The influence of child day care on cognitive and phychosocial functioning in adolescence

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ~-~~~--~---~---~~--~---------~---- ~ ---~--~--------~~-~~~ I THE INFLUENCE OF CHILD DAY CARE ON COGNITIVE AND PSYCHOSOCIAL FUNCTIONING IN ADOLESCENCE by William Michael Brock 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 (Education) May 1981 L __ ---~---- - UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90007 This dissertation, written by William Michael Brock under the direction of h .. is. .. DisSertation Com mittee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of requirements of the degree of DOCTOR OF PHILOSOPHY ................. .l.__{.(k~-~-rd ................ . Dtan i I I ACKNOWLEDGMENTS This research was supported in part by the National Institute of Child Health and Human Development, Contract Number NOl-HD-82807, 11 Consequences of Family Structure and Maternal State for Child and Mother 1 s Development .•• Without the encouragement and support of Robert Baker, my advisor, Birgitte Mednick, and the other members of my committee, this study could not have been done. am grateful to all the individuals who have provided technical advice and assistance for this disserta- tion. Thanks to my parents for their support in a variety of ways, and many thanks to my wife, Christine, and my family for their help and monumental patience. i I L ___ _ --------------·---------- ------- ----------- _ _ ___ _ __ I_ d TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i i LIST OF TABLES. . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . • . . . . . . . . . . . . . . . . . . . . v LIST OF FIGURES ..•...............••.....••......•................. vii ABSTRACT. . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . • . . . . . . . vi i i Chapter I . INTRODUCTION ......•.•.............•...................... Purpose of the Study Conceptual Background Definitions II. REVIEW OF THE LITERATURE................................. 9 The Psychosocial Effects of Day Care The Mother-Child Relationship Summary of Day Care Effects of the Mother-Child Relationship Day Care Effects on Other Psychosocial Development Summary of Psychosocial Effects of Day Care Cognitive Effects of Day Care Cognitive Effects of Day Care for High Risk Children Summary of the Cognitive Effects of Day Care Discussion II I . METHODOLOGY.............................................. 45 Sample The Danish Perinatal Study The Danish Perinatal Followup Study Sample Characteristics Generalizability to Other Samples Instrumentation and Data Collection The Perinatal Data Mother's Interviews Teacher's Questionnaire Other Data Sources iii ~------------~~----~~~-------- ---------~~-- ! I I IV. Data Analysis The Analytic Strategy Stage 1: Variable Definition Stage 2: Discriminant Function Analyses of Dependent Variables Stage 3: Model Development and Testing Ad Hoc Analyses RESULTS ..........•....................................... Stage 1: Variable Definition Stage 2: Discriminant Function Analyses of Dependent Variables Predicting Toward Type of Child Care Predicting Toward Psychosocial Related Variables Predicting Toward Cognitive Related Var i ab 1 es Stage 3: Model Development and Testing Model Development Model Testing and Refinement 66 V. SUMMARY AND DISCUSSION ................•................•. 103 Summary and Interpretation Child Care Outcomes Psychosocial Outcomes Cognitive Outcomes Methodological Comments The Usefulness of Molar Measures Degree of Factor Specificity Limitations of Multivariate Method for Analyzing Interactions Analysis Under Conditions of Periodic Treatment Conclusions REFERENCE NOTES .................................................. . REFERENCES ....••........•.....•.....•..•..•..............•........ 121 124 I APPENDICES .............................•.......................... 130 A. Summary Tables of Studies I, I I, and I I I for the Danish Perinatal Followup Study .......................... 131 B. Variables Used in the Present Study ...................... 135 i I L-----~-~--------------------------- ,--------------------------------------------------- - ---------------- ------- LIST OF TABLES Table Page 1. Day Care Effects on the Mother-Child Relationship.......... 12 2. Day Care Effects on Other Psychosocial Development......... 21 3. Day Care Effects on Cognitive Development.................. 32 4. Variables Defined During Stage One Analyses................ 57 5. Stage Two Discriminant Function Analyses................... 59 6. Variables Derived from Items in the Mother's Interview Schedule................................................... 67 7. Variables Derived from Items in the Social Worker's Appendix to the Mother's Interview......................... 68 8. Variables Derived from Items in the Teacher Questionnaire.. 69 9. Discriminant Function Analysis Predicting Toward Type of Chi 1 d Care Ages 0 to 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 10. Discriminant Function Analysis Predicting Toward Type of Child Care Ages 3 to 6..................................... 74 11. Discriminant Function Analysis Predicting Toward Type of Chi 1 d Care Ages 6 to 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 12. Discriminant Function Analysis Predicting Toward Type of Chi 1 d Care Over Age 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 13. Discriminant Function Analyses Predicting Toward Psychosocial Outcomes in Adolescence....................... 81 14. Discriminant Function Analyses Predicting Toward Cognitive Outcomes in Adolescence.......................... 86 15. Number of Measures Predicting to Dependent Variables in the Discriminant Function Analyses...................... 91 16. Correlation Matrix for Day Care Males...................... 99 17. Correlation Matrix for Day Care Females.................... 100 ~b];-----------------~~--- ----- ------------- -.-- ----------------- --------- 118. Summary of Findings Related to Child Care Outcome Variables ................................................. . 104 ! 19. Summary of Findings Related to Psychosocial Outcome Variables.................................................. 108 20. Summary of Findings Related to Cognitive Outcome Variables .................................................. 111 I L _________________ ----------- ,------------------- 1 I Ll ST OF FIGURES I Figure Page I I 1. Time Frames for Variables in the Discriminant Function I Ana 1 yses .................................................. . 58 2. Initial Model of Day Care Effects Based on Discriminant Function Analyses and Literature........................... 93 3. Final Model Predicting Cognitive and Psychosocial Outcomes for Day Care Males................................ 98 4. Final Model Predicting Cognitive and Psychosocial Outcomes for Day Care Females .............................. 101 I I I I L ________ -- ---- - vi ---------~ ABSTRACT This study analyzed the long-term effects of day care and other types of child care on cognitive and psychosocial functioning in adolescence. Current day care 1 iterature provides scant information about the effects of average quality day care or the effects of various alternative types of in-home and out-of-home care (such as full-time sitter care or half-time day care). The Danish Perinatal Followup Study provided a context for examining the long-term effects of various types of child care. Individuals in the sample were born between 1959 and 1961. Long-term outcome data were collected when subjects were in mid- to late-adoles- cence using mother interview protocols, teacher rating scales, and government data sources. Antecedent measures for the sample were SES- related variables (including family income, family instability, and family dissension), maternal variables (mother•s orderliness and contentment), child health measures (neonatal physical and neurological status), and the subject 1 s sex. Five different child care variables were analyzed at ages 0 to 3, 3 to 6, 6 to 11, and over 12 years old. Below age 6 the child care variables included maternal care, full- and half-time day care, grandmother care, and care by other adults. Over age 6 the focus was on the type of supervision children received after school (e.g., maternal care, after school center, siblings, and I no supervision). I The psychosocial outcome variables included peer I L~- ~~-~--~-~----~~~~-------~- -~~----~- ~teraction, ad~lt inte~action~~~~~t~~t~~~ction, !cognitive outcomes included verbal skills, reading and math proficiency I land whether the child had experienced remedial instruction or other i [special services. I 1 A three-stage data analysis included 1) variable definition I [via reliability analyses, 2) discriminant function analyses of all dependent variables, and 3) development and testing of a model of day care effects. Type of child care was predicted by the type of care in the previous age category. Stable families with greater economic resources tended to use maternal care and half-time day care at early ages. Less stable families with fewer economic resources utilized full-time day care at earlier ages. This study helped document the variety and fluidity of child care arrangements at all age levels. Additional research is needed to further describe the variety of child care a 1 ternat i ves. Day care influenced both psychosocial and cognitive outcomes in adolescence. Full-time day care from 0 to 3 was associated with increased aggressiveness, poor math achievement, and a higher incidence of special education. Full-time and half-time day care from 3 to 6 exerted a positive influence on peer, adult, and parent relations; verbal skills; work organization; and grade retention. Child care in an after-school center was associated with increased aggressiveness, special education, and remedial instruction. After-school care by older siblings predicted good peer relations. I i l _________________ -------- --- Two distinct SES-related factors were important for understand- I i i ng day care I environment. effects: economic resources and the family's home Study results tended to indicate the limited usefulness ! lof molar outcome measures, the limitations of current multivariate I 1 methods for interpreting interactions, and the difficulty of prediction under conditions of periodic treatment. The results of this study provide support for the idea that average quality day care at too early an age may produce harmful results. However, in this study, day care after age 3 predicted positive outcomes, supporting the beneficial effects of appropriate quantities of day care. Overall, few negative outcomes were associated with day care, supporting the need for further research to analyze the specific child care environmental features that promote healthy development. I l ____________________ . ------------- I I I CHAPTER I INTRODUCTION Prior to the 1970 1 s there was little research evaluating day care programs, particularly for children under three. At the policy- making level there were few resources allocated for day care, probably due to an assumption that separation of young children from their mothers was the equivalent of maternal deprivation. Maternal depri- vat ion, as many studies have shown (see Casler, 1961; and Bowlby, 1969 for reviews of the maternal deprivation literature), creates serious and long-standing effects, and, as a result, publicly supported infant care programs were virtually non-existent. Etaugh (1980) has docu- mented the persistent popular belief that non-maternal care is harmful to children. However, public policy with regard to day care has been changing over the past decade. According to Caldwell (1972) the primary reason for this is the vastly increased number of mothers of young children who are entering the labor force and a corresponding increase in day care attendance. In 1958 there were approximately 800,000 full-time working mothers of children under three (Caldwell, 1972). By 1976 that number had grown to over three million (National Council of Organizations for Children and Youth, 1976). All of these parents must make child care provisions and the trend has been away '------~------ ---------- ------·--- -----·------ [:: i :::~a~~:-:~~~ ;~ 9 : 6 ~~me to some type of g ~~p daY~~ set t i ~g~ I \N There are other reasons why day care services have come to be I increasingly demanded and more closely scrutinized. The relatively greater poverty levels of young families with children, especially single-parent and minority families, who are primary consumers of day care services, indicates the extreme importance of our national day care policies. Another factor is the increased number of teenage mothers, many of whom are in need of further schooling and vocational training before they will be prepared to enter the job market and become self-supporting. The children of these subpopulations may present a high risk for later public intervention and support, and it is important to understand day care effects for these groups. Because the advent of large scale day care services in this country has been recent, there is 1 ittle research related to the long- term effects of day care services. In addition, appropriate outcome evaluation for day care and early intervention programs has been extremely difficult (Belsky & Steinberg, 1978; Beller, 1974). How- ever, both taxpayers and public policy decision-makers are legitimately concerned with the relative success or failure of various child care systems. If funding for child care services is to continue to expand, it is crucial that a thorough understanding exist as to the cognitive, psychosocial, and physical consequences of such programs. Purpose of the Study The present study examined the long-term cognitive and psychosocial effects of child day care, utilizing a sample from ~nma~k:-;-co-~~~ry th~-t ha; publl~~-sup~~-~t~d -~h-ild care services for I I !some time. Wagner and Wagner (1976) have presented some of the reasons I researchers in the United States should look to countries such as I I 1 Denmark for information about the effects of child care. This study examined the long-term effects of several child care ~alternatives, with a special focus on day care settings. The specific 1 0utcome measures examined were cognitive and psychosocial variables measured in late adolescence. These outcome measures included both constructed measures and certain other "molar•• measures, such as grade retention, following Zigler and Trickett 1 s (1978) recommendations. This study 1 s significance includes the fact that the effects of child care were examined, not for one or two years, but into mid- to late- adolescence when an individual •s overall social competence is becoming apparent. The relationships between the type of child care provided and the cognitive and psychosocial functioning of the child were depicted in a descriptive model of day care effects. An additional feature of this study was that it analyzed long-term outcomes, not only for children with exclusive mothering or full-time day care, but also for children with various types of intermediate child care arrangements. This is a topic for which there has been scant infor- i mat ion. i Conceptual Background Many authors have commented on the changing structure of the U.S. family. There is 1 ittle doubt that the manner in which adults are striving to meet the needs of today•s children is changing. i ! Etzioni (1977) reported that fewer than two-thirds of the households L_ ------·--·---·--- I in this country are-headed by a married coupl;~-~d--th~-thls I p~rZent=l I age has continued to decrease. Etzioni recognized that political, I economic, and socl a I changes are affecting chi I drear i ng practices, j although opinions differ as to the consequences of these changes. I The National Council of Organizations for Children and Youth ! (1976) has published statistics related to the growing need for, and I use of, child care services. One third of children from zero to five I years old have working mothers, a seventy-one percent increase from 1970 to 1975. After a child reaches school age there can still be as much as a 1500 hour differential per year between the care a school provides and the child care requirements of a full-time working mother. In 1975 there were approximately 7.5 million children under fourteen in licensed and unlicensed day care centers, a fraction of the 24.5 million children whose mothers were employed. The percentage of children in day care is rising steadily. I Developmental psychologist Urie Bronfenbrenner has attempted to bring some perspective to present day child care services (Bronfen- brenner, 1974; Note 1). Bronfenbrenner (Note 1) provides a statistical summary of the changing U.S. family and argues that childrearing today is distinguished by increasing segregation according to age. Citing his own research, Bronfenbrenner states that over the past twenty-five years there has been a continual decrease in all spheres of interaction between parents and their children, and that children are growing up with their own age-mates rather than with persons younger or older than themselves. Parents, because of their many obligations, are I creating a situation in which children are exposed to alternate I I I L~----------------------------- --- l i i ~retakers -~gre~-ter percentage --~f-th~ti~;-~;.:;d-~p;~--1 es;--time with I their parents. Il l Bronfenbrenner (Note 1) states that the decreased number of I adults in the lives of children pervades not only the home situation [but the school and community as well. Children are generally experi 'encing less contact with adults and are concurrently exhibiting greater dependency on their peers than in the past. Although he recognizes the difficulties involved in evaluation of early childhood programs, Bronfenbrenner strongly recommends that we study the types of family support systems that are presently utilized and the effects of these child care systems on children and their families. With this informa- tion it will be possible to generate public policies designed to provide incentives for the development of optimal child care and family support systems. Zigler and Trickett (1978) have documented the limited helpfulness of process and short-term evaluation studies of early intervention. They note that, although it is important to know how effectively a program is installed and what short-term cognitive and social gains the participants demonstrate, the true measure of a program's success are the long-term psychosocial consequences of participation. Zigler and Trickett recommend the study of ''social competence" as a desirable long-term outcome measure. They state that physical health, cognitive ability, school and employment achieve- ment, achievement motivation, self-image, andcreativity could all be measures of a social competence factor. Zigler and Trickett also recommend other "molar" measures be used. Molar measures should be [used becaus;-they-can be concretely-;;-;~~~-~;;d~~a-;=~-~h-i-ghly-~elated to lsocial adjustment, and are readily interpretable by both policy makers 1 and taxpayers. They recommend the following appropriate molar mea- sures: incidence of juvenile delinquency, teenage pregnancy, child abuse, 11 dropping-out, 11 special education, and receipt of income maintenance. Lazar, Hubbell, Murray, Roche, and Royce (Note 2) have conducted a long-term outcome evaluation study of experimental early intervention programs conducted in the sixties and early seventies. Utilizing outcome measures of grade retention, incidence of special education, and several simple questionnaire items posed to students and their parents (collected about ten years after the programs 1 termination), Lazar and his colleagues argue for the effectiveness of early intervention. The authors claim that they have found positive long-term effects despite the fact that the short-term positive effects for most of the programs included in the study had faded with time. Several sources indicate, however, that the programs included in the Lazaret al. study were considered exemplary, and probably should not be characterized so simply (Brown, 1978; Ryan, 1972). In the sole long-term followup study of day care effects, Moore (1975) found that prediction for groups selected according to child care arrangements early in life was better for adolescents than for younger ages. Taken together, Lazaret al. (Note 2) and Moore (1975) provide some evidence that latent or sleeper effects may exist for child care I and/or early interventl~n progra~~----Th~;~~-t~d-i-~~-~~phasi;~ the I importance of appropriate longitudinal research on the effects of i ch i 1 d care. ! Definitions For the purposes of this study several definitions are presented (adapted from Kamerman & Kahn, 1976). Child care refers to all types of public and private arrange- ments for child care and may be designated as either full time or part time; public or private. Rendered by someone other than the immediate family, child care services may be either inside or outside the home of the child. Day care refers to programs, usually full time and outside the home, that provide child care under adult supervision. Day care may be provided in a group setting, which is called a day care center, or in a family setting, which is called family day care. Early intervention programs refer to a variety of organized programs for children under compulsory school age. They may take the form of day care or pre-school programs depending upon the amount of time child care is provided and the extent of administrative commitment to an ••educational" or 11 developmental 11 model of program service. Head Start is an example of an early intervention program with both educational and child care provisions (e.g., health care, a nutritional program, full day coverage, etc.). It is also useful to distinguish between child care provided to children under school-age (0-5 years) and those attending school [who require child care services after normal school ho~~-;:~n-(usually i 1 6 to 12 years o 1 d). It is often assumed that those over 12 are I capable of caring for themselves after school. i I I I ~I ----------------~~-- ------------- CHAPTER II REVIEW OF THE LITERATURE The purpose of this review is to examine previous research on the effects of day care on psychosocial and cognitive development. Research on psychosocial effects has traditionally focused on the mother-child relationship, and this area will be reviewed first. Following that will be a critique of all other psychosocial effect research. The review will end with an analysis of the cognitive effects of day care, including the effects of day care for high- risk infants. Because the present study does not examine child health as an outcome variable, the effects of day care on health are not reviewed. Relevant reviews of this literature can be found in Ki 1 mer ( 1979) and Sj01 und ( 1973) . The Psychosocial Effects of Day Care The Mother-Child Relationship One of the earliest and continuing concerns regarding day care effects has been the possible negative influence of mother-child separation. The mother-child bond has traditionally been viewed as critical to healthy development. The ability to form attachments, to be socialized by parents and other adults, and to form a capacity for trust and confidence are considered the direct result of a success ful mother-child relationship (Bowlby, 1969). If the mother-child L_ ____________ ~ - -·-··--· ·---- ,------------~---- ----· -----------------·-- ------- - is not adequately formed at an early stage, it has been assumed! there would be profound negative consequences for the child's psycho- social development. Spitz (1945), Goldfarb (1943), and Bowlby (1953) each studied maternal deprivation under different circumstances but with similar negative conclusions. Since these early studies there has been a tendency to equate maternal separation with maternal deprivation. Despite the fact that few day care settings resemble the extreme deprivation of the studies mentioned above, there has been little interest until recently in defending day care. Casler (1961) and Yarrow (1964) were among the early writers to point that the day care environment need not always be associated with the consequences of severe maternal deprivation. They also noted that many children who do experience severe maternal deprivation recover with time. These writers suggested that maternal deprivation, per se, may not be the critical child care variable for predicting negative psychosocial outcomes. The quality of substitute care that is provided, and whether, in fact, some type of sensory, cognitive, or emotional deprivation has occurred should be far better predictors of negative outcomes. For example, a study by Rheingold (1960) indicated that children in institutions received less visual and verbal stimulation although they did not spend more time alone than home-reared children. The gradual acknowledgement that day care may not be the equivalent of maternal deprivation, although maternal separation surely accompanies day care, has caused a refocusing of questions regarding the effects of day care on the mother-child bond. The i L_ --------~------- --------------- ~imary ! questions today have become: I I • Does having an alternative caretaker a large portion of the time have a negative influence on the parent-child relation ship? • Do repeated daily separations from the mother have any negative effect on the child if high quality substitute care is provided? Table l provides an overview of recent research on the effects of day care on the mother-child relationship. A review of research conducted prior to 1976 can be found in Ricciuti (Note 3). One of the early studies of maternal attachment was conducted by Caldwell, Wright, Honig, and Tannenbaum (1970) comparing 30-month- old home care children to a matched group of children who had been attending a day care program at Syracuse University for an average of nineteen months. This early study is typical of the day care literature in several respects. The total sample size was small (~ = 41) and the children were described as 11 disadvantaged." Social class was not clearly defined and the mean IQ scores at the time of data collection were well within the average range (the matched groups were within two points of each other). Most typical of later studies was the fact that the day care program was conducted in a university setting and specifically designed to provide 11 critical environmental experiences" to the children. Caldwell et al. found no differences in maternal attachment between the home and day care groups that could be attributed to type of care. They concluded that child care can vary tremendously before negative consequences will occur. Day care should be viewed, they felt, not as institutional care, but as a return to extended 11 ~~~-- 1 I ! Author Blanchard and Main : B l eha r I Brookhart and Hock l Caldwell , Wright, Honig, ' and Tannenbaum Cochran Cornelius and Denny Table l Day Care Effects on the Mother-Child Relationship Sample Date Size 1979 21 1974 20 1976 15 1970 41 1977 120 1975 64 Age Level 12 to 25 months 24 to 36 months 11 months 30 months 6 months to 6 years 4 and 5 years Results/Comments Avoidance of a stranger was negatively related to time in day care (opposite finding from Blehar) 3-year-olds exhibited "anxious, ambivalent" and 2-year-olds a "detached" relationship with their mother in an Ainsworth strange situation if they had been in day care No difference in maternal attachment for day care and home group; sex by type of care interaction Syracuse Program, no differences found between home and day care groups Few differences found between home and day care children on a number of measures of child-caregiver interaction Sex by type of care interaction, no differ ences in proximity seeking and attention seeking measures of dependency in Ainsworth 1 s situation ,----- 1 I I I Author l Doyle I Far ran and Ramey Kagan, Kearsley, and Zelazo Keister (Note 7) Moskowitz and Schwarz Portnoy and Simmons Ragozin (Note 4) Date 1975 1977 1978 1970 1977 1978 1975 Sample Size 24 160 24 24 35 28 Table ]--Continued Age Level 5 to 30 months 3 to 29 months 3 to 48 months 42 months 3 1/2 to 4 years 17 to 38 months Results/Comments No differences found in response to Ainsworth situation (see Ramey and Farren, 1976) No difference day vs. home care, SES by care interaction for Chinese sample Day care group had faster development as measured by the Preschool Attainment Scale and the Vineland, day care also more willing to separate Day care children significantly more distressed by mother leaving in strange situation (opposite of Blehar) No differences in attachment patterns among home, day care, and family day care children; some sex differences noted Day care children engaged in less give and take with a stranger, no differences on other variables ,----- Table !