchange patterns are complex, since different resources require explanations from a dissimilar perspective. Yet, given the moderate similarity among resource networks, the resource dependence perspective appears to be a relatively strong perspective for the client and information networks. Moreover, as a common condition to inter-organizational linkages, domain similarity remains an important contributor to money and staff exchanges. 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. One major insight that emerged from the MRQAP analysis is the relatively small impact of MANDATE on the presence of these resource exchanges. The multiple-perspective model appears to be more applicable to describing the client and information networks and less so to the money and staff network. Given that few studies have investigated what factors influencing the development of various resource exchanges, it is difficult to suggest any reason for such difference. However, what is implied in the differential explained variance from the MRQAP results on these resource networks further affirm that one should not treat network relationship of these resources as they are of the same kind. General Conclusion on the Organization of Care Care Systems Are More Than the Commonly Suggested Aging Networks The organization of care systems for older adults, as identified by service providers, is not just an aging network delineated by gerontological service researchers or aging policy makers. Although major programs serving older adults remain core service providers in these care systems, the network encompasses a diverse set of service providers. From the results, providers identified themselves as serving older adults may include private foundations, emergency services, and other general service providers. While they are the major services for community-dwelling older adults, these are supportive services for the entire care system in that 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. community. Hence, it would be important to take this diversity and community-based characteristic into account, when community-wide intervention is desired. In fact, according to Adams and Nelson (1997), organizing care for older adults may also involve family and other important components in the community as well. Care Systems Are Made Up Of Exchanges Of Not Single But Multiple Resources Another aspect of understanding the complexity of these community- based care systems is its multiplexity of resources being exchanged in the network. As described in the result chapter, the majority of exchanges between service providers are multiplex in nature, mainly made up of client, information, and staff. The multiplexity of resource exchanges suggests an already relatively high level of integration/connection among providers in the network. Since service providers from a diverse set may have more difficulty in working with one another to promote common understanding, the development of multiplex relations may also suggest their long-term commitment of working toward network development for serving older adults. Most of these multiplex ties are associated with exchange of clients. This finding indicates that establishing relations with other providers is mainly for client concerns. Service providers may not develop relationships 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. solely for exchanging information, and most of the information exchanges are client-related. The scarcity of four-tie exchanges in community-based care systems merits two comments. First, it could relate to the area yet to be integrated for a more cohesive care network for service providers in a particular community. From the result chapter, money is found to be the resource least likely to generate a dense network. Since money necessitates more trust in building relations between service providers, it is easier to understand why the current network consists mostly of multiple ties composed mainly from client and information, and occasionally from staff. Moreover, perhaps because ties composed of money and staff (administrative function) are much fewer than those of the delivery function, true/extensive collaboration/integration has not been realized in these communities. Following Bolland and Wilson’s (1994) argument and the hypothesis developed in this study, client and information tend to have similar functions for community-based care systems and are found dominating the resource exchange landscape of the networks. Second, the lack of comparable money exchanges among service providers could be related to the nature of program funding in the United States. For many services for older adults, the flow of money resources is mainly vertical in nature: from the federal or state or local government to individual vendors. As a result, horizontal money exchanges are relatively scattered in community-based care systems. Nevertheless, recent rapid 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. development of managed care may encourage more service providers to establish horizontal links with private charities and managed care organizations at the community level, hence re-shaping the money exchange network to be much denser than what is found in this study. Implications of Similarity and Differences across Resources One major insight from this study concerns the similarity and differences across the four resource exchanges. The four resources chosen in this study represent important elements at various levels in the organization of care for older adults. Few studies have examined how similar they are in organizing care in a service system. The comparison of various resources being exchanged among service providers in community-based care systems suggests that client and information serve a similar function in the operation of the care system than money and staff. One may argue that the similarity between client and information may be related to a limited definition of “information” to “client- related” information. While the original questionnaire does not specify what kind of information is being exchanged between service providers, the close similarity found between these two resources infers that information may be closely client-related. The mixed findings on the similarity between money and staff exchanges may be related to the fact that these are administrative resources operating at different levels. Since the current financing system 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. emphasizes the vertical flow of funds from the federal or state government to individual vendors, the absence of money exchange may be more related to this categorical type of financing. The relatively under-developed staff exchange network, however, could be associated with organizational barriers to communicate across organizations with different disciplines and gradual realization and dedication of the importance of committing to care system development. To the extent funding may restrict or facilitate more effective exchanges of staff use, the latter may also be considered as a further step after coordinating clients through facilitating information exchanges towards care system development. This interpretation may suggest the reason for a closer relation of staff exchange to client and information exchanges. The differences in these resources suggest that service delivery innovation needs to target carefully the type of resources with the strongest impact. Since these resources are not totally different from one another, policy makers designing innovation/intervention targeted at one type of resource may need to consider how such intervention may affect exchange patterns of the other resources. Moreover, success of an intervention at one level may require concurrent intervention at another level as supporting element. For example, innovation targeted at facilitating client exchanges has to take information integration into consideration. Without integrating or unifying the information system or collection, coordinating clients care is likely to be less effective. 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Integration in the Organization of Care In this section, the extent of fragmentation in the organization of care is discussed- Given that many previous studies have described the community-based care systems as fragmented, this section provides some insights into the issue and by revisiting the conceptualization of community- based care systems presented in the background chapter. Then, community differences in regard to resource exchange patterns will be discussed. Extent of Integration Versus Fragmentation in Community-Based Care Systems The extent of integration versus fragmentation in community-based care systems depends on which aspect of the care systems is examined. One major finding from this study is that different exchange patterns result from different resources, hence the level of resource integration varies. The differences in the extent of integration by resources are manifested not only in the number of service providers (actors) establishing exchanges in that resource network but also in the number of partners (ties) that have maintained exchanges with in that network. There are fewer service providers involved in the money and staff networks. Moreover, service providers refer clients and exchange information with more providers in the network than those in the money and staff networks. 