--Continued ~------------------------------------------------------- Sample Size Age I Author I Ramey and Farran I (Note 8) Ramey and M i 1 1 s Ricciuti Ricciuti and Poresky (Note 9) Will is and Ricciuti Date 1976 1977 1974 1973 1974 23 15 9 and home- rea red controls 10 10 Level Results/Comments 9 to 31 Impact of day care on maternal attachment months less important than the mother•s behavior and relationship with the child 3 1/2 to 9 High-risk day care infants vocalized more months and showed more interactive behavior with their mothers than home group 12 1/2 months 4 to 12 months 4 to 15 months Day care group showed significantly more negative reactions to being left alone with a stranger The presence of a stable, familiar caregiver lessens problems with separation from mother Few distress reactions were noted when parents left their child daily, reactions were greater after nine months [living arrang;me~-t;--;:;~t- uncommo~prf;~--u;--th-e era of the nuclear I family. I The Caldwell et al. study highlighted the importance of I I considering other maternal and family variables and their possible I interaction with type of care. For example, an association was found I i between strength of the child's attachment and both maternal stimula- 1 tion and the child's developmental level. Caldwell et al. found no difference in maternal attachment in day care and home-reared infants. However, this finding did seem contradictory to the maternal deprivation literature and a number of investigators continued to scrutinize the effects of day care on mother-child relationships. Many, perhaps most, of the mother-child relationship research has used variations on the "strange situation'' experimental procedure described by Ainsworth and Wittig (1969) and Ainsworth and Bell (1970). This procedure involved having a mother, child, and stranger together in a room. The child's behavior is recorded during several types of episodes. For example, the child is observed while the mother leaves the room. The stranger attempts to comfort the child if the child becomes distressed. The child's behavior is rated periodically for the amount of crying, proximity or avoidance of the stranger, and I searching for the mother. 1 during reunion with the mother Ainsworth, Bell, and Stayton (1971) I described various child strate~ies and reunion behaviors. The child's behavior is also observed I I I defined by Ainsworth and her associates, Blehar (1974) conducted a Following both the terminology and experimental procedures I L-~------- --------- 15iudy that demonstrated 11 qualitative disturbances 11 in day care I children's ability to separate from their mothers. Comparing 40 I and 3-year-old children who had attended day care for about five with a matched group that were home reared, Blehar characterized the younger group, when separated from their mothers, as 11 detached 11 in their relationship with a stranger, and the older children as resis- tant, avoiding the stranger. Belsky and Steinberg (1978) have critiqued both the strange situation procedure and Blehar•s findings. Blehar recommended that future research focus on possible interactions between children 1 s experiences prior to day care and their reaction to day care. Blehar also noted that, whether or not short-term adverse effects are supported by research, long-term studies should be conducted to determine the cumulative effects and/or 11 sleeper 11 effects of day care. The contrast between the findings of Caldwell et al. (1970) and Blehar•s study has stimulated a great deal of research on possible day care effects on the mother-child relationship. Most of these studies have failed to replicate the negative findings of Blehar. In a study designed to examine differences between home rearing and day care on the behavior of 30 10- and 12-month-old infants, Brookhart and Hock (1976) found no main effects attributable to day care. Although there were no effects due to the rearing condition (home vs. day care), a significant rearing condition by sex interaction was found. This implies that boys and girls may respond differentially to the day care experience. I Clear differences were found between the behavior of infants I • 1n a home versus a laboratory setting, casting some doubt on the ecological validity of the strange situation paradigm. Brookhart and Hock stated that day care and home rearing, as globally defined variables, do not affect infant social behavior or the mother-child bond. They recommend that future research examine contextual vari- ables in greater detail than has been done in the past. Kagan, Kearsley, and Zelazo (1978), in their book Infancy, report the results of their infant studies, including early signs of SES-related effects on cognitive development and cross-cultural comparisons of infant cognitive development. Kagan et al. also report the results of their study of thirty-two pairs of children matched for social class, ethnicity, and sex. The day care children, who had been enrolled in a high quality center from 3 1/2 to 30 months, exhibited no significant differences from home-reared children in their crying behavior on separation. Kagan speculated that separa- tion anxiety in children is highly dependent on developmental level and is probably not an appropriate variable for studying the child 1 s ability to form attachments. Kagan•s experimental situation, however, differed from the strange situation procedure in that no stranger was introduced. This factor probably decreased the amount of anxiety experienced by the child. Portnoy and Simmons (1978) examined the attachment behavior of thirty-five children ages 3 1/2 to 4 years old. The children were observed in several strange situation episodes. The children were classified as reared exclusively at home, until age three after which '-------· -~-- -----" --- ·-- [they were enrolled in full-time day -~are,---;n-d ~hlld~~~~~rolled i~-~ I family day care at age one and in a group center at age three. No I significant differences were found between children raised at home I I and the two day care conditions. Some sex differences were noted among low SES children who attended day care. The authors also reported that the sex differences that were uncovered tended to indicate that the strange situation procedure may not have been appropriate for this age group. In a study of children who had been enrolled in day care for an average of twelve months, Blanchard and Main (1979) did not support the findings of Blehar. Twenty-one children, ranging in age from 12 to 25 months, were observed during their reunions with their mothers and rated on the Ainsworth scale for avoidance in the strange situation. Avoidance was negatively related to months in day care, a finding that suggests Blehar•s results were due to the shorter time her subjects had spent in day care. In addition, more time in day care was associated with better social-emotional adjust- ment. Many other studies have not supported Blehar•s findings of a negative effect on the mother-child relationship. These include Moskowitz, Schwarz, and Corsini (1977), Doyle (1975), Roopnarine and Lamb (1978), Ragozin (Note 4), Cornelius and Denny (1975), Saunders (Note 5), and Farran and Ramey (1977). A limited number of studies have supported Blehar•s finding that day care experience impairs children•s maternal attachment. Several studies by Henry Ricciuti at Cornel University can be !interpreted in favor of home-reared children-;--ho-wever the studies focus on the child's adjustment to day care rather than the effects of day care. Ricciuti (1974) studied children's reactions to being left by their mothers in a strange room. Although Ricciuti acknow- ledged that day care children cried more on separation from their mothers, he noted that daily separation to day care, with a familiar care giver, would evoke much less crying than being left with a stranger. Ricciuti's work should probably be regarded more as research on children's reactions to brief separation rather than studies of day care effects. Cochran (1977) and Ragozin (Note 4) only found a small number of differences in favor of home care from among many dependent vari- ables, clearly pointing out that the similarities between groups are far greater than observed differences in attachment. Summary of Day Care Effects on the Mother-Child Relationship In summary, only a few studies have found evidence that day care, as a global variable, negatively affects the mother-child relationship. The overwhelming majority of studies have detected no differences. For this reason, studies of day care effects have begun focusing on a wider range of psychosocial variables. Other psychosocial effects are considered in the next section. A recent article by Stroufe and Waters (1977) has described many of the limitations inherent in using "attachment" as a dependent measure in studies of day care effects. Portnoy and Simmons (1978) I I and Ricciuti (Note 3) have both criticized the use of the strange I I I L--------~-------------------- ~tuation paradigm to the virtual exclusion of other experimental I i methodo 1 og i es. I i Day Care Effects on Other Psychosocial Development I In recent years there has been a noticeable shift in the i emphasis of research examining the social-emotional effects of day care. This research is summarized in Table 2. Earlier studies focused almost exclusively on the mother-child bond. Today a wider range of social and emotional variables are under study. This new direction in the study of day care effects is beneficial in that researchers are beginning to understand the limitations of the traditional outcome measure (mother-child attachment) and explore other psychosocial vari- ables and other, equally important, organizational characteristics that may influence day care effects. Despite the shift in emphasis of psychosocial studies of day care effects, the fundamental question has remained the same: Is day care harmful to children's development? Ironically, one of the early studies to find negative effects was Schwarz, Strickland, and Krolick (1974), who studied children at the Syracuse University day care center. This was the same program for which Caldwell et al. (1970) found no negative effects on maternal attachment several years earlier. Schwarz et al. studied 19 matched pairs of 4-year-olds and found that children in day care since infancy were more verbally and physically aggressive (both with peers and adults) and less coopera- tive with adults. Meyer (Note 6) presented the most recent progress report on the children in the Syracuse University day care program. Meyer ,-~- -~- 1 i Author Braun and Calwell Cochran Honig, Lally, and Mathieson (Note 12) Kagan, Kearsley, and Zelazo Lippman and Grote (Note 1 0) Macrae and Herbert Jackson i :N Table 2 Day Care Effects on Other Psychosocial Development Date 1973 1977 1979 1978 1974 1976 Sample Size 120 160 198 16 Age Level 1/2 to 6 years K and 1st grades 3 to 29 months 1 to 4 years 2 years Results/Comments Day care was not harmful to the social emotional development of subjects Swedish population, minimal differences between day care and home reared, differences seemed to be related to variations in the constraints within different settings Followup indicated day care center graduates had more positive social skills both with peers and adults and were more verbally communicative No differences between day and home care groups, significant care by SES interaction~~~ for Chinese subjects No differences among day care center, family day care and home care for cooperativeness, reaction to stress, and sex role typing Day care exhibited better peer relations L-:::_ ____ " --~-----~--------- ---------------"-""--~------"- r-~---~---- ~---------·- ------- --·-------------·- Author McCutcheon and Calhoun Meyer (Note 6) Moore Robertson (Note 11) i Rubenstein and Howes Date 1976 1979 1975 1978 1979 Sample Size 18 40 167 121 30 Table 2--Continued Age Level 5 to 30 months 36 to 52 months 1 year of day care prior to age 5 5 and 8 years 18 months Results/Comments Children successfully adjust to day care by decreasing their crying, sitting alone, and interacting more with both peers and adults Day care children had more positive social peer interaction (especially verbally), and more negative interactions; no differences in interactions with adults Only followup study to mid-adolescence; care by sex interaction; day care boys were more peer oriented Care by SES, care by sex interactions found; day care influenced responsiveness to social reinforcement, imitation, attention seeking and cooperativeness Day care children had more adult-infant play, reciprocal smiling, higher developmental level of play with toys; home care demonstrated more crying, maternal restrictiveness, and verbal response to mother's talking Author Schwarz, Krolick, and Strickland Date 1973, 1974 Sample Size Table 2--Continued Age Level 42 to 46 months Results/Comments No major differences in '73 study; day care less cooperative with adults, more active, more aggressive in '74 study I analyzed 20 children in day care since the age--or-6months, and a I matched comparison group. The original day care sample was selected I i as a 11 high risk'' group. As in the Schwarz et al. study the day care I children exhibited more physical and verbal aggressiveness, more I I activity, and less cooperation with adults. I I I Meyer's interpretations of these findings are similar to I Bronfenbrenner's (Note I) theoretical speculations about the effects I of day care. Meyer notes that day care children receive more stimula tion and more peer interaction than their home-reared counterparts. This results in more interactive activity of every type, both positive and negative. Meyer's study indicated that day care children were more interactive, and he provides some evidence that the day care children had elevated positive interactions as well as negative interactions. -These findings raise several questions about the psychosocial effects of day care. How can it be determined whether observed differences are positive or negative? It seems possible to interpret increased levels of both positive and negative interactions either way. To what extent are variables such as peer interaction desirable? For example, peer interaction would seem to be desirable until it becomes so intense that it negatively impacts adult socialization. It can readily be seen that the issues become complex when maternal attachment is no longer the sole focus of psychosocial effect research. Lippman and Grote (Note 10) compared 198 children in three groups (matched by sex, race, and number of parents in the home) representing home care, family day care, and day care centers. The 24 '--------~----------- ··-··- ----- ----·----- ~t-of-home care groups averaged two years of day care. Minimal I differences were observed among the children in sex typing, impulse I control, parental relationships, and cooperation. Day care children, I in contrast to the home care group, appeared to have a more realistic I I assessment of their own skills. In contrast, home-reared children I [were more likely to take turns in a game requiring cooperation than i I children in day care. Lippman and Grote do not regard their findings as evidence of negative day care effects, but urge that the simple contrast between day and home care be re-examined. Moore (1975) conducted the only longitudinal study through mid-adolescence of children who received either exclusive mothering or day care for at least one year prior to the age of five. Because the sample size was relatively large (~ = 167), Moore was able to analyze several subpopulations separately, e.g., boys, girls, and children with both parents 1 iving at home. Moore•s dependent variables were derived from mother interviews • as well as various psychological tests. Moore found a type of care by ~~·sex interaction, noting that prediction for boys was considerably better than for girls. Children who received exclusive mothering I 1 tended toward self-control and school achievement, indicating that I they had internalized adult standards. Children in substitute care I tended to be more peer-oriented, active, and aggressive. Moore•s I findings are interesting although the psychological interpretations I I of his findings in this ex-post-facto study seem tenuous at times. I i I Moore•s study is significant in at least two respects. First, in contrast to most studies of high quality centers, a variety of day I L_ ___________________________ -- --· - --- --·---- l]care centers were included in the study.--Thls permits some general- 1 ization of Moore 1 s results to the 11 average 11 day care setting. Moore i ]speculates that optimal child care arrangements could well include I [part-time substitute care after the age of three. Secondly, this I lstudy has followed children to mid-adolescence. The author states I I \that predictions from type of care to outcomes during adolescence Jare superior to short and midrange predictions, a finding supportive of possible sleeper effects. Studying 8 matched pairs of 2-year-olds, Macrae and Herbert- Jackson (1976) examined the same behavioral traits that Schwarz et al. (1974) (discussed earlier) had examined. Macrae and Herbert-Jackson•s findings were essentially in favor of day care, exactly the opposite of the negative findings of Schwarz et al. The day care group was significantly different from the home care group in their ability to get along with peers, problem-solving, and 11 planfulness.•• What can account for the essentially opposite findings of the Schwarz et al. and the Macrae and Herbert-Jackson studies? Several important variables may play a part in these findings. First, the ages of the two groups differed. The children in the Macrae and Herbert-Jackson study averaged two years younger than in the other study (average age of 2 as opposed to 4). Second, the home-reared group in the Macrae and Herbert-Jackson study had actually been in day care for one to six months (no average length of time in day care \was reported). This group could possibly be more appropriately I !considered a 11 transitional 11 group rather than home reared. Finally, iMacrae and Herbert-Jackson point out that different programs may I ~I ----~-----·-------·------------ -- !produce di-fferent-~res~l~ts, and that neg-;tl~~-fl~-dln~s in one day car~~! I setting should not be generalized to other settings without empirical I 1 ve r i f i cat i on . I [ Robertson (Note 11) studied the effects of day care center I i attendance on IQ, school achievement, and several measures of social behavior and motivation. Type of child care did not produce a signi- ficant main effect on any of eleven dependent measures; however it interacted with other variables in several ways. Day care interacted with SES to influence the children's responsiveness to adults, with lower SES day care children less responsive to adults than others. Type of care did not effect the school achievement of lower SES subjects, while home-reared children had relatively higher achievement ratings. Robertson suggests that numerous studies remain to be done to define appropriate constructs for studying day care effects. She also recommends more long-term studies that analyze specific attri- butes of day care centers as well as consideration of child character- istics such as SES-related variables. In a study of two matched groups of 18-month-old infants, Rubenstein and Howes (1979) found several differences in social and play behavior favorable to day care. The day care children engaged in more reciprocal smiling and played with adults more frequently than the home-reared sample. The authors noted that there seemed to be no negative effects due to daily mother-child separation and that peers played an important role for day care infants in stimulating play and positive affect. ! L-~-----------------~-----·-- ,------·----------------------------------·-------- ----- -------- ------- The Rubenstein and Howes study focused on community-based day care rather than high quality (often university-based) day care. In accounting for the differences between the home and day -~ I care infants, Rubenstein and Howes rely heavily on their observations of adult and peer interactions among day care infants. In a separate ad hoc analysis the authors compared day care children with home-reared children who had frequent interaction with other peers at home, and found many of the day care/home care differences disappeared. Also noted was that the contingencies governing peer and caregiver inter- action in day care may differ substantially from home care. Rubenstein and Howes suggest that child care in a social setting may be more pleasurable for both caretaker and child than care provided by an i so 1 a ted parent. The favorable results of the Rubenstein and Howes study should be interpreted cautiously. The total sample size(~= 30) is small and the average length of time in day care was only 4.7 months. If the results are accepted, several important questions remain. How can the effects of an altered home environment and the center be differentiated for day care children? Are the results of greater peer influence always positive? And, what are the lasting effects of differences between day and home care? Cummings (1980) studied the effect of environmental context on caregiver preference, that is, whether the laboratory or day care center setting would affect a child 1 s preference for their caregiver. In the laboratory setting children showed no preferences for any caretaker other than their mothers. This study certainly supports '----- -- ------ -------- ---------------- ·- --------- ------ I v:: i : r ::::::: mp~:·~::~ t ~:n: t ~::: ~:g~~Y~ :~:~ ~f~:~x~:p :: ::o :::::::I y I efforts to move beyond the study of global day care effects and ! 