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A third indicator is the presence of cliques in the client and information networks and not in the money and staff networks. Since cliques reflect close exchanges among a group of at least three service providers, its presence in the client and information networks suggests that, compared to money and staff exchanges, service providers in the former two networks are more integrated. Such presence of cliques, however, does not indicate a particularly high level of integration in these networks. Findings in the result chapter show that not only is the number of cliques is very limited in the client and information networks, but the size of the cliques is also small, limited to no more than four service providers that have close client and information exchanges with one another. These small cliques may suggest two possibilities in regard to care system development. First, it could be that the presence of small-sized cliques indicates a need for greater system development. Current findings suggest that service providers be at an embryo-like stage of forming alliances or joint ventures. Therefore, the size of cliques may increase as service providers develop consensus and collective goals in serving older adults through forming alliances and coalitions. Second, the small size may be the end stage of system development. Given the high cost of coordination (Van de Ven and Ferry, 1980), service providers may find it very costly to maintain close relations with many other service providers and hence are more likely to keep the size of their clique to a minimum. This possibility is supported by another 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. finding that a majority of resource exchanges are in the form of informal understanding, especially in term of client and information exchanges. Since informal relationship requires intense personal exchanges over time, it is extremely difficult to extend the size of the cliques. The overlap of cliques of the same resource may imply that the service providers who belong to at least two cliques acts as broker for these cliques. Clique overlap may present another issue to the integration- fragmentation discussion. In the result chapter, Social Services performs as a key member in all three information-cliques and two client-cliques in Tulare. The absence of Social Services would disintegrate the structure of all the cliques in Tulare. Another example would be Adult Day Health Care and Health Clinics in the client-clique and information-clique in San Francisco. Their “dual” membership in these cliques characterizes the interaction of the two cliques and promotes further integration of the care networks. When comparing the multiple affiliation of Adult Day Health Care and Health Clinics to the cliques in San Francisco with that of Linkages/MSSP Programs, Adult Day Health Care, Home Health Care, and Hospitals in San Mateo, the integration in terms of resources is greater in the latter because all the client-clique members are also members of the same information cliques. No service-providers are excluded from exchanging both information and clients to one another in these cliques. Another implication for high degree of clique overlap lies in the potential these cliques create for individual service-providers in developing 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. future inter-organizational linkages in other cliques. As Gulati (1996) argues, organizations are more likely to make new ties with potential organizations that have existing relationships with their current partners. It is the network that cultures/nurtures potential future relationships among service-providers. This point can be well elaborated by using Social Services in Tulare as an example. The position of Social Services allows its other clique partners to obtain more accurate information about each other through their close relations with Social Services, and hence generates high potential for further relationships between members of different cliques. Besides overlapping cliques, non-overlapping cliques may also provide additional insights to the extent of integration in community-based care systems. If cliques are found to be non-overlapping with each other, service providers of each clique may not have common goals with members in other cliques. This situation can be found in the two information-cliques in San Francisco. One information-clique is composed of Senior Centers, Information and Referral, and Transportation, and the other of Adult Day Health Care, Health Clinics, and Case Management. The presence of these two cliques suggests that they serve different segments of the older population, with the Senior Centers-lnformation and Referral-Transportation clique serving a younger and healthier group and the ADHC-HEACLIN- CASEMGT clique possibly serving a more frail group of older adults. Moreover, the non-overlapping nature of these two cliques indicates the 208 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. existing organization among service-providers in serving two groups of older adults. The absence of close information exchanges may imply that while the various services are available in the community, accessibility from one group of services to another may prove to be low. Integration, therefore, could be difficult to achieve when groups of service providers serve different populations. The purpose of integrating services for various populations should be questioned. A fourth indicator of the integration of community-based care systems is the interaction among structurally equivalent groups across different resources. As described in the conceptual chapter, structurally equivalent groups are composed of service providers with approximately the same exchange relationships with other service providers in the network. The differences in exchange dynamics across resources suggest again that the complexities of integration cannot be subsumed under a simplistic description that characterizes community-based care systems as duplicative, fragmented, or uncoordinated (Bolland & Wilson, 1994), or that portrays the organization of various services as a non-system (Hennessy & Hennessy, 1990). The blockmodels in the result chapter indicate few structurally equivalent groups are related to all the other groups. However, given that some service providers are traditionally major providers for caring for older adults, the lack of interaction of these groups to other structurally equivalent groups may imply that they operate in a closed, separate subsystems of their own. For example, skilled nursing facilities are often 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. found as an isolate service-provider-type, implying that relocating residents from these facilities back to the community could be difficult. Future Research Findings from this dissertation provide new opportunities for research in understanding in care systems. First, the increased popularity and penetration of managed care is likely to reshape the existing community- based organization of care for older adults. “To what extent do various managed care arrangements reorganize the network of resource exchanges among service providers?” would be an important area of research. Community-based organization of care can be studied longitudinally between several time points to assess the process of reorganization in resource exchange patterns among service providers by managed care. Future research could also include other important care system components such as participating federal and state units as well as families. The inclusion of federal and state-level government units in understanding the local organization of care systems adds the vertical dimension to networks of resource exchanges and may provide a fuller description on how a particular resource exchange pattern comes to place. Adding federal and state-level linkages to the community-based system also may strengthen our understanding of how changes in federalism impact the local configuration of service delivery. 210 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As mentioned in the background chapter, despite the reduced availability of adult children as potential caregivers, family remains a very important component in the overall care system. Changes in family composition and amount of caregiving time provided by family members would affect the demand for formal care, although it does not infer that the latter would substitute the specific role of the former. The interface between the informal and the formal care systems would be an important area for future research. Important research questions may include “to what extent families can help older adults access to the formal care systems?" and “how do care systems ensure families that provide caregiving to older adults be supported?” Another area of future research focuses on the development of taxonomy for care networks developed across various communities. Fortinsky (1990) described many care systems are led by one of these three organizations: Area Agencies on Aging, acute care hospitals, and residential facilities. An important finding from this dissertation is the complexity of care systems which a single organization is less likely to be prominent in all aspects or levels of resource exchanges. Consequently, depending on the level of resource exchanges as well as the local community dynamics, care networks are likely to be diverse across communities. Nevertheless, the need to develop a taxonomy is important to further understand how care systems in general. 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The motivation of designing efficient and effective systems of care leads us to another important area of future research: linkage of structure to outcomes. Currently, little is known about how and to what extent care organization is related to various health and functional outcomes of older adults. The argument that integration would benefit frail older adults has been too vague that it is difficult for policy makers, researchers, and providers to specific if benefits exist and who is benefited from integration (Branch, 1999). In order to clarify the relationship between care systems and individual’s care outcomes, one potential research area can be examining various levels of care systems and how one intervention at a particular level impacts the other levels as well as care outcomes of the consumer of the system. Limitations Several limitations have emerged as this dissertation attempts to describe and examine the community-based organization of care for older adults. The first one is related to the nature of the data set used for the study. Although the response rate for community-wide data collected in this study is relatively high, collecting information for the entire population of service providers and creating complete sociomatrix for analyses proved to be too time- and resource-consuming. This dissertation aggregated the responses from individual service providers into service-provider types for two reasons. First, it was to create 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. complete sociomatrices on the four resources so that exchange patterns could be revealed. Without aggregating from the individual provider level to provider-type level, each service provider was considered possessing its own ego-network. Data on how various service providers might be incomplete. Second, aggregating into the provider-type level allowed comparison across communities. Generic provider-type levels were predetermined and found common across the four communities. As a result, community comparison was made possible. Furthermore, while details of how individual service providers developed resource exchanges might be lost, the interaction among various service-provider types appeared to be more relevant to policies of service implementation across communities. Instead of focusing on one service provider with peculiar organizational characteristics, the aggregation rendered findings more useful to community-level planning and innovations. A second limitation concerns the ability of this dissertation to provide statistically meaningful findings for community-based network analyses. Although the relational and positional approaches could enrich the understanding of network structure, both approaches were short of standardized significance tests. For relational analysis, the reliability and validity of cliques lied in its interpretability and substantive value based on the theoretical framework. In this study, there was no standard on what size of a clique was acceptable and how many cliques should be present in a system in order to consider the system connected and integrated. 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Answers to these normative questions were beyond the scope of this study. Nevertheless, this study provides an initial glance on these issues to the community-wide network development of service delivery. For positional analysis, the lack of consensus on its goodness-of-fit measures created difficulties for network researchers to advance in theory development (Faust & Wasserman, 1994). Given these constraints, findings in this dissertation might require more careful consideration before generalizing toward care systems of other communities. 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Adams, P., & Nelson, K. (1997). Reclaiming community: An integrative approach to human services. Administration in Social Work. 21(3/4), 67-81. Agranoff, R. (1991). Human services integration: Past and present challenges in public administration. Public Administration Review. 51(6). 533-542. Aldrich, H. (1976). Resource dependence and interorganizational relations: Local employment service offices and social services sector organizations. Administration and Society. 7(41. 419-453. Aldrich, H. (1982). The origins and persistence of social networks. In P. V. Marsden & N. Lin (Eds.), Social Structure and Network Analysis (pp. 281-293). Beverly Hills, California: Sage. Alexander, E. R. (1995). How Organizations Act Together: Inter- organizational Coordination in Theory and Practice. Amsterdam: Gordon and Breach. Alter, C., & Hage, J. (1993). Organizations Working Together. New Delhi: Sage. Alter, C. F. (1988). The changing structure of elderly service delivery systems. Gerontologist. 28(1), 91-98. Austin, C. D. (1983). Case management in long-term care: Options and opportunities. Health and Social Work. 8(1). 16-30. Austin, C. D. (1993). Case management: A systems perspective. Journal of Contemporary Human Services. 74(8). 451-459. Austin, M. J. (1997). Service integration: Introduction. Administration in Social Work. 21(3/4). 1-7. Baker, F. B., & Hubert, L. J. (1981). The analysis of social interaction data. Sociological Methods & Research. 9(31. 339-361. 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Baker, W. E. (1992). The network organization in theory and practice. In N. Nohria & R. G. Eccles (Eds.), Networks and Organizations: Structure. Form, and Action (pp. 397-429). Boston, Massachusetts: Harvard Business School Press. Bazzoli, G. J., Stein, R., Alexander, J. A., Conrad, D. A., Sofaer, S., & Shorten, S. M. (1997). Public-private collaboration in health and human service delivery: Evidence from community partnerships. Milbank Quarterly. 75(4), 533-561. Berkowitz, S. D. (1982). An Introduction to Structural Analysis: The Network Approach to Social Research. Toronto: Butterworths. Bolland, J. M., & Wilson, J. V. (1994). Three faces of integrative coordination: A model of interorganizational relations in community-based health and human services. Health Services Research. 29(3). 341 -366. Boult, C., Boult, L., & Pacala, J. T. (1998). Systems of care for older populations of the future. Journal of American Geriatrics Society. 46. 499- 505. Boult, C., & Pacala, J. T. (1999). Integrating Care. In E. Calkins, C. Boult, E. H. Wagner, & J. T. Pacala (Eds.), New wavs to care for older people: Building systems based on evidence (pp. 196-247). New York: Springer Publishing Company. Brachman, J. Y. (1999). State issues in the integration of acute and chronic care. Generations. 23(2). 9-13. Branch, L. G. (1999). Introduction: Integrating acute and chronic care. Generations. 23(2). 5-8. Branch, L. G., Coulam, R. F., & Zimmerman, Y. A. (1995). The PACE evaluation: Initial findings. Gerontologist. 35(3). 349-359. Breiger, R. L., Boorman, S. A., & Arabie, P. (1975). An algorithm for clustering relational data with applications to social network analysis and comparison with multidimensional scaling. Journal of Mathematical Psychology. 12. 328-383. Bruner, C. (1991). Thinking Collaborativelv: Ten Questions and Answers to Help Policy Makers Improve Children's Services. Washington, DC: Education and Human Services Consortium. 216 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Burt, R. (1980a). Innovation as a structural interest: Rethinking the impact of network position on innovation adoption. Social Networks. 2. 327- 355. Burt, R. S. (1980b). Models of network structure. Annual Review of Sociology. 6. 79-141. Callahan, J. J. J. (1981). A systems approach to long-term care. In J. J. J. Callahan & S. S. Wallack (Eds.), Reforming the Lonq-Term-Care System: Financial and Organizational Options (pp. 219-235). Lexington, Massachusetts: LexingtonBooks. Callahan, J. J. J., & Wallack, S. S. (Eds.). (1981). Reforming the Long-Term Care System: Financial and Organizational Options. Lexington, Massachusetts: LexingtonBooks. Carcagno, G. J., & Kemper, P. (1988). The evaluation of the National Long Term Care Demonstration. Health Services Research. 23(11-22). Chisholm, D. (1989). Coordination Without Hierarchy: Informal Structures in Multioroanizational Systems. Los Angeles, California: University of California Press. Coberly, S., & Wilber, K. (1990). Developing a long-term care system of services: Where do we go from here? In P. S. Liebig & W. W. Lammers (Eds.), California Policy Choices for Long-Term Care (pp. 99-118). Los Angeles, California: University of Southern California, Ethel Percy Andrus Gerontology Center. Coleman, B. (1996). New Directions for State Long-Term Care Systems. Volume I: Overview (9602). Washington DC: Public Policy Institute, American Association of Retired Persons. Coleman, B., Kassner, E., & Pack, J. (1996). New Directions for State Long-Term Care Systems. Volume 2: Addressing Institutional Bias and Fragmentation (#9606). Washington, DC: Public Policy Institute, American Association of Retired Persons. Cook, K. S. (1977). Exchange and power in networks of interorganizational relations. Sociological Quarterly. 18(winter), 62-82. Cook, K. S. (1982). Network structures from an exchange perspective. In P. V. Marsden & N. Lin (Eds.), Social Structure and Network Analysis (pp. 177-199). New Delhi: Sage. 217 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Counte, M. A. (1998). The Emerging role of the client in the delivery of primary care to older Americans. Health Services Research. 33(2). 402- 423. DiMaggio, P., & Powell, W. W. (1988). The Iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. In C. Milofsky (Ed.), Community Organizations: Studies in Resource Mobilization and Exchange (Vol. 77-98,). New York: Oxford University Press. Ebers, M. (1997). Explaining inter-organizational network formation. In M. Ebers (Ed.), The formation of inter-organizational networks (pp. 3-40). New York: Oxford University Press. Eleazer, G. P. (1996). Integrating acute and long-term care: the PACE project. In J. C. Romeis, R. M. Coe, & J. E. Morley (Eds.), Applying Health Services Research to Long-Term Care (pp. 208-217). New York: Springer Publishing Company. Eng, C., Pedulla, J., Eleazer, P., McCann, R., & Fox, N. (1997). Program of All-inclusive Care for the Elderly (PACE): An Innovative model of integrated geriatric care and financing. Journal of American Geriatrics Society. 45. 223-232. Estes, C. L., & Close, L. (1998). Organization of health and social services for the frail elderly. In S. M. Allen & V. Mor (Eds.), Living in the Community with Disability: Service Needs. Use and Systems (pp. 73-91). New York: Springer. Evashwick, C. (Ed.). (1996). The Continuum of Long-Term Care: An Integrated Systems Approach. Albany, New York: Delman. Faust, K., & Wasserman, S. (1992). Blockmodels: Interpretation and evaluation. Social Networks. 14. 5-61. Fortinsky, R. H. (1990). The National Elder Care Systems Project: A National STudv Comparing Successful Community-Based SYstems of Care for Older People . Portland, Maine: Human Services Development Institute. Fried, B. J., Johnsen, M. C., Starrett, B. E., Calloway, M. O., & Morrissey, J. P. (1998). An empirical assessment of rural community support networks for individuals with severe mental disorders. Community Mental Health Journal. 34(1). 39-56. 218 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gage, B. (1998). The history and growth of Medicare managed care. Generations(Summer). 11-18. Galaskiewicz, J., & Krohn, K. R. (1984). Positions, roles, and dependencies in a community interorganization system. Sociological Quarterly. 25. 527-550. General Accounting Office. (1992). Integrating Human Servies: Linking At-Risk Families with Services More Successful Than System Reform Efforts (GAQ-HRD-92-108). Washington, DC: U.S. General Accounting Office. General Accounting Office. (1995a). Long-Term Care: Current Issues and Future Directions . Washington, DC. General Accounting Office. (1995b). Long-Term Care: Diverse. Growing Population INIcudes Millions of Americans of All Ages. (GAO- HEHS-95-26). Washington, DC: U.S. General Accounting Office. Gerlach, M. L. (1992). The Japanese corporate network: A blockmodel analysis. Administrative Science Quarterly. 37(11. 105-139. Grandori, A., & Soda, G. (1995). Inter-firm networks: Antecedents, mechanisms, and forms. Organization Studies. 16(2). 