1 isolate specific environmental factors that influence the day care I child. One of these factors, caregiver stability, received support I I in Cummings' study. Numerous other factors, including others that I can be influenced at the policy-making and regulatory levels, remain to be analyzed. The studies of psychosocial effects of day care have also given rise to studies that have a broader perspective than a strict focus on the child's response to day care. Important issues such as the reasons for type of care selection, and the effects of day care on parents, siblings, and the community still need to be examined. Fowler and Khan (Note 13) and Ramey and Mills (1974) found that child- ren's day care attendance influences the type of parental stimulation they receive at home. Steinberg and Green (Note 14) compared families using different types of child care and found no differences in SES, father absence, or parental age. Not surprisingly, the Pennsylvania Day Care Study (Elliot, 1973) reported that maternal employment and income was greater for day care children than home-reared children. Several studies have also shown that satisfaction with child care is associated with marital and employment satisfaction (Meyer, Note 3; Harrell & Ridley, 1975). Summary of Psychosocial Effects of Day Care The 1 iterature related to the psychosocial effects of day care is complex and often contradictory. While researchers should L ________ ----- ------- -- -- I be -~ommended for beginning to focus--;~-~--~~d~r--~a-ng~ ~-fd~~pendent variables (other than maternal attachment), it has been only very recently that covariables (such as SES, age, sex, and length of time in care), as well as the day care construct, have begun to be system- atically examined. The studies that could be considered unfavorable to day care include Schwarz et al. (1974), Meyer (Note 3), Rubenstein and Howes (1979), Moore (1964), Raph, Thomas, Chess and Korn (1974), and Lippman and Grote (Note 10). Most other studies report no differences ion findings for children attending day care. These studies include Kagan et a 1. ( 1978), Corne 1 ius and Denny ( 1975), Moore ( 1975), McCutcheon and Calhoun (1976), and others. The methodological weaknesses of many of the above studies are serious. There have been very few controlled studies, and those have often been discredited because of their lack of ecological validity. The typical study involves an ex-post-facto comparison of several matched groups. While these studies shed some light on day care effects, the variables and study context are seldom des- cribed in sufficient detail to permit strong inferences from the data. Often the number of subjects in such studies is 1 imited, also. I Moore's (1975) long-term followup study is the only study I with positive : centers. The effects that utilized average or typical day care majority of studies, including those finding negative effects, examined children in high quality day care. The variables of greatest interest to researchers have been peer and adult interaction. These have usually been defined in terms I I L-~-~------------- ~------- I : r ~::::: ~: v~~~~a :~:~:::• :~::-::o:~: r::: i ~r~~~~:s 0 :o 0 :::; ne I I the psychological environment of day care. Cognitive Effects of Day Care Studies of the cognitive effects of day care constitute a major portion of the day care literature. There has been concern that day care centers, like large institutions, may provide decreased cognitive stimulation. Others have speculated that, particularly for disadvantaged children, a day care setting may provide greater stim- ulation than a single caretaker at home would provide. Research related to the cognitive effects of day care is summarized in Table 3. The methodological problems involved in studying cognitive effects of day care are similar to many studies that use standardized test results as dependent measures. The content of the tests is in no way tied to the curriculum of the center. While this may be acceptable or even desirable in a large scale effort to determine total effects, these tests are extremely limited in their sensitivity to specific aspects of the day care curriculum, as the experiences of Head Start program evaluators have clearly indicated (see Zigler & Trickett, 1978; Zigler & Valentine, 1979). Two studies, Peaslee (Note 15) and Cochran (1977), have reported negative effects on cognitive functioning for children in day care, and Cochran discounts his findings as only one significant difference among numerous findings of no difference between home and day care children. : I 311 " ___________ :__j ~--- 1 I Author I Caldwell, Wright, Honig, and Tannenbaum : Cochran Doyle i Fowler and Khan (Note 13) Garber et al. i Kagan et al. Date 1970 1977 1975 1974 1976 1978 Table 3 Day Care Effects on Cognitive Development Sample Size 40 120 24 39 20 116 Age Level 30 months l /2 to 6 years 5 to 30 months 5 to 70 months l /2 to 6 years 3 to 29 months Results/Comments Cognitive gains for children enrolled in group care six months or more as measured by the Stanford-Binet No differences home vs. day care on the Griffiths Mental Development Scale No differences home vs. center care (for an average of 7 months) on the Cattell Infant Intelligence Scale Griffiths, Binet measures showed "enriched" day care can be beneficial; no regression apparent at 70 months, advantaged children ! benefitted more from the program than disadvantaged children Low income, black children ("high risk") showed substantial gains on the Binet after 22 months Bayley and other measures; no substantial differences, evidence of positive day care effect when it enhances the child 1 s environment ---~-·------------------------·- --------· ~- ---------~---------~--~----~~-------~-~~----~ --------- --------~~---~~ 1 Table 3--Continued 1 I I I I i Samp 1 e Age I Author Date Size Level Results/Comments I Keister (Note 7) Lally (Note 18) , Macrae and Herbert . Jackson Peaslee (Note 15) Ramey and Mi 11 s 1970 1974 1976 1976 1977 24 129 16 50 60 3 to 48 months 36 months 2 years 2 years 3-1/2 to 9-1/2 months No negative effects of high quality day care, University of North Carolina Project Syracuse Children•s Center, Binets of day care children were better than matched control group of low income/ education non-clients Day care group performed better than home group in problem solving, ability to abstract and planfulness Home-reared infants more competent than day care; measures of abstract ability, language ability, social competency, discriminative ability, and mental development (Bayley) High-risk day care infants more closely resembled low-risk than high-risk home reared infants in Bayley motor develop ment, vocalizations, visual discriminations -------------------------- - --------------, Author : Ramey and Smith (Note 19) Robertson (Note 11) Robinson and Robinson Taylor (Note 17) W i 11 is and Ricciuti Winett, Fuchs, Moffatt and Nerviano Date 1976 1978 1971 1975 1974 1977 Sample Size 121 31 75 10 124 Table 3--Continued Age Level 1-1/2 to 18 months followup at 6 and 9 years 1 to 48 months 4 years 4 to 15 months 3 to 5-1/2 years Results/Comments Day care children same at 7 months on Bayley, 16 points higher at 18 months; day care required fewer trials to learn two choice visual discrimination task Care by SES interaction, middle SES home care had higher academic achievement High quality day care enhanced Bayley (mental and motor), language, WPSSI, and Frostig performance No differences between home, part-time, and full-time center care No harmful effects found of high quality day care on cognitive development SES by care interaction, no day care vs. home care differences on Peabody Picture Vocabulary Test or Illinois Test of Psycholinguistic Abilities, mixed care group exhibited the highest scores ~-----P~~~~~~TNot;Tsr-exam i ned -th~ e f-{ects-of ho~e-a-~d day I I on the competence of preschoolers. Building on the definition of I I competence presented by White, Watts, Barnett, Kaban, Marmor, and I ] Shapiro (1973), Peaslee hypothesized that day care children would i j score lower on tests that measured aspects of competence: ability to abstract, language ability, social competence, discriminative ability and mental development. In her matched sample of 50 2-year-olds the home-reared children were significantly more competent than day care infants. This study, focusing on the effects of day care on children who had been enrolled at less than 8 weeks of age is important because the day care centers that were included were not university-based, and could be considered average day care settings. Unfortunately, the author used individual t-tests to analyze the data, increasing the probability of finding significant differences when none existed. This may not have been a problem as the home-reared group scored better on every dependent measure; however this does cast some doubt on the methodology. Cochran (1977) studied the social and cognitive development of Swedish children who were raised in their own homes, a family day care setting, or a day care center. The subjects were 12, 15, and 18-month-old children who were matched for sex, family size, SES, and the location of their home. A variety of observational and developmental data were collected. Few differences existed among the groups with regard to their common experiences and environmental surroundings. Cochran found that day care children tended to be less involved in exploring I their surround-lng~;nd--seemed ];~ -l~v~;;di~-~~g~l t i ve-verba l I situations with adults. Cochran stressed that the similarities I Iacross rearing conditions were much greater than the differences I land, as noted above, tended to disregard the negative findings for , day care. I The studies reporting beneficial effects of day care on cognitive development include Robinson and Robinson (1971), Vroegh (Note 16), and Winett, Fuchs, Moffatt, and Nerviano (1977). In addition, the overwhelming majority of studies with typical children detected no cognitive differences between day care and matched controls (Caldwell et al., 1970; Kagan et al., 1978; Moore, 1975; Doyle, 1975; Keister, Note 7; Taylor, Note 17). One of the early studies to suggest the possible positive cognitive effects of day care was done by Robinson and Robinson (1971) at the University of North Carolina at Chapel Hill. The Bayley scales were administered to 31 infants enrolled in the University•s comprehensive day care center. The results indicated that high quality day care may produce beneficial results both for normal and disadvantaged children. Robinson and Robinson stated that the benefits of quality day care seemed to focus on verbal rather than motoric or other skills. The authors raise several questions about beneficial effects of day care: Will positive effects be persistent or trans- itory? What are the elements within a day care program that provide beneficial effects? And, could these critical elements be packaged and delivered more efficiently to children? Winett, Fuchs, Moffatt and Nerviano (1977) conducted a relatively large (~ = 124) cross-sectional study of day care center, home, babysitter and mixed care children. A large number of measures related to the children's cognitive development were made, including standardized tests, behavioral measures, and parent and center indices. The subjects were separately analyzed according to subpopulations defined by such variables asSES, race, birth order, age, and maternal employment. The mixed care group in this study exhibited both higher SES and higher scores on the cognitive measures. The other groups were similar on most child measures. Maternal employment was associated with child care outside the home, and also with greater paternal involvement. A number of SES differences were found that indicate the importance of considering SES-related variables in day care research. Winett et al. made a number of recommendations for future research in day care. They noted that global measures of day care, such as those employed in their study, are unlikely to help tease lout interactions between 11 family-social types 11 and child care groups I I or between child care and outcome measures. Precise variable defini- \ tion of either family types or child care are difficult because of I I the heterogenity of family patterns. I I , Winett et al. also noted the importance of social class as a prime determinant of day care effects. Following Deutsch's (1973) I I concept that mediating variables associated with poverty must be I 1 1 inked to child outcomes, Winett et al. encourage the identification, I I L-~-----~-------------~------- -- \t"h ro-z;gh fu-tur;-~~-;;~r~h~-SES- r;l-;t;d~~~l-;bl-es associ a ted with I child care that can be manipulated to ameliorate negative outcomes. I Cognitive Effects of Day Care for High-Risk Children Some of the research that has reported cognitive benefits associated with day care attendance have studied 11 high-risk 11 children. High-risk may be defined in many ways, but most of these studies are referring to children from very poor families or children whose parents would be classified as mildly retarded. Unfortunately, some of these studies define high-risk very generally and this casts some doubt on the actual characteristics of their sample. For example, Meyer (Note 6) describes his sample as high risk, yet both the home and day care groups averaged above 100 on an IQ pretest. Studies of day care for high-risk children are conceptually related to outcome studies for the Head Start program. Positive findings for day care and Head Start programs were scarce until researchers determined which subpopulations and dependent variables were critical for analysis. Robinson and Robinson (1971) were among the first in the day care literature to make this point. They noted that, because of the use of standardized instruments as dependent measures, specific program features could not be associated with specific outcomes. They also raised the question of whether bene- ficial effects of day care were transitory (i.e., they would '~ash out 11 ), a problem that plagued early Head Start evaluations (see Brown, 1978). Half of the Robinson and Robinson (1971) sample were as 11 disadvantaged, 11 with income data reported that supported the r:· c-~ iPt ion. -Th;-d i sadvantaged group in this s t~dYbe~~f i t ted more 1 from their day care experience than the middle class group. Although ! the 11 advantaged 11 group scored higher on the dependent measures, the \disadvantaged group made greater gains. High-risk infants have been studied by Ramey and Mills (1974), 1who studied three matched groups totaling sixty infants averaging 6-1/2 months old. The dependent variables for this study included both mental and psychomotor scores on the Bayley Scales of Infant Development, maternal attitude measures, and observation of the infant- mother interaction and the home environment. Ramey and Mills found that the high-risk group attending day care did as well as the general population on most measures, and that the home-reared high-risk group did less well. The high-risk children in day care scored higher than their home-reared counterparts on the psychomotor portion of the Bayley, vocalization, and social relation- ships. Ramey and Mills state that the characteristics of day care programs must be further studied to determine the exact types of social behavior that are facilitated by day care. Fowler (1972) studied thirty-nine infants age 2 to 30 months, including nine disadvantaged children. Using the Bayley Scales, the Stanford-Binet (after two), and interviews, Fowler found average rises of 20 points in IQ scores. In a followup study, Fowler and Khan (Note 13) found no regressions to preprogram levels of function- ing, a highly atypical finding. The New York City Infant Day Care Study (Golden, Rosenbluth, Grossi, Policare, Freeman and Brownlee, 1978) provided the first [evidence that average quality day care may also be~;flt~lgh-risk !children. In a sample of four hundred children from poor families I in New York, the day care children performed better than home-reared I children on the Bayley, Stanford-Binet, and Peabody Picture Vocabulary Test. While the practical significance of an average ?-point differ- ence in IQ scores may be questioned, the New York study is important because it indicates the possible effectiveness of modal rather than model day care centers. Summary of the Cognitive Effects of Day Care Little evidence has been presented that day care attendance can have a deleterious effect on children's cognitive functioning. In fact, there is evidence to support the opposite view. While most studies have reported no difference between day care and home-reared middle class children, several studies have shown positive results. Of those studies, only Fowler and Khan (Note 13) have reported follow- up results that did not show a regression or washout effect. Of the few projects that have studied the effects of day care on the cognitive functioning of high-risk children, most have demon- strated beneficial results. Thus, there is evidence that day care, particularly for disadvantaged children, may aid cognitive development. The effects of day care on cognitive variables other than IQ have been less frequently reported in the literature, and the picture is 1 ess clear. Discussion I Throughout the day care literature evidence for negative i L ________________________ ------------- --- !effects has appeared only sporadically. While this h~ndf~-l~f-~t~-di~ I finding harmful effects is a cause for concern, the majority of evi- l I dence indicates that day care children do not fare worse than children remaining at home. Many, perhaps most, studies report no differences between day care and home-reared children on psychosocial and cognitive dependent measures. Some studies report beneficial effects of day care on a limited number of dependent measures or with certain types of children. These findings lead to two conclusions: • Day care, as a global construct, should be used only under certain circumstances. Studies that purport to make general izations about day care as a whole should include many centers, including centers that are of average quality. There is evidence that day care by sex, day care by SES, and other interactions should be carefully considered. • Studies that occur in only one day care center should describe the program's methods and content in sufficient detail to relate specific features of program content to child outcomes. The study methods and purposes of the two types of research proposed above are different. Many of the apparent controversies in the day care literature could be resolved if researchers acknowledged the limitations of their study methodology. For example, Schwarz et al. (1974) found increased aggressiveness in day care children, while Macrae and Herbert-Jackson (1975), using similar dependent measures, found decreased aggressiveness. Schwarz et al. (1974) suggested that day care effects are program specific, yet there is insufficient data in either study to determine the specific program features that may produce differential results. A number of studies have found type of care by SES and type I of care by sex interactions, and it is clear that sex and SES I L _______ ------------- ----------------~-------- --------~J [~ f~f e ~e nc;~-;, h.;~Jd~be ana 1 y zed i n f u t ~ r0 da_y_;;a;:-.;-St~d i e~. ~ Severa I I studies have also noted that children's individual development levels I seem to influence day care effects. Another important drawback of day care studies to date has been their failure to consider a wide range of effects in addition to child outcomes. Belsky and Steinberg (1978) have stated that studies of day care effects should consider the effects of day care on the parents, siblings, other family members, the community, businesses, and other populations that may be influenced. The decision to uti! ize child care services is made in a social context, and that context must be understood before day care effects can be deciphered. Very few studies to date have included a sufficiently large sample of children and/or centers to draw inferences about total day care effects. Exceptions to this rule are Golden et al. (1978), Moore (1975), and Robertson (Note 11). Moreover, the majority of studies have been conducted in university-based, high quality centers. These studies demonstrate what can be done but do not reflect the effects of an average day care center. With a single exception (Moore, 1975) there are no long-term followup studies of day care effects. The followup studies that have been conducted (e.g., Robertson, Note 11; Fowler & Khan, Note 13; Meyer, Note 6) have been relatively short-term followups of not more than three years. Part of the reason for this may be that the advent of large scale day care in the United States has been relatively L __ ------------------ I recent. At least two sources of evidence indicate that this is a I ~~·serious shortcoming. Moore (1975) reported a followup study to mid-adolescence I . lof chtldren in day care prior to age 5 and children cared for [exclusively by their mothers until school age. Moore found that, I ! although the selection criteria was type of care before age five, I I he had better prediction at age fifteen than at younger ages. Moore I I found a more consistent picture of day care effects in adolescence I than at any previous time in the subjects' I i ves. Moore states that lhis findings support the notion that marginal differences and trends I in early experience often cannot be observed until adolescence when I I their ultimate effect on development can be seen. i I This hypothesis has also received support in the Head Start evaluation literature. Lazar, Hubbell, Murray, Rosche, and Royce (Note 2) have published the results of a long-term followup of a i group of Head Start programs. These programs, which began in the mid-sixties, now have program graduates who are in early adolescence. The programs included in the study were considered exemplary during the late sixties, and represented a variety of teaching methods. Lazaret al. state that the effects of these programs appeared to have faded by the earl~to mid-seventies; however other sources indicate the situation was more complex (Brown, 1978; Kilmer, 1973). Lazaret al. argue that the children have had lasting effects from program participation. Lazar et al. report that program partici- pants had a lower rate of special education in school than matched 1 controls and were less likely to repeat a grade. Although there were I ! L_ _______ -------- --- I no differen~es in IQ between the group~-:--attlt~din~-1 differences w~re-~ 1 I clear. The graduates of Head Start programs exhibited better attitudes toward school, their peers, and their vocational opportunities accord- ing to the measures adopted for the study. The studies by Moore (1975) and Lazaret al. (Note 2) seem to indicate a need for long-term followup studies of the effects of day care. These authors each suggest the possibility of sleeper effects, a finding that would have major implications for many types of evaluation studies. The study of day care effects may be on the verge of increased 1 specificity. Prospective and/or cross-sectional ex-post-facto studies I of the long-term effects of day care should be conducted. In addition, smaller studies should utilize present knowledge to define variables within individual day care settings that will help tie specific pro- gram features to program-related child outcomes. I The present study directly addressed the need for long-term followup studies of day care effects. The outcome variables included the child 1 s effectiveness in peer and adult interactions, as well as several cognitive measures. Following Zigler and Trickett (1978), a number of molar outcome measures were included, such as the existence of an arrest record, incidence of special education, and incidence of I school psychology evaluations. Both the factor structure for SES- 1 related variables and an analysis of SES-related effects were con I sidered to determine the significance of SES in the study sample. I Separate analyses were conducted for males and females because of I frequently found sex differences in studies of day care effects. I I L ___________________________ ---- ---- ,------------------------------------- ----- CHAPTER Ill METHODOLOGY The purpose of this study was to evaluate the differential effects of various types of child care during childhood on cognitive and psychosocial outcomes in adolescence. Type of child care, child health, maternal adjustment, and several SES-related constructs were considered important antecedent and/or mediating variables. The dependent variables were measures of cognitive and psychosocial functioning during mid- to late adolescence. The Danish Perinatal Study and Followup Project provided a unique opportunity to obtain the information needed to conduct this study. The first part of this chapter describes this sample. The second section describes the instrumentation and data collection procedures for the measures utilized in this study. The final section of this chapter describes the data analysis, which consisted of three distinct sets or stages of analyses. The first-stage analysis focused on defining the study variables through scale construction and reliability analysis. The second stage focused on analyzing day care effects through a series of discriminant function analyses of antecedent variables predicting toward each of the out- 1 come measures. The stage-three analysis involved developing a model I of day care effects based on the results of stage two and extant I I literature and testing that model on the Danish Perinatal Followup L ____________________ -··· ·-·-·-·· .. ~::~b ~:~::·~f ~::::::t ::n h::f::::y:::w:::~f~:~~t~::::i :: :::i Jy I instability and child care arrangements. I 1 1 The present study was an extension of the type of longitudinal 1 evaluation strategy that has been used to study the effects of day I I care (Moore, 1975) and early intervention programs (Lazaret al., I 1 Note 2). In addition, this study also developed and tested a model ! of day care effects. Sample The Danish Perinatal Study The subjects in this study were part of a major longitudinal research project known as the Danish Perinatal Study. This study has been described by Zachau-Christensen and Ross (1975) and Mednick, Baker, and Sutton-Smith (1979). The Danish Perinatal Study was begun in Copenhagen, Denmark in 1959 with the intention of analyzing the short- and long-term effects of selected variables related to child development. The variables of interest included maternal, socio- demographic, and family structure measures as well as medical and health data. The original Danish cohort, employed in the Danish Perinatal Study, consisted of 9,125 infants resulting from 8,949 consecutive deliveries (of over 20 weeks gestation) which took place at the State University Hospital (Rigs hospitalet) in Copehagan, Denmark between September, 1959 and December, 1961. Data collection for the Danish Perinatal Study began when i ! each potential mother in the study sample was contacted as early as L ______________________ -·--·--------·- fssl~d~;:T~-p~eg~an-cy for a medic;l~ex-;~T~~tion and ongoing iprenatal care. During delivery, obstetricians, midwives, and midwife I trainees i were present and assisted in data collection related to the I del Ivery and neonatal status of the mothers and their children. In cases where the general condition permitted, the infants