183-214. Gulati, R. (1995). Social structure and alliance formation patterns: A longitudinal analysis. Administrative Science Quarterly. 40. 619-652. Gulati, R. (1998). Alliances and networks. Strategic Management Journal. 19. 293-317. Hall, R. H., Clark, J. P., Giordano, P. C., Johnson, P. V., & Van Roekel, M. (1977). Patterns of interorganizational relationships. Administrative Science Quarterly. 22. 457-474. Hansen, J. (1999a). The impact of Medicare changes on paversand integrated delivery systems. Paper presented at the Health Care Execuitive of Southern California Annual Seminar: Moving Medicare into the New Millennium, Long Beach, California. Hansen, J. C. (1999b). Practical lessons for delivering integrated services in a changing environment: The PACE model. Generations. 23(2). 22-29. 219 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Harrington, C., Lynch, M., & Newcomer, R. J. (1993). Medical services in Social Health MaintenanceOrganizations. Gerontologist. 33(6). 790-800. Hassett, S., & Austin, M. J. (1997). Service integration: Something old and something new. Administration in Social Work. 21(3/4). 9-29. Heflinger, C. A. (1996). Measuring service system coordination in managed mental health care for children and youth. Evaluation and Program Planning. 19(2). 155-163. Hennessy, C. H., & Hennessy, M. (1990). Community-based long term care for the elderly: Evaluation practice reconsidered. Medical Care Review. 47(2L 221-259. Hoffman, C., Rice, D., & Sung, H.-Y. (1996). Persons with chronic conditions: Their prevalence and costs. Journal of American Medical Association. 276(18). 1473-1479. Johnson, J. C., & Miller, M. L. (1986). Behavioral and cognitive data: A note on the multiplexity of network subgroups. Social Networks. 8. 65-77. Johnsen, M. C., Morrissey, J. P., & Calloway, M. O. (1996). Structure and change in child mental health service delivery networks. Journal of Community Psychology. 24(3). 275-289. Justice, D. (1988). State Long-Term Care Reform: Development of Community Care Systems in Six States . Washington, DC: Health Policy Studies, Center for Policy Research, National Governor's Association. Justice, D. (1990). State community-based care systems. Generations. 14MV 45-48. Kagan, S. L., & Neville, P. (1993). Integrating servics for children and families: Understanding the past to shape the future. Pennsylvania: Yale University Press. Kaluzny, A. D., & Fried, B. J. (1985). Interorganizational coordination of services to the elderly. In H. I. Phillips & S. A. Gaylord (Eds.), Aging and Public Health (pp. 267-292). New York: Springer Publishing Company. Kaluzny, A. D., Zuckerman, H. S., & Rabiner, D. J. (1998). Interorganizational factors affecting the delivery of primary care to older Americans. Health Services Research. 33(2). 381-401. 220 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kane, R. A. (1988). The noblest experiment of them all: Learning from the National Channeling evaluation. Health Services Research. 23(1). 190-198. Kane, R. A. (1993). Dangers lurking in the 'continuum of care’. Journal of Aging and Social Policy. 5(4). 1-7. Kane, R. L. (1998). Managed care as a vehicle for delivering more effective chronic care for older persons. Journal of American Geriatrics Society. 46. 1034-1039. Kane, R. L. (1999). A new model of chronic care. Generations. 23(2). 35-37. Kane, R. L., Kane, R. A., & Finch, M. D. (1995). Once and future SHMOs. Gerontologist. 35(3). 294-295. Kane, R. L., Kane, R. A., Ladd, R. C., & Veazie, W. N. (1998). Variation in state spending for long-term care: Factors associated with more balanced systems. Journal of Health Politics. Policy, and Law. 23(2). 363- 390. Kemper, P. (1988). The evaluation of the National Long Term Care Demonstration. Health Services Research. 23(1). 161-174. Krackhardt, D. (1987). QAP partialling as a test of spuriousness. Social Networks. 9. 171-186. Krackhardt, D. (1988). Predicting with networks: Nonparametric multiple regression analysis of dyadic data. Social Networks. 10. 359-381. Krackhardt, D., Blythe, J., & McGrath, C. (1994). KrackPlot 3.0: An improved network drawing program. Connections. 17(2). 53-55. Lago, D., & Zarit, S. H. (1992). Empirical perspectives on services integration and aging: A review. Journal of Aging & Social Policy. 4(3/4L 91- 109. Laumann, E. O., Galaskiewicz, J., & Marsden, P. V. (1978). Community structure as interorganizational linkages. Annual Review of Sociology. 4. 455-484. 221 permission of the copyright owner. Further reproduction prohibited without permission. Leutz, W. (1997). The Social HMO: Flexible models to enhance, finance, and integrate services. In K. H. Wilber, E. L. Schneider, & D. Polisar (Eds.), A Secure Old Aae: Approaches to Long-Term Care Financing (pp. 57-84). New York: Springer. Leutz, W., Greenlick, M. R., Ripley, J., Ervin, S., & Feldman, E. (1995). Medical services in social HMOs: A reply to Harrington et al. Gerontologist. 35(1). 6-8. Leutz, W. N., Greenlick, M. R., & Capitman, J. A. (1994). Integrating acute and long-term care. Health Affairs. 13(4L 58-74. Litwak, E., & Hylton, L. F. (1962). Interorganizational analysis: A hypothesis on coordinating agencies. Administrative Science Quarterly. 6. 395-420. Luke, R. D., & Wholey, D. R. (1999). Commentary: On "a taxonomy of healthcare networks and systems: Bringing order out of chaos". Health Services Research. 33(6). 1719-1725. Manton, K. G., Corder, L. S., & Stallard, E. (1993). Estimates of change in chronic disability and institutional incidence of prevalence rates in the United States elderly population from the 1982, 1984, and 1989 National Long Term Care Survey. Journal of Gerontology. 48(4). S153-166. Marsden, P. V. (1990). Network data and measurement. Annual Review of Sociology. 16. 435-463. McAuley, W. J., & Safewright, M. P. (1992). Formal and informal interaction between area agencies on aging and departments of social services: implications for practice. Journal of Gerontological Social Work. 18(1/2), 97-115. McKinney, M. M., Morrissey, J. P., & Kaluzny, A. D. (1993). Interorganizational exchanges as performance markers in a community cancer network. Health Services Research. 28(4). 459-478. Mizruchi, M. S. (1993). Cohesion, equivalence, and similarity of behavior: a theoretical and empirical assessment. Social Networks. 15. 275-307. Mizruchi, M. S., & Galaskiewicz, J. (1993). Networks of interorganizational relations. Sociological Methods & Research. 22(1), 46- 70. 222 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Morrissey, J., Calloway, M., Johnsen, M., & Ullman, M. (1997). Service system performance and integration: A baseline profile of the ACCESS demonstration sites. Psychiatric Services. 48(3). 374-380. Morrissey, J. P., & Calloway, M. (1994). Local mental health authorities and service system change: Evidence from the Robert Wood Johnson Program on chronic mental illness. Milbank Quarterly. 72(1). 49- 80. Morrissey, J. P., Tausig, M., & Lindsey, M. L. (1985). Network Analysis Methods For Mental Health Service System Research: A Comparison of Two Community Support Systems (DHHS Publication # (ADM)85-1383). Rochville, MD: U.S. Department of Health and Human Services. Mulford, C. L., & Rogers, D. L. (1982). Definitions and Models. In D. L. Rogers & D. A. Whetten (Eds.), Interorganizational Coordination: Theory. Research, and Implementation (pp. 9-31). Ames, Iowa: Iowa University Press. Myrtle, R. C., & Wilber, K. H. (1994). Designing service delivery systems: Lessons from the development of community-based systems of care for the elderly. Public Administration Review. 54(3). 245-252. National Association on Aging. (1993). Long-Term Care: Options for the Future. Paper presented at the 40th Annual Gerontological Society of America Meeting, New Orleans. National Long-Term Care Resource Center. (1996). Managed Care Handbook for the Aging Network . Minneapolis, Minnesota: University of Minnesota. Nelson, R. E. (1986). The use of blockmodelling in the study of organization structure: A methodological proposal. Organization Studies. 7(1), 75-85. Nelson, R. E. (1988). Social network analysis as intervention tool. Group & Organization Studies. 13(1), 39-58. Newcomer, R. J., Harrington, C., Manton, K. G., & Lynch, M. (1995). A response to respresentatives from the Social HMOs regarding program evaluation. Gerontologist. 35(3). 292-293. 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Oliver, C. (1990). Determinants of interorganizational relationships: Integration and future directions. Academy of Management Review. 15(2). 241-265. Ory, M. G., Cooper, J., & Siu, A. L. (1998). Toward the development of a research agenda on organizational issues in the delivery of healthcare to older Americans. Health Services Research. 33(2). 288-297. O'Toole, L. J. J. (1997). Treating networks seriously: Practical and research-based agendas in public administration. Public Administration Review. 57(1 V 45-52. O'Toole, L. J. J., &Montjoy, R. S. (1984). Interorganizational policy implementation: A theoretical perspective. Public Administration Review(November/December), 491-503. Panning, W. H. (1982). Fitting blockmodels to data. Social Networks. 4, 81-101. Perrow, C. (1972). Complex Organizations: A Critical Essay (Second ed.). Glenview, Illinois: Scott, Foresman and Company. Polivka, L. (1998). The science and ethics of long-term care. Generations. 22(3). 21-25. Provan, K. G., & Milward, H. B. (1994). Integration of community- based services for the severely mentally ill and the structure of public funding: A comparison of four systems. Journal of Health Politics. Policy, and Law. 19(4). 865-894. Provan, K. G., & Milward, H. B. (1995). A preliminary theory of interorganizational network effectiveness: A comparative study of four community mental health systems. Administrative Science Quarterly. 40. 1- 33. Provan, K. G., & Sebastian, J. G. (1998). Networks within networks: Service link overlap, organizational cliques, and network effectiveness. Academy of Management Journal. 41 (4). 453-463. Randolph, F., Blasinsky, M., Leginski, W., Buckley, L., & Goldman, H. H. (1997). Creating integrated service systems for homeless persons with mental illness: The ACCESS program. Psychiatric Services. 48f3). 369-373. 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ranson, S., Hinings, B., & Greenwood, R. (1980). The structuring of organizational structures. Administrative Science Quarterly. 25. 1-17. Redburn, F. S. (1977). On "Human Services Integration". Public Administration Review(Mav/June), 264-269. Reitan, T. C. (1998). Theories of interorganizational relations in the human services. Social Service Review. 72(3). 285-309. Riley, T. (1998). States' perspective on managed care for the elderly. Generations. 22(2L 69-73. Ring, P. S., & Ven, A. H. V. d. (1994). Developmental processes of cooperative interorganizational relationships. Academy of Management Review. 19(1). 90-118. Rochefort, D. A., Rosenberg, M., & White, D. (1998). Community as a policy instrument: A comparative analysis. Policy Studies Journal. 26(3). 548-568. Rosenheck, R., Morrissey, J., Lam, J., Calloway, M., Johnsen, M., - Goldman, H., Randolph, F., Blasinsky, M., Fontana, A., Calsyn, R., & Teague, G. (1998). Service system integration, access to services, and housing outcomes in a program for homeless persons with severe mental illness. American Journal of Public Health. 88(11). 1610-1615. Schmidt, S. M., & Kochan, T. A. (1977). Interorganizational relationships: Patterns and motivations. Administrative Science Quarterly. 22, 220-234. Schopler, J. H. (1987). Interorganizational groups: Origins, structure, and outcomes. Academy of Management Review. 12(4), 702-713. Scott, W. R. (1993). The organization of medical care services: Toward an integrated theoretical model. Medical Care Review. 50(3). 271- 303. Shah, P. P. (1998). Who are employees' social referents? Using a network perspective to determine referent others. Academy of Management Journal. 41 (3L 249-268. Shortell, S. M., Gillies, R. R., Anderson, D. A., Erickson, K. M., & Mitchell, J. B. (1996). Remaking Health Care in America: Building Organized Delivery Systems. San Francisco: Jossey-Bass. 225 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sofaer, S. (1994). Applying organizational theory to improve the continuity of health services for older persons. Journal of Applied Gerontology. 13(1), 73-85. Sofaer, S. (1998). Aging and primary care: An overview of organizational and behavioral issues in the delivery of healthcare services to older Americans. Health Services Research. 33(2). 298-321. Sofaer, S., & Myrtle, R. C. (1991). Interorganizational theory and research: Implications for health care management, policy, and research. Medical Care Review. 48(4). 371-409. Stone, R. I., & Katz, R. E. (1996). Reflections on the future of integrating acute and long-term care. Annuals of Gerontology and Geriatrics. 16. Tausig, M. (1987). Detecting "cracks" in mental health service systems: Application of network analytic techniques. American Journal of Community Psychology. 15(3). 337-351. Van de Ven, A., & Ferry, D. L. (1980). Measuring and Assessing Organizations. New York: John Wiley & Sons. Van de Ven, A. H., & Walker, G. (1979). Coordination patterns within an interorganizational network. Human Relations. 32(1). 19-36. Van de Ven, A. H., & Walker, G. (1984). The dynamics of interorganizational coordination. Administrative Science Quarterly. 29. 598- 621. Wacker, R. R., Roberto, K. A., & Piper, L. E. (1998). Community Resources for Older Adults: Programs and Services in an Era of Change. Thousand Oaks, California: Pine Forge Press. Waldfogel, J. (1997). The new wave of service integration. Social Service Review. 463-484. Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and Applications. New York: Cambridge University Press. Weiss, J. A. (1987). Pathways to cooperation among public agencies. Journal of Policy Analysis and Management. 7(1). 94-117. 226 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Weissert, W. G. (1988). The National Channeling Demonstration: What we knew, know now, and still need to know. Health Services Research. 23(1). 175-187. Weissert, W. G., Cready, C. M., & Pawelak, J. E. (1988). The past and future of home- and community-based long-term care. The Milbank Quarterly. 66(2). 309-389. Wellman, B. (1988). Structural Analysis: From Method and Metaphor to Theory and Substance. In B. Wellman & S. D. Berkowitz (Eds.), Social Structures: A Network Approach ( p p . 19-61). New York: Cambridge University Press. White, H. C., Boorman, S. A., & Breiger, R. L. (1976). Social structure from multiple networks. I. Blockmodels of roles and positions. American Journal of Sociology. 81 (4). 730-780. Wiener, J. M., & Skaggs, J. (1995). Current Approaches to Integrating Acute and Long-Term Care Financing and Services (#9516). Washington DC: Public Policy Institute, American Association of Retired Persons. Wiener, J. M., & Stevenson, D. G. (1998). State policy on long-term care for the elderly. Health Affairs. 17(3). 81-100. Wiewel, W., & Hunter, A. (1985). The interorganizational network as a resource: A comparative case study on organizational genesis. Administrative Science Quarterly. 30(4). 482-496. Wilber, K. H., Coberly, S., & Myrtle, R. (1991). Developing Community Based Systems of Care. Los Angeles, California: University of Southern California, Andrus Gerontology Center. Wilber, K. H., & Myrtle, R. C. (1998). Developing community-based systems of care: Lessons from the field. In J. McCroskey & S. D. Einbinder (Eds.), Universities and Communities: Remaking Professional and Interprofessional Education for the Next Century (pp. 119-138). Westport, Connecticut: Praeger. Wolinsky, F. D., & Johnson, R. J. (1992). Perceived health status and mortality among older men and women. Journal of Gerontology. 47f6). S304-S312. 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Woodard, K. L. (1994). Packaging effective community service delivery: The utility of mandates and contracts in obtaining administrative cooperation. Administration in Social Work. 18(2). 17-43. Woodard, K. L. (1995). Introduction and acceptance of inter organizational agreements: The experience of seventy-five administrators in one county. Administration in Social Work. 19(4). 51-83. Wright, E. R., & Shuff, I. M. (1995). Specifying the integration of mental health and primary health care services for persons with HIV/AIDS: The Indiana integration of care project. Social Networks. 17. 319-340. Zawadski, R. T. (Ed.). (1984). Community-based Systems of Long- Term Care. New York: Haworth Press. Zoroboros, C., & LeMasurier, J. D. (1997). Medicare and managed care. In P. R. Kongstvedt (Ed.), Essentials of Managed Health Care (Second ed., pp. 405-431). Gaithersburg, Maryland: Aspen. 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix A 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Excerpt of the SEED Questionnaire III. Most programs interact with other programs and organizations in order to accomplish their goals and responsibilities. Below is a diagram that represents different types of programs and organizations that may interact with your program. IHSS ADHC Senior Housing Linkages Public ^ Guardian Police Your Program APS AAA f&R MSSP Home Health Mental Health Center Hospital Now, we want you to think of the programs by name, both inside and outside your organization, that you interact with in order to accomplish the goals and responsibilities of your program. In the space below, please write the names of these programs. From this list, select at least 5 programs that are the m ost im portant to accomplishing the goals and responsibilities of your program and write their names at the top of the 5 columns on the next page. If you selected more than 5, additional pages are attached at the end of the questionnaire. EXAMPLE: Yosemite Kondike APS_______ Sunshine Restful Meals______ Hospital________________ADHC______ Board & Care 230 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In the column below each program you have listed, circle the response to each question as it pertains to that program. 1. Why does your relationship with this program exist? (Name) (Name) (Name) (Name) (Name) a) physically close to organization Yes No Yes No Yes No Yes No Yes No b) a hassle to switch Yes No Yes No Yes No Yes No Yes No c) only provider available Yes No Yes No Yes No Yes No Yes No d) high quality provider Yes No Yes No Yes No Yes No Yes No e) low cost provider Yes No Yes No Yes No Yes No Yes No f) provides needed service Yes No Yes No Yes No Yes No Yes No g) part of the network Yes No Yes No Yes No Yes No Yes No h) ability to serve clients with multiple needs Yes No Yes No Yes No Yes No Yes No 0 eligibility limitations of our program Yes No Yes No Yes No Yes No Yes No j) eligibility limitations of their program Yes No Yes No Yes No Yes No Yes No k) other Yes No Yes No Yes No Yes No Yes No I) other Yes No Yes No Yes No Yes No Yes No m) other Yes No Yes No Yes No Yes No Yes No 2. What resources do you receive from this program? a) clients Yes No Yes No Yes No Yes No Yes No b) money Yes No Yes No Yes No Yes No Yes No c) staff Yes No Yes No Yes No Yes No Yes No d) space Yes No Yes No Yes No Yes No Yes No e) equipment Yes No Yes No Yes No Yes No Yes No f) information Yes No Yes No Yes No Yes No Yes No g) technical assistance Yes No Yes No Yes No Yes No Yes No 231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3. What resource do you give to this program? (Name) (Name) (Name) (Name) (Name) a) clients Yes No Yes No Yes No Yes No Yes No b) money Yes No Yes No Yes No Yes No Yes No c) staff Yes No Yes No Yes No Yes No Yes No d) space Yes No Yes No Yes No Yes No Yes No e) equipment Yes No Yes No Yes No Yes No Yes No f) information Yes No Yes No Yes No Yes No Yes No 9) technical assistance Yes No Yes No Yes No Yes No Yes No 4. H o w is y o u r relationship with this program m aintained? a) required by government or funding regulations Yes No Yes No Yes No Yes No Yes No b) contractual agreement Yes No Yes No Yes No Yes No Yes No c) memorandum of understanding Yes No Yes No Yes No Yes No Yes No d) formal verbal agreement Yes No Yes No Yes No Yes No Yes No e) informal, interpersonal communications Yes No Yes No Yes No Yes No Yes No f) advisory board membership Yes No Yes No Yes No Yes No Yes No 9) joint program Yes No Yes No Yes No Yes No Yes No h) part of same Yes No Yes No Yes No Yes No Yes No organization 232 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Name) (Name) (Name) (Name) (Name) Not at Great Not at Great Not at Great Not at Great Not at Great All Deal All Deal All Deal All Deal All Deal 5. How much do you 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 need this program to accomplish your program’s goals? 6. How much does 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 this program need you to accomplish its goals? 7. To what extent do 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 you share the same goals? 