were given a thorough physical and neurological examination on the first and fifth days of life. If any type of intervention was called for during the examinations, appropriate treatment was initiated. Upon discharge from the hospital the mothers were given a self-administered developmental progress questionnaire to be filled out during the child's first year. When the children were one year old, their parents were requested to bring them to the hospital for a developmental examination. When the parents were not able to bring the children to the hospital, home visits were arranged. During the first year, information was obtained as to attendance at infant health examinations, intercurrent diseases, admission to hospitals and other institutions, and immunization records. Selected variables from the Danish Perinatal Study are discussed in the section describing instrumentation and data collec- t ion. The Danish Perinatal Followup Study A longitudinal followup study of the Danish Perinatal Study has been conducted for approximately ten percent of the total sample. The Danish Perinatal Followup Study, which is still in progress, has been described by Mednick, Baker, and Sutton-Smith (1979). It is ~: 5 ::: I =~:Y ~ t udy subjects who-camp i'isOd the-Samp if, USOd in the I The Danish Perinatal Followup Study sample(~= 857) appears largely representative of the original sample. Attrition did occur in cases of subject or mother mortality, nonmaternal custody, subject I inaccessbility due to leaving the country, and maternal refusal to I participate. Discussions of the followup study sample attrition and representativeness can be found in Sutton-Smith and Mednick (Note 20) and Mednick and Baker (Note 21). The total Danish perinatal Followup Study was conceived as three separate but closely related studies. These three studies are described below. • Study One was an analysis of the entire Danish Perinatal sample through one year of age. A variety of child, maternal, and familial variables were examined to determine their effects on neonatal and one-year outcome measures for the subjects, includ ing their physical health, motor development, and neurological functioning. • Study Two utilized the ten percent followup sample. The long-term outcomes for the subjects in mid- to late adolescence were analyzed. The data collection for this project occurred from 1976 through 1979. • Study Three used the same subsample as Study Two, but focused on the mother•s health and psychosocial outcomes. Data for Studies Two and Three were collected concurrently. Summary tables describing Studies One, Two, and Three can be found in Appendix A. Although child care was not the primary focus of either the Danish Perinatal Study or the Danish Perinatal Followup Study, data was collected for the followup sample regarding the predominant types 1 of child care at various age levels. Given the paucity of appropriate ! I i L ________________________________ - ·----- -··-------· -·---------- ongltudin~l ___ data on day care effects (Ricciuti, Note 4; Belsky & einberg, 1978), the unique value of 1 Study sample can readily be seen. the Danish Perinatal Followup li I Sample Characteristics I The original sample of 9,125 was not representative of pregnant ! [Danish women. The Danish birth cohort differs from the total Danish I 1 • • 1 • popu at1on 1n severa maJor ways: i • Medical Risk. The original sample contained a large number of older women who were referred to the university hospital setting because of prenatal difficulties or previous obstetrical problems. • Inner City Service Area. The service area of the hospital at which the study was initiated was located in the inner city and, consequently, had a high incidence of lower-SES mothers who were frequently young and unmarried. • High Quality Prenatal Care. Medical screening and intervention began early in pregnancy, continued until the subject 1 s birth, and was of relatively high quality for that era. These sample characteristics have been dicussed in detail in Mednick, Baker and Sutton-Smith (1979). Detailed descriptions of the Danish Perinatal sample 1 s characteristics can be found in Mednick and Baker {Note 21). Generalizability to Other Samples It is apparent from the above description of the Danish Perinatal sample characteristics that the sample cannot be considered a representative sample suitable for a population study. For this reason the Danish Perinatal sample should be considered a selected sample. A number of comparisons have been made between the Danish Perinatal sample and other perinatal study samples. L ----------------~-------- ----------- ~---- Medni~-kand Baker (Note 21) have co~p~-~;d--th~--[)~-;,-i_s_h Perinata I sample to other perinatal studies, in particular the New York City I Infant Death study (Kessner, Singer, Kalk, & Schlesinger, 1973), the I ! British Birth Study (Chamberlain, Chamberlain, Howlett, & Claireaux, I 1975) and the American Collaborative Perinatal Project (Niswander & I l Gordon, 1972). None of these comparisons have demonstrated outstanding :differences among the samples in terms of pre-, peri-, and postnatal child measures, nor among various maternal and sociodemographic variables. Mednick, Baker, and Sutton-Smith (1979) compared the Danish perinatal sample to the white subsample of the American Collaborative Perinatal Project (ACPP) (Niswander & Gordon, 1972). Both studies are prospective, longitudinal studies that drew their subjects from university hospital settings that offered higher-than-average levels of pre- and perinatal care. The ACPP sample was similar to the Danish sample in the elevated frequencies of younger, lower-SES mothers as well as mothers with past or present obstetrical problems. Instrumentation and Data Collection This study utilized data from both the original Danish Perinatal Study and the followup study. The data sources that were used to construct the study variables were: • The Perinatal Data. Information about the child's health was taken from the original data base collected for the entire Danish Perinatal sample from the prenatal period through the end of their first year. • Mothers' Interviews. Interviews were conducted by a social worker with each subject's mother as part of the Danish Peri nata 1 Fo llowup Study. Information from these interviews was used to-consTruct materna 1 adj us-tmerit~-aria-S ES- re 1 a ted variables. The interviews were also used to identify of child care the subjects received. antecedenfl the types ] • • Teacher Questionnaires. Questionnaires were completed by the subject's primary teacher. This questionnaire data was used to construct child outcome variables. Other Data Sources. Data related to the subject's arrest record was derived from a national registry of arrests and convictions. School psychology data was obtained from an examination of school records for those subjects referred to a school psy chologist. The Perinatal Data The original data base for the Danish Perinatal sample was extremely large, and documentation of the data base has generally been published only as it relates to specific studies that utilized the data. Zachau-Christensen and Ross (1975) and the Center for Inter- actional Research in Human Development and Learning (Note 22) provide the most comprehensive descriptions of the perinatal data base. The variables that were utilized for the present study related to the subject's health status at birth. A number of measures were combined to create two variables measuring infant physical health and neurological status. The construction of these variables is described in Appendix B, Part 1. Mother's Interview A structured interview was conducted with the mother of every subject in the final Danish Perinatal Followup Study sample. The interview protocol and a description of the protocol development and interview procedures can be found in Sutton-Smith and Mednick (Note i 20). i i L _______________________ ----- ~------- - 51 I I ~-----Th;-or lgl ~~~--l;;te rv i ew p rotoc~l~nde rw;~-t-~ -d-~ ~~~~p~~~ t and I field-testing cycle prior to the beginning of data collection. I interviewers were qualified social workers who were trained in a i standardized interview procedure. The interviewers conducted two 'pilot interviews with feedback on their performance and met weekly to resolve problems in administration, data coding, etc. The mother•s interview covered a large amount of data related to the subject 1 s health, school history, social behavior and upbring- ing. The interview also included information about the mother•s work history, health, family problems, and other perceptions of the children and family. Typically, interviews were 1-1/2 to 3 hours long. Appendix B, Part 2 contains the items drawn from the mother•s interview that were used in the present study. Individual items from the mother•s interview were considered for inclusion in a number of study variables including type of child care, SES-related variables, and maternal adjustment. In addition to completing the mother•s interview protocols, each social worker completed a series of items in an 11 appendix 11 to the mother•s interviews. These items consisted of the social worker•s ratings of several SES-related family environment variables. Several of these items were included in theSES-related variables. The items can be found in Appendix B, Part 3. The entire social worker•s rating form can be found in Sutton-Smith and Mednick (Note 20). Teacher•s Questionnaire I The purpose of the teacher 1 s questionnaire was to obtain outcome I schoo 11 I data for the subjects related to their academic and psychosocial 521 ·--- ________ _j ' I abilities. A description of the development of the teacher question- naire, the conceptual 11 blueprint, 11 and a copy of the instrument may be found in Mednick and Sutton-Smith (Note 23). The teacher 1 s questionnaires were distributed to each subject 1 s room teacher, who was the teacher most familiar with the child and had generally been with him or her for several years. Teacher ques- tionnaires were only distributed to a teacher after permission had been obtained from the subject 1 s parents. For this reason the total number of teacher questionnaires (~ = 439) was smaller than the number of mother interviews (~ = 857). The teacher 1 s questionnaire was designed, for each of administration, as a simple rating scale. Each item consisted of a word or short phrase descriptor about the child followed by a scale ranging from l ( 11 well below average 11 ) to 5 ( 11 well above average 11 ). In addition, the teachers responded to several questions related to whether or not the subject received special education or was retained. The teacher questionnaire items included in this study can be found in Appendix B, Part 4. Other Data Sources A number of other data sources were utilized to obtain informa- tion about the followup sample. These sources included national registries of mental health and criminality. In addition, school psychology data were obtained for those subjects referred to a school psychologist. L_ -------- ------- --- -- --- 53 ~~---~For--purp~-~~~~of this study on 1 y -t-wo ;~]~~ meas-~~~s were taken !from these sources. One of these was whether or not the subject has an arrest record (taken from the national criminality files). The collection of criminality data has been described in Sutton-Smith and Mednick (Note 20). The other measure taken from these data sources was simply whether or not a school psychology report existed for the subject. This molar measure was a simple means of identifying whether subjects experienced some kind of academic and/or behavioral problems in school. Data Analysis ,The Analytic Strategy I I The analytic strategy was designed both to evaluate the i I I longitudinal effects of day care and, simultaneously, to develop a descriptive model of day care effects. To accomplish this goal three stages of analyses were undertaken: • Stage 1: Variable Definition. A number of individual measures were utilized to construct study variables. Individual data items were combined into scales created on the basis of a post hoc reliability estimate for each scale. • Stage 2: Discriminant Function Analyses. A series of discriminant function analyses were conducted to determine the prediction of antecedent variables toward all outcome variables in each time frame. The results of these analyses were interpreted directly, and used as part of the rationale for a descriptive model of day care effects proposed in Stage 3. • Stage 3: Model Development and Testing. The third-stage analyses involved the development and testing of a descriptive path analytic model of day care effects. An initial model was developed based on the results of Stage 2 analyses and the extant literature. This model was then tested and refined to develop a final model describing day care effects in the Danish Perinatal Followup Study sample. L-~-- --------- ------ ---- ~;g;-1 :-v-;;:-i~b-1-~-o-~1=-i-~it~------ -------~------------ 1 Several scales selected for inclusion in the present study !were defined during the first stage of the analysis. Re 1 i a b i 1 i t y analysis was used to define variables from measures available for the Danish Perinatal Followup Study sample. Variables were constructed by developing equal interval, additive scales related to the subject's SES and psychosocial and cognitive functioning. This follows the recommendation of several studies (Wainer, 1976; Winer, 1978) which note that the low reliability of coefficients in linear models greatly reduces the usefulness of using factor loadings when constructing scales. Each scale was sub- jected to a test of its internal reliability subsequent to construe- tion. Thus, for each constructed scale, an internal consistency reliability analysis, and subsequent modification when necessary, was completed. The variables excluded from the reliability analysis were family income, the child care measures, and the molar outcome measures {such as the existence of an arrest record, participation I in special education, etc.) that were not scales but observable measures. The scales were analyzed using the reliability subprogram of SPSS (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1976). Subsequent to an initial analysis, items were considered for exclusion from the scale if they were detrimental to the homogeneity of the scale as a whole. Item-to-scale correlations and Cronbach's alpha were used as measures of each scale's homogeneity. In those cases where 1 a number of measures empirically contributed to a scale, the scale I I L__ ___________ --- -·· ~-----·- [was limited to the most homogeneous items because additional measu.=-~~-~ I !would no:a:~:t:i~~::::::t::::::::.t:h~:c::::e:e:~:::b:::t:~e present lstudy. Note that sets of variables are related to hypothetical group- 1 ings It was these groups of variables that made up the factors for I . 1 1 the model development analyses. The results section lists the specific, I measures that were included in the reliability analyses and the results of those analyses. Stage 2: Discriminant Function Analyses of Dependent Variables The second stage of the analytic strategy was a series of temporally organized discriminant function analyses of each dependent va r i ab 1 e. In these ana 1 yses the dependent va r i ab 1 es were the chi 1 d care variables and each of the outcome measures. Child care vari- ables were included as dependent variables because the antecedent variables (particularly SES) were considered likely predictors toward them even early in the child 1 s 1 ife. This type of analysis has previously been described by Baker, Mednick, and Brock (in press). Figure 1 depicts the time frames within which the variables were arrayed, and will help to identify exactly which variables were included in the discriminant function analyses. Note that 11 Time 2 11 was the first time frame preceded by (potential) predictor variables. Thus the first discriminant function analysis consisted of SES, child health, and maternal adjustment variables predicting toward child care in Time 2. The categorical groupings of the child care 1: generated. L-------~----------------------- ----- I I I 561 ----·---~ variable served as the groups for which discriminant functions were Table 5 describes the seventeen discriminant function ~---~-------~-- -~ I Table 4 1 1---------------v_a_r_i_a_b_l_e_s __ D_e_f_i_n_e_d __ D_u_r_i_n_g __ s_t_a_g_e __ o_n_e __ A_n_a_l_y_s_e_s ______________ ~ Variable Defined Stand 11 Molar'' 1 1 by Rel iabi 1 ity Alone Outcome 1 I Variable Analysis Variables Variables 1 I, _____ S_u_b_J_.e_c_t_'_s __ s_e_x------------------------------------X------------~~ 1 I I SES-related variables: I Family income X Family instability X I Family Dissension X Mother's neighborhood X rating Maternal orderliness Maternal contentment Child's health at birth: Neonatal physical health Neurological status Child care (predominant): Zero to three Three to six Six to eleven Twe 1 ve and over Psychosocial status in adolescence: Peer interaction Adult interaction Parent interaction Aggressiveness Arrest record Cognitive status in adolescence: Verbal skills Reading achievement Math achievement Work organization Grade retention Special education Remedial instruction School psychology report X X X X X X X X X X X X X X X X X X X X X I L-~~---~-~------------~-- -------- 571 --- _______ "__j ~- 1 I SES Related Vari•bles ::~:;;~' j ment ( Child ~ Health { Time 1 Beginning at Birth Family Income family 0 i ssen- s ion Materna I Order I i ness F and ly 1 nsta- b iIi ty Materna I Neighbor- hood Rating Materna I Content ment L----------------- Neonat a I Physical Neuroloqical Health Status I [] Time 2 Ages 0 to -- Day Care 0 to 3 Time 3 Ages 3 to --- Duy Lare } to L Time 4 Ages 6 to 11 --- Day Care b to 11 Figure 1. Time Frames for Variables in the Discriminant Function Analyses. Time 5 Ages 12 Plus -- Day Care 12 PI us Peer Inter- act ion Aggres- s i ve- ness Verbal Ski lis Work Organ i .... zat ion Remedial Ins true- t ion Time 6 Mid- to late Ado 1 escence Parent Inter- act ion Psychosoc i a I Variables Reading Achieve- ment Grade Ret en- t ion Schoo I Psych- ology Report Cognitive Relat~d Variables Adult Inter- act ion Arrest Record Plath Achieve- ment Special Educa- t ion ! I I Table 5 Stage Two Discriminant Function Analyses Analysis Dependent Variable Predictor Variables 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Notes: Chi 1 d Care 1 (zero to three) Chi 1 d care 1 (three to six) Chi 1 d care 1 (six to eleven) Chi 1 d care 1 (twelve and over) Peer interaction 3 Adult interaction 3 Parent interaction 3 Aggressiveness 3 Arrest record 4 Verbal ski lls 3 Reading achievement 3 Math achievement 3 Work organization 3 d . 4 Gra e retention h Special education R d . 1 . . 4 eme 1a 1nstruct1on 4 School psychology report Subject's sex, family income, family instability, family dissension, Mother's neighborhood rating, Maternal orderliness, Maternal contentment, Neonatal physical health, Neurological status Same as number 1 plus child care 0-3 2 Same as number 2 plus child care 3-6 2 Same as number 3 plus child care 6-11 2 Same as number 4 plus child 2 care ages twelve and over - The categorical dependent variable defined the groups 2- Various categories of child care are dummy coded 3- The continuous dependent variable was recoded for maximum separation into high and low scoring groups 4 - Dichotomous variable ~----·---~----~-------------------- ·----------- ,-----·----·-- -----·---· lyses necessary to analyze the effects of all antecedent variables on each dependent va r i ab 1 e. In addition to inc 1 us ion of the common set of predictors in each analysis, each subsequent analysis also included all of the outcome variables from the preceding time frames as predictor variables. Table 5 also summarizes the coding that was used during the discriminant function analyses. When the categorical child care variables defined the outcome groups no recording was necessary. When the child care variables were included in an analysis as pre- dieters, the categories of interest were dummy coded. The continuous variables that were used to define outcome groups were receded into two separate groups. Approximately the top and bottom 40% defined groups of high and low scorers for each outcome variable. For this reason, the subjects for each of the analyses consisted of a different subsample. Those subjects who were considered high scorers on one variable were not necessarily the same as the high scorers on another variable. This grouping procedure was utilized so that relationships would be apparent that might otherwise have been confounded within total sample variance. The results of the Stage 2 analyses were analyzed in two ways. First, the results of the discriminant function analyses were inter- preted directly with regard to the antecedents and outcomes of several categories of child care. Second, the results were considered in the model development stage described below. Stage 3: Model Development and Testing 'i In recent years there has been an increased interest in path _I ------ -----·- ---- - --·- - ,---------·-------- ------------- lysis and modeling procedures applied to the social sciences (Bentler, 1980). This section describes the development of a descrip- tive causal model of day care effects. Model Development. An initial model of day care effects was specified on the basis of the discriminant function analyses and the review of the literature. The variables and time frames for the model remained similar to those indicated in Figure 1, except that the child care variables for ages 6 to 11 and 12 and over were dropped because they were not directly relevant to full-time day care. A causal path was considered for inclusion in the initial model in those cases where a variable successfully predicted to another variable in a later time frame, or the 1 iterature suggested a causal relationship. In specifying the model it was helpful that, in spite of its lack of comprehensive or mathematically specified theories, the day care literature is rich in suggestions of testable hypotheses. SES is an example of a variable that, based on the 1 iterature (see Ramey and Mills, 1974; Robertson, 1978), would be expected to have both a direct effect on child outcomes and an indirect effect (via child care) to certain outcomes. Thus the relationships found in Stage 2 and the literature were used to make an approximation of a model of day care effects. It is clear that a model developed in this manner can only be considered speculative for at least two reasons. First, the discrim- inant analyses were based on different subpopulations defined by the outcome variables. High- and low-scoring groups on one outcome measure would not necessarily be expected to be high- and low-scorers on other 6 I out comes . Simi 1 a r 1 y, paths p red i c ted--~n -th~-b~~-i-s -;-f-the lwere derived from other samples and child care systems. However, this I initial model served as the starting point from which a descriptive model was refined and tested for the Danish Perinatal Followup Study sample. In complex models such as the one presented here, the distinction between independent and dependent variables frequently is blurred. In the present model the dependent variables (child care, psychosocial-related outcomes, and cognitive outcomes) were firmly embedded in a known time frame (those of Figure 1). The independent variables had an undetermined time of impact and, in many cases, probably had ongoing impact. The independent variables were con- sequently entered first. They are shown in the left-hand column of Figure 1. While this solution is not entirely satisfactory, particu- larly because several of the variables are lagged, it does follow the ''least is last" recommendation of Cohen and Cohen (1975) for multiple regression procedures. With the time frames clearly established, the utility of the sequentially organized discriminant function analyses for model development becomes apparent. Model Testing and Refinement. The procedural framework utilized in the study was that of J5reskog and Sorbom (1978). They developed a model for the analysis of linear structural equations via a maximum 1 ikelihood technique. The computer program (LISREL IV) is capable of handling both measured and unmeasured (or latent) variables and is sufficiently versatile to permit some degree of j model manipulation to improve the "fit" of a model to a sample I I L_ _____________________ ------· --------. ]{Maruyama & McGarvey, 1980). This capability allows exploratory !building and testing strategies to be superimposed upon what is I essentially a confirmatory factor analytic procedure. Application of path analytic procedures involves initial development (described above) of a generalized model to be tested by the available data. Ideally, the model should be developed on the basis of existing theory. However, although the published literature is voluminous, it is not mathematically specified and the model was generated in part by the inquiry itself, i.e., through the sequence of preceding analyses leading to the initial specification of a day care model. The model was analyzed for the Danish Perinatal Followup Study sample using a missing data correlation matrix as defined by Cohen and Cohen (1975) with pair-wise deletion of missing data. Tests for Goodness of Fit. Several statistics and measures were utilized to determine the extent to which the specified model was an accurate descriptor for the input data. These indicators have been described by McGarvey (Note 24) and Bentler and Bonett (1980). In this study, the measures of goodness of fit were a likelihood ratio chi square test (testing the goodness of fit for the entire model) and the standard errors of individual parameter specifications I (for the significance of individual path coefficients). The closer !the chi square value was to the degrees of freedom for the model the I !better the indication of a good fit, i.e., the ratio of the chi square lvalue to the degrees of freedom was minimized. i I I An examination of the standard errors of individual path I I coefficients revealed those that were statistically nonsignificant ! I L _______________________________ --- land considered to detract from the 1 subsequent tests of the model these avera 11 fit of the mode 1. In nonsignificant parameters were dropped in an effort to improve the X 2 /df ratio. Two methods were utilized that have been suggested for the rational addition of new paths to models, both involving information derived from the LISREL IV solution. The residuals of the solution were examined to determine the specifications of the model that were poorly fitted to the data (Bentler, 1980), and the first order derivatives of the solution also provided information about specific patterns of lack of fit (Sorbom, 1975). Model modifications were undertaken cautiously since there are a number of pitfalls possible in such a post hoc procedure. When modifying a model to improve its fit to the study sample, paths may be added or deleted. The elimination of paths increases the degrees of freedom for the model. The addition of paths may decrease the overall chi square when that addition improves the fit of the model. Thus, adding and deleting paths can contribute to minimization of the X 2 /df ratio, which is the best overall measure of an improved model. The primary danger involved in model modification is one of overfitting the model to the sample, that is, developing a model that precisely describes random variations in a sample. The path coefficients computed in this analysis were unstandardized, since the use of a correlation matrix, as opposed to the variance-covariance matrix, effectively standardized the variables as a function of collapsing the variance to unity. ~ I Hoc Analyses Several analyses were performed subsequent to the model development and testing stage. These analyses were conducted to screen for possible interaction or sensitization effects that may have had a bearing on interpretation of the Stage 2 or Stage 3 analyses. Both family income by care and family instability by care interactions were examined for all outcome measures. In no cases was an interaction effect found. r--------------------------- ---------- 1 CHAPTER IV RESULTS This chapter presents the results of the analyses described in the methodology section. The results are presented separately for each analytic stage. Stage 1: Variable Definition Twelve of the twenty-six variables included in the present study were derived scales and were subjected to a reliability analysis. One of the scales, the quality of the family's neighborhood, was not reliable and was dropped. A single data item, the mother's judgment of the quality of the neighborhood, was used. The remaining fourteen variables (the subject's sex, family income, family instability, neonatal physical health status, neonatal neurological status, four measures of the type of child care, existence of an arrest record, grade retention, receipt of special education, receipt of tutoring, and referral to a school psychologist) were single, observed measures. This section presents the reliability analysis results for each of the eleven derived variables. Individual scale items were combined to create a reliable indicator of each subject's functioning on a particular factor or variable of interest. Tables 6 through 8 present each of the scales 1making up the derived variables. Each measure is listed and a measure I ' L __________ ----- --------- -----· I i I Variable Family Dissension -l Table 6 Variables Derived from Items in the Mother's Interview Schedule Cronbach's Alpha . 