8. To what extent do 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 you share the same clients? 9. To what extent do 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 you share the same funding source? 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 10. To what extent does your program influence this program? 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 11. To what extent do they influence your program? 233 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix B 234 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.1. Centrality Correlation Results for San Mateo fndegrees Outdegrees Indegrees (Client) (Client) (Information) Indegrees 1.000 .850 .881 (Client) Outdegrees .850 1.000 .969 (Client) Indegrees .881 .969 1.000 (Information) Outdegrees .874 .921 .937 (Information) Indegrees .657 .587 .601 (Money) Outdegrees .591 .377 .463 (Money) Indegrees (Staff) .620 .753 .773 Outdegrees .711 .651 .632 (Staff) ro 03 0 1 Outdegrees (Information) .874 .921 .937 1.000 .456 .479 .757 .653 Indegrees (Money) .657 .587 .377 .601 .463 .456 .479 1.000 .412 .412 1.000 .634 .365 .424 .453 (Staff) (Staff) .620 .711 .753 .651 .773 .632 .757 .653 .634 .424 .365 .453 1.000 .498 .498 1.000 Outdegrees Indegrees Outdegrees (Money) .591 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.2. Centrality Correlation Results for Long Beach Indegrees (Client) Outdegrees (Client) Indegrees (Information) Outdegrees (Information) Indegrees (Money) Outdegrees (Money) Indegrees (Staff) Outdegrees (Staff) Indegrees (Client) 1.000 .918 .949 .964 .505 .426 .623 .482 Outdegrees (Client) .918 1.000 .966 .921 .390 .593 .698 .406 Indegrees (Information) .949 .966 1.000 .961 .512 .487 .591 .551 Outdegrees (Information) .964 .921 .961 1,000 .545 .462 .577 .611 Indegrees (Money) .505 .390 .512 .545 1.000 -.079 .037 .670 Outdegrees (Money) .426 .593 .487 .462 -.079 1.000 .762 .121 Indegrees (Staff) .623 .698 .591 .577 .037 ,762 1.000 .132 Outdegrees (Staff) .482 .406 .551 .611 .670 .121 .132 1.000 w o > Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.3. Centrality Correlation Results for Tulare Indegrees (Client) Outdegrees (Client) Indegrees (Information) Outdegrees (Information) Indegrees (Money) Outdegrees (Money) Indegrees (Staff) Outdegrees (Staff) Indegrees (Client) 1.000 .818 .965 .935 .175 .119 .262 .457 Outdegrees (Client) .818 1.000 .894 .914 .313 .246 .550 .480 Indegrees (information) .965 .894 1.000 .975 .249 .186 .354 .516 Outdegrees (Information) .935 .914 .975 1.000 , .338 .262 .432 .540 Indegrees (Money) .175 .313 .249 .338 1.000 .044 .192 -.178 Outdegrees (Money) .119 .246 .186 .262 .044 1.000 .390 .329 Indegrees (Staff) .262 .550 .354 .432 .192 .390 1.000 .501 Outdegrees (Staff) .457 .480 .516 .540 -.178 .329 .501 1.000 ro c o ■vj Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.4. Centrality Correlation Results for San Francisco Indegrees (Client) Outdegrees (Client) Indegrees (Information) Outdegrees (Information) Indegrees (Money) Outdegrees (Money) Indegrees (Staff) Outdegrees (Staff) Indegrees (Client) 1.000 .898 .937 .913 .404 .339 .545 .463 Outdegrees (Client) .898 1.000 .947 .974 .288 .478 .683 .407 Indegrees (Information) .937 .947 1.000 .982 .417 .428 .673 .454 Outdegrees (Information) .913 .974 .982 1.000 .325 .447 .664 .384 Indegrees (Money) .404 .288 .417 .325 1.000 .340 .526 .574 Outdegrees (Money) .339 .478 .428 .447 .340 1.000 .478 .420 Indegrees (Staff) .545 .683 .673 .664 .526 .478 1.000 .585 Outdegrees (Staff) .463 .407 .454 .384 .574 .420 .585 1.000 ro c o 00 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.5. Bivariate Correlation. San Mateo: Independent Matrices with Each Other Matrices Same Net work Same Organ ization Proximate Low Cost Only Available Mandate Same Goal Same Client Same Fund My In fluence Their My Their Influence Need Need Same Network 1.000 Same Organ 0.476* 1.000 ization Proximate 0.139 0.176 1.000 Low Cost 0.707* 0.462* 0.166 1.000 Only 0.571* 0.320* 0.031 0.616* 1,000 Available Mandate 0.400* 0,092 -0.037 0.281* 0.355* 1,000 Same 0.873* 0.509* 0.135 0.801* 0,691* 0.389* 1.000 Goal Same 0.890* 0.509* 0.110 0.801* 0.712* 0.423* 0.984* 1,000 Client Same 0.880* 0.513* 0.137 0.768* 0.676* 0.427* 0,959* 0.975* 1,000 Fund My 0.890* 0.509* 0.110 0.801* 0.712* 0.423* 0.967* 0,984* 0.959* 1.000 Influence Their 0.890* 0,509* 0.110 0,801* 0.712* 0.423* 0,984* 1,000* 0.975* 0.984* 1.000 Influence My Need 0.890* 0,509* 0.110 0,801* 0.712* 0.423* 0,984* 1.000* 0.975* 0.984* 1,000* 1,000 Their 0.880* 0.484* 0.112 0.788* 0,697* 0.392* 0.975* 0,992* 0,967* 0,975* 0.992* 0.992* 1.000 Need * p<0.05 to C O C D Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.6. Same Same Proxi Low Only Mandate Same Same Same My In Their In- My Their Matrices Network Organ mate Cost Available Goal Client Fund fluence fluence Need Need ization Same 1.00* Network Same 0.277* 1.00* Organ ization Proximate 0.74* 0.438* 1.00* Low Cost 0.856* 0.277* 0.772* 1.00* Only 0.275* 0.133 0.392* 0.519* 1.00* Available Mandate 0.301* 0,335* 0.207* 0.423* 0.422* 1.00* Same 0.868* 0.414* 0.827* 0.922* 0.517* 0,39* 1.00* Goal Same 0.852* 0.419* 0.808* 0.907* 0.483* 0.395* 0.987* 1.00* Client Same 0.795* 0.404* 0.732* 0.824* 0.535* 0.435* 0.895* 0.907* 1.00* Fund My 0.847* 0.375* 0.799* 0.903* 0.496* 0.405* 0,961* 0.974* 0.903* 1.00* Influence Their 0.753* 0.339* 0.72* 0.811* 0.527* 0.429* 0.909* 0.92* 0,866* 0.917* 1.00* Influence My Need 0.868* 0.414* 0.827* 0.922* 0.517* 0,39* 1.000* 0.987* 0,895* 0.961* 0.909* 1.00* Their 0.836* 0.370* 0.788* 0.891* 0,489* 0.4* 0.974* 0,987* 0.891* 0,987* 0.933* 0.974* 1.00* Need * p < 0.05 N > O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.7. Bivariate Correlation. Tulare: Independent Variables Matrices Same Network Same Organ ization Proximate Low Cost Only Avail able Mandate Same Goal Same Client Same Fund My Influence Their My Influence Need Their Need Same 1.00* Network Same 0.456* 1.00* Organ ization Proximate 0.778* 0.494* 1.00* Low Cost 0.777* 0.399* 0,629* 1.00* Only 0.79* 0,333* 0,538* 0.753* 1.00* Available Mandate 0.469* 0,32* 0.38* 0.475* 0.547* 1.00* Same 0,918* 0.419* 0.174* 0.774* 0,823* 0.518* 1.00* Goal Same 0.918* 0.419* 0.174* 0.774* 0.823* 0.518* 1.00* 1.00* Client Same 0,899* 0.382* 0.691* 0.753* 0,803* 0.518* 0,982* 0.982* 1.00* Fund My 0.908* 0,386* 0,697* 0,781* 0,831* 0.523* 0.991* 0,991* 0,973* 1.00* Influence Their 0.927* 0.423* 0.72* 0.781* 0.810* 0.523* 0.991* 0,991* 0.973* 0,982* 1,00* Influence My Need 0.918* 0.419* 0.714* 0,774* 0.823* 0.518* 1,000* 1,000* 0.982* 0.991* 0,991* 1.00* Their 0.918* 0.419* 0.714* 0.774* 0.823* 0,518* 1.000* 1.000* 0.982* 0.991* 0,991* 1.00* 1.00* Need *p < 0.05 N ) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table B.8 Bivariate Correlation. San Francisco: Independent Variables.___________________________________________ Same Same Proximate Low Only Mandate Same Same Same My In- Their My Their Matrices Net- Organ- Cost Avail- Goal Client Fund fluence Influence Need Need ___________ work ization_______________________ able_________________________________________________________________ Same 1.00* Network Same 0.431* 1.00* Organ ization Proximate 0.576* 0.419* 1.00* Low Cost 0.743* 0,369* 0.577* 1.00* Only 0.538* 0,256* 0,299* 0.635* 1.00* Available Mandate 0.469* 0.44* 0.171* 0.442* 0.436* 1.00* Same 0.849* 0.404* 0.568* 0.824* 0.676* 0.496* 1.00* Goal Same 0.825* 0.384* 0.539* 0.8* 0.649* 0.51* 0.959* 1.00* Client Same 0,831* 0.381* 0.549* 0.782* 0,66* 0.499* 0.945* 0.931* 1.00* Fund My 0,831* 0.372* 0.536* 0.784* 0.685* 0.488* 0.959* 0.945* 0,958* 1,00* Influence Their 0.837* 0.408* 0.56* 0.801* 0.67* 0.501* 0,973* 0.959* 0.944* 0.968* 1.00* Influence My Need 0.825* 0.384* 0.539* 0.769* 0.638* 0.494* 0.941* 0.964* 0.912* 0,927* 0.914* Their 0.831* 0,382* 0.536* 0,796* 0.658* 0,508* 0.964* 0.986* 0,936* 0.95* 0.964* Need * p <0.05 Appendix C 243 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 244 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure C1. S a n M ateo Client Network. IREGCTRI SNF X _____ REHAB SOCSEC C O U N TYH D a PUBGUR HOSPITAL M O N E Y M G T LINKM SSP I l e g a l a id M OW OM BUDS RSVP ADHC TRANSPRT HHC C H A R ITY e m b r g ] o t h e r Figure C2. San Mateo Information Network. a c t B S b £ O 2 o B O 0 ] 8 CO O c a 3 a, c o B I £ V . X X a I 5 o 5 t C D z > 0 3 c o o a ) ro 2 c C D CO C O o 0 3 3 0 3 246 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. | MOW | H H c| LINKMSSP HOSPITAL DPIHAPS I REHAB [----- i — , 1 ------ r^O U N T Y H D [ ADHC I s e n io r c t r I TRANSPRT | s n f | jEMERG I l e g a l a id ISQCSECl Ip u b g u r lOMBUDS Ir s v p I |OTHER m o n e y m g t I Figure C4. San Mateo Staff Network. to -L -o 248 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. F igure C5. L o n g B ea ch Client Network. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. [ m o w ] ADHC HHCI , . 7 F \ H O S P IT A L \ / \ \ I -DTIOrf-TP u i -1 D P X H ^ | - ^ Kt 1 S S P l ISENIORCTR 1 * 1 " --------------- 1 \iT T U " * 1 Figure C6. Long Beach Information Network. [t r a n s p r t ] [ o t h e r ] N i C O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. [m h ] I H O S P IT A L I T |A D H C | | A L Z ) | h h c Figure C7. Long Beach Money Network. | h s o 1 | s n f ] |SOCSBc] | H E A C L In I | o t h e r ] to O l o 251 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure C8. L o n g B ea ch Staff Network. 252 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. HSG CHARITY 1 MOW AAA COUNTYHD SENTORCTRI ------- I d p ih a p s / OMBUDS HHC HOSPIT iq PUBGUR | e m e r o Figure C10. Tulare Information Network. to O l w G O 0 - a CO O e S 254 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure C11 . Tulare Money Network. in Q 3 u < u u w w 8 w c a D D C Q § 255 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure C12. Tulare Staff Network. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. I m o n e y m g t F] 1 PUBG UR[ ]C A S E M G T P t SOCSECj | R E H A B | | L IN K M S S P D P IH A P S A D H C OMBIJDS | ►-Vat H E A C L IN C H A R IT Y M O W O T H E R | H O S P IT A L . 