730 Items Item Descriptors Vl70 Disagreement about child- rearing Vl71 Disagreement about sharing housework Vl72 Disagreement about economy Vl73 Problems regarding jealousy Vl74 Sexual problems Vl75 Problems with in-laws Vl76 Disagreements about spare- time activities Vl77 Other problems Note: Actual data items may be found in Appendix B, Part 2. L _____________ ---------------------- ------ ~?- j I I I ! I Variable Mother•s Contentment Mother• s Orderliness Table 7 Variables Derived from Items in the Social Worker•s Appendix to the Mother•s Interview Cronbach 1 s Alpha .712 .787 Items V274 V281 V282 V283 V284 V262 V263 V265 V267 I tern Descriptors Mother 1 s attitude toward chi 1 d Mother•s over a 11 content- ment Acceptance of situation Mother•s apparent isolation Father•s apparent isolation Home furnishings Home appearance Mother 1 s dress Mother•s grooming Note: Data items may be found in Appendix B, Part 3. L-~------------------ -----------------~------ -- - - - - -------- - -------~ ~ ~--~---~-----~~ Table 8 -, ! I Variables Derived from Items in the Teacher Questionnaire 1------------------------------------------------------- Cronbach 1 s Variable Alpha Peer Interaction .750 I Items 45 48 49 50 51 52 53 Item Descriptors Teased often Acts immature Prefers younger children Changes friends frequently Has many friends Popular with peers Leader among peers !Adult Interaction .978 24 25 Interaction with adults Interaction with teachers I Parent Interaction .720 Aggression .806 Verbal Skills .790 Reading Achievement .899 Math Achievement .953 Work Organization .880 73 74 75 76 78 40 57 58 I 2 6 7 8 10 I I I 3 14 15 16 20 Good parent relations Critical parents Parents too strict Fearful of parents Parents overindulgent Violent outbursts Bullies other children Frequently fights Understands verbal material Oral expression Vocabu I a ry Reading comprehension Reading speed Understands math concepts Computation skills Works to ability Work quality [ Work well organized / Neatness Functioning in unstructured/' situation --------------1 Note: Data items may be found in Appendix B, Part 4. I I i I I I 691 ~~~ ~~-~-~- ~:_j internal consistency reliability (Cronbach 1 s Alpha) is included I I each scale. Stage 2: Discriminant Function Analyses of Dependent Variables Predicting Toward Type of Child Care Four discriminant function analyses (DFA) were conducted to predict to the major types of child care in four age groupings: 0 to 3, 3 to 6, 6 to 11, and 12 years and older. The groups for each discriminant function analysis were defined as the various types of child care, providing a sufficient number of subjects were within a group to permit a meaningful analysis. A cutoff sample size of forty subjects within a group (prior to listwise deletion of cases with missing data) was used to determine which types of child care would be included. Five distinct child care variables were identified at each age level, although the five variables differed among age groups. For each age level the five outcome groups will be described, including the size of each group after listwise deletion. Each analysis will then be described, including the number of significant discriminant functions, and an interpretation of the functions predict- ing to each type of care. For five groups (the number of outcome groups at each age level) the number of possible discriminant functions is four. However, it may not have been possible to discriminate among all five groups, in which case fewer functions were generated during the analysis. These functions discriminated among subsets of the total number of outcome groups. For example, it may be possible to discriminate successfully between home care and three types of out-of-home care, but not be possible to successfully predict to any of the three out- f ---------~--~----~------- ------------------ ---------- lof-home care situations. This situation occurred when predicting I toward child care from 0 to 3. I, I I When multiple functions are generated (as in the remaining .DFAs to child care) interpretation can become difficult. Combinations ~of functions predict to the child care groups. For example, it may be that membership in one group is successfully predicted by a high score on one function and a low score on another. For this reason, all functions must be considered in the interpretation, and tables have been included that summarize the DFA completely. At the bottom of each of the tables summarizing a DFA there is a matrix indicating the interpretation of positive and negative scores for a particular function. In a sense each function can be interpreted as a factor related to the outcome variable. A particular outcome may be predicted by either a positive or negative function score. Predicting Toward Child Care, 0 to 3. Table 9 summarizes the results of the DFA predicting toward type of child care from ages 0 through 3. The five outcome groups were: maternal care (~ = 347), full-time day care (~ = 79), a day nurse(~= 41), grandmother (~ = 55), and a full-time in-home babysitter(~= 32). Although five outcome groups were included in the analysis, only two functions were generated that had canonical correlations greater than . 10, and only four variables were successful predictors. Function one discriminated between home and out-of-home care. Subjects !with high family income, family stability, neonatal neurological I I L---~----- ----------------- ------- 1 -~----~-~-----~~ --------~---~- ~~ --~- ---- Table 9 Discriminant Function Analysis Predicting Toward Type 1 of Child Care Ages 0 to 3 (Wilks 1 A= .83) I I I Standardized Discriminant Function Coefficients Predictor Function Function 2 Subject 1 s sex 7\ .. k Fam i 1 y income .48 .86 Family instability -.60 .69 Neonatal physical health ;~ ;'' Neonatal neurological .24 -.04 status Neighborhood rating ;"; ·k Materna 1 order] iness ;f.:: i'\ Maternal contentment ·-k .. ;';_ Family dissension -.29 -.01 Canonical correlation .33 .24 Note: An asterisk indicates a variable did not contribute to the function and was dropped. Group Membership Maternal care Full-time day care Day nurse Grandmother In-home sitter Function Score Function Function 2 + + + Note: A blank cell indicates a function did not predict membership for that row. r :oblems, and-low--famiJYdissension tended-tor-emain at--home; they re cared for by their mother or an in-home sitter. Subjects in -, t-of-home care tended to have the opposite characteristics. I I The second function discriminated between maternal and sitter lcare, and the major discriminating variable was family instability. Subjects in more stable families tended to receive maternal care. Overall, the prediction to maternal and sitter care from 0 to 3 was good but it was not possible to discriminate between three types of out-of-home care. No differences existed among the day care, grandmother care, and other adult care groups on the predictor vari- ables. Furthermore, the subject's sex, the neighborhood rating, and two maternal variables did not predict toward type of child care under age 3. Predicting Toward Child Care, 3 to 6. The five child care groupings for this analysis were: maternal care (~ = 287), full-time day care (n = 140), half-time day care (n = 61), father care (~ = 26), and grandmother care (~ = 31). Four functions, shown in Table 10, were generated by this analysis. Overall prediction to type of child care at the 3 to 6 age level was quite good (Wilks' A= .47). Maternal care was predicted for subjects with negative scores on both functions one and two. These subjects received maternal care, day care or day nurse care from 0 to 3, had stable families, and had good neonatal health. The day care group was similar, except that they exhibited high family instability and absence of maternal care during the 0 to 3 age range, as indicated by positive scores on function one. Day care i I I 731 ----~ I r-------------------- Table 10 Predictor Discriminant Function Analysis Predicting Toward Type of Child Care Ages 3 Through 6 (Wilks' A= .47) Standardized Discriminant Function Coefficients Function Function 2 Function 3 Function 4 Subject's sex ;'~ : Fam i 1 y income -.07 • 41 -.09 -.17 -.24 .08 . 10 . 51 -.44 ! Family instability Neonatal physical health Neonatal neurological Neighborhood rating Maternal orderliness Maternal contentment Family dissension 0 to 3: Mother care Full-time day care Day nurse Grandmother In-home sitter Canonical Correlation • 19 • 13 .05 -.83 .06 • 12 -. 19 -.07 .59 -.29 -.24 -.05 .16 -. 14 -1 .05 -1 . 01 -. 73 .00 -. 31 .43 -.17 -.18 .04 .94 .55 .68 l. 15 .03 .28 -.02 * .01 -.114 -. 12 . 39 .69 . 36 .68 .55 . 18 Note: An asterisk indicates a variable did not contribute to the function and was dropped. Function Score Group Membership 2 3 4 Maternal care Full-time day care + Half-time day care + + Father care + + Grandmother care + + + Note: A blank cell indicates a function did not predict membership for that row. L __ ---------------------------------- ,--------------- ·--···------·--- ifrom 0 to 3 did not predict to day care from 3 to 6. i Part-time day care from 3 to 6 was predicted by positive 1 scores on functions two and four and a negative score on function I three. These children had stable families, good neonatal health, a mixed pattern of previous child care, and relatively high family income. Care by the father was predicted by positive scores on functions one and two, and negative scores on functions three and four. This group was typified by a lack of previous care by the subject's mother or grandmother. There was also low maternal contentment and high family instability in this group. Care by the grandmother from ages 3 to 6 was predicted by positive scores on all four functions. The most powerful predictor was previous care by the grandmother. This group also showed high family instabi 1 ity. Predicting Toward Child Care, 6 to 11. The prediction to type of child care for subjects 6 to 11 years old was very good (Wilks' A= .46). The five outcome groups were mother care(~= 346), father care (~ = 46), grandmother care (~ = 21), another adult out side the home (n = 38), and an after-school center(~= 95). The results of this analysis are shown in Table 11. These groups described the predominant child care arrangements for each subject after regular school hours. Function one predicted to maternal care, a positive score indicating a higher likelihood of maternal care. This group had L ____ -------- --- ----------- r-----~----------------~--~-------------- ---------- 1 Table 11 I I I I Discriminant Function Analysis Predicting Toward Type of Child Care Ages 6 to 11 (Wilks' A.= .43) Standardized Discriminant Function Coefficients Predictor Function Function 2 Function 3 Function Subject's sex "k ;'~ ·k ;'~ Fam i 1 y income . 21 -.21 .00 . 15 Family instability -.49 .02 .20 .22 Neonatal physical health ·;':. ;~ ;~ ;'\ Neonatal neurological .05 -. 12 . 14 -.20 Neighborhood rating ;~ * ;'\ ;'\ Maternal orderliness ;': ;'\ ;'c ;'c Maternal contentment -.13 . 15 . 16 .04 Family dissension .06 -.18 . 38 -.16 0 to 3: Mother care . 17 -.37 .03 .00 Full-time day care . 11 -.25 . 13 -.37 Day nurse -.08 -. 11 .04 -.30 Grandmother ·k ;'(. ·k. ;': In-home sitter -.03 -.25 .62 -.13 3 to 6: Mother care .05 -.37 -.91 -.48 Full-time day care -.70 -.53 -.64 -. 17 Half-time day care -.02 -.21 -.41 -.01 Father care -.14 .44 -.27 -.74 Grandmother -.17 . 36 -.38 . 34 Canonical Corre 1 at ion .59 .46 .34 .27 Note: An asterisk indicates a variable did not contribute to the function and was dropped. Group Membership Materna 1 care Father care Grandmother care Another adult outside the home After school center + + Function Score 2 + 3 + + 4 i Note: I A blank cell indicates a function did not predict membership for that row. l _____ -~--~-------------- 4 ~;family-~stability, was not in day care fro~-3-t~-6~--and had ---~ I relatively higher family income. I The group experiencing after-school day care was predicted by 1 1 negative scores on functions one and two. These subjects had high I family instability, were in day care from 3 to 6, and tended toward I lhigh family dissension and low family income. The father care group had positive scores on function one and negative scores on function four. These children were frequently cared for by family members (other than the mother) rather than child care arrangements outside the family. Previous care by the father I lor the grandmother predicted toward group membership. There \'las a slight tendency toward low family dissension in this group. Positive scores on functions two and three and a negative score on function one predicted toward care by the child's grandmother from 6 to 11. This group had high family instability, experienced day care from 3 to 6, did not have substantial maternal care from 1 o to 6, and had received previous care by the grandmother. There lwas some indication of high family dissension and father absence in the grandmother care group. Prediction toward care by an adult outside the family was , indicated by a positive score on function three. This group was I I characterized by high family dissension and a mixed pattern of !child care at previous age levels, particularly child care alterna- 1 tives other than maternal care or organized day care. I Predicting Toward Child Care, Over Age 11. Prediction to child i i care I I after age 11 was very good (Wilks' A= .30). L------------~---- -------- -------- ------ The four outcome groups were: maternal care (~ = 293), father care (n 38), after school center (~ = 40), older siblings (~ = 34), and no caretaker designated after school (~ = 147). The results of this analysis are shown in Table 12. The mother care group was described by a negative score on function one, which can be interpreted essentially as mother care after school from ages 6 to 11. Father care, too, was predicted almost exclusively by father care during the previous time frame, indicated by a positive score on function two. Care at an after-school center was predicted for subjects with a positive score on function one, and negative scores on functions two and three. These subjects may have had maternal or father care during the 6 to 11 age range, but did not receive care from an unrelated adult outside the home (except in after-school centers). The best predictor of care in an after-school center after age eleven was center care during the 6 to 11 age range. Two types of child care became sufficiently frequent after age eleven to be included in this analysis: care by older siblings after school, and the absence of any caretaker after school. The older sibling care group was predicted by positive scores on functions three, one, and four; and a negative score on function two. This group experienced all types of child care from 6 to 11, and many appear to have been cared for by older siblings from 6 to 11, although the number of subjects cared for by siblings in that age group did not permit their inclusion in the DFA. Participation in full-time day ' 1 care from 3 to 6 predicted to care by older siblings as did maternal i L-----~---~- -----~-~--~~-- ---- ------ -- -~---- ! I r-- 1 I I I I i Table 12 Uiscriminant Function Analysis Predicting Toward Type of Child Care Over Age 11 (Wilks' A= .30) Standardized Discriminant Function Coefficients Predictor Function Function 2 Function 3 Function Subject's sex .06 .09 .01 -. 16 Family income -.10 -.06 .00 -.09 Family instability .20 -.07 -. 11 -.54 Neonata 1 physical health .... , ·'· ·'· .L Neonatal neurological ·'· ~·, * .. Neighborhood rating ·'· of: .... ,. .L Maternal orderliness ~ ... " -.'t '" Materna 1 contentment .01 -. 01 . 18 . 19 Family dissension .18 .01 .08 .08 0 to 3: Mother care .03 -.06 .25 .48 Full-time day care . 15 -. 15 .04 .27 Day nurse . 10 -.09 .07 .33 Grandmother .08 -' 10 .27 . 10 In-home sitter ·'· ·'· ;\ ·'· 3 to 6: Mother care -.07 -.07 . 10 .48 Full-time day care -.17 -.05 .05 .75 Half-time day care .03 -. 13 .00 .28 Father care -.03 .07 .09 . 19 Grandmother ·'· 'i~ ·'· .... , 6 to 11: Mother care -.82 . 13 -1.21 -. 32 Father care -.09 .96 -.78 . 01 Grandmother -.22 .11 -.42 -.28 Another adult .02 . 17 -. 17 -.33 After school center . 19 -. 14 -1.22 -.09 Canonical Correlation .64 .60 .42 .22 Note: An asterisk indicates a variable did not contribute to the function and was dropped. Function Score Group Membership 2 3 Maternal care Father care + After school center + Older siblings + + No caretaker + + Note: A blank cell indicates a function did not predict membership for that row. 4 + L_-------~--- 4 rca~----~-~---------~-------~-~~--------- ~~-~~- ------------- - I_ re from 0 to 6. The older sibling care group was also predicted b-y-l 'family stability. The no caretaker group was predicted by positive scores on functions one and three and a negative score on function four. This group had mixed care in the 6 to 11 age range. They also were associated with high family instability and the absence of maternal care or organized day care during early childhood. The successful prediction in this analysis of child care arrangements after age 11 was accomplished almost exclusively by child care arrangements from 6 to 11. Child care at earlier ages and SES-related variables were only important in predicting toward care by older siblings and the absence of supervision. Family income and maternal variables did not predict child care after age 11. Predicting Toward Psychosocial Related Variables Five outcome variables related directly to the psychosocial functioning of the subjects in adolescence. High- and low-scorers lwere used to define two groups for discriminant function analysis of leach outcome var"1able. Th f" · bl · e 1ve outcome var1a es were peer Inter- action, parent interaction, aggressiveness, and the existence of an arrest record. Table 13 summarized the DFA for the five psychosocial variables. Each column represents a different analysis, with column entries equal to the discriminant function coefficients for the final solution. In the discussion that follows each column (representing prediction toward an outcome variable) will be discussed separately. ! iThe meaning of positive and negative scores on the functions is defined I L--~-------~-~--- ~-~~--~-- ----- --~-~-- ~-~~- ~-------- ------~------------------ -- Table 13 Discriminant Function Analyses Predicting Toward Psychosocial Outcomes in Adolescence Standardized Discriminant Function Coefficients Predictor Subject's sex Family income Family instability Neonatal physical health Neonatal neurological Neighborhood rating Maternal orderliness Maternal contentment Family dissension 0 to 3: Mother care Ful 1-time day care Day nurse Grandmother In-home sitter 3 to 6: Mother care Full-time day care Half-time day care Father care Grandmother 6 to 11: Mother care Father care Grandmother Another adult After school center Over 11: Mother care Father care After school center 0 1 de r sib 1 in gs No one Wi 1 ks' Canonical C6rrelation Function score to indicate positive outcome --------- ---- Peer Interaction ·'· - 32 32 ·'· ·'· ·'· ·' ·'· ·'· •'· " ·'- ·'· " ·'· ·'· - 51 ·'· " -·· ·'· ·'· - .28 ,., ·'· ·'· " - .64 ;';. .90 31 Adult Interaction - 23 "· .63 ;';; ·'· ·'· ·'· " •'• ·'· " ·'· ·'· - 37 - 33 ·'· - .42 ·'· ·'- ·'· ·'· ·' ·'· " " - 32 ;'r ·'· ·'· .90 31 Parent Interaction ·'· 36 - 18 -.48 - .24 -.55 ·'· ·'· ·'· ~·< ·'· 30 ·'· ·'· "· - 14 .24 ·'· ·'· .... , .29 ;'r " ·'- ·'· ·'· - .27 ·'· -.22 .83 . 41 + Aggression ··~ - 16 .49 .27 .24 ·'· ·'· ·' ·'· ·'· " ·'· " - .24 ·'· " ~·, ·'· ·'· ,., ·'· .24 - 35 ·'· ·'· ...... 32 "· ;';. .90 31 Arrest Record - .54 ·'· 39 .24 " ·'· - 23 ·'· .23 ·'· ·'· ·'· ·'· ·'· ·'· .07 ·'- ·'· " ·'· - 1 1 ·'· " . 03 - .55 - .23 ;'r. ·'· -.44 .83 41 ze rc. Note: An asterisk indicates a variable did not contribute to the function and was dropped. ! L ____________ --------------------- -------- 8 --~----- --------~---------- -- ---------- - at the bottom of Table 13 and will be described in the text. The outcome groups will be described in terms of the variables that successfully predict group membership. In most cases either high- or I low-scorers will be described, but not both groups. The description of the contrast group is exactly the opposite when only one function is used for prediction. Thus, if the high-scoring group is predicted by family stability, the low-scoring group is predicted by family instability. Repetitive descriptions have generally not been included except in those cases where a finding is of special interest. In several cases, child care variables were better predictors than was indicated by the final equation. In those cases the child care variable was entered in the equation later than would have been done had the sole criterion been predictive power. The order of entry for all variables was controlled to remain consistent with their actual temporal sequence. However, in those cases where a child care variable had greater predictive power than the final solution indicated, this finding will be presented. Overall, the prediction of high and low scores on the psycho- social variables was not good. The lowest Wilks' A was .83 for parent interaction and child criminality. The other three AS were .9, , indicating relatively poor predictive power. Predicting Toward Peer Interaction. The prediction to high- and low-scorers on peer relations was poor (Wilks' A= .9) but significant. The number of high-scorers (good peer relations) was 134, the number of low-scorers was 107. I - ~--··- ----- -------------------------- sc~.=-el Successful predictors of good peer relations (a negative for the function), were the types of child care that exposed the chi 1 d to other chi 1 dren. The single best predictor was care by older siblings after age 11. However, the second best predictor was half- time day care from 3 to 6. Those with good peer relations were cared for outside the home from 6 to 11, and tended to have more stable families and higher family income. Predicting Toward Adult Interaction. The number of subjects in the good adult relations group was 145, the number of poor relaters was 100. Negative scores for the function predicted toward good adult relations, positive scores to poor adult relations. Subjects with good adult relations tended to have stable families, and more individual child care by adults and parents. Half- time day care from 3 to 6 predicted toward good adult relations, and females had better adult interactions than males. Poor adult relaters had unstable families and were cared for outside the home. Both full-time day care from 0 to 3 and an after- school center from 6 to 11 predicted well toward poor adult relations, but were not included in the final equation. Predicting Toward Parent Relations. A positive score on this discriminant function predicted toward the group with good parent interaction (~ = 109), while a negative score indicated poor parent relations (n = 90). The prediction was good to this outcome measure (Wilks 1 \ = .83). Children who were rated as having poor parent relations had I <a worse neighborhood rating by their mother, had more neonatal physical i I ! L __ ~ ----- --~~~-~------~--------~- ----~ 1 and neuro log i ca 1 prob 1 ems. had lower fam i 1 y i nco~es~- and were not cared for by a day nurse from 0 to 3 or the father from 6 to 11. Many of the children in this group were in after-school centers or unsuper- vised after age 11. While full-time day care from 3 to 6 was a good predictor of poor parent relations in adolescence, its discriminant function coefficient was small in the final solution. The subjects with good parental relations had better neighbor- hood ratings, fewer neonatal problems, higher family incomes, and were often cared for by a day nurse from 0 to 3, by a half-time day care center from 3 to 6, or by the father from 6 to 11. Predicting Toward Aggressiveness. High-scorers on this function (~ = 136) were rated aggressive, while low-scorers (n = 97) were not. The Wilks' A for the final function was .9. Subjects who were rated aggressive were characterized by family instability, less frequent care by an adult outside the home ages 6 to 11, and more frequent use of an after-school center after age 11. The aggressive group also tended to have more physical and neurological problems at birth. The low aggression group had stable families, more care by an adult outside the home from 6 to 11, less care in an after-school center over age 11, and fewer neonatal health problems. Full-time day care from birth through 6 years and care in an after school center from 6 to 11 predicted toward the aggressive group, but were not included in the final solution. Predicting Toward an Arrest Record. Prediction toward the igroup of subjects with an arrest record was good (Wilks' A= .83). i I I arrest group (~ = 43) had high positive scores on the discriminant function, while the nonarrest group (~ = 555) had scores that were negative and close to zero. The subjects with an arrest record did not receive maternal care after age 11, but did receive some type of care (i.e., the presenc of supervision predicted toward criminality). The group with an arrest record had higher family instability, and were male. They had higher family dissension, more neonatal physical health problems, and their mothers rated lower on orderliness. Attendance at an after-school center over age ll predicted toward an arrest record, but was not included in the final function. Predicting Toward Cognitive Related Variables A series of Discriminant Function Analyses were also conducted to predict high- and low-scorers on eight variables related to the subjects• cognitive achievement in adolescence. Prediction toward cognitive outcomes was, in general, more accurate than prediction of psychosocial outcomes. The Wilks• AS for the cognitive variables ranged from .76 to .91, with better prediction for the teacher rating scales than the molar outcome measures. The results of all the discriminant function analyses for cognitive outcomes are shown in Table 14. In the discussion that follows each of the cognitive outcomes will be discussed separately, including the size of the outcome groups and the variables predicting toward high and 1 ow scores. Predicting Toward Verbal Skills. Subjects predicted to have Iabove average verbal skills (~ = 126) had negative scores for the I ! I i L_ ___ ------------~---- -------- 851 ------- _J Predictor Subject's sex Family income Family i nstabi 1 i ty Neonatal physical health Neonatal neurological Neighborhood rating Maternal orderliness Maternal contentment Family dissension 0 to 3: Mother care Full-time day care Day Nurse Grandmother In-home sitter 3 to 6: Mother care Full-time day care Half-time day care Father care Grandmother 6 to 11: Mother care Father care Grandmother Another adult After school center Over 11: Mother care Father care After school center Older siblings No one Wilks' Canonical Correlation Function score to indicate positive outcome Table 14 ·-·-·--~-----, Discriminant Function Analyses Predicting Toward Cognitive OtJtcomes in Adolescence [ Verbal Skills -.16 -.39 .45 ~·( -.2 -. 41 -.38 -.28 .,., -.38 -. 31 ;'( -.26 .,., .29 .86 .38 Reading Achievement .62 .42 .26 .02 -.28 • 21 .19 .26 -.39 -.45 -.27 . 81 .43 + Standardized Discriminant Function Coefficients Math Achievement -.28 .74 -.19 .39 .43 .27 .22 .,., .32 -:, -.20 -.2 . 76 .49 + Work Organization -.53 -.33 .43 ~·, . 19 -.25 -.26 . 17 .83 .42 Special Education .47 -.44 .57 .29 -.22 -.24 . 18 .9 .32 + Grade Retention -.36 .49 -.45 .32 .22 .25 .34 ;': -.43 -.36 .9 .32 zero Remedial Instruction . 16 .34 -.34 -.37 .25 .43 . 10 .35 .30 . 15 .20 .35 .56 :;'\ -.39 -.39 -.27 . 19 .79 .46 + School Psychology Report -.37 -.23 .53 -.24 -.22 -.18 -:. '~ -. 17 .37 .08 ··~ .40 .27 .28 .28 .001 .91 .30 I I I Note: An asterisk indicates a variable did not contribute to the function and was dropped. _j discriminant function, while subjects with positive scores had poorer verbal skills (~ = 119). Wilks' A for the function was .86. Above average verbal skills were predicted for subjects with stable families, poor maternal neighborhood ratings, high family income, high maternal contentment, and a high incidence of half-time ,day care from 3 to 6. Predicting Toward Reading Achievement. Prediction to reading achievement was good (Wilks' A= .81). The group with good reading skills (~ = 130) scored positively on the function, while poor readers scored negatively (~ = 146). The best predictor of reading success was family income: higher family income was consistently associated with better reading. Better readers also were less frequently cared for by older siblings, had lower maternal neighborhood ratings, and were not cared for after school by the mother after age 11. Predicting Toward Math Achievement. Math achievement was predicted for subjects with a positive score for the discriminant function. Low math achievement was predicted by negative scores. There were 67 subjects in the high-achievement group and 70 in the low-achievement group. The Wilks' A was .76 for predicting high- and low-scorers, indicating that prediction was good. Subjects with good math skills came from families with higher incomes, had mothers who were more content, had mothers who gave their neighborhood a poor rating, and often had an unrelated adult provide child care from 6 to 11. High math achievers were also usually female. L_--------~-----~----------- Predicting Toward Work Organization. Subjects• ability to organize their work appropriately was predicted by a negative score on the discriminant function. The size of this group was 134. Subjects with poor work organization skills had positive function scores (~ = 112). Prediction toward work organization skills was good (Wilks• A= .83). Those with good work organization tended to be female, have stable families, and had higher family income. Half-time day care from 3 to 6 also predicted toward good work organization, as did in-home care from 0 to 3. Poor work organization skills were predicted for those who were male, had high family instability, low family income, and did not attend half-time day care from 3 to 6. Predicting Toward Special Education. A total of 248 subjects were included in the discriminant function analysis designed to predict normal vs. special education. Fifteen children in this group were attending special classes, while 233 were in regular classrooms. Predictions of special education placements in this sample were relatively unsuccessful (Wilks 1 A= .9). Highly negative discrim- inant function scores indicated special education, while small, positive scores predicted to a regular classroom. Subjects who were in special education tended to have mothers who were rated low in orderliness and contentment, were male, had many neonatal health problems, and received father care or full-time day care from 3 to 6. Attendance at an after-school center from 6 to 11 also predicted toward special education, although this predictor was suppressed 1 in the final solution. ! L~~ ~--~-~-~---------- ------- ---~---------- -- ~------------ ------------ ~I Predicting Toward Grade Retention. Of a total of 222 subjects ! included in this analysis, 210 were not retained and 12 were retained. 1 Highly negative function scores predicted grade retention, while zero and positive scores predicted nonretention. Wilks' A for the predic- tion of retention was .9. Subjects who were retained had lower family income, more neonatal neurological problems, were cared for by their mother from 6 to 11 and/or their father ages 12 and over, were female, and did not attend half-time day care from 3 to 6. These children also had mothers who were rated lower on orderliness. Predicting Toward Remedial Instruction. Forty subjects of the 240 included in the analysis of remedial instruction received remedial instruction. The subjects that received negative function scores were predicted to have received remedial instruction, while those with positive scores were not. Prediction of remedial instruc- tion was good (Wilks' A= .79). The subjects who received remedial instruction did not have mother care from 3 to 6, but did have mother or after-school center care from 6 to 11. This group had low maternal neighborhood ratings, more neonatal physical problems, higher family instability, lower family incomes, an absence of grandmother care from 0 to 3, and greater family dissension. Full-time day care from 3 to 6 also predicted to remedial instruction, but was not included in the final discriminant function. Predicting Toward School Psychology Report. The Wilks' A for prediction to the availability of a school psychology report L ________________________ ----- r---------------- . ·----------------------~------- ·- ---------- .90. Four hundred and eighty-eight subjects in the analysis did not have psychology reports, while 110 did. Subjects with school psychology reports tended to have positive scores on the discriminant function. These children were from unstable families, attended an after-school center or received mother care from 6 to 11, attended an after-school center over age 11, and were male. Stage 3: Model Development and Testing Model Development This section describes the specification of a model of day care effects suitable for testing utilizing the Joreskog and Sorbom (1978) method for analyzing systems of linear equations. Such models must specify the relationships among a discrete set of variables of interest. While it is not necessary to predict the magnitude of such relationships (although it would be desirable), the sequencing and directionality of all associations must be specified. The model examined in the present study was developed on the basis of the literature and the discriminant function analyses. How- ever, it should be kept in mind that prior analyses were based on high- and low-scoring subjects, and these groups may not have always resembled the less restricted sample that the model was tested on. Table 15 summarizes the discriminant function results related to the effects of day care. The left column contains the antecedent variables, while the dependent variables are displayed across the page. All of the variables in the table are measures with several indicators {with the exception of the subject's sex). There were two measures of neonatal health, six SES-related measures, and four measures each for ~----~---------- 1 I I i I Table 15 Number of Measures Predicting to Dependent Variables in the Discriminant Function Analyses (Expressed as a Ratio of Successful Predictions to Total Possible Predictions) Dependent Variables Independent Day Care Day Care Psychosocial Cognitive Variables 0 - 3 3 - 6 Cluster Cluster Subject's sex 0/2 0/2 1/4 3/4 SES-related variables 6/12 10/12 8/24 16/24 Neonatal health 2/4 2/4 4/8 1/8 Day care 0 to 3 4/4 0/8 0/12 Day Care 3 to 6 5/12 5/12 i I I the psychosoc-J;j- and--;;;,gn it i ve cl us te rs . Th;-d;;y- ~a r~ r-ia b I es were for the mother care and day care groups only; all other forms I of child care were not included in the present analysis. Each entry in Table 15 represents the ratio of successful to total possible predictors for a particular cell in the table. For lexample, with six SES measures and four cognitive measures there existed a total of twenty-four possible predictions. The table entry for these variables indicates that there were sixteen successful predictions for these two variables. Because this percentage is relatively high, it would be anticipated that a causal relationship or intercorrelation may exist between SES and the cognitive outcome variables. And, because theSES variable is a temporal antecedent, the likelihood of a causal relationship is increased. Table 15 represents a summarized version of the discriminant function analysis results presented in Tables 9, 10, 13 and 14, with measures eliminated that could not be included in a model of day care effects. The excluded measures were unrelated forms of child care and dichotomous molar measures not appropriate for outcome measures in the model testing statistical procedure. The model that was tested and further refined to develop a final model is shown in Figure 2. The following discussion will describe the considerations involved in proposing this model. The subject's sex did not predict to either day care variable, but did predict to the psychosocial and cognitive variables. Because of the sex differences found in the discriminant function analyses, sex was considered an important variable to include. However, because L.__ ___ --------- ·~-~- ------- --.------·- ----------- - II Figure 2. Initial Model of Day Care Effects Based on Discriminant Function Analyses and Literature. L _______________________ _ [ se~l~--;-d i ch~to~;~~-va r i ab 1 e, and b~-~~~~e -~~vera-1 ;uthors have I I suggested a sex by type of care interaction (e.g., Robertson, 1978), I it was considered appropriate to develop two models, one for females I [and one for males. This procedure would highlight male/female ! ! differences. I SES-related variables successfully predicted to each of the dependent variables, and each effect was specified in the initial model. The literature provides strong support for SES effects (e.g., Robertson, 1978; Moore, 1975; Kagan et al., 1978; Winett et al., 1977), but 1 ittle information regarding the substantive components of SES. The subject•s neonatal health predicted to each dependent variable, and was included in the initial model with the exception of the effect of neonatal health on cognitive achievement, where prediction was weak. The literature provided no guidance in hypothe- sizing the effects of neonatal health. Day care from 0 to 3 predicted to day care from 3 to 6, but not to either of the outcome variables. Peaslee (Note 15) and Cochran (1977) have suggested that, where day care represents an environment that does not provide as much cognitive stimulation as the home, negative cognitive effects will be present for children who experience early childhood day care. In the present study, day care can be characterized as average, relatively custodial child care as provided in Denmark during the early- and middle-sixties. For that reason a negative influence of day care from 0 to 3 on cognitive achievement was hypothesized. I L__ ___ ------------ I Day care from 3 to 6 predicted to both psychosocial and ~cognitive outcomes, and has been included in the model. Meyer (Note ~6) noted that day care children have more positive peer interaction I ! than home-reared children, a finding that supports an effect from day care to psychosocial functioning. Moore (1975) and Schwarz et al. (1974) found day care boys less cooperative with adults. Several other features of a model must be considered before the model can be tested. One such consideration is the proposed factor structures for each of the variables (also referred to as factors) in the model. Each variable, illustrated by a circle, was defined by a number of individual measures, illustrated in the final models as rectangles. In specifying the model it was necessary to indicate the sign of each factor loading and to define the factor as most closely associated with one measure, indicated by a factor loading of 1 .0. TheSES-related factor initially consisted of six measures, and family income was given a factor loading of one. Child health, indicated by two measures, was defined by neonatal neurological problems. The psychosocial and cognitive factors were each indicated by four measures. The psychosocial factor was defined as most closely associated with peer relations. The cognitive factor was defined as reading achievement. In addition to defining the factors, possible noncausal associations among measures and factors were also considered. All measures indicating SES and child health were allowed to correlate I free 1 y' wh i 1 e other-measures werecons i derea- notu-cor-refated. The !psychosocial and cognitive factors were also considered correlated. I I !Model Testing and Refinement I : The initial model described in Figure 2 was tested using the I iLISREL IV procedure developed by Joreskog and Sorbom (1978). Two I samples were utilized (males and females) with teacher data available. Sample size was also determined by selection only of those subjects who received either exclusive maternal care or full-time day care on the child care measures. This restriction helped obtain the maximum contrast between the home and center care groups. This resulted in final sample sizes of 145 for males and 152 for females. This section will first describe the day care model for males, followed by the day care model for females. Model of Day Care Effects for Males. When the initial model was tested for males the LISREL solution presented several ways in which the model could be modified to improve the model •s fit to the data. The SES factor structure was changed to incorporate two inter- correlated factors. One factor consisted of family income and mother's orderliness, and has been named economic resources. The other factor included family instability, neighborhood rating, family dissension, and maternal contentment. This second SES-related factor will be referred to as home environment. A change in the factor structure of the outcome variables was also indicated. Work organization, which was included as a cognitive outcome measure also loaded on the psychosocial adjustment factor. \fhe facto~~;t-;=-uct~;:-e was changed to allow ;~rk organization to serve las an indicator of both cognitive and psychosocial adjustment. I Paths were dropped from neonatal health to all dependent \variables (thus the factor was dropped). The paths from economic jresources to both day care variables and psychosocial skills, and I !from home environment to psychosocial skills and cognitive skills I I were dropped. The day care variables did not predict directly to either of the outcome variables, but were associated with individual outcome !measures. Day care from 0 to 3 had a negative association with the !teacher's rating of math achievement, one of the measures used to estimate cognitive functioning. Day care from 3 to 6 had a positive association with peer relations, one of the measures used to estimate psychosocial functioning. Home environment predicted to the type of day care at both age levels. A negatively rated home environment predicted toward the child receiving day care. The family's economic resources did not predict to type of care, except indirectly as the resources affected the home environment factor. Economic resources was the only antecedent variable with an effect on an outcome variable, exhibiting a strong influence on the child's academic achievement. The final day care model for males is shown in Figure 3. The correlation matrix on which this model is based is shown in Table 16. The x 2 value with 98 degrees of freedom was 210 for the final model I for males. I I l-~--~~~ ~-- -~~ -~-~-~--~~~~~~-~-------- \..0 00 .63 Figure 3. .37 1.0 -. 15 .64 ment 2 = X df=99 210 1.0 Day Care 3 to 6 ' '/ ~ .191 .284 .·66 •ent Final Model Predicting Cognitive and Psychosocial Outcomes for Day Care Males. .65 -l I .38 .26 .16 .45 Table 16 Correlation Matrix for Day Care Males DC03 DC36 PEERREL ADULTREL AGGRESS PARNTREL VERBAL READ MATH WRKORG DC03 1.000 DC36 0.684 1.000 PEERREL -0.117 0.061 1.000 ADULTREL -0.065 -0.021 0.431 1.000 AGGRESS o. 116 -0.015 -O.Itl2 -0.564 1.000 PARNTREL -0.044 0.019 0.441 0.441 -0.544 1.000 VERBAL -0.091 0.020 0.357 0.270 -0.301 0.310 1.000 READ -0.091 -0.057 . 303 0.230 -0.285 0.261 0.804 1.000 MATH -0.212 -0.136 0.365 0.357 -0.327 0. 352 0.605 0.670 1.000 WRKORG -0.109 -0.043 0.462 0.609 -0.431 0.474 0.548 0.512 0.595 1.000 SESNOW -0.021 -0.059 0.024 0.044 -0.258 0.217 0.240 0.364 0.325 0.183 MOTH NEAT -0.034 -0.179 0.102 0.038 -0.126 0.076 0.098 0.237 0.210 0.206 NEIGHBOR 0.077 0.110 -0.027 0.008 0.071 -0.148 0.089 0.100 0.048 0.010 NOCONST 0.286 0. 342 -0.199 -0.406 0.229 -0.189 -0.136 -0.134 -0.283 -0.283 FAMDISS 0.213 0. 387 -0.022 -0.092 0.041 -0.046 -0.102 -0.137 -0.142 -0.138 MOTHCONT -0.083 -0.096 -0.042 0.072 -0.077 0.082 0.117 0.179 0.130 0.155 SESNOW MOTHNFAT NEIGHBOR NOCONST FAMDISS MOTHCONT SESNOW 1.000 MOTH NEAT 0. 336 1.000 NEIGHBOR -0.074 -0.163 1.000 NOCONST -0.189 -0.245 0.069 1.000 FAMDISS -0.112 -0.235 o. 161 0.512 1.000 MOTH CO NT 0.215 0.261 -0.309 -0.343 -0.575 1.000 r------------------ ----------------- ----- ------------ --- ------- Model of Day Care Effects for Females. The model of day care effects for females was developed similarly to the model for males. The initial model was exactly the same, and the sample selection criteria was identical. The sample size for testing this model was 152. The final model was very similar to the model for males, except for the effects to and from day care. The final X 2 /df ratio was 158.6/98. The final factor structure was identical to the model for males. Work organization was an indicator of both academic achievement and social skills. Economic resources predicted only to academic achievement. Home environment predicted to day care from 0 to 3, but did not predict type of care at age 3 to 6. Day care at a younger age did predict to day care at a later age, as expected, but neither day care variable predicted to outcome variables. Home environment did predict toward the psychosocial factor, with a better home environment leading to better psychosocial functioning. The final day care model for females is shown in Figure 4. Table 17 contains the correlation matrix that was used to develop this mode 1. r-~ ~--~-- -~~--- 1 ! .68 .]6 .94 .69 .B7 Figure 4. .32 ment 1.0 -.407 Day Care 1 . 73 0 to 3 ~ 1.0 Day Care 3 to 6 ------· 7 5 .. 1 xz df=98 = 158.6 .47 .67 me-nt Final Model Predicting Cognitive and Psychosocial Outcomes for Day Care Females. ___ j 0 N ~--~--------~-------~----- 1 CHAPTER V SUMMARY AND DISCUSSION This chapter summarizes the study results presented in Chapter IV. All results pertaining to each of the groups of outcome variables (child care, psychosocial, and cognitive outcomes) are summarized and discussed. Further comments are then presented that relate to the study methodology. Finally, several conclusions and recommenda- tions are presented. Summary and Interpretation Tables 18, 19 and 20 summarize the results of both the discriminant function and path analyses for child care outcomes, psychosocial outcomes, and cognitive outcomes, respectively. Within each table, the left column contains the dependent measures. The right column lists the most important predictors. The text summary presented below does not highlight each of these findings, but attempts to extract only the most significant findings related to day care effects. Child Care Outcomes Prediction to child care arrangements was very good for all age levels. These results are summarized in Table 18. Typically, the most important predictor of child care in one time frame was the subject's child care status during the previous time frame. Out-of- 10 '-----~------~--~---------- -----~- ---- -~-~ - . ~------------~------ Tabl~~~8---- -- ------- 1 Summary of Findings Related to Child Care Outcome Variables Outcome Variable Predicted by: I I ~------------------------------------------------------------------------ ' CHILD CARE--0 TO 3 Maternal care Day care In-home sitter care Out-of-home care CHILD CARE--3 TO 6 Maternal care Full-time day care Half-time day care Father care Grandmother care CHILD CARE--6 TO 11 Maternal care Father care High family income, family stability, neonatal neurological problems, low family dissension. Poor home environment. High family income, neonatal neuro logical problems, low family dissension and family instability. Low income, family instability, fewer neonatal neurological problems, and high family dissension. Maternal care; day care or day nurse 0 to 3, family stability, good neonatal health status. Day care or other non-maternal care from 0 to 3, poor home environment. Family stability, good neonatal health, mixed child care, 0 to 3, high family income. Little previous care by mother or grandmother, low maternal contentment, family stability. Previous care by grandmother, family instabi 1 ity. Family stability, no day care 3 to 6, high family income. Previous care by relatives other than the mother, particularly father and grandmother, low family dissension, family instability. Outcome Variable Grandmother care Adult outside the home After-school center CHILD CARE--11 AND OVER Maternal care Father care After-school center Older siblings No caretaker L----~-~--------- ------------------------- ~---- Predicted by: Family instability, day care 3 to 6, little previous maternal care, and previous care by grandmother. High family dissension, mixed child care 0 to 6, absence of maternal care and day care. Family instability, day care from 3 to 6, family dissension, low family income. Maternal care 6 toll. Father care 6 toll. After-school center 6 to 11, absence of care by an unrelated adult 6 to 11. Mixed care ages 6 to 11, full-time day care 3 to 6, maternal care 0 to 3, family stability. Mixed care ages 6 to 11, family instability, absence of maternal care and day care 0 to 6. lhome care (including day care) was usuallyp7edul~ted-fo~-families I I ]lower incomes. Thus, lower SES children were exposed more frequently 1 ito child care situations with substantial peer contact. High-income ! families tended to utilize more in-home, individualized child care arrangements. Although day care from 0 to 3 could not be predicted differentially from other out-of-home care arrangements, it was predicted more by the child's home environment than the family's economic resources. Families with greater instability, dissension, and maternal discontent tended to use infant/toddler day care. Once children began attending day care, they typically \continued. Day care from 0 to 3 was the best predictor of day care i from 3 to 6. The tendency also continued for children from poorer home environments to enter day care earlier and to receive more full- time day care. The use of half-time day care was infrequent at ages 0 to 3. From 3 to 6, half-time day care was used by stable families with relatively high family income. Thus, families with better home environments and greater income did use day care, but with less intensity and with a later age of onset. Maternal care for all ages was consistently predicted by better family economic resources and home environment. This is, apparently, a fairly stable group of families who provide virtually exclusive maternal care, particularly prior to age 6. The only exception to this picture was that from 0 to 3 a number of children 'with early neurological problems received home care. ~---~~th_e_r-gr~~p- existed that resembled I the maternal care group I !very much. A number of children received in-home care from a baby- i i :sitter or young woman from 0 to 3. Higher family instabi 1 ity was the only variable on which this group differed from the maternal care group. Prediction to after-school child care arrangements was very good, accomplished almost exclusively by knowledge of the type of child care in the previous time frame. Care by older siblings and the absence of supervision were common over age 11. Subjects cared for by older siblings often received full-time day care from 3 to 6 and had stable family backgrounds. Subjects with no caretaker after age 11 tended to have full-time day care age 0 to 6 and less stable fami 1 ies. Psychosocial Outcomes The results of this study indicate that day care, and other forms of child care, do affect psychosocial outcomes in adolescence. These results are summarized in Table 19. The effects of SES-related variables on adolescent psychosocial outcomes are strong. Child care arrangements are successful predictors, although the effects are complex. Type of child care, usually in conjunction with family stability, predicted peer interaction, parent interaction, an arrest record, adult interaction, andaggressiveness. Individualized child care arrangements positively influenced adult relations. Child care in settings with other children tended to influence social behavior i !strongly, increasing both peer relations and aggressiveness. L _______ ------ -------------·-- ------------ Table 19 Summary of Findings Related to Psychosocial Outcome Variables Outcome Variable i :(Good) Peer interaction (Good) Adult interaction (Good) Parent interaction Aggressiveness Arrest record Predicted by: Half- to full-time day care 3 to 6, care by older siblings after age 11, out-of-home care 6 to 11, family stability, high family income. Family stability, more individualized child care throughout childhood, half-time day care 3 to 6, sex (females were better adult relaters). Better perception of the neighborhood by the mother, few neonatal problems, high family income, day nurse from 0 to 3, half-time day care from 3 to 6 and father care from 6 to 11. Family instability, less care by an unrelated adult 6 to 11, after-school center over age 11, more physical and neurological problems at birth, full-time day care 0 to 6, and after school center 6 to 11. Non-maternal care over 11 (but were not unsupervised), family instability, sex (males tended to have an arrest record more than females). [---Th;-;ffu~-t~ of ha 1f-t ime and full-=-ti~~day--~~~~ from 3 to 6 1 were positive for the sample. Good peer, adult, and parent relations lwere all associated with half-time day care from 3 to 6. This finding ! :supports those of Moore (1975) and Meyer (1979) who found day care associated with good peer relations. Half-time day care did not predict toward negative psychosocial outcomes. Full-time day care from 0 to 3 was associated with aggressive- I lness in adolescence. From 3 to 6 full-time care predicted both I aggressiveness and good peer relations, a finding very similar to Meyer (1979) who found full-time day care associated with both good peer relations and aggressiveness. The effects of full-time day care 0 to 3 were not strong enough to influence psychosocial outcomes in the path mode 1 . Care in an after-school center was associated with several negative outcomes in this sample. These negative outcomes included aggressiveness, as well as the presence of a school psychologist's evaluation, remedial instruction, and special education. There are several possible explanations for this finding. Children attending an after-school center may have received additional supervision because they were already exhibiting problem behavior. Alternatively, some factor associated with after-school care (such as increased peer influence) may promote increased problem behavior. It is interesting that, in the Danish Perinatal Followup Sample, after-school care is more closely associated with negative outcomes than either care by older siblings or the absence of supervision. ~--~~b-y-old~-r-sibl ings after age 11 pre;Ti-ct-~-c:l-9;od peer I j relations. The absence of supervision was only a predictor for non ! !criminality--an apparently paradoxical finding. As suggested above, subjects who later were arrested may have displayed behavior earlier than resulted in the presence of supervision over age 11. The most influential SES-related variable was family instability, which predicted four of five psychosocial outcome measures. In all cases, the family instability measure operated in an expectable direction. Family instability was associated with negative outcomes, and family stability was associated with positive outcomes. Cognitive Outcomes Prediction to cognitive outcomes was successful, but accomplished more by SES-related variables than child care variables. These results are summarized in Table 20. In this sample, cognitive outcomes were effected more by economic resources, while psychosocial outcomes were effected by SES-related measures more closely associated with home environment. Full-time day care, particularly from 0 to 3, was associated with negative cognitive outcomes. Early full-time care was associated with poor math achievement and an increased incidence of special education. Half-time day care from 3 to 6 exerted a positive influence on several outcome measures. Half-time day care predicted to good ,verbal skills, good work organization skills, and appropriate grade ]placement for age. I L___ _______________ --------- ------------------ 11 ~----~------------~------~----Ta-b-1 e -2-0---~-- Summary of Findings Related to Cognitive Outcome Variables '! ------------------------------------ Outcome Variable (Good) Verbal skills (Good) Reading achievement 1 (Good) Math achievement (Good) Work organization Special education Grade retention Remedial instruction L ________________ --------- --- - Predicted by: Family stability, poor maternal neighborhood ratings, economic resources, maternal contentment, and half-time day care 3 to 6. Economic resources, not cared for by older siblings or the mother over age 11, and poor maternal neighborhood ratings. Economic resources, absence of early day care, maternal contentment, poor maternal neighborhood ratings, sex (females had higher achievement), and unrelated adult care from 6 to 11. Sex (females had better work organization), stable families, economic resources, half-time day care from 3 to 6, and in-home care from 0 to 3. Low maternal orderliness and contentment, sex (males were more frequently in special education), neonatal health problems, father care or full-time day care 3 to 6, and after-school care 6 to 11. Low family income, neonatal neuro logical problems, mother care 6 to 11, female, and did not attend half-time day care from 3 to 6. Non-maternal care 3 to 6, maternal or after-school center care 6 to 11, low maternal neighborhood ratings, poor neonatal health, unstable families, and low family income. 111 ~-~-----~---- Table 20--Continued 1 1------------------------------------------------------------------------ l I I I Outcome Variable School psychology report L_ __ ----------------- ---------~---- Predicted by: Unstable families, maternal or after school center care 6 to 11, after school center care over age 11, sex (males more frequently had a school psychology report). 11 ,---------~---- ------~· ··- ·-- ·------, 1 The positive influence of ha 1f-t ime day care is presumab 1 y the i result of exposure to increased social and cognitive stimulation in appropriate dosages for the developing child. The cognitive effects of child care arrangements after the child enters school provide additional support for this interpretation. Reading achievement was predicted by the absence of maternal care or care by older siblings over age 11. Math achievement, too, was predicted by care by an unrelated adult from 6 to II. Both maternal care and care in an after-school center were associated with several negative cognitive findings. Maternal care 3 to 6 predicted to both grade retention and school psychologist examinations. Maternal care over age 11 predicted to poor reading achievement. Attendance at an after-school center influenced the rate of special education, remedial instruction, and school psychology reports. The SES-related variables, particularly those associated with the families• economic resources, showed strong effects on cognitive measures. Families that were relatively stable also exhibited maternal contentment and poor maternal neighborhood ratings. These home environment measures usually influenced the cognitive outcome measures. The families• economic resources, as indicated by family income and maternal orderliness, consistently influenced the child's cognitive functioning. The fact that families judged to have good home environments also had poor maternal neighborhood ratings may indicate the mothe~s L ___ ------ -------------------------------· [awareness of a~dp-~l-ority for raising the family in a good neighbor- hood. The subject•s sex was frequently a predictor of cognitive outcomes. In the Danish sample, females generally performed better than males. Males tended to have worse math skills, more frequently had a school psychologist report, had poorer adult relations, and more frequently had an arrest record. Females were more frequently retained in schoo 1. Methodological Comments The results of this study indicate the usefulness of examining day care effects for a wide range of long-term outcome measures. The methodology and data analysis strategy employed support the feasibility of developing models of day care and/or child care effects. The sequential analyses progressing from variable definition through discriminant function analysis to path analysis has illustrated a procedure for data reduction and model development. This section describes several limitations in the analytic strategy that may be important in defining future research. The Usefulness of Molar Measures Zigler and Trickett (1978) strongly recommended the use of molar measures as outcome variables for defining the long term effects of early childhood programs. Their recommendation was based on their opinion that molar outcomes could be concretely measured and that such measures are both highly related to social adjustment and readily ! interpretable to policy makers and taxpayers. I L--------~--- ------------------ JJ In the present study several molar measures were included: the existence of an arrest record, the existence of a school psychology report, grade retention, receipt of remedial tutoring, and receipt of special education. Three of these measures were specifically recom- mended by Zigler and Trickett. The utility of molar measures in this study was limited for several reasons. The molar measures employed could not be character- ized as 11 powerful 11 variables--the teacher ratings virtually always had a greater percentage of their variance explained by antecedent variables, despite the fact that the ratings had been artificially dichotomized for inclusion in the discriminant function analyses. i Because they were dichotomous, the molar measures could not be included' in the model development analyses. It is uncertain why the molar measures were not powerful outcome variables. However, some of Zigler and Trickett 1 s observations can be called into question. Molar measures may not be any more reliably or val idly measured than other measures (such as the teacher•s ratings). For example, arrest records and other confidential informa- tion was made available for this research from highly reliable government registries. Such confidential information would probably not be available for a similar sample in the United States. The validity of molar measures is questionable also. The very fact that molar measures are concrete and observable may mean that subtleties associated with such measures are lost: What type of arrest record? How long was a child in special education? Given these problems, ;molar measures may be only second rate substitutes for carefully I ! L--~-------------------~---·---------------- ---- -- ---- --~-- - ---- [c;11ected--~s;~h~met~l~ data or~i;-th;-~~~-;-~.f thi~ study, a I teacher's considered opinion about a child's functioning in specific I !areas. The results of this study seem to support these observations. I Degree of Factor Specificity In the present study an effort was made to develop a model of day care effects on cognitive and psychosocial outcome factors. Such an attempt was probably overly ambitious. While there was some support in the literature and ample evidence in the Stage 2 discrim- inant function analyses for day care effects, it soon became evident that such effects were too complex to support day care effects on such general factors. The factor structure that was developed was supported, however, and represents an interesting result of this study. Thus, an attempt to develop a model that predicted to more specific factors or observed dependent variables would represent a tradeoff between information derived from the factor structure and the "fit" of the model. Limitations of Multivariate Method for Analyzing Interactions One of the primary difficulties of analyzing nonexperimental, longitudinal data is the problem of finding and interpreting inter- active effects. In the case of day care effects, the literature clearly indicates the probability of SES by care and sex by care interactions (see, e.g., Moore, 1975; Robertson, Note 11; Meyer, Note 3). In the present study the probability of a sex by care interaction led to the development of separate models for males and 11 !fe~a 1 es -:--ThesEs by-ca-re intera-ct ion was -anal yzecf-by- screening for I interactions between family instability or family income and child I care. These potential interactions could be anticipated and explained~ although they do indicate the need to develop methods for finding and incorporating interactions in multivariate models. Analysis Under Conditions of Periodic Treatment One of the most perplexing findings of this study was the association of after-school center attendance with several negative outcomes, including delinquency. The explanations proposed in relation to this finding hypothesize either an effect (e.g., increased peer involvement in the after-school center resulting in delinquency) or an association due to prior intervention (i.e., placing the child in an after-school center to increase supervision). The interpretive problem arises when deciding when to hypothesize an intervention/treatment effect. Without collecting data related to interventions, it is only possible to assume that interventions were initiated when negative outcomes were becoming apparent and resources were available for dealing with such problems. Considering the tremendous variety of naturally occurring treatments, collecting data for all possible contingencies may be difficult, but such an effort would greatly aid interpretation in longitudinal studies. Conclusions In Chapter I I it was stated that two types of research could L _____________________________ ------- ]be initiated based on previous day ca-;~--inqulry~-Fl;st, studies ar-;n-l I I ]needed that microscopically analyze child care environments and link I !them to specific child outcomes. I Day care environments vary tremen- i :dously and need to be described and differentiated along useful ,dimensions. A second type of study would continue to define general '1day care effects, but would do so with average day care across multiple settings. The study of global day care effects should be conducted across settings, in an effort to characterize the effects of an entire class of child care arrangements on functioning later in life. The present study was of the second type. Average quality day care was examined to determine its effects on cognitive and psychosocial functioning in late adolescence. The results of this study provide support for the idea that too much day care at too early an age may produce harmful results. In the Danish Perinatal Followup Sample, children who received full- time day care from 0 to 3 exhibited some negative effects. They were rated lower in math achievement and more frequently received remedial instruction. There was also a negative effect of early day care on psychosocial functioning: early full-time care predicted to aggressiveness in adolescence. There were several outcome measures for which half-time day care or full-time day care after age three showed positive benefits. This finding supports the concept that, in appropriate quantities, day care will exert a positive influence on children's psychosocial and cognitive functioning in late adolescence. This study showed a ! positive effect of day care on peer interactions, adult interactions, L~--~~----- -- -- ----- -- 11 lparen~ in~eraction, verbal organ1zat1on. skills, reading achievement, and work All types of child care can be rated for the exposure to peers that they represent. Out-of-home care arrangements usually involve greater peer interaction. Settings with increased peer interaction 1 seem to positively influence later peer relations. However, settings I !with greater peer influence were also associated with increased !aggressiveness. The critical variables in determining the positive I or negative effects of greater exposure to peers seemed to be family instability and age of enrollment in day care. Children from stable families who were exposed to peers in day care benefitted. Children from unstable families or children enrolled in full-time day care under age 3 were rated more aggressive in adolescence. This study has helped to document some of the variety and fluidity of child care arrangements at all age levels. Although there is a continuous increase in the number of children in day care with increasing age, there were more than five distinct types of care represented within each age grouping studied. Much additional research is needed to clarify what types of children are cared for under various arrangements, and to further examine the results of such arrangements. SES-related variables played an important role in predicting both cognitive and psychosocial outcomes. At least two distinct SES factors should be incorporated in future research: economic resources and home environment. In the present study, these factors were 1 correlated, but operated very differently. Economic resources were I L-~~---~~-~~~~~-~-~ -~~~---~--~-~---~-----~~-- ----- 11 cognitive outcomes, while home environment predicted toward psychosocial outcomes. Home environment also predicted to day care, while economic resources did not. The Danish Perinatal Followup Study sample has provided a unique opportunity to examine the long term effects of day care on cognitive and psychosocial functioning in adolescence. This sample allowed the study of day care effects for a large group that experienced average, institutional day care. In this study rela- tively few negative outcomes were found, providing support for continued research in day care, both to document beneficial effects and to analyze the specific environmental features that promote healthy development. ! I L-~----~--~----~--- -~ ----~-~-~-----~~-- --~------~--~- -- I 1201 """ ~---~-- __j ,-------~·-----·~ --------·- ··--··---------- ----- ····----------· ---- -- Reference Notes 1. Bronfenbrenner, U. The challenge of social change to public policy and developmental research. Paper presented at the President's Symposium on Child Development and Public Pol icy at the Society for Research in Child Development, Denver, Colorado, Apri 1, 1975. 2. Lazar, I., Hubbell, V., Murray, H., Rosche, M., & Royce, J. The persistance of preschool effects: a long term follow-up of fourteen infant and preschool experiments. Washington, D.C.: U.S. Department of HEW, Final report of the Consortium on Develop mental Continuity, Education Commission of the States, September, 1977. 3. Ricciuti, H. Effects of infant day care experience on behavior and development: research and implications for social policy. Unpublished report, Cornell University, 1976. 4. Ragozin, A. Attachment in day care children: field and laboratory findings. Paper presented at the meeting of the Society for Research in Child Development, Denver, Colorado, April, 1975. 5. Saunders, M. Some aspects of the effects of day care on infants' emotional and personality development. Unpublished doctoral dissertation, University of North Carolina, 1972. 6. Meyer, W. Developmental effects of infant day care. Paper presented at the annual meeting of the American Educational Research Association, San Francisco, April, 1979. 7. Keister, M. A demonstration project: group care for infants and toddlers, 1967-1970. University of North Carol ina, Final Report, 1970. 8. Ramey, C., & Farran, D. Infant day care and attachment behaviors toward mothers and teachers. Revision of paper presented at the American Psychological Association meeting, Chicago, 1975. Unpublished manuscript, 1976. 9. Ricciuti, M., & Poresky, R. Development of attachment to caregivers in an infant nursery during the first year of life. Paper presented at meetings of the Society for Research in Child Development, Philadelphia, March, 1973. L---------~- -------- 121 110-. -L i ppma~--:- -M.-~-&G~~-te, B. Soc i-~-em~t-l~~~l eff~~-t~-~-f day care. -~ Project report, Western Washington State College, 1974. ' ' ' I I I 11. Robertson, A. deve 1 opment. 1978. Group day care and children•s social-motivational Unpublished doctoral dissertation, Yale University, 12. Honig, A., Lally, J., & Mathieson, D. Personal social adjustment of school children after 5 years in a family enrichment program. Paper presented at the annual meeting of the American Educational Research Association, San Francisco, April, 1979. 13. Fowler, W., & Khan, N. A follow-up investigation of the later development of infants in enriched day care. Paper presented at the annual meeting of the American Educational Research Associa tion, Chicago, I 11 inois, Apri 1, 1974. 14. Steinberg, L., & Green, C. concerns, and consequences. of California, Irvine, 1978. Three types of day care: causes, Unpublished manuscript, University 15. Peaslee, N. The development of competency in 2 year old infants in day care and home reared environments. Unpublished doctoral dissertation, Florida State University, 1976. 16. Vroegh, K. Infant day care: some research findings. Unpublished manuscript, Institute for Juvenile Research, Chicago, Illinois, 1976. 17. Taylor, L. Outcome and process evaluation of a day care center. Institute for Research in Human Abilities, Memorial University of Newfoundland, Research Bulletin Number 75-004, June, 1975. 18. Lally, R. The family development research program, Progress Report. Syracuse University, 1974. 19. Ramey, C., & Smith, B. Learning and intelligence in disadvantaged infants: effects of early intervention. Paper presented at the Council on Exceptional Children, Chicago, April, 1976. 20. Sutton-Smith, B., & Mednick, B. The consequences of family structure and maternal state for child and mother•s development, Progress report. Columbia University, June, 1977. 21. Mednick, B., & Baker, R. Consequences of family structure and maternal state for child and mother•s development, Progress report (Volumes 1 & 2). Center for Interactional Research in Human Development and Learning, University of Southern California, July, 1979. !...-.........---~----------~--~-·-·---- . - -·---------- 24. Center for Interactional Research in Human Development and Learning, Information on Plum Tape (revised), University of Southern California, June, 1977. Mednick, B., & Sutton-Smith, B. The consequences of family structure and maternal state for child and mother•s development, Progress report. Center for Interactional Research in Human Development and Learning, University of Southern California, Feb r ua ry , 1 9 79 . McGarvey, W. Longitudinal Factors in School Desegregation. Unpublished doctoral dissertation, University of Southern California, September, 1976. r--------· ·-----··-·- 1 i ! References Ainsworth, M., & Bell, S. Attachment, exploration and separation · illustrated by the behavior of one-year-olds in a strange situation. Child Development, 1970, ~. 49-67. ;Ainsworth, M., & Wittig, B. Attachment and exploratory behavior of one-year-olds in a strange situation. In B. M. Foss (ed.), Determinants of Infant Behavior, Volume 4. London: Methuen, 1969. Ainsworth, M., Bell, S., & Stayton, D. Individual strange situation behavior of one-year-olds. (ed.), The Origins of Human Social Relations. Press, 1971. differences in In H. R. Schaffer London: Academic Baker, R., Mednick, B., & Brock, W. The application of causal modeling techniques to prospective longitudinal data bases. In S. Mednick and M. Harway (eds.), Longitudinal Research in the United States. Boston, Massachusetts: Niehoff Publishing Company, (in press). Beller, E. Research on organized programs of early education. In R. Travers (ed.), 2nd Handbook of Research on Teaching. Chicago: Rand McNally and Co., 1973. Belsky, J., & Steinberg, L. The effects of day care: a critical review. Child Development, 1978, 49, 929-949. Bentler, P. Multivariate analysis with latent variables: causal modeling. In M. R. Rozenzweig and L. W. Porter (eds.), Annual Review of Psychology, 1980, l!_, 1•19-456. Bentler, P., & Bonett, D. Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 1980, 88, 588-606. Blanchard, M., & Main, M. Avoidance of the attachment figure and social-emotional adjustment in day-care infants. Developmental Psychology, 1979, l2• 445-446. Blehar, M. Anxious attachment and defensive reactions associated with day care. Child Development, 1974, 45, 683-692. L ________________________________________ _ ~, wlbych,.IJld. seSop-a~~r--apt~lotnho.logical proce~~~~ngendered by early mother- 1 - In M. J. Senn (ed.), Infancy and Childhood. I New York· Josiah Macy, Jr., Foundation, 1953. I . :Bowlby, J. Attachment and Loss, Volume 1: Attachment. New York: 1 Basic Books, 1969. ! !Braun, S., & Caldwell, B. Emotional adjustment of children in day care or after the age of three. Early Child 1973, ~. 13-21. ' who enrolled prior to Development and Care, Bronfenbrenner, U. Is early intervention effective? In Bronfenbrenner U., and Mahoney, M. (eds.), Influences on Human Development, 2nd edition, Hinsdale, Illinois: The Dryden Press, 1975. Brookhart, J., & Hock, E. The effects of experimental context and experiential background on infant•s behavior toward their mother and a stranger. Child Development, 1976, 47, 333-340. Brown, B. Found: Long-term gains from early intervention. Boulder, Colorado: Westview Press, 1978. Caldwell, B. What does research teach us about day care for children under three? Children Today, 1972, l• 232-240. Caldwell, B., Wright, C., Honig, A., & Tannenbaum, J. Infant day care and attachment. American Journal of Orthopsychiatry, 1970, 40, 397-412. Casler, L. Maternal deprivation: a critical review of the 1 iterature. Monographs of the Society for Research in Child Development, 1961, 26, Number 2. Chamberlain, R., Chamberlatn, G., Howlett, B., & Claireaux, A. British Births, 1970. London: W. Heinemann Medical Books, Ltd., 1975. Cochran, M. A comparison of group day and family child rearing patterns in Sweden. Child Development, 1977, 48, 702-707. Cohen, J., & Cohen, P. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. New York: John Wiley & Sons, 1975. Cornelius, S., & Denny, N. Dependency in day-care and home-care children. Developmental Psychology, 1975, ll• 575-582. Cummings, E. Caregiver stability and day care. Developmental Psychology, 1980, ~. 31-37. I 1251 ------ - ___ J I Dawes, R. Th~-~~bu~t-beauty of imprope~-~~~~~r--;~d~ls in decision- , making. American Psychologist, 1979, 34, 571-582. :Deutsch, C. Social class and child development. In B. Caldwell and , , H. Ricciuti (eds), Review of Child Development Research. Chicago:; Chicago University Press, 1973. Doyle, A. Infant development and day care. Developmental Psychology, 1975, ll' 655-656. Elliott, V. Impact of day care on economic status of the family. In D. Peters (ed.), A Summary of the Pennsylvania Day Care Study. University Park: Pennsylvania State University, 1973. Etaugh, C. Effects of nonmaternal care on children: research evidence and popular views. American Psychologist, 1980, ]2, 309-319. Etzioni, A. Science and the future of the family. Science, 1977, 196, 487. Farran, D., & Ramey, _C. Infant day care and attachment behaviors toward mothers and teachers. Child Development, 1977, 48, 1112-1116. Fowler, W. A developmental learning approach to infant care in a group setting. Merrill-Palmer Quarterly, 1972, ~' 145-175. Garber, H., Heber, R., Hoffman, C., & Harrington, S. Preventing mental retardation through family rehabilitation. TADS Infant Education Monograph. Chapel Hill: TADS, 1976. Golden, M., Rosenbluth, L., Grossi, M., Policare, H., Freeman, H., & Brownlee, E. The New York City Day Care Study. New York: Medical and Health Research Association of New York City, 1978. Goldfarb, W. The effects of early institutional care on adolescent personality. Journal of Experimental Education, 1943, ~' 106- 129. Harrell, J., & Ridley, C. Substitute child care, and the quality of mother-child interaction. and the Family, 1975, 12, 556-565. maternal employment, Journal of Marriage Joreskog, K., & Sorbom, D. LISREL IV: Analysis of linear structural relationships by the method of maximum likelihood. Chicago: Nat i ona 1 Educat i ona 1 Resources, Inc., 1978. Kagan, J., Kearsley, R., & Zelazo, P. Infancy: Its Place in Human Development, Cambridge, Massachusetts: Harvard University Press, 1978. l ____ ----- -------·. -·------ -------- ~;-~rman, -S-~-&-Kahn, A. Social services in the United States: I policies and programs. Philadelphia: Temple University Press, 1976. i '!'Kessner, D., Singer, J., Kalk, C., & Schlesinger, E. Infant death: An analysis by maternal risk and health care. (Contrasts in I Health Status, Volume 1). Washington, D.C.: National Academy of Sciences, 1973. Kilmer, S. Infant-toddler group day care: A review of research. In L. Katz (ed.), Current Topics in Early Childhood Education, Volume I I. Norwood, New Jersey: Ablex Publishing Corp., 1979. Lyons, M. Techniques for using ordinal measures in regression and path analysis. In H. Costner (ed.), Sociological Methodology, San Francisco: Jossey-Bass, 1971. Macrae, J., & Herbert-Jackson, E. Are behavioral effects of infant day care program specific? Developmental Psychology, 1976, ~. 269-270. Maruyama, G., & McGarvey, B. Evaluating causal models: An application of maximum-likelihood analysis of structural equations. Psychological Bulletin, 1980, 87, 502-512. McCutcheon, B., & Calhoun, K. Social and emotional adjustment of infants and toddlers to a day care setting. American Journal of Orthopsychiatry, 1976, 46, 104-108. Mednick, B., Baker, R., & Sutton-Smith, B. Teenage Pregnancy and Perinatal Mortality. Journal of Youth and Adolescence, 1980, ~. 343-357. Meyers, L. The relationship between substitute child care, maternal employment, and female marital satisfaction. In D. Peters (ed.), A Summary of the Pennsylvania Day Care Study. University Park: Pennsylvania State University, 1973. Moore, T. Exclusive early mothering and its alternatives: the outcome to adolescence. Scandanavian Journal of Psychology, 1975, ]i, 255-272. Moskowitz, D., Schwarz, J., & Corsini, D. Initiating day care at three years of age: effects on attachment. _Child Development, 1977, 48, 1271-1276. National Council of Organizations for Children and Youth, America's Children: 1976, Washington, D.C., 1976. I i L ___________ ·- -- ----- ----- 1271 ---- - _ _j l le, N., Hull, c-:-~Jenkins, J., Steinbren~ner, K.-, & Be~t, o: SPSS: Statistical package for the social sciences, 2nd edition, New York: McGraw Hill Book Company, 1975. Niswander, K., & Gordon, M. The women and their pregnancies: The collaborative perinatal study of the National Institute of Neurological Diseases and Stroke. Philadelphia: W. B. Saunders Company, 1972. Portnoy, F., & Simmons, C. 1978, 49, 239-242. Day care and attachment. Child Developmen~, I Ramey, C., & Mills, P. care for high risk in Childhood: Day Hopkins University Social and intellectual consequences of day infants. In R. Webb (ed.), Social Development Care Programs and Research. Baltimore: Johns Press, 1977. Raph, J., Thomas, A., Chess, A., & Korn. S. The influence of nursery school on social interactions. American Journal of Ortho psychiatry, 1964, ~. 144-152. Rheingold, H. The measurement of maternal care. Child Development, 1960, }!, 565-575. Ricciuti, H. Fear and the development of social attachments in the first year of life. In M. Lewis and L. Rosenbaum (eds.), The Origins of Human Behavior: Fear. New York: John Wiley a~ Sons, Inc., 1974. Robinson, H., & Robinson, N. Longitudinal development in very young children in a comprehensive day care program: the first two years. Child Development, 1971, 42, 1673-1683. Roopnarine, J., & Lamb, M. The effects of day care on attachment and exploratory behavior in a strange situation. Merrill Palmer Quarterly, 1978, £i, 85-95. Rubenstein, J., & Howes, C. Caregiving and infant behavior in day care and in homes. Developmental Psychology, 1979, ~. 1-24. Ryan, S. Overview. In S. Ryan (ed.), A Report on Longitudinal Evaluations of Preschool Programs. Washington, D.C.: Office of Child Development, 1972. Schwarz, J., Krolick, G., and Strickland, R. Effects of early day care experience on adjustment to a new environment. American Journal of Orthopsychiatry, 1973, ~. 340-346. Schwarz, J., Strickland, R., & Krolick, G. behavioral effects at preschool age. 1974, lQ, 502-506. Infant day care: Developmental Psychology, i .. " ..... .. __ , .......... : . .;, I Sj01und, A. Day-Ca-re~l-nstitutions and ct1fldren 1 _s_Development. · (Translated by W. Glyn Jones). Westmead, England: Saxon House, D.C. Heath, 1973. Sorbom, D. Detection of correlated errors in longitudinal data. British Journal of Mathematical and Statistical Psychology, 1975, 28' 138-151. Spitz, R. Hospitalism: an inquiry into the genesis of psychiatric conditions in early childhood. Psychoanalytic Study of the Child, 1945, .!_, 53-74. Stroufe, L., & Waters, E. Attachment as an organizational construct. Child Development, 1977, 48, 1184-1199. Wagner, M., & Wagner, M. The Danish National Child Care System. Boulder, Colorado: Westview Press, 1976. Wainer, H. Estimating coefficients in linear models: It don't make no never mind. Psychological Bulletin, 1976, ~. 213-217. White, B., Watts, J., Barnett, 1., Kaban, B., Marmor, J., & Shapiro, B. Experience and Environment--Major Influences on Development of the Young Child. New Jersey: Prentice-Hall, Inc., 1973. Willis, A., & Ricciuti, H. A Good Beginning for Babies: Guide] ines for Group Care. Washington, D.C.: National Association for the Education of Young Children, 1975. Winer, B. Statistics and data analysis: Trading bias for reduced mean square error. Annual Review of Psychology, 1978, 29, 647- 681. Winett, R., Fuchs, W., Moffatt, S., & Nerviano, V. A cross-sectional study of children and their families in different child care environments. Journal of Community Psychology, 1977, ~' 149-159. Yarrow, L. Separation from parents during early childhood. In M. Hoffman and L. Hoffman (eds.), Review of Child Development Research, Volume 1. New York: Russell Sage, 1964. Zigler, E., & Trickett, P. IQ, social competence, and evaluation of early childhood intervention programs. American Psychologist, 1978, ]l, 789-798. Zigler, E., & Valentine, J. (eds.) Project Head Start. New York: Free Press, 1979. -- --------- ---------~------------ ------------ APPENDICES i L _____________ _ 130 I I i -------------- APPENDIX A Summary Tables of Studies I, I I, and I I I for the Danish Perinatal Followup Study L _________________________ --------- -------- 1 3_ '~ T A B L E The Effects of Maternal and Child Spacing Variables on the Status of the Neonate and the 1-Year-Old: Antecedent Variables Social Data on Family SES of family at birth of child Mothers marital stat us Mother's age at first birth Mother's age at birth of child Father's age at birth of child Wantedness of Present Pregnancy Perinatal Factors Medical data on previous pregnancies Mother's pre-pregnancy health Pregnancy complications Delivery complications Birth weight Multiple birth Family Structure Variables Number of older siblings Spacing of older siblings Sex of older siblings Variables to be Included in Analyses of Study Mediating Variables Sociological Factors SES at child's one year birthday Childs postpartum placement l~others work Day care experience Urban/rural locale Home environment Parental Health Mothers postpartum condition Mothers mental illness Family Structure Variables Spacing of older siblings Number of older siblings Sex of older siblings Status of Child Immediately After Physical examination score Neurological examination score Congenital malformations Birth weight Birth Outcome Variables Status of Child Immediately After Birth Physical examination score Mortality Neurological examination score Birth weight Status of Child at One Year Neurological examination score Physical examination score Motor developmental score r--- 1 VJ TABLE II Effect of Family and Maternal Variables on Child Status at Late Adolescence Variables to be Included in Analyses of Study II Antecedent Variables Familial Variables (i.e., birth order, sex of subject and siblings, spacing of siblings) Maternal Variables Was child wanted Pregnancy complication score Mother's age at first child Delivery complication score Mother's post partum condition Infant Variables Neonatal physical condition Neonatal neurological condition Child's neurological, physical and motor developmental status at 1 year Mediating Variables Sociological Variables SES of head of family during rearing of child Father's education Father's presence or absence Father's work (including night shift work) Mother's Career Mother's education Mother's occupation Mother's work history full-time/ ha 1f-t ime work Developmental Milieu Stability of family constellation Stresses on the family, i.e., economic, marital, health Mental health of parents Criminality of parents Childrens' home experience Amount of time in day care centers and/or nursery school Urban or rural locale Consequent Variables School Performance Evaluation by school psychologist Evaluation by teachers Physical Health Evaluation by school physicians De 1 i nquency Report of delinquency Psychological and Social Functioning Reports by mothers and teachers about the subject's personality and social functioning Identification of sex role ~---- 1 I I Antecedent Variables Family Variables Mother's age at first delivery Number of chi 1 dren Sex and spacing of children Marital status at birth of first chi 1 d Family Variables in Subjects Own Family Subject's birth order Sex of sib I i ngs Number of siblings Spacing of siblings TABLE Ill Effect of Family and Child-bearing Variables on the Mother Variables to be Included in Analyses of Study I I I Mediating Variables SES -:Husband's education Husband's occupation Subject's education Subject's occupation Subject's work history (full-time/ part-time) Husband Difficulties Apprehended criminality Mental health Poor work history Marital instability Environmental Factors Rural-urban locale Availability of child care i nst i tut ions Children's Status School performance Level of career aspirations De 1 i nquency Consequent Variables Career ~1 attained Level attained/level aspired Income Health ~ical Mental Marriage Separation, divorce Happiness Child-rearing Experienced failures/success Characteristics of Subject's Family Number of children Sex of children Spacing of children Age at first child I I I APPENDIX B VARIABLES USED IN THE PRESENT STUDY This appendix contains the individual data items that were used in the present study. The variables are presented in four parts: Part 1: Information taken from the or i gina 1 Danish Perinatal Study data. Part 2: Information taken from the Mother's Interview schedule. Part 3: Information taken from the interviewing social worker's Appendix to the Mother's Interview. Part 4: Information taken from the Teacher's Quest i onna i re. L~------------------- -·---------- .. -- ·-------·· ------ ---- Appendix B, Part This part describes the information used in the present study that was taken from the original Danish Perinatal Study data base. L ___ ------~--------------- ------------- ---- - -----------~---- ---- r--------·----------------------------------- ----·- - - ' Two variables were used that were taken from the original I Danish Perinatal data base. In both cases the variables were scales developed to provide indications of the subject's health at birth. The variables are briefly described below. Complete documentation of the measures used to construct the variables is documented in the Center for Interactional Research in Human Development and Learning (Note 22). • Neonatal Physical Examination Scale. The neonatal physical score was constructed of measures, taken during the first five days of life, that described the child's physical condition Measures that were included were, for example, the child's need for oxygen at birth, or degree of jaundice. Over thirty-five measures were rated both for severity and frequency to calculate a physical score. The subject's scores ranged from zero to twenty-seven with a mean of 5.4. • Neonatal Neurological Examination Scale. The neonatal neurological score was constructed of measures that described the neonatal status in regard to probable neurological injury at birth. Reflex tests, conducted on the first and fifth days of life, were included in the measure. Over fifty-nine measures were used to construct the scale. Scores ranged from 1 through 83, with a mean of 16.4. Both of the variables described above are severity scores, i.e., the higher the score the greater the indication of abnormal functioning in the subject. L ____________________________ --------------- -- --------- I I Appendix B, Part 2 This part contains the items from the Mother's Interview that were used in the present study. L ________________ -----------------------.- ,------------ ~----------·------------~------------ ·------------ !V29 Total number of constellations: 1 oo-98. ' 99 = no information I IV35 Are you working now? As what? I 00 =unskilled 01 =semi-skilled worker (cigar vendor, kiosk owner, hauler with one vehicle, nurse's assistant) 02 = skilled 03 =office, laboratory assistant 04 = independent businessman, middle-examination, kindergarten teacher 05 = lower academic, teacher 06 = academic and big businessman 07 =academic in responsible position 08 housewife, no work 09 = involuntarily unemployed 10 = disabled pensionist 11 =helper in husband's business 12 = student 99 = no information . V80 What kind of day care did you use most frequently when the index child was under 3 years old? 0 = not applicable 1 =full-day institution 2 = half-day institution 3 = day nurse 4 = father 5 = 6 = 7 = 8 9 = grandmother, other relative young girl various babysitters other no information V81 What kind of day care did you use most frequently when the index child was between 3-6 years old? 0 =not applicable 1 =full-day institution 2 = half-day institution 3 = day nurse 4 = father 5 = 6 = 7 8 = 9 = L __ ~ -- --- ------ -- grandmother, other relative young girl various babysitters other no information I I 1391 . ----~-_j V83 Vl68 Vl69 While you were working, who looked after the child when he/she was between 6-11 years old? 00 = not applicable 01 = boarding school 02 = father 03 = grandmother, or other relative 04 another adult 05 = after-school center 06 = had a part-time job--was home when the children home from school 07 = varying arrangements; at times, no babysitter 08 = older siblings 09 no one watched him/her 99 = no information came Who took care of the child when he/she was over 11 years old? 00 = not applicable 01 = boarding school 02 father 03 = grandmother, or other relative 04 = another adult 05 = after-school center 06 = had a part-time job--was home when the children came home from school 07 = varying arrangements; at times, no babysitter 08 = older siblings 09 no one watched him/her 99 = no information Has your family ever experienced periods of unemployment? 1 no, never 2 once, or a few short periods 3 = many times, or one long period 4 = almost always 9 = no information What kind of work does/did your husband have? 00 01 = 02 = 03 04 = 05 = 06 = 07 = unskilled semi-skilled (cigar vendor, kiosk owner, hauler with one vehicle, nurse's assistant) skilled office, laboratory assistant independent businessman, middle-examination, kindergarten teacher lower academic, teacher academic and big businessman academic in responsible position G/l69 (continued) i I ' 08 = no work i 09 involuntarily unemployed 10 = disabled pensionist . ' i I I I •v170 11 = helper in wife's business 12 = student 99 = no information Has there been disagreement concerning child-raising? 0 =not applicable 1 = no 2 =a little 3 = problems 4 = big problems 9 = no information Vl71 Has there been disagreement concerning the sharing of household chores? 0 =not applicable 1 = no 2 =a little 3 = problems 4 =big problems Vl72 Has there been disagreement concerning the economy? Vl73 Have there been problems regarding jealousy? Vl74 Have there been sexual problems? Vl75 Have there been problems with the in-laws? Vl76 Have there been disagreements about what to do on your vacation and in your spare time? Vl77 Have there been other problems? (describe) V258 Are you happy with your neighborhood? 1 = very happy 2 satisfied 3 unhappy 9 = no information ------~--------~----·----- 141 I I Appendix B, Part 3 This part contains the items from the social worker's Appendix to the Mother's Interview that were used in the present study. ' I i i 1421 ---- _J I I V263 V265 V267 V274 V281 2) 3) 5) 7) 14) 21) Was the home reasonably furnished? 1 = very bad 2 = bad 3 = average 4 = very good 9 =no information Did the home appear neat and clean? 1 =very messy and dirty 2 = somewhat messy 3 = average 4 = perfect 9 = no information How was the mother dressed? 1 = very poorly dressed 2 = somewhat poorly dressed 3 = average 4 =well-dressed 9 = no information Did the mother appear well-groomed? 1 =very slovenly 2 = somewhat slovenly 3 average 4 =very well-groomed 9 = no information How did the mother•s attitude towards the child seem? 1 very negative 2 = somewhat negative 3 = somewhat positive 4 = positive 9 no information Does the mother seem to live in isolation in relation to her husband, children and the outside world? 0 =not applicable 1 = 2 = 3 9 = no somewhat markedly no information L-----------------~------- -------- -- ~--------~-------~----~-~-~-------------------~--- ----~-- ----- ! vL82 22) Does the family seem to live in isolation? ! ! V283 23) V284 24) 0 = not applicable 1 = no 2 = somewhat 3 = markedly 9 = no information Mother's overall degree of contentment? discontent content --- Mother's acceptance of her situation and her ability to function in it: poor acceptance good and functioning·------------ ---aacceptance and func tioning r---------- -------·-·-----·------------------~---- ! I Appendix B, Part 4 This part contains the items from the Teacher 1 s Questionnaire that were used in the present study. ~---------------~1----;, we_l_l_beT ow -a-verage __ _ : 2 = below average : 3 = average I I 4 5 = above average =well above average ! !Column Item i 01 02 06 07 08 10 11 13 14 15 16 20 24 25 40 45 48 49 50 51 52 Understands material presented orally by others • . • . . . . . . • . . . • Orally expresses ideas and concepts c 1 ear 1 y • . • . • . . • • • Vocabulary or word power •• Reading Comprehension • Reading speed Understanding of mathematical concepts. Numerical and computational skills. Works up to level of ability ..•• Work products are of good quality •• Work is well-organized .• Work products are neat. Functions well in unstructured learning situations ..•. Interacts well with teachers. Interacts well with adults. Has violent outbursts ••. Is often teased by other children • Acts younger than his age .••. Prefer~ to be with younger children . Changes friends frequently •• Has many friends ... Is popular with peers L~-------------------------~---- ------~---------- 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 2 3 4 2 3 4 1 umn I tern ~----~---------------- ------------------ ------------------------ ---------- I I 53 57 58 13 14 15 16 18 52 Is identified as a leader by peers. Bullies other children. . . Frequently gets into fights . Relationship with parents seems good. Parents seem critical of child. Parents seem to be too strict with child. . Chi 1 d seems fearful of parents. . . . Parents seem to be overindulgent with chi 1 d . . . . . . . . . . . . . . . . Did the child attend a normal class? Yes (1) Don 1 t know (3) (D) 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 2 3 4 5 No (2) 53 Did the child ever remain an extra year in one grade? (E) Yes (1) ; No (2) ; Don't know (3) 54 Did the child receive any kind of remedial instruction? Yes ( 1 ) ; No ( 2) ; Don 1 t know ( 3) 147

Asset Metadata

Creator Brock, William Michael (author)

Core Title The influence of child day care on cognitive and phychosocial functioning in adolescence

Contributor Digitized by Interlibrary Loan Department (provenance)

Degree Doctor of Philosophy

Degree Program Education

Defense Date 05/01/1981

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

Tag OAI-PMH Harvest

Format application/pdf (imt)

Language English

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-490718

Unique identifier UC11286673

Identifier etd-Brock-578190.pdf (filename),usctheses-c3-490718 (legacy record id)

Legacy Identifier etd-Brock-578190.pdf

Dmrecord 490718

Document Type Dissertation

Format application/pdf (imt)

Rights Brock, William Michael

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

Linked assets

University of Southern California Dissertations and Theses