1 C O N S O R T H H C A A A | I t r a n s p r t I \ \ if/ I - • ■ • A m \ I------------- / v L \ \ V | L E G A L A 1 D | I----------^----- » | COUNTYHDI |SNF| {H O M E L E S S 1 1 1 Figure C13. San Francisco Client Network. N ) C l O ) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. C ASEMGT |CONSORT[ EMERG AAA | d p ih a p s | PUBGUR OTHER SENlORCTR HEALCIN COUNTYHD TRANSPRT HOMELESS ADHC CHARITY MOW MONEYM GM T HOSPITAL l in k m s s p T J / i c n A i A1 [h h c | I l e g a l a id RSVP [r e h a b OMBUDS H LEGPOL SOCSEC Figure C14. San Francisco Money Network. to O l Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. IpTrRnTTP I l A A £ | t r a n s p r t | lPUBP ^ l ----------------L - w t ^ ^ |h o m e le s s 1 ^ C O U N T Y H D | SNF H E A C LIN LINK M SSP M O N E Y M G T A D H C IA Y lH H C H nTpTTA T^ SEN1QRCTR \r~\ CONSORT 1 M O W | |k s v p | s z DPIHAPS l e g a l a id ] C H A R ITY EM ERG OM BUDS LEGPOL Figure C15. San Francisco Information Network. N J Ol 00 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. HHC CONSORT | C O U N TYH D 5 j C H A R ITY EMERG H E A C LIN AAA I& R TRANSPRT | c a s e m g t | [LEGPOL I p u b g u r I HOMELESS L E G A L A ID |\\ \ X \ _ OTHER ADHC HOSPITAL DPIHAPS OMBUDS LINKM SSP z RSVP |M O N E Y M G t ] | SOCSEC1 »| REHAB Figure C16. San Francisco Staff Network. to Ol C O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Group 4 Emergency Services Social Services Adult Day Health Care Group 3 Skilled Nursing Facilities Social Security Public Guardian Group 7 Senior Meals Program Mental Health Money Management Group 1 Linkages/MSSP Home Health Care Hospitals Ombudsman etired Senior Volunteers Group 2 Housing County Health Department Regional Center 1 r i Group 6 Legal Aid Transportation Group 8 Rehabilitation Group 5 Senior Centers Information & Referral Charity Other Figure C17. The Structure of San Mateo Care System: Client Exchange Pattern.(Relations with isolate "Rehabilitation" marked with broken lines) to O ) o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Emergency Services Social Services Adult Day HeaMkQtre Group 3 Skilled Nursing Facilities Social Security Public Guardian Group 7 Senior Meals Program Mental Health Money Management Group 1 Linkages/MSSP Home Health Care Hospitals Retired-Senior Volunteers Group 2 Housing County Health Department Regional Center Group 5 Senior Centers Information & Referral Charity Group 6 Legal Aid Transportation Group 8 Rehabilitation Figure C18. The Structure of San Mateo Care System: Information Exchange Pattern.(Relations with isolate "Rehabilitation" marked with broken lines) ro O ) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Emergency Services Social Services Adult Day Health Care Group 2 Housing County Health Department Regional Center Group 1 Linkages/MSSP Home Health Care Hospitals Ombudsman ired Senior Volunteers Group 3 Skilled Nursing Facilities Social Security Public Guardian Group 5 Senior Centers Information & Referral Charity Other Group 6 Legal Aid Transportation Group 8 Rehabilitation Group 7 Senior Meals Program Mental Health Money Management Figure C19. The Structure of San Mateo Care System: Money Exchange Pattern.(Relations with isolate "Rehabilitation” marked with broken lines) to C D Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Emergency Services Social Services Adult Day Health Care Group 3 Skilled Nursing Facilities Social Security Public Guardian Group 1 Linkages/MSSP Home Health Care Hospitals Ombudsman Group 2 Housing County Health Department Regional Center Group 5 Senior Centers Information & Referral Charity Other Group 6 Legal Aid Group 8 Transportation Rehabilitation Group 7 Senior Meals Program Mental Health Money Management Figure C20. The Structure of San Mateo Care Svstem:Staff Exchange Pattern.(Relations with isolate "Rehabilitation" marked with broken lines) to O ) w Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 7 Transportation Group 1 Linkages/MSSP Home Health Care Group 3 Skilled Nursing Facilities Other Group 2 Housing Retired Senior Volunteers Group 4 Senior Centers Hospitals Alzheimer’s Services Group 6 Information and Referral Social Security Mental Health Group 5 Senior Meals Program Social Services Adult Day Health Care Health Clinics Charity Figure C21. Structure of Long Beach Care Systems: Client Exchange Pattern. to O ) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 6 Information and Referral Social Security Mental Health Group 4 Senior Centers Group 7 Transportation / Group 1 Linkages/MSSP Hospitals Home Health Care Alzheimer’s Services A j i Group 5 Senior Meals Program Social Services Adult Day Health Care Health Clinics Charity Group 3 Skilled Nursing Facilities Other Group 2 Housing Retired Senior Volunteers Figure C22. Structure of Long Beach Care Systems: Information Exchange Pattern. O l Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Senior Centers Hospitals Alzheimer’s Services Group 6 Information and Referral Social Security Mental Health Group 7 Transportation Group 1 Linkages/MSSP Home Health Care Group 5 Senior Meals Program Social Services Adult Day Health Care Health Clinics Charity Group 3 Skilled Nursing Facilities Group 2 Housing Retired Senior Volunteers Figure C23. Structure of Long Beach Care Systems: Money Exchange Pattern. ro o > 0) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Senior Centers Hospitals Alzheimer’s Services Group 6 Information and Referral Social Security Mental Health Group 7 Transportation Group 1 Linkages/MSSP Home Health Care Group 5 Senior Meals Program Social Services Adult Day Health Care Health Clinics Charity Group 3 Skilled Nursins Facilities Group 2 Housing Retired Senior Volunteers Figure C24. Structure of Long Beach Care Systems: Staff Exchange Pattern. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Group 4 Emergency Servic Hospitals County Health De es )artment ' r Group 5 Senior Centers Legi si ator/Pol i ti ci an Charity Group 1 Area Agencies on Aging -ReliredSenior Volunteers I Group 7 Social Services Social Security Public Guardian Group 3 Group 6 Skilled Nursing Facilities W Legal Aid Home Health Care Senior Meals Program Mental Health Information & Referral * Group 2 Housing Group 8 Adult Day Health Care Ombudsman Figure C25. Structure of Tulare Care Systems: Client Exchange Pattern. (Relations to isolate "Housing" marked with broken lines) N > O ) 0 0 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Emergency Services Hospitals County Health Department a Group 1 Area Agencies on Aging Retired Senior Volunteers 7 j ------------ Group 5 Senior Centers Legislator/Politician Charity Group 2 Housing Group 7 Social Services Social Security Public Guardian Group 3 Group 6 Skilled Nursing Facilities w Legal Aid Home Health Care Senior Meals Program Mental Health Information & Referral Group 8 Adult Day Health Care Ombudsman Figure C26. Structure of Tulare Care Systems: Information Exchange Pattern. (Relations to isolate “Housing" marked with different color) to 0 ) < o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Group 4 Emergency Services Hospitals County Health Department Group 1 Area Agencies on Aging Retired Senior Volunteers Group 2 Housing Group 5 Group 7 Senior Centers I W Social Services Legislator/Politician Social Security Charity Public Guardian Group 3 Group 6 Skilled Nursing Facilities Legal Aid Home Health Care Senior Meals Program Mental Health Information & Referral Group 8 Adult Day Health Care Ombudsman Figure C27. Structure of Tulare Care Systems: Money Exchange Pattern. (Relations to isolate "Housing" marked with broken lines) to o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 4 Emergency Services Hospitals County Health Department Group 1 Area Agencies on Aging Retired Senior Volunteers ......... * Group 2 - / ................................... Housing Group 5 4 -V ........ / Group 7 Senior Centers Social Services Legi si ator/Pol i ti ci an Social Security Charity Public Guardian k W Group 3 Group 6 Skilled Nursing Facilities Legal Aid Home Health Care Senior Meals Program Mental Health Information & Referral Group 8 Adult Day Health Care Ombudsman Figure C28. Structure of Tulare Care Systems: Staff Exchange Pattern. (Relations to isolate "Housing" marked with broken lines) to - n I Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 1 Linkages/MSSP Area Agencies on Aging Information & Referral Charity Consortium Group 6 Transportation Rehabilitation Money Management Group 2 Housing Home Health Care Senior Centers Senior Meals Program Social Services Adult Day Health Care Mental Health Legal Aid Group 4 Emergency Services Hospitals Health Clinics Case Management County Health Department Homeless Group 3 Skilled Nursing Facilities Public Guardian Retired Senior Volunteers Group 7 Ombudsman Other Grouo 5 ^ W Legislator/Politician Social Security Figure C29 .The Structure of San Francisco Care System: Client Exchange Pattern. N > ro Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 1 Linkages/MSSP Area Agencies on Aging Information & Referral Charity Consortium Group 6 Transportation Rehabilitation Money Management Group 2 Housing Home Health Care Senior Centers Senior Meals Program Social Services Adult Day Health Care Mental Health Legal Aid Group 4 Emergency Services Hospitals Health Clinics Case Management County Health Department Homeless Group 3 Skilled Nursing Facilities Public Guardian Retired Senior Volunteers Group 7 Ombudsman Other Group 5 Legislator/Politician Social Security Figure C30.The Structure of San Francisco Care System: Information Exchange Pattern. to -n J c o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 1 Linkages/MSSP Area Agencies on Aging Information & Referral Charity Consortium Group 6 Transportation Rehabilitation Money Management Group 2 Housing Home Health Care Senior Centers Senior Meals Program Social Services Group 4 Emergency Services Hospitals Health Clinics Case Management County Health Department Homeless I Group 3 Skilled Nursing Facilities Public Guardian Retired Senior Volunteers Group 7 Ombudsman Other Group 5 Legislator/Politician Social Security Figure C31 .The Structure of San Francisco Care System: Money Exchange Pattern. N > 4 ^ Reproduced with permission o f th e copyright owner. Further reproduction prohibited without perm ission. Group 1 Linkages/MSSP Area Agencies on Aging Information & Referral Charity \ Group 2 Consortium * Housing i i Home Health Care Senior Centers Senior Meals Program Group 6 ------------ ► Social Services Transportation Adult Day Health Care Rehabilitation Mental Health Money Management ~J<egal Aid Group 4 Emergency Services Hospitals Health Clinics Case Management County Health Department Homeless V Group 3 Skilled Nursing Facilities Public Guardian Retired Senior Volunteers Group 5 Legislator/Politician Social Security Figure C32.The Structure of San Francisco Care System: Staff Exchange Pattern. to C J 1

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evaluation of the Care Advocate Program: Bridging managed care and home community -based services

PDF

Out -of -pocket health expenditures by older adults in relation to age, race, and insurance

PDF

The extent and effectiveness of nursing home regulation in the 50 states

PDF

Evidence of age-based rationing of health care to the elderly: The effect of age and death on the use of health care services by elderly enrollees in an HMO

PDF

Marriage and health: The effect of marital status on health and the mechanisms by which marital status affects health among elderly people

PDF

Effects of a formulary expansion of the use of SSRIs and health care services by depressed patients in the California Medicaid program

PDF

The construct validity of the Health Utilities Index in patients with chronic respiratory disease in a managed care population

PDF

Physician adherence to national hypertension guidelines in an elderly Medicaid population

PDF

Heart disease among middle -aged and elderly persons in the United States: Trends and a multistate model

PDF

Assessing the cost implications of combined pharmacotherapy in the long term management of asthma: Theory and application of methods to control selection bias

PDF

Toward a theory of networks and network effectiveness: An analytical framework for the study of multilevel networks in health care service delivery

PDF

The effects of case mix, hospital competition, and managed care penetration on hospital costs and revenues

PDF

A new paradigm to evaluate quality-adjusted life years (QALY) from secondary database: Transforming health status instrument scores to health preference

PDF

The effects of Medi -Cal managed care on access to mammography screening for women residing in Los Angeles County

PDF

A quantitative and qualitative evaluation of the least burdensome provisions of the Food and Drug Administration Modernization Act of 1997

PDF

Service evaluation and quality improvement of a geriatric evaluation and management program

PDF

Variations in physician practice patterns for eye care under the National Health Insurance of Taiwan

PDF

An analysis of health risk selection and quality of care under Medicare fee -for -service and Medicare managed care health care systems

PDF

Assessment of prognostic comorbidity in hospital outcomes research: Is there a role for outpatient pharmacy data?

PDF

Recruitment of family physicians into rural California: Predictors and possibilities

Asset Metadata

Creator Yip, Judy Yun (author)

Core Title Using network perspective to examine the organization of community -based elder care systems across four communities

Degree Doctor of Philosophy

Degree Program Gerontology and Public Policy

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Gerontology,health sciences, health care management,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-70409

Unique identifier UC11337714

Identifier 3018046.pdf (filename),usctheses-c16-70409 (legacy record id)

Legacy Identifier 3018046.pdf

Dmrecord 70409

Document Type Dissertation

Rights Yip, Judy Yun

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA