Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Strategies Of Marital Communication
(USC Thesis Other)
Strategies Of Marital Communication
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This material was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted. The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction. 1.The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity. 2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame. 3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again — beginning below the first row and continuing on until complete. 4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding of the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced. 5. PLEASE NOTE: Some pages may have indistinct print. Filmed as received. Xerox University Microfilms 300 North Zeeb Road Ann Arbor, Michigan 48106 75-1049 BUSCH, Lewise Langston, 1943- STRATEGIES OF MARITAL COMMUNICATION. University of Southern California, Ph.D., 1974 Psychology, clinical Xerox University Microfilms, Ann Arbor, Michigan 48106 THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED. STRATEGIES OF MARITAL COMMUNICATION by Lewise Langston Busch 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 (Psychology) August 1974 UNIVERSITY OF SOUTHERN CALIFORNIA T H E G R A D U A TE SC HO O L U N IV E R S IT Y PARK LO S A N G E LE S , C A L I F O R N IA 9 0 0 0 7 This dissertation, written by Lewise Langston Busch under the direction of h3X... Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillm ent of requirements of the degree of D O C T O R O F P H IL O S O P H Y Dean DISSERTATION COMMITTEE Chairman, ACKNOWLEDGMENTS I wish to express my gratitude to Dr. Milton Wolpin, Chairman of my Dissertation Committee, for all that he has offered in our relationship and for his pa tience and support. I also wish to thank Dr. Steven Fran- kel and Dr. Carlfred Broderick for serving as members of my committee. There are several other people who have helped par ticularly with this dissertation. My husband, Dick, of fered assistance and encouragement, day in and day out. Without him, I might never have completed this work. Elizabeth Eugenie Hartmann, my research assistant, Hunter and Colleen Beaumont, and James Lakehomer have each gone beyond the calls of duty or friendship in helping with this research. I also thank my parents, Beach and Catherine Langston, for their very real support in this venture. ii TABLE OF CONTENTS Page ACKNOWLEDGMENTS ..................................... ii LIST OF TABLES..................................... V Chapter I INTRODUCTION ................................ 1 Purpose of the Study Importance of the Study Limitations of the Study Methods of Research Hypotheses Definition of Terms Used II REVIEW OF FAMILY AND MARITAL INTERACTION RESEARCH ...................... 16 Contributions from Small Group Research Reviews of Family Interaction Research Survey of the Methods of Family Interaction Research Evaluation of Research Methods III REVIEW OF THE SPECIFIC THEORETICAL AND EMPIRICAL BACKGROUND OF THE RESEARCH................................... 74 Theoretical Background Empirical Background Rationale Hypotheses IV M E T H O D .................................. . . 131 Subjects Topic Selection Procedure Procedure Coding Procedure iii Chapter Page V RESULTS 140 Demographic Characteristics Discussion Topics Interrater Reliability Numbers of Messages Used Strategies of Communication VI DISCUSSION................................... 172 Codification Subject Selection Procedures Procedures for Generating Interaction Results Experimental Results Marital and Family Interaction Research VII SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS . 186 Conclusions Implications for Further Research BIBLIOGRAPHY 193 APPENDICES 209 LIST OF TABLES Page Table 1 A COMPARISON OF SECOND AND THIRD DIGIT CODES: MARK (1971) VERSUS ERICSON AND ROGERS (1973)............................ 99 2 A COMPARISON OF CONTROL DIMENSION CODES: MARK (1971) VERSUS ERICSON AND ROGERS (1973) 100 3 GILL'S RELATIONSHIP CODING GUIDELINES........... Ill 4 DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLES........................................ 141 5 DISCUSSION TOPIC RANKS FOR CLINIC AND NON-CLINIC GROUPS ............................ 143 6 TOPICS DISCUSSED BY CLINIC AND NON CLINIC COUPLES..................................144 7 INTER-RATER AGREEMENT ......................... 147 8 NUMBER OF MESSAGES U S E D ........................ 149 9 SUMMARY TABLE FOR THE ANALYSIS OF VARIANCE FOR NUMBER OF MESSAGES PER DISCUSSION .................................... 150 10 MEANS AND STANDARD DEVIATIONS OF COMMUNICATION STRATEGIES ...................... 154 11 SUMMARY TABLE OF MULTIVARIATE ANALYSIS OF COMMUNICATION STRATEGIES ................. 156 12 CELL MEANS FOR ONE-DOWN AND SYMMETRICAL RESPONSES TO ONE-UP AND SYMMETRICAL STIMULI........................................ 163 13 SUMMARY TABLE FOR THE ANALYSIS OF VARIANCE COMPARING ONE-DOWN AND SYMMETRICAL RESPONSES FOLLOWING ONE-UP STIMULI........................166 V Table Page 14 RESULTS OF SCHEFFE'S TEST FOR SIGNIFICANT CONTRASTS IN THE CLINIC STATUS X DIS CUSSION X RESPONSE CLASS INTERACTION, U.D. VERSUS U.S ...................... 15 SUMMARY TABLE FOR THE ANALYSIS OF VARIANCE COMPARING ONE-DOWN AND SYMMETRICAL RE SPONSES FOLLOWING SYMMETRICAL STIMULI . . Vi . 167 . 170 CHAPTER I INTRODUCTION Married couples provide an opportunity for the study of behavior in the context of chosen, prolonged, in terdependent relationships between men and women. No other human groups so fully evince these characteristics. The current general systems theory recognizes the circular nature of causality in marital and family interaction. The action of one person is understood to be both a stimulus for, and a response to, the behavior of another. Investigators from the Mental Research Institute in Palo Alto have been among the leaders in developing this theoretical framework. Their discussions have concentrated on understanding those aspects of communication which have implications for the control of relationships. However, they have not operationalized the strategies of communica tion which they have delineated and to which they have at tributed a great deal of influence on mental health. The present study was addressed to this problem. 1 2 Purpose of the Study It was the purpose of this study to: 1)develop a satisfactory approach to coding communication strategies in contexts of marital agreement and disagreement; and 2) use this procedure in an investigation relating the strate gies to the variables of clinic status and gender, and to the conditions of agreement and disagreement over various topics related to marital satisfaction. Importance of the Study Interest in the study of marital and family inter action is increasing steadily. Beginning in the 1950's with Strodtbeck's (1951) development of the Revealed Dif ferences Technique for the study of conflict resolution, and gaining momentum in the sixties, investigators ap proached the field from a number of disciplines. The study of small group behavior developed within sociology and con tributed important concepts like "power," "role," "domi- nance-submission," and "decision-making." Methodology from this field often has been adapted for the study of familiesJ Bales' Interaction Process Analysis (1950a) has been a par-' 3 ticularly influential technique. From the discipline of experimental psychology, work has appeared on learning, perception, and cognition in family contexts. A great deal of the research and theory in this field has come from the clinical tradition. Spurred by the hope that family thera py might prove helpful, especially in treating the schizo phrenic, clinicians primarily have concentrated on discern ing differences between "normal" and "disturbed" families. Riskin and Faunce (1972) reviewed the field and pointed out that investigators from the different disciplines seldom have been familiar with the research generated within other disciplines. Much repetition of effort, methodological naivete, and- conceptual confusion have resulted. One of the most irritating problems plaguing the field is the plethora of intriguing concepts with inade quate, incomparable, or absent operational definitions. Even such basic ideas as "conflict" have been measured in so many different ways as to make results from one study irrelevant to the results of other investigations of the same phenomena. This dissertation was designed in part as a re sponse to the problem posed by inadequate operational defi nitions of some strategies of communication deemed impor tant for understanding the ways in which family members 4 continually define and redefine their relationships. Ruesch and Bateson (1951) pointed out that communication is composed of a relationship and a content component. Theo rists from the Mental Research Institute at Palo Alto, in cluding Haley (1963) and Watzlawick, Beavin and Jackson (1967), asserted the importance of the relational component in the development of psychopathology. They interpreted psychiatric symptoms as messages used to constrict the be havior of others without the sender's accepting responsibil ity for the ever-present directives defining the relation ship. The most frequently used concept proposed by these theorists has been the double bind. Whereas much effort has gone into studies of the double bind in relation to schizo phrenia, no research has appeared which adequately defines the concept at the operational level. Olson (1972) attribu ted much of the difficulty of operationalization to the in consistent and esoteric qualities of the theoretical defi nitions . Although the other strategies of communication dis cerned by the Palo Alto group are more straight-forward, they have received less attention. Jackson, Riskin and Satir (1961) suggested that patterns of symmetrical and com plementary responses could be used to categorize families, and that classifications based on these patterns might "be 5 correlated with specific manifestations of mental illness." (p. 31). In this paper, they also stressed the importance of observing whether statements were qualified in a congru ent manner or disqualified in a paradoxical manner. Obviously, terms such as "symmetrical," "complemen tary," and "paradoxical" required operational definition to permit movement from theoretical speculation to empiri cal investigation. A few investigators have worked on this task. Sluzki, Beavin, Tarnoplosky and Veron (1967) began with a theoretical paper suggesting that stimulus and re sponse pairs of statements could be classified as symmetri cal or complementary transactions. They presented examples of types of interchanges which could be grouped in each category. Mark(1971) developed an interaction coding sys tem using the work of Sluzki et. al. as a base. However, I his system had many weaknesses, among which were a failure i to maintain an emphasis on the relationship component of I communication and an indefensibly narrow definition of sym- , metrical. Ericson and Rogers (1973) revised Mark's ap proach to coding. While they did improve the definition of symmetrical, eliminate overlapping categories, and make the system more logically coherent, they introduced some very troublesome problems of their own. They established the category of "one-across" messages, which were considered i to be transitory statements, suggesting that control of the f | 6 relationship was being "neutralized" when they were used. Yet, the Palo Alto theorists have argued that one cannot not define a relationship when communicating. If one is neu tralizing control asserted by one's partner, one is taking control oneself. Gill(1971) developed a more adequate approach to coding, and he included "paradoxical" as a category. His guidelines provided for coding with a sustained emphasis on the relationship level of communication, and they were quite consistent with the theroetical definitions. How ever, they were designed specifically for use in his ex perimental context, in which the experimenters adopted a complementary one-up style of communication with the sub jects, who were college students. The present research was designed using a revision of Gill's approach. His guidelines were modified for use in coding discussions between husbands and wives about im portant areas of their relationships over which they: (1) agreed, or (2) disagreed. This task provided the op portunity to test several predictions about which strate gies would be used most frequently in each of the two con ditions. The use of "disturbed" couples in marital thera py and "normal" couples as subjects permitted the assess ment of several theoretically-based predictions relating the use of particular communication strategies to the 7 level of disturbance in the relationships. The results, in turn, were used in assessing important aspects of the theo retical base within which the strategies of communication were delineated. Thus, this study represents an attempt to fill a gap in the literature by providing adequate operational definitions of the strategies of marital communication, and seeing if they can be utilized reliably in the context of agreement and disagreement between spouses. It also pro vides a partial test of the theoretical framework developed by the Palo Alto theorists for understanding marital and family communication. Limitations of the Study This study was not proposed as a definitive test of the Palo Alto theorists' beliefs about the strategies of communication. Rather, the results were to be construed as lending support to, or casting doubt on, the validity of their ideas. Even as a clear test of the predictions gen erated from the theory, this research has several limita tions . Some limitations resulted from the choice of sub jects. First, this study used no measures of marital dis turbance or individual psychopathology other than whether or not the subjects were in marital therapy and were con 8 sidered by the referral sources to be "disturbed" or "nor mal." It was considered possible, but not likely, that the selection procedures did not generate two distinct groups of subjects with different levels of marital disturbance. Second, the subjects were not a random sample of the popu lations compared. Subject willingness to participate may have biased the sample in some way affecting the results. Third, the different ways of recruiting subjects for the two groups could have generated different sets for the therapy and the non-therapy couples. Being referred by one's therapist probably is different from being referred by a minister or a graduate student. No better approach to recruiting was found. Another limitation was introduced by the sample size. Larger absolute differences in out comes were required to reach statistical significance with ten couples (20 people) in each group than if the sample had been larger. The experimental procedures introduced other limi tations. The sessions were aonducted with an experimenter present. One could argue that communication strategies would be different in such a context from those used when the same issues emerged at home. The fact that the dis cussions were directed toward particular topics also might have changed the character of interaction from what it would have been if it had emerged spontaneously. No solu 9 tion to these problems were found. Rather, it was assumed that the influence of the setting and the semi-structured nature of the task would not be enough to obscure complete ly the differences which would emerge in a more natural setting. Another limitation of the study relates to the topic selection procedure. Since different couples dis cussed different topics, one could argue that the content of the topics influenced the communication strategies used. The alternative of standardizing topics would have run the risk of generating discussions of subjects with little im portance to the couples, if the agreements and disagree ments were rigged, or of having couples discuss topics with very different meanings in their relationships. The final limitation to be mentioned was produced by using audio-tape instead of video recordings, which would have captured another channel of communication. Methods of Research This dissertation presents the results of a search of the literature on family interaction and an empirical study. The empirical research first required the develop ment of a scheme for coding strategies of marital communi cation. For this purpose, major modifications were made in the coding guidelines utilized by Gill (1971). A pilot 10 study investigating the inter-rater reliability of the scheme was conducted in the summer of 1973. Recordings of three couples discussing a total of six area of agreement and disagreement taken from the Marital Agreement Question naire were used. Over-all agreement between raters was 94%, while agreement between raters on coding the separate stra tegies ranged from 83% for paradoxical responses to 90% for symmetrical responses. These levels of agreement were deemed high enough to justify proceeding with the larger study. Twenty couples, subjects for the study, were re cruited and participated in the experimental sessions be tween March and May, 1974. Ten couples were in marital therapy; ten were not. Each couple was seen for about an hour at the Pastoral Counseling Center of Claremont. The procedure included two ten-minute discussions between the partners. An area of agreement and an area of disagree ment about issues considered important to the marital re lationship were discussed. The resulting samples of inter action were recorded on tape, then transcribed and coded. The data were analyzed with the use of an IBM 360 terminal at the computer center of the Claremont Colleges. 11 Hypotheses The specific hypotheses examined here were: Paradoxical Responses 1. The percent paradoxical responses will be greater in the disagreement condition than in the agree ment condition. 2. The percent paradoxical responses will be greater among clinic subjects than among non-clinic sub jects . 3. The difference in the percent of paradoxical re sponses between the two conditions will be greater for the clinic subjects than for the non-clinic subjects. Transactional Disqualifications 4. The percent transactional disqualifications will be greater in the disagreement condition than in the agreement condition. 5. The percent transactional disqualifications will be greater among clinic subjects than among non clinic subjects. 6. The difference in the percent of transactional dis qualifications between the two conditions will be 12 greater for the clinic than for non-clinic sub jects . Symmetrical Responses 7. The percent symmetrical responses will be greater in the disagreement condition than in the agree ment condition. 8. The percent symmetrical responses will be greater among the non-clinic subjects than among the clinic subjects. 9. The difference in the percent of symmetrical re sponses between the two conditions will be greater for the non-clinic subjects than for the clinic subjects. Metacommunicative Responses 10. The percent metacommunicative responses will be greater or smaller among the non-clinic subjects than among the clinic subjects. One-Down Complementary Responses 11. The percent one-down responses will be greater in the agreement condition than in the disagreement condition. 13 12. The percent one-down responses will be greater among the clinic subjects than among the non-clinic subjects. 13. The difference in the percent of one-down responses between the two conditions will be greater for the non-clinic subjects than for the clinic subjects. One-Up Complementary Responses 14. The percent one-up responses will be greater among the non-clinic subjects than among the clinic sub jects . Husband-Wife Differences No directional hypotheses. The data will be evalu ated for husband-wife differences in the use of all strate gies across and between experimental groups and conditions. Definition of Terms Used The following terms were used to label categories of responses having different implications for the control of relationships. The definitions are essentially those proposed by the theorists from the Mental Research Insti tute in Palo Alto. Strategies of communication. The different re sponse classes defined below, when considered together, are 14 called the strategies of communication. Symmetrical. A message indicating that the sender is contending with the receiver for control of the rela tionship is a symmetrical response. Symmetrical interac tions tend to be competitive in nature, as whe one person responds with a boast to the boast of another. Both people are equal. They are both trying to control the definition of the relationship. Complementary. Complementary responses include one-up, one-down, and metacomplementary messages. In com plementary exchanges, one person is in charge of the re lationship while the other accepts his being in charge. The behavior of the partners is different in nature, but fits together, as when one person gives advice and the other accepts it. Metacomplementary. A response explicitly permit ting or directing another person to define the relationship is metacomplementary. The person using a metacomplementary strategy is in control of determining who defines the re lationship, as when one person says to another, "You de cide what we should do." Metacommunicative. Messages about the nature of communication taking place are metacommunicative. Although there is considered to be a metacommunicative aspect to all communication, this term is used in this study to denote 15 explicit, overt metacommunication. Paradoxical. A response in which some aspect of the communication is disqualified or contains conflicting directives is paradoxical, as when a person says, "Don't follow my advice." Transactional Disqualification. A transactional disqualification occurs when "the response is incongruous with the frame created by the previous statement, and meta communicative indicators labeling the incongruity are ab sent" (Ransom, 1970, p. 115). Since transactional dis qualifications involve disqualification of the context in which they occur, they are considered as a subset of para doxical responses. Organization of the Remainder of the Study Chapter II contains a review of the methods of re search on family and marital interaction. Chapter III in cludes discussions of the specific theoretical and empiri cal background of this study. Chapter IV describes the experiment in detail. The results are presented in Chapter V, and are discussed in Chapter VI. The last chapter in cludes a summary, conclusions, and recommendations for further study. CHAPTER II REVIEW OF FAMILY AND MARITAL INTERACTION RESEARCH This chapter deals with the empirical literature on family and marital interaction, and is made up of four sections: (a) a specification of important contributions to this field from the area of small group research; (b) a brief overview of reviews of family interaction studies; (c) a survey of the methods used in such investi gations, including sources of data, techniques of genera ting interaction, and coding procedures; and (d) an evalua tion of these methods. The literature on the family and on marriage abounds in hypotheses relating group interaction to family member or marital disturbance. This richness of concept ualization within the field is almost matched by the vari ety of investigative methods employed. However, the rela tionships of the measures used to the concepts for which they are operational definitions often is obscure; and the field contains a quagmire of overlapping theoretical con structs. At this point in history, the body of research is growing rapidly, but the reliable information gained is 16 17 mounting slowly. Although there is a prodigious general literature in the field, the purpose of this review is to examine the field of research dealing with families and married couples as they interact. Contributions from Small Group Research Research investigating the behavior of small groups has dealt with many of the issues relevant to fam ily studies, but the results cannot be considered to shed light on behavior within families. Framo (1965) discussed several of these areas, including leadership, power, co alitions, group pressure, and communication. He concluded that the similarity of the ad hoc or short term groups to families was dubious because the relationship bonds proba bly would be much stronger in families. Although feelings of love, hate, guilt, loyalty, and disloyalty may develop in any group, they have a particular intensity in the fam ily. One hardly could expect the results of family and small group investigationsto be similar. A classical replication of a small group study in a family context demonstrated the differences. Mills (1953) found that three-person groups tended to break into a coalition and a third party— who became a scapegoat, a common enemy, or someone who tended to strengthen the co alition as much as he threatened it. Strodtbeck (1951) used families as subjects in his replication. In his ex periment, in which the Revealed Differences Technique was used to manipulate coalitions within three-person families, he found generally that the division of a triad into a coalition of two against one was much attenuated in family, as compared to ad hoc, groups. Although the findings from small group research have added little to our understanding of families, the ex perimental procedures developed for this field have contri buted a great deal. Most procedures of collecting and analyzing data have been borrowed directly from the small group field. These include the use of control groups, standard stimulus situations, direct observation and tape recording of interactions, and coding systems. Two methods taken from the small group field have been particularly influential in shaping research on family and marital interaction. Both Strodtbeck's Revealed Dif ferences Technique (1951) and Bales' Interaction Process Analysis (1950) have been used in many of these studies, and each has generated modifications specifically for use with families. Interaction Process Analysis probably was used in more studies of family interaction than any other method of analyzing communication. This system permits the in 19 vestigator to categorize interaction into four classes of activity basic to Parsons and Bales' (1955) theory of group and family interaction. The twelve categories primarily differentiate between positive and negative task-oriented and social-emotional acts. The Revealed Differences Technique is used to gen erate discussions. Family or group members are asked to come to agreement about problem situations over which they independently indicated disagreement. This approach, and modifications thereof, were used in numerous studies of conflict resolution. Besides generating data for analysis with the Bales or other coding systems, it makes possible several win-loss type outcome measures. The magnitude of the contributions of small group research methodology to the study of family interaction 4 and particularly the contributions of Bales and Strodtbeck, become clearer as the field is reviewed. Reviews of Family Interaction Research As the number of review articles in a field in creases, it can be assumed that the field is beginning to coalesce and develop. Studies of family interaction slow ly are coming to be related to each other, and a signifi cant quantity of reliable information is becoming avail able. Most of this research has attempted to discover dif 20 ferences in family interaction between "normal" families and those with at least one "disturbed" member, or between types of disturbed families. The reviews included in this section focussed mainly on this type of investigation. The first major review was presented by Spiegel and Bell in 1959. They summarized work related to the family of the patient. They attempted to delineate the general historical background and sources of influence on the early research into the role of the family in the etiology of mental illness. Foudraine (1961) focussed his review on research from 1956-1960 dealing with the role of the family in the etiology of schizophrenia. A somewhat narrower area was reviewed by Watzlawick (1963), as he summarized concep tual developments and research on the double bind theory and schizophrenia. Focussing still on studies relating family charac teristics to schizophrenia, Meissner produced a major re view article in 1964. Criticizing particularly the con ceptual sloppiness of many of the studies produced before this time, he urged researchers (1) to clarify the kinds of questions they are investigating; (2) to delineate the specific aspects of the family under study; and (3) to examine the relevance and limitations of their methods. In 1965, the Psychological Bulletin published two reviews, granting family interaction research status as a 21 field within psychology. Handel developed his review around a conceptual framework presented in an earlier book (Hess and Handel, 1959). He traced the development of the field and evaluated several major approaches. Frank fol lowed with a pessimistic review of the literature, empha sizing the retrospective nature of the research and ignor ing many important studies. For example, he cited none of the work done by Bateson, Cheek, Framo, Jackson, Haley or Wynne. He concluded that the literature did not support the assumption that the family plays a significant role in the development of psychopathology. The year 1965 also marked the emergence of two other theoretical reviews. Zuk and Rubenstein discussed a variety of concepts used in the study and treatment of schizophrenia, as revealed in Boszormenyi-Nagy and Framo's collection, Intensive Family Therapy (1965). They approach ed the conceptual problems from a historical perspective. Mishler and Waxier (1966) summarized the theoretical per spectives of the Bateson (Palo Alto), Lidz (Yale), and Wynne (National Institute of Mental Health) research groups. They took a much needed step in trying to inte grate the very divergent concepts developed by these groups. Later reviews focussed on theoretical issues in- 22 eluded Aldous1 (1970) work on the requirements for theory building in the field and Hill's (1971) work on systemiza- tion (mentioned in Riskin and Faunce, 1972). Olson, in a highly focussed review (1972) came close to dealing a death-blow to double bind theory and research, concluding that: "until more creative and valid measures are obtained, it will be impossible to empirically unbind the double bind." (p. 91). Reviews examing methodological problems as opposed to theoretical difficulties also began to appear in the 1960's. Sanua (1961, 1967) dealt primarily with methodo logy in his reviews of sociocultural factors in studies of "schizophrenic" families. Rabkin (1965) and Framo (1965) both elucidated the inadequacy of much of the work in this field, pointing out such common shortcomings as the lack of adequate control groups and disregard for socio-economic variables. Framo's review dealt specifically with studies of family interaction. Only one major reviewer chose to deal specifically with issues relating to the appropriateness of the methods used in family research and to examine the assumptions underlying them. Fontana (1966) emphasized the problems of reducing, assembling and analyzing data in this field where operational definitions and hypothetical constructs abound, but the two are often loosely related. A very extensive and thorough review of theoretical and methodological issues in family interaction research appeared in 1972. Riskin and Faunce reviewed empirical studies appearing between 1960 and 1971. Their article also reflected the results of discussions with the major investigators in the field. The present review was influenced to a great ex tent by the work of Fontana and Riskin and Faunce. No at tempt is made to cover all the work in the field. Rather, a general outline is presented, and problem areas are evaluated with the use of examples. Survey of the Methods of Family Interaction Research The next part of this review does three things: (a) surveys the kinds of data used in family interaction research and the methods used to generate the data; (b) presents the procedures used in reducing the data for analysis; and (c) evaluates the methods used in this re search. 24 Types of Data and Techniques of Data Collection The type of data used in any investigation should depend on the questions addressed by the study. Unfortu nately, the advantages of any data source are accompanied by disadvantages. In family and marital interaction re search, four general sources of data have been used. Ob servational notes are the simplest to obtain, but are highly selective. Audio-tape recordings are fairly inex pensive and easy to obtain and yield an objective record. However, they exclude non-verbal events from analysis. The major handicaps of film are its high cost and the relative ly intrusive nature of the procedures. Its use would seem appropriate only if a wide spectrum of communicational variables were being analyzed. Outcome scores, measuring how many times who did what things, provide a very limited range of data, but are strong in precision and comparabil ity of results. Investigators in this field have covered the spec trum in their preferences for different sources of data. At one extreme, Bowen (1959) and Brodey (1959) relied heavily on notes of ward observations in putting together a broad picture of the interaction in hospitalized families of schizophrenics. At the other extreme, Haley (1962) used 25 only a precisely documented, non-inferential source of data; who spoke after whom, as recorded electronically by voice-activated throat microphones connected to a small computer. Riskin and Faunce (1972) pointed out the bargains which must be struck between the advantages and disadvan tages of different data sources when an experimental design is chosen. They observed that: Highly reliable, tightly controlled laboratory conditions may be achieved, but the price may be the loss of the complex human aspects of the fami ly interaction. Precision may lead to sterility. (p. 377) The different means of gathering data have been used in a variety of settings, ranging from observations of ongoing interaction in the homes, to therapeutic inter views in the clinic, to problem solving in the research institute. Numerous procedures to generate data also have been utilized. In presenting the methods used in this process, the more naturalistic approaches will be presented first. Studies using the most structured and artificial tasks will come last. 1. Techniques using "natural" events. The most natural setting for observation is the home, and a few investigators have carried out their research in homes. 26 Henry (1963) lived for weeks at a time with several fami lies containing psychotic children. He made minute anthro pological observations of behavior. After all the data were collected, he analyzed his notes and interpreted the findings. In a giant study, McCord, Porta, and McCord (1962) reported on observations of 255 boys and their fam ilies at home. The observations were made every two to three weeks for five years. Behrens and Goldfarb (1958) and Lennard, Beaulieu, and Embrey (1965) analyzed data from tape-recorded dinner table conversations. Families waiting for clinic appointments or visit ing in hospitals have also been used in investigations of natural interaction. Layman and Lourie (1958), for ex ample, had two observers take notes on behavior in families waiting for psychotherapy. The Goldfarb group recorded and filmed mothers visiting their children in hospitals. Levy, Goldfarb, and Meyers (1963) reported data covering twenty such visits. The concept of "parental perplexity" develop ed from this work. Although the concept of natural events is strained by the inclusion of observations of hospitalized families, Bowen (1960) and Brodey's (1959) work sponsored by the National Institute of Mental Health is included here be cause it utilized voluminous notes on family interaction made by members of the hospital staff. Recordings of 27 daily therapy sessions also were used. These studies re ported on five hospitalized families containing adolescent schizophrenics. The family triads were hospitalized for periods lasting as long as two and one half years. Several concepts such as "externalization," "narcissistic relation ship," and "emotional divorce" evolved from observations of these families. 2. Techniques using interviews and psychotherapy sessions. The least structured, most "natural," kind of discussions were used as sources of data by Titchener and Emerson (1958). They used recordings of "family meetings," in conjunction with individual interviews and family re lations inventories, in their study of families applying for psychiatric help. Recordings of therapy have provided samples of family interaction for numerous investigations. Some ex amples will illustrate the variety of ways in which such recordings have been employed. Haley (1960) used therapy protocols and films of loosely structured family discuss ions in his earliest research. From such work, he develop ed his views stressing the difficulty of coalition forma tion and maintenance in families of schizophrenics. Ry- ckoff, Day, and Wynne (1959) used taped interviews and clinical material from psychotherapy sessions in their in vestigation of stereotyped role playing in families of 28 schizophrenics. Morris and Wynne (1965) predicted the diagnoses of hospitalized offspring from coded excerpts of family therapy. Jones (1964) studied supposed double binds; and Sluzki, Beavin, Tarnoplosky, and Veron (1967) and Ransom (1970) explored transactional disqualifications found in excerpts of family therapy recordings. Jackson and Yalom (1966) found still another use for observations taken from family therapy, which they related to data on family member activity outside the home, in their study of families containing patients with ulcerative colitis. Standardized interviews have been used to generate clinical information and to provide samples of communica tive behavior. Investigators from the Mental Research In stitute (MRI) in Palo Alto have developed the most popular standardized approach, which comprises their family intake procedure. Rosenbaum (1961) , for instance, examined pa tient family similarities using tapes from MRI. Watzlawick (1966) described in detail the whole structured interview. Other standardized interviews, too, have been util ized in family studies. Baxter, Arthur, Flood, and Hedge peth (1962) and Baxter and Arthur (1964) developed such an interview to explore conflict patterns in the families of schizophrenics. For this research, families were seen to gether during the last halves of three-to-four hour inter views about family history. Notes from the proceedings 29 were transcribed after the sessions. Another use for structured interviews was developed by Beavers, Blumberg, Timkin, and Weiner (1965). They coded recordings of inter views with mothers of hospitalized schizophrenic and pas sive-aggressive males. The sessions focussed on the mothers' feelings about the sons, but were coded only for variables other than content. 3. Techniques using discussions of specified topics. Discussions of particular topics have been used numerous times to generate samples of family communication in which all members would seem to have an interest in participating. The MRI interview segment in which the family is asked to "plan something you could do together," falls in this category. Riskin (1964) utilized his Family Interaction Scales using excerpts from these discussions. Terrill and Terrill (1965) also used this technique to demonstrate the usefullness of their revised scheme for coding Leary's dimensions of interpersonal behavior. Len- nard, Beaulieu, and Embrey (1965) had their families talk about three topics relating to the duties, plans, and schooling of the schizophrenic and "control" sons. These Subjects were discussed first by the mothers and sons, and then by both parents and the sons. They were coded with still another coding system, in which the labelling of be havior held a prominent place. Other work also has used 30 structured discussions to try out new coding procedures. Mark (1971) and Ericson and Rogers (1973) utilized record ings of such conversations as the testing grounds for meth ods of coding the MRI strategies of communicaiton. The Revealed Differences Technique (RDT) and its modifications have been used in numerous studies of marital and familial conflict resolutions. It has proven to be relatively simple to use and easily adaptable for different purposes. Strodtbeck, after a study of power relations in a variety of ad hoc groups, developed the RDT first for use with married couples (1951). Each couple picked three reference couples they knew well, then husbands and wives were separated and asked to choose the couples which best filled each of 26 conditions, such as being most religious or having the happiest children. Then the couples were brought together and asked to reconcile their differences. The discussions were recorded, and some outcome scores and Bales interaction coding were used in analyzing the re sults. This study reported on the use of the RDT in three distinct subcultures in the southwest United States. Strodtbeck (1953) used the RDT in what Handel (1965) called the first family experiment. He saw adoles cent sons with their parents at home and asked them to com plete forced-choice questionnaires relating to the son's occupational choice. He asked the triads to discuss and | 31 come to agreement on their answers to nine of the items. The topics were chosen for representing each possible com bination of two member coalitions with one member original ly dissenting. Protocols of these discussions were coded later with Bales' twelve categories. Mishler and Waxier (1968), Schuham (1970), and Smith (1971) provided other examples of the use of the RDT. Several modifications of this procedure have been introduced. Farina (1960) and Farina and Dunham (1963) used the RDT with questions taken from Jackson (1956). The questions related to the ways parents should handle possible misbehavior by teenagers. The participants in the study were parents of good and poor premorbid schizo phrenic and tubercular sons. Lerner (1965) used the RDT with parents of schizophrenic sons rated as having dif ferent levels of thought disorder. The topics related to family problem situations. Titchener, D'Zmura, Golden, and Emerson (1964) revealed differences in questionnaire responses to each participant by writing all responses on each response sheet. They also filmed two three-minute samples of the interaction resolving differences over some controversial issues in American life. Stabenau, Tupin, Werner and Pollin (1965) followed suit. Caputo (1963) varied the number of choices available to the sub jects by asking them to rate concepts on the Semantic Dif- 32 ferential. Ferreira introduced a more substantial modification of the RDT with what he called the technique of unrevealed differences. In this paradigm (see, e.g., Ferreira, 1963; Ferreira and Winter, 1965, 1966, 1968a; and Haley, 1967), the family members were brought together to reach a deci sion about which response is best for a given situation, but they were not made aware of the private opinions of the other individuals first. Ferreira argued that this method approached the natural conflict situation more closely than did the RDT because no outsider revealed the differences to the participants. Ferreira's approach also introduced two other changes from the RDT. He used rela tively "neutral" questions and required subjects to rank- order three preferred responses and eliminate three re jected choices from a list of prepared answers. Another technique apparently using unrevealed dif ferences was developed by Cheek (1964a, 1964b, 1965a, 1965b) for her investigations of the convalescent adjust ment of young adult schizophrenics living at home. The twenty topics all dealt with problems of such young people; half related to patient behavior and half related to paren tal behavior. The participants checked off those problems which had come up in their own situations, thus providing a measure of topic relevance. Two discussions of questions 33 on which there had been disagreement were recorded. Three more variations of the RDT were developed specifically for use with married couples. The Color Matching Test (CMT) represented an attempt to build a conflict resolution task in which the stimuli were neutral and differences of opinion could be manipulated. This yielded gains in methodological precision accompanied by reduced importance of the conflicts to the participants. Couples were instructed to pick colors from separate in dividual displays which would match the colors held up by the experimenter. After each subject signalled that he had found a match, the reference color was removed and the cou ple was asked to decide together on a match without seeing each others1 displays. Half the decisions were rigged to produce disagreement; agreement was expected on the others. The discussions were recorded, then usually analyzed with the Bales codes. Studies by Goodrich and Boomer (1963), Ryder (1966, 1968, 1970a), and Ryder and Goodrich (1966) used this technique. Olson and Ryder (1970) developed the Inventory of Marital Conflicts to overcome the relevance problem of the CMT and the fact that couples disagree on and discuss dif ferent items and different numbers of items on the RDT. Choosing items considered relevant in data from couples married about two years, Olson and Ryder had the partners 34 indicate whether the problem presented was similar to one they had experienced or to one people they knew had experi enced. To control the items on which disagreement would occur, the problems were slanted differently in the ex positions for the husbands and for the wives two-thirds of the time. The topics were presented as difficulties of hypothetical couples. Tapes of the discussions were analy zed with their own coding system, and win-loss and rele vancy scores were calculated. Recently, Olson and Ryder (1973) developed the Inventory of Family Conflicts for use with couples in the child-rearing years of marriage. The format and procedures were the same as for the Inventory of Marital Conflicts. However, the problems related to child-rearing, and rele vance was measured by asking how important each choice was for the subject. 4. Techniques using traditional psychological tests. Psychodiagnostic instruments have been introduced into the study of family interaction by several investiga tors. The work of Singer and Wynne (1966) with the Rorsch ach is perhaps the best known. Loveland, Wynne, and Singer (1963) described two approaches; one in which individual Rorschach responses were obtained before conjoint responses and one in which only conjoint responses were obtained. Each family member was asked to write down the responses 35 on which all could agree. The discussion format was simi lar to the unrevealed differences techniques described above, but included measures of each subject's perceived agreement in a task with unlimited possible responses. In addition, the Rorschach responses could be interpreted from within a more traditional psychodynamic framework. The Thematic Apperception Test has been particular ly popular with investigators in this field. The stimulus pull toward family themes and the projective nature of the test have attracted attention. Some variations exist in the methods of using the cards, the numbers of cards used, and whether or not individual stories are elicited before the family group makes up joint stories. The variations will not be described here. Investigators using the family TAT have included: Fisher, Boyd, Walker, and Sheer (1959); Levinger (1963); O'Rourke (1963); Haley (1964); Stabenau, Tupin, Werner, and Pollin (1965); Fereira, Winter, and Poindexter (1966); and Winter and Ferreira (1967). Turning to the more structured psychological tests, the Comprehension and Similarities subscales of the Wech- sler Adult Intelligence Scale were used in a difference resolution format by Sharan (1966), though the technique was described earlier by Roman and Bauman (1960). This approach generates efficiency ratios of correct answers obtained together to correct answers offered by any of the 36 participants. Bauman and Roman (1966) used this technique, then revised it to utilize the full WAIS for measuring "marital intelligence" (Roman, Bauman, Borello, and Melt- zer, 1967). 5. Techniques using structured family tasks. Per haps the least structured method falling in this category is the Referential Communication Technique, which requires subjects to instruct others (usually children). Measures of compliance with directions given by the subject were ob tained. Studies by Levin (1966), Haley (1968), and Anandam and Highberger (1972) utilized this sort of task. Family games also have been used to generate out come and communication data. Perhaps the best known is SIMFAM (simulated family interaction game), which has been used in a variety of settings by Straus (1964, 1968) and Straus and Tallman (1971). The SIMFAM requires participants to play a laboratory game involving flashing lights, pucks, and pushers in simulated crisis and non-crisis conditions. The subjects reach joint decisions about strategies as they play. Reiss (1967, a, b; 1968; 1969; 1971 a, b, c, d) went further in developing family games. He designed three puzzle solving tasks which yield records of concurrent group and individual thinking as each solution progresses. These records allowed Reiss to distinguish and relate indi- 37 vidual and family approaches to dealing with environmental tasks. The games were designed to yield non-verbal, ob jective data on: pattern recognition and hypothesis test ing, categorization, and communication. Participants com municated with each other by passing sequentially numbered notes. Reiss developed his theory of "consensual experi ence" from work using these games. Perhaps the most structured family interaction task yet developed is Haley's (1962) Coalition Game. This pro cedure excludes verbal content from the interaction and analysis. It consists basically of a situation in which family members cannot see each other, and must contact each other by pressing buttons. To win, each participant must shift coalitions (signalled by pressing the correct button for a partner who is pressing a button for oneself) and try to gain scores from other players. The scores are kept electronically. Haley used this procedure in an examina tion of coalition formation in families with and without schizophrenic off-spring. Several other investigators have developed other structured interaction tasks for use in family studies. Examples of such techniques are: Wild's (1965) Object Sorting Test; Bing's (1970) family drawing technique; and the Marbles Test used with families by Usanivadras, Grim- son, Hammond, Issaharoff, and Romanos (1967) . 38 6. Techniques using multiple methods of generating interaction. Many investigators have used more than one technique in sampling interaction. This is particularly true in the larger studies, especially Bowen (1960) and Brodey's (1959) work. Other writers using more than one of the approaches mentioned above will be listed in the approximate order of their publications. Cleveland and Longaker (1957) gathered data from psychotherapy sessions and home visits. Fisher, Boyd, Walker, and Sheer (1959) made "interaction evaluations" from family discussions of the son's assets and faults and stories told to TAT cards. Meyers and Goldfarb (1961) observed dinner table conversa tions at home and used a semi structured interview called the "Maternal Interview Technique." Bell (1962) recorded clinical material and home vists. Levinger (1963) used ten different tasks including discussions of where to go on a holiday and telling stories to TAT cards. Drechsler and Shapiro (1964) administered a 20-item questionnaire which the family discussed, in combination with a family diagnostic interview. Wynne and Singer (1963a, b, 1965) and Singer and Wynne (1965) used several techniques in cluding the RDT, taped waiting-room conversations, and fami ly Rorschach protocols. (Other techniques were not speci fied) . Haley (1964, 1967) used family responses to a questionnaire in conjunction with, stories told to TAT 39 cards. Borke (1967) recorded dinner table conversation and asked her subjects to do four family tasks: telling stor ies to three TAT cards chosen out of five; deciding what to do with $1,000 received unexpectedly; choosing between sets of construction materials and agreeing on, and building a structure; and choosing a gift from a selection of gifts differing in attractiveness to different family members. Rosenthal, Behrens, and Chodoff (1968) used recordings of family interviews and the Relation Rorschach in their study focussing on family interaction in the homes of low er-class schizophrenics. Now that the major methods of generating family interaction for investigation have been presented, the dif ferent approaches to reducing such data for analysis will be identified and enumerated. The Process of Coding Interaction for Analysis In family or marital interaction research, the in vestigators must decide which aspects of the data they will take as their focus. Coding, or focussing, is the process of reducing interactional information to second order vari ables for analysis. Codification and analysis are two separate steps, which should be related to each other. One could, for example, use Bales* codes and analyze the 40 data in several different ways. Role induction, agreement-disagreexnent, and ratios of instrumental and ex pressive acts all have been used in studies utilizing these codes. In this section, an attempt is made to organize and present the major approaches used to code and classify family interaction data. The task is complicated because the approaches vary in the units of behavior coded and in the levels of inference required for their use. The fol lowing code types have been identified for purposes of this presentation: structural, individual, transactional, content or thematic, outcome, other types, and uncoded ana lyses. Brief descriptions and examples of these types are given below. 1. Structural codes. Structural coding consists of operations performed on data which require no inferences about relational behavior or the content of communication. Interpretations related to content or relationships are then based on frequency, duration, or sequences of objec tively identifiable events. These are the most "objective" measures of interaction other than strict outcome codes. The best known and most mechanical structural mea sure is Haley’s (1964) index of who speaks after whom. His subjects wore voice-activated microphones connected to a Family Interaction Analyzer which recorded this variable, 41 and to an event recorder which kept a sequential record of the conversation. No inferences at all were required in making these measurements. The other structural measures require human inter vention, and sometimes inference, in coding the material. As Haley (1964) pointed out, the question of who is speak ing to whom can require rather complex judgments. Titchener and Emerson (1958) first used this variable in discussions among families applying for psychotherapy. Neither they nor Drechsler and Shapiro (1964) used con trol groups when they used this measure, Riskin and Faunce (1970a, b) and O'Conner and Stachowiak (1971) also used this variable and did use control groups. Fisher, Boyd, Walker, and Sheer (1959) compared verbatim notes on discussions between parents of schizo phrenic, neurotic, and normal veterans on tallies of: who spoke first, who spoke most, and who initiated new topics most after the partner had finished speaking. Although Ferreira's emphasis generally has been placed on content variables in decision-making processes, he directed one study using structural variables as a prime source of data (Ferreira, Winter, and Poindexter, 1966). The authors compared normal and abnormal families on measures of time spoken, amount of overlapping speech, who talked after whom, and silence. They found significant differences between groups only on silence. Reiss (1967) and O'Conner and Stachowiak (1971) joined numerous others in coding interruptions and over lapping speech. Leighton, Stollak, and Fergusson (1971) used these behaviors in indices of "dominance" and "power heirarchies." Studies directed by Farina used the number of times people spoke, total time and percent of time spoken, fre quency of simultaneous speech, and the number of inter ruptions. Farina (1960) compared parents of good and poor premorbid schizophrenics and tubercular sons on these and other measures. Farina and Dunham (1963) assigned good and poor premorbid schizophrenics and their parents to experi mental and control conditions in their study of the impact of exposure to parents on the son's efficiency in the pres ence of cues related to parental scolding. Farina and Holzberg (1968) compared three groups of sons hospitalized for psychiatric reasons and their parents in their study of interaction patterns. Lennard, Beaulieu and Embrey (1965) measured "in trusions" and "successful intrusions," defined in struc tural terms requiring inferences about whether or not the third person's entry into the conversation was requested or elicited by the two people talking together before the speech of the third party. Intrusions were interpreted as 43 indices of transactional styles, and the conclusions indi cated that the members of families of schizophrenics were more "impervious" to actions of others which were not in response to behavior initiated by themselves, than were members of control families. 2. Codes of individual behavior. For purposes of this exposition, an individual code is defined as a code of an individual act or acts for which the coding does not rely on structural qualities or the specific behavior of another person. This definition contrasts with that of transactional behavior, in which the coding of an act is impossible without reference to the behavior of someone with whom the subject is interacting. Coding these behav iors requires decisions by raters about whether or not certain behaviors belong in certain categories. Further, they require decisions about whether or not certain behav iors fulfill the requirements of whatever definitions of scoreable units of behavior are being used. Some of the dimensions of personal behavior designated for coding and analysis are presented below. Systems for coding communicative behavior have held a central position in much of the research on family inter action. One study stands out as requiring a high level of inference on the part of the raters, and as demonstrating 44 the combination of loosely connected subscales as measures of more embracing concepts. Meyers and Goldfarb (1961) developed scales for the global rating of three hour seg ments of family interaction and of material behavior in interviews. In rating family behavior, a total of 44 scales were used. Eight seven-point scales, for example, comprised a measure of the mother's "perplexity," or "abil ity to structure the child's environment:" (1) spontaneity of interaction, (2) decisiveness, (3) consistency of emo tional relatedness, (4) appropriate mode of relating to child, (5) appropriate control of child, (6) appropriate imposition of routines, (7) appropriate anticipation of physical needs, and (8) appropriate meeting of child's de mands. The mother's behavior in the interview was rated on nine scales considered together to comprise a measure of "parental functioning." Although no data on interrater reliability were presented, significant differences on the global measures of parental functioning were found between mothers of normals and of organically impaired and non- organically impaired schizophrenics. Later, Levy, Goldfarb, and Meyers (1963) revised this approach and developed guidelines for rating somewhat less inferential aspects of communication. They reported rating films and audio-tapes of interaction for the degree and frequency of clear and relevant communication and the 45 extent to which mothers provided anchors for defining suc cess or failure in communication. Other studies have focussed more directly on the clarity of communication. For example, Riskin's Family Interaction Scales (1964) included a scale of clarity of communication; and Goodrich and Boomer (1963) indicated that they used an unspecified measure of "effectiveness of communication." A strong emphasis on clarity also was present in Singer and Wynne's (1966) scoring system for conjoint Rorschach protocols and in Wynne and Singer's (1963a, b) work on "fragmentation." Finally, Riskin and Faunce (1970a, b) compared normal and troubled families on clarity of communication. Special consideration also has been given to "at tention." Titchener and Emerson (1958) coded their data for "degree of attention." Wynne and Singer did extensive work pointing to the importance of parents' sharing a focus of attention. This dimension was found to be most valuable in predicting which test protocols came from parents of schizophrenics and which did not. See Wynne and Singer (1965) and Morris and Wynne (1965) for examples. Codes of affect and affective expression have been used widely. Stabenau, Tupin, Werner, and Pollin (1965) used such a dimension, apparently without rigorous guide lines, in developing profiles of families of schizophren- i 46 ics, delinquents, and normals. Riskin (1964) included a three-point scale of affective intensity in his Family Interaction Scales. Morris and Wynne (1965) differentiat ed four aspects of affective expression in their study of communication patterns in parents of schizophrenics. The aspects were: (1) the variety of affect expressed and recognized; (2) the clarity with which affects could be identified; (3) the intensity of affective expression and energy level; and (4)"the patterns of empathy with the emotional states of others." Ryder (1968) used a combination of scores to rate couples' "propensity for affectivity." Measures of affec tive relationship also were included in the following: all studies using Bales' system; Straus (1964); Mishler and Waxier (1968); Riskin and Faunce (1968); and Goldstein, Gould, Alkire, Rodnick and Judd (1970). Several other individual characteristics have been gauged in family interaction studies. Goodrich and Boomer (1963) indicated that "involvement in discussion of color" formed a useful part of their profiles of conflict resolu tion behavior during the Color Matching Test. Titchener and Emerson (1958) also reported rating involvement. Com mitment was rated in two studies. Riskin (1964) used a three category commitment scale, while Riskin and Faunce (1970a, b) separated response and stimulus statements in 47 their eleven category scale. Both Titchener and Emerson (1958) and Goodrich and Boomer (1963) used ratings of ac tivity level in their studies; and the latter also used a measure of capacity to readh agreement. Reiss (1967, 1968, 1969) developed a scale of risk taking for his puzzle solv ing tasks. These are prominent examples of studies using individual codes. 3. Transactional codes. Transactional codes deal with events occurring between people. These events cannot be understood in terms of their individual components alone. These codes can be used only in situations where at least two people are interacting, and the level of inference re quired of raters using them is high. They must judge what is happening between two or more other people. Some of the attempts to judge interpersonal events have used vaguely defined, esoteric categories. Even Haley (1960), in his early work, indicated that he tried to take into account double binds and alliances occurring during family discussions, but he did not operationalize these variables. Levy and Epstein (1964) analyzed family Ror schach responses for "how one member uses the others to overcome individual conflicts." Singer and Wynne (1965) focussed on patterns of erratic and inappropriate inter personal distance and closeness, but did not specify how such patterns were defined in practice. 48 Somewhat less abstract transactional categories, calling for less esoteric inferences, were incorporated in several studies. Levy and Epstein (1964), for instance, analyzed communication in a family Rorschach task in terms of who assumed leadership, the quality of leadership, and which responses were accepted by whom. Riskin (1964) used a Relationship Scale for indicating whether a speech demon strated a friendly, negative, or neutral attitude toward the previous speaker. He also included a scale of topic continuity. Codes of agreement and disagreement require the comparison of one person1s statements with the remarks of someone else. Most measures of conflict have included some index of disagreement. Examples of transactional codes of agreement-disagreement can be found in the Bales' system and in Riskin's (1964) Family Interaction Scales. A few investigators have developed systems for coding specific transactional aspects of communication in families. Lennard, Beaulieu, and Embrey (1965) operation alized the definitions of content and person-process ori ented communication and used them in examining interaction in families of schizophrenics. The authors focussed on se quential shifts between these types of communication. Ransom (1970) further developed procedures for coding "labeling," which is the equivalent of person-process com- 49 munication. Sluzki, Beavin, Tarnoplosky, and Veron (1967); Mark (1971); and Ericson and Rogers (1973) have all at tempted to define categories of relational communication using transactional definitions. The problems posed by overlapping and inconsistent definitions have been especially evident where transaction al codes have been used. Riskin and Faunce (1972) present ed a case history of "acknowledgment" to demonstrate the difficulty of comparing the results of studies where the same concept is defined in different ways, and in which the operational definitions of several terms overlap. Several of the examples mentioned here were cited in that review. Mishler and Waxier (1968) defined five categories of "ac knowledgment, stimulus" and six categories of "acknowledg ment, response." Reiss (1967) used the same stimulus codes, but changed the response codes to "high," "medium," and "low." In his work on transactional disqualification, Ransom (1970) developed categories for coding the demand potential and acknowledgment dimensions of communication which were inspired by Mishler and Waxier's work, but he used his own modifications of their codes. Riskin and Faunce (1972) listed several concepts with definitions overlapping Mishler and Waxier's (1968) codes of acknowledgment. Some examples are: Lennard and Bernstein's (1969) codes of disagreement and responses to 50 preceding "presentation of self" statements; McPherson's (1970) codes of "intent"; and Riskin and Faunce's (1970a, b) Commitment Scale. The problem of multiple definitions of transaction al events is demonstrated in extreme form by the work on the double bind. In his review of the history of the con cept, Schuham (1967) pointed out that the theoretical def inition had changed over time. Olson (1972) reviewed empirical research on the double bind. Of the studies he cited, only Beakel and Mehrabian (1969) used measures of incongruence between verbal and non-verbal messages. The other studies cited employed a wide variety of definitions of double binds. For example, Beavers, Blumberg, Timkin, and Werner (1965) coded interviews with mothers of mental patients for the numbers of definite responses, shifts, and evasions. In contrast, Ringuette and Kennedy (1966) had three experts involved in the development of the concept, and other raters, rate letters on a seven-point scale of the degree of double binds present. They found interrater reliability to be only.19. The "slipperiness" of the con cept is suggested by the multiple theoretical and opera tional definitions which have been developed. 4. Codes of content or themes. Codes dealing with the contents or themes of family interaction have appeared infrequently. This fact reflects a general bias that the 51 relational or formal aspects of communication are more im portant in delineating family types and in contributing to psychopathology. Titchener and Emerson (1958) named "con tent" as a category, but presented no data using it. Len nard, Beaulieu and Embrey (1965) reported coding their data for the interpersonal referent of each statement, but pre sented no data involving these specific codes. Friedman and Friedman (1970) rated joint TAT-type stories for pres ence or absence of thematic focus. To the extent that agreement and disagreement can be seen as quasi-content dimensions, studies in which they were used could be in cluded in this category. While using no direct measures of family interac tion, a series of articles on recurrent themes in projec tive tests given several generations of family members stressed the possible role of familial preoccupation with particular themes in the development of personality and psychopathology. Fisher and Mendell (1958); Mendell and Fisher (1956); and Mendell, Cleveland and Fisher (1968) presented conclusions from several intensive studies of this nature. The last one includes conclusions drawn from testing 27 members of five generations of one family. These studies suggest that common themes may provide a useful approach to exploring family interaction data. 5. Outcome codes. Outcome codes are used for 52 tallying the results of marital or family interaction. They have been used most frequently in studies of family conflict resolution and in games generating scores. Farina (1960) first used outcome measures of "agreement" and "disagreement and aggression" in his study of role dominance and conflict in parents of good and poor premorbid schizophrenics and tuberculosis patients. He also counted the following events: passive acceptance of solutions, yielding to partner's solution, and failure to reach agreement. His experimental paradigm was similar to the unrevealed differences technique. Farina and Dunham (1963) used the same measures, except for yielding, in a study involving two similar groups of schizophrenics in dis cussions with their parents. All of the original measures were used again by Farina and Holzberg (1968) comparing dis cussions in triads made up of good and poor premorbid schiz ophrenics and their parents, and of patients hospitalized with other psychiatric diagnoses and their parents. Relying primarily on outcome measures, Ferreira studied decision making in normal and abnormal families. His unrevealed differences technique was used to generate interaction for all these studies. In the first study in this series (Ferreira, 1963), he assigned decisions to one of the following categories: unanimous, majority, dicta torial, or chaotic. In later work, he and Winter changed 53 the categories to: spontaneous agreement, decision time, and choice fulfillment. These codes were defined in terms of outcomes and grouped into an Index of Normality. The results from the four studies in this group (Ferreira and Winter, 1965, 1966, 1968; and Winter, Ferreira and Bowers, 1973) were quite consistent, This finding indicates the reliability of the dimensions measured. Other authors, too, have focussed on variables re lated to who wins in conflict resolution paradigms. Bauman and Roman's (1966) measure of "dominance" was based on winning. Lerner (1965, 1967) used the categories of com promise, yielding, and passive acceptance of solutions. Work on relative spousal power, too, has used measures re lated to winning. See, for example, Strodtbeck (1951), Goodrich and Boomer (1963), Ryder and Goodrich (1966), Ryder (1968), Olson and Ryder (1970, 1973b), and Schuham (1970). Another kind of outcome code was used by Watzlawick, Beavin, Sikorski and Mecia (1970) in a study of the "Blame" section of the MRI interview. They developed measures of protection and scapegoating from ratios of blame statements attributed correctly or incorrectly to different family members. Several techniques obviously generating outcome scores include the tests of marital or family intelligence 54 and the interaction games. Examples of these techniques appeared earlier in this review and will not be repeated here, since the outcomes follow directly from the game scores. 6. Other coding systems. Some coding systems resist inclusion in the groupings above, though particular categories do overlap. For instance, the Interaction Pro cess Analysis (IPA) system is composed of twelve catego ries relating to positive and negative social-emotional and task orientation. It was presented in brief form by Bales (1950). Some modifications have been introduced for the study of family interaction; and data have been analyzed using various combinations of the original codes. Investi gators who have used the IPA or its modifications include: Strodtbeck (1953), Caputo (1963), Levinger (1963), Cheek (1964a, b; 1965a, b), Sharon (1966), and Winter and Fer reira (1967) . Leary's Interpersonal Diagnostic System formed the basis for Terrill and Terrill's (1965) system for aoding family interaction. These authors refined the original categories and added four neutral codes for speeches in which the interpersonal impact was not clear to the rater. The other categories essentially reflect ratings of the dimensions of dominance-submission and affiliation-hostil- ity. 55 Borke (1967) developed a somewhat eclectic coding system specifically for use with observations of family interaction. This system focussed on the intent of the speaker. The primary codes were based on Horney's classi fication of behavior into acts going toward, going against, or going away from, the other. The secondary codes, re flecting subgroups within the primary categories, seem to have been inspired by Bales and Leary. They included such things as "evades," "seeks support," "instructs," and "of fers information." In Borke1s system, each "behavioral unit" was coded as a response and as a stimulus. Both out come and process analysis were envisioned with the use of this system. Olson and Ryder (197 3a) developed a coding system and wrote a manual for use with their Inventory of Marital Conflicts (IMC). A similar procedure now is being used with the Inventory of Family Conflicts (197 3b). The IMC interaction coding system contained 29 codes divided into assertive, supportive, and structural acts. The codes were established specifically for use with the IMC, though they could be used with any conflict resolution tasks. The man ual also specified procedures for calculating win-loss style outcome scores and relevancy scores, when used with the IMC. 7. Studies without formal coding systems. A num- 56 ber of influential studies have used macroanalytic ap proaches to reducing family interaction data for analysis. In these studies, clinical intuition was used in the an alyses; the coding process was omitted— or, at least, not reported. Some of the most germinal work in the field has used this kind of approach. Wynne's early investigations of "pseudo-mutuality" and stereotyped roles in the families of schizophrenics were examples. (See, for instance, Wynne, Ryckoff, and Day, 1959). Other prominent studies using no coding systems included Jackson and Yalom's (1966) characterization of the families of patients with ulcera tive colitis, Bell's (1962) analysis of the psychological functions of extended-family relations, and Bowen (1960) and Brodey's (1959) investigations of family process in hospitalized families. Unfortunately, such studies defy replication. They are justified only by the fertility of the hypotheses generated. Thus far this review has presented a survey of the methods of research on family interaction. The next sec tion deals with these methods from an evaluative point of view. 57 Evaluation of Research Methods The evaluation presented here focusses first on the strengths and weaknesses of the various approaches to collecting data on family and marital interaction. Then it turns to an appraisal of the processes of codification and data analysis. Several criteria are used to organize the discussion in each section. Techniques of Collecting Data In discussing how data are generated, three basic issues will be used in the evaluation: (1) How well does the method fit the purpose for which the data are gener ated? (2) How relevant are the data to understanding interaction? (3) Is the method sufficiently comprehensive to yield valid results? How well does a chosen method of generating inter action fit the purpose for which the data are gathered? Why should one method be picked in preference to another? If an investigator wishes to use outcome measures, a highly structured and well controlled task would seem to be pref erable to insure the comparability of data samples. If, on the other hand, one is using some method only for the sake of producing a sample of interaction, a premium should be put on sampling behavior in as natural a context (or 58 contexts) as possible. Or, it may be that the experimen ter wishes to produce samples of interaction over specific issues, or over issues of high or low "relevance" to the subjects. In these cases, some structure must be given to the task, but hopefully not so much as to obscure the pro cesses being measured. A review of the uses of the RDT and its derrivatives illustrates the seriousness of this ques tion. Strodtbeck (1951) designed the RDT to study power relations between husbands and wives in different sub-cul tural groups. He chose to have them discuss how well other couples fit certain criteria, hoping that both part ners would be equally-well informed on the topics to be discussed. Strodtbeck chose topics of some interest and little threat. His primary interest was in outcome mea sures, though he did use Bales IPA categories in coding, too. In studying communication in three-person families, Strodtbeck (1953) used the RDT to promote coalition forma tion. The discussions dealt with family attitudes toward vocational choices for the son. These topics can be as sumed to have been more relevant to these subjects than to the subjects in the earlier study. Again, outcome mea sures provided the major focus, though IPA coding was used in this research too. 59 The RDT has come to be used primarily to generate interaction in standard stimulus situations where each family member supposedly has an equal right to participate. The emphasis with this technique has shifted away from out comes. Rather, the interaction is generated for its own sake and examined from other perspectives. The discussion topics have ranged from very low to very high apparent (but unassessed) relevance. It would seem inappropriate to use a conflict resolution paradigm when a sample of usual family communi cation is desired, especially with topics with unknown levels of importance to the subjects. Even if discussion of disagreements is desired, the implication that conflicts should be resolved would seem to pressure participants to do so. This kind of conversation probably is quite atypical. It would seem reasonable that findings about communication processes generated in such research should not be generalized as if they were characteristic of usual family interaction. Even as samples of conflict resolution behavior, such interaction could be influenced to a great extent by the importance and content of the issues dis cussed. The variety of discussion topics used in differ ent studies has been so great as to make the comparability of results questionable. Even standardization of discus sion topics would n<bt seem to solve the problem where im- 60 portance or relevance is not measured or equated. The provision of a standardized list of alternative decisions, after the manner of Ferreira (e.g., 1963), would tend to standardize discussions to some extent, but only at the cost of limiting the participants' freedom. Two major revisions of the RDT reflected attempts to deal with these problems. The Color Matching Test (CMT) and the Inventory of Marital Conflicts (IMC) have standardized the number and content of items involving disagreement, and have provided topics of agreement for contrast. The CMT has equated item relevance by minimizing it, while the IMC includes procedures for assessing rele vance and requires forced-choice decisions between alter native problem solutions incorporated in the inventory. However, the procedures for assessing relevance in the IMC seem inadequate. Since it deals with problems presented as if they related to unknown couples, since the problems are cast in rather idiosyncratic contexts, and since the discussion time for 18 problems is expected to be half and hour; one must assume that the relevance of the topics to the subjects is not high. The IMC relevance measure, "Have you, or someone you know, ever had a simi lar problem?"also seems inadequate. The measure used with the Inventory of Family Conflicts seems preferable. It requires subjects to rate the conflicts on how important 61 they seem. The increased methodological rigor reflected in the CMT and the IMC has rendered the outcome measures more important than the process measures. The interaction they require could hardly be taken as typical marital communica tion. The investigators have developed systems for coding this interaction, but one must wonder why. The problems of using the RDT, CMT, and IMC in studies of family interaction have been explored to em phasize the importance of choosing an experimental task which generates data relevant to the purpose of any inves tigation . The extent to which an interaction sample is suited to a particular process of codification and data analysis also should be considered in designing research. Some codification systems fit best certain kinds of data. An extreme example is the coding system designed particu larly for use with the Inventory of Marital Conflicts (Olson and Ryder, 1973a). An example of the importance of this problem can be found in considering Bales' Interaction Process Analy sis (IPA) system of codification. It was developed through the study of small groups in the process of "task- solving.” The original emphasis was on finding a balance of instrumental and expressive activity during such pro- 62 cesses. It has been used in many RDT-style studies of family and marital interaction. Such tasks appear to be relatively comparable to the tasks for which the IPA was originally designed. However, Winter and Ferreira (1967) concluded that "the Bales IPA system, in its present form, is not suited for work with families" (p. 170). In their study of three- person family interaction with the task of creating joint TAT stories, they estimated interrater disagreement to be about 33%. They commented that this system is more clearly related to the behavior of the child than the parents and is inadequate for differentiating families with normal and maladjusted children. Although Winter and Ferreira's article included criticisms of the IPA on the basis of Borgatta's (.1964) finding that correlations of scores for the same subjects doing different tasks were very low (-.12 to +.23), they did not relate this observation to the discussion of their own findings. One cannot decide whether the task or the IPA was responsible for their dif ficulties . Yet, the number of revisions of the IPA (Waxier and Mishler enumerated six in their critical appraisal in 1966.), suggests dissatisfaction among a variety of inves tigators. The reasons include the belief that the IPA may not be subtle enough for comparisons of families. Another 63 prime source of dissatisfaction has been the questionnable levels of interrater reliability obtained. Waxier and Mishler (1966) used the IPA or RDT data from three-person families. They found act-by-act interrater agreement to be 61%; it fell to 43% when chance agreements due to mar ginal distributions were removed. They concluded by recom mending revision of the IPA to separate coding of instru mental and emotional dimensions, in a manner similar to Leik's (1963) . They also recommended using marginal totals for calculating interrater agreement when an investigator is interested in over-all comparisons of the number of acts falling in different categories. They advised using act-by-act agreement when transactional codes are used. Actually, act-by-act agreement is the only kind of inter rater agreement which measures whether raters agree on the same items. This approach, then, would seem preferable in all cases of nominal coding. This discussion has emphasized the importance of choosing codification systems and methods of assessing interrater reliability which are appropriate for the inves tigations undertaken. Although the IPA was discussed, many concerns relating to its use also apply to other coding systems. The question of how relevant the data generated are to understanding family interaction, seems to have 64 three facets. The first concerns the extent to which the interaction sampled resembles behavior occurring in other contexts. The second questions the comparability of the sets of experimental and control subjects; and the third relates to the influence of the observation procedures on results. The assumption of stability of interactional vari ables across settings and tasks is made in most of the re search in this field. However, the assumption of variabil ity may be more nearly correct. Mischel's (1968) review of the experimental literature on personality cast doubt on the validity of "trait" theories. One must recognize the possibility that studies of family interaction comparing findings within families across contexts and tasks might cast doubt on assumptions of stable familial differences. Only three studies relevant to this problem were located. O'Rourke (1963) compared IPA family interaction scores on the same decision-making task using TAT cards at home and in a laboratory. He found differences in results for the two settings. Ransom (1970) reported data taken from two families in each of three contexts, but his sub jects were engaged in different tasks in each setting. Contextual variability was demonstrated, but task and set ting differences were confounded. Smith (1971) explored the different kinds of information shared in family dis 65 cussions and note passing tasks. Again, differences were found. Given the state of the art, Riskin and Faunce (1972) argued that contextual stability across tasks would at least control for variability due to the influence of setting. Differences in "set" for normal and abnormal fami lies have also been suggested as a source of variation be tween groups. Haley (1962) argued that a family partici pating in research in exchange for treatment, or because one member has some disorder, probably comes with a more highly defensive set than a family chosen because of their "normality." Lennard, Beaulieu and Embrey (1965) expressed a different point of view. In speaking of families of schizophrenics, they pointed out that such families are assumed to be quite inflexible in their interrelationships, so that they would not be able to shift their style even if they so desired. Clearly, these variables require fur ther investigation. Such work might help theorists in the field to recognize the problem and deal with it in their models. Riskin and Faunce (1972) also mentioned observation procedures as variables having potential biasing effects on experimental results. They indicated the importance, for the time being, of specifying the extent to which ex perimenters participate in the research procedures, the 66 relationship existing between the experimenters and sub jects, the likely influence of experimenter gender on the interaction or ratings, and the extent to which raters are "blind." No studies relevant to these concerns were located. The issue of the representativeness of an interac tion sample has been raised by several investigators. Some have attempted to "solve" the validity problem with in vivo sampling. Straus (1964) presented a discussion of the multiple meanings of validity in relation to family stu dies. Some of the studies are astounding in the extent to which the question of how representative the sampled inter action is, is ignored. If the behavior required by the IMC or Haley's coalition-formation task, for example, is high ly unusual, then how sure can one be that the results cor respond with more "natural," or frequent, behavior? Valid ation studies were not found in the literature search for this review. Few, if any, exist. One last issue seems important to consider when choosing a task and mode of recording interaction in fami lies. This issue is the extent to which the data gathered and coded are comprehensive enough to permit meaningful inferences to be drawn. Do the channels of communication analyzed permit valid inferences? Verbal messages are confirmed, altered, or disqualified by non-verbal behavior, 67 including such things as intonation, facial expression, and other "body language." Generally, to the extent the problem has been considered at all, it seems that most in vestigators have assumed that there is enough redundancy in information carried by different channels to justify the use of only one channel as a source of data for coding. The fallibility of this assumption is demonstrated in three studies. Waxier and Mishler (1966) found that when tapes and transcripts were used in coding, the dis tribution of one-third of the acts in IPA categories was different from ratings done using only transcripts. More acts fell into "negative affect" categories with the for mer method than with the latter. In related work, Schuham and Freshley (1971) compared IPA ratings using typescripts plus videotape recordings; they also found differences. Anandam and Highberger (1972) found that the congruity of verbal and non-verbal channel's of communication was in fluenced by task differences apparently related to stress or difficulty. Unfortunately, much of the research in this field has used only typed transcripts in the coding procedure; sometimes the form of the material coded has not been specified. Another aspect of this problem is the extent to which the complexity of the data analyzed is sufficient to 68 support the inferences made from the analysis. Haley's (1962) research on coalition formation is an example of an extreme mistake made in this area. Coalition formation and maintenance in families would seem to be a long concep tual leap away from button-pressing tasks with goals of maximizing scores accumulated by pressing buttons at the same time as other family members. The problems discussed above relate to the choice of experimental procedures. The overriding question for use in evaluation is, "Can the procedures support the con clusions which might be drawn from the study?" In the next section, some considerations of issues relating to codification and measurement are raised. Codification and Measurement Decisions about the rationale and content of vari ables, and the procedure for measuring them, greatly in fluence the rest of any research study. In the field of family interaction, there is a proliferation of concepts and overlapping definitions. For discussion purposes, problems in codification have been grouped around three criteria for evaluation: (a) the explicitness of the coding procedures; (b) the adequacy of the discussion of the relationship between coding variables and the con- 69 ceptual framework of the study; and (c) the adequacy of the consideration of the coding procedures. The first question is whether or not the coding procedures are made explicit. Some studies in this field simply list the categories used, omitting descriptions of coding procedures. Even when the length of a publication must be restricted, a few examples should be included to convey an impression of the relationship of observed be havior to coding variables. Another problem related to making the coding procedures explicit is that of making clear which codes are used on which portions of interac tion, and to whom the coded behavior is attributed. Un usually positive examples of coding systems and instruc tions presented in detail include: Bale's IPA (1950) , Mishler and Waxler's coding system (1968); Straus and Tallman's SIMFAM (1970); Olson and Ryder's Inventory of Marital Conflicts (1970-1973); Riskin and Faunce's Family Interaction Scales (1968); and Singer and Wynne's Ror schach Scoring Manual (1966). One problem afflicting the field may be attributed in part to the obscurity of coding procedures. If, for example, "conflict" or "dominance" is measured one way in one study, and measured in a different way in another study, then the comparability of results is questionable. Nevertheless, such comparisons are made and findings are 70 regarded as contradictory if they differ. For example, Farina and Dunham (1963) measured "conflict" as the sum of the frequencies of simultaneous speech, interruptions, dis agreements and aggressions, and failures to agree. Baxter and Arthur (1964) noted that, in their study, "Care was... taken to select items [from their interviews with parents of schizophrenics] which bore a general correspondence to the types of behaviors which had been used earlier by Farina in direct observations of parental interactions." (p. 276). Ratings of conflict on these items, however, were made on a four-point scale reflecting "the extent to which the mother and father disagree; express opposing ideas, attitudes, values, or the like; or express disapproval or disparage ment of the other's attitudes or behaviors." (p. 276). Al though comparability of findings obviously was desired, the measurement procedures used were quite different. Several other versions of the definition of "conflict" also may be found in the field. When no clear operational definitions are provided, the difficulty of comparing findings is com pounded even further. The second criteria for evaluation of coding proce dures is whether or not the variables are adequately re lated to a conceptual framework. Several studies in the field of family interaction leave one wondering, "So what?" after reading the report. For instance, Watzlawick, Bea- 71 vin, Sikorski, and Mecia (1970) conducted a study of "Pro tection and Scapegoating in Pathological Families," but did not discuss any theoretical relationship of pathology to these variables. Ryder (1970) presented evidence suggest ing that men who remember dissatisfaction with their fami lies of origin tend to be ambitious, while women with such dissatisfaction tend to generate stormy marriages. He used a great deal of data, but none of the data sources appear to have been chosen with any conceptual framework in mind. The results also were discussed without benefit of such a frame of reference. In another example, Farina and Holzberg (1968) showed close attention to the usual methodological problems, such as the use of appropriate control groups and large sample size. This study of domi nance and conflict in families with offspring hospitalized for the treatment of schizophrenia or other psychiatric disorders, included no discussion of the place of conflict or dominance in any theory of schizophrenia or psychopath ology. In fact, the authors merely pointed out that their results were "very consistent with contemporary psychogenic theories of schizophrenia which assign an important role to the family" (p. 117). The question of "which theories?" remains unanswered. In a field as scattered and uncoordi nated as the study of family interaction is, it would be helpful if investigators would relate what they are doing to what they and others think about the variables they are 72 studying. One final point will be made about coding proce dures. That is, the mechanics of coding often are not adequately considered and reported. Several questions often are not answered. What were the qualifications of the coders? Were they trained; and, if so, how? Were they "blind" with regards to the group of membership of sub jects? Was more than one rater used? Was interrater re liability reported? Was the way reliability was calculated appropriate for the uses to which the data were put? Was the level of reliability obtained high enough to justify further use of the coding system without modification? Since the possibility of reading biases into interaction data is so great, it seems reasonable to demand that the issue of reliability be taken seriously. It seems sad that this reminder needs to be included. Once the coding is completed, statistical analysis can be undertaken. Thus far, almost all of the family interaction research has utilized statistical methods for analyzing differences between marginal totals (i.e., the incidence for a group or an individual of a particular variable). Mishler and Waxier (1969) and Waxier and Mish- ler (1970) have suggested that correlational and sequential analyses might be useful in this field. However, high levels of interrater agreement on coding individual com 73 munications must be assured before sequential or correla tional analyses can yield valid results. Such high levels of agreement are not yet the rule in this field. This overview of methodological considerations re lated to the choice of experimental tasks and codification procedures by no means covers the methodological problems encountered in doing a study using family interaction data. Other reviews, particularly Riskin and Faunce (1972), do this. The next chapter will present a more intensive look at the research which has studied interaction through mea suring the communication strategies developed by Bateson and his co-workers at the Mental Research Institute. CHAPTER III REVIEW OF THE SPECIFIC THEORETICAL AND EMPIRICAL BACKGROUND OF THE RESEARCH This chapter deals with the specific background, rationale, and hypotheses of the present research. It is composed of four sections. The first section presents the theoretical foundation and includes definitions of the strategies of communication used in this study. Section two is an examination of the empirical work which has uti lized these strategies. The rationale and hypotheses for this study are presented in sections three and four, re spectively. Theoretical Background In this section, the theoretical background for the present study is laid. The general approach of the communication theorists to understanding communication is presented, and the strategies of communication to be used in this research are delineated. Bateson and others generally associated with the Mental Research Institute in Palo Alto have built the the oretical base. The purpose of communication was elabo- 74 75 rated by Ruesch (1953). He stated: The functions of communication are to maintain contact with other biological beings and to avoid isolation— a tendency which is basic and inborn— to receive and transmit messages and to retain information, to reconstruct the past and to antic ipate future events, to perform operations with the existing information for the purpose of deriving new aspects, which were not directly perceived, to initiate and modify physiological processes within the body, and to influence and direct other people and external events [italics not in the original], (p. 242) The subtle nature of human communication becomes apparent when one recognizes that human communication con tains two major elements. Ruesch and Bateson (1951) rec ognized that content is communicated only in the company of instructions about interpretation of the message. They observed that: When A speaks to B, whatever words he uses will have these two aspects: they will tell B about A, conveying information about some perception or knowledge which A has; and they will be a cause or basis for B's later action. In the case of language, however, the presence of these two meanings may be obscured by syntax [semantic meaning].... For example, A may say "Halt!" and B may obey the command ignoring the informational aspects— e.g., the fact that A's words indicate some perception or other mental process of which his command is an indication. Or A's words may have the syntax [semantic meaning] of report, and B may fail to notice that this report has influ enced him in a certain direction.(p. 180) Thus, communication is seen as being composed of a content 76 component and a directive component. Watzlawick, Beavin, and Jackson (1967) call these the "report" and "command" aspects of a message. The difference between the two aspects becomes clear if one considers that "I'm glad to see you." may be used to mean the speaker is glad to see his friend, or it may be said sarcastically to mean just the opposite. At any rate, the directive component is present any time a message is sent. Haley (1963) contends that "...it is impossible for a person to avoid defining or taking con trol of the definition of his relationship with another..." (p. 9) The distinction is crucial to the Palo Alto theo rists and others concerned with communication and psycho therapy (e.g., Beier, 1966). While the report aspect is likely to be clear to an observer who is not aware of the context, the command aspect depends on context and seman tics. The command aspect conveys information about how the sender wants to be treated by the receiver and what rules for communication the sender wants to establish in the relationship. Thus, this component represents an at tempt to define the relationship between two or more per sons . The importance of the control dimension in ordi nary life was highlighted in this statement by Haley: 77 In any interchange between two people they must deal not only with what kind of behavior is to take place between them but how is that behav ior to be qualified or labeled. A young lady may object if a young man places his arm around her, but she may not object to this behavior if she has first invited him to place his arm around her. When she invites it, she is in control of what behavior is to take place, and therefore she is in control of the definition of the rela tionship. (1963, p. 9) With the first appearance of Naven in 1936, Bate son began to develop the systems theory approach to under standing relational aspects of human communication. The seminal nature of this work was recognized with the publi cation of a revised second edition in 1958. In describing the Iatmul culture of New Guinea, Bateson discussed the process of role differentiation, which he termed, "schis- mogenesis." He defined this as "a process of differentia tion in the norms of individual behavior resulting from cumulative interaction between individuals." (1958, p. 175) Thus, the single person was seen as a participant in forming the nexus of forces to which his own behavior was also a response. A response was seen as a stimulus for responses which, in turn, were stimuli for further ac tion on the part of the subject. The concept of schismogenesis foreshadowed an understanding of what now is popularly understood as homeostasis in social groups. Bateson pointed out that 78 role differentiation could be based on complementarity, as when one person responds with submission to another's assertiveness; or on symmetry, as when one person re sponds with boasts to the boasts of another; or on a mix ture. Bateson argued that at least a temporary shift from either type of behavior is necessary to prevent a rela tionship from disintegrating. The changes in the rela tional aspect of communication, thus, would act as a gov ernor, holding the relationship within bounds which permit its continuation. Besides this general prediction that schismo- genesis must be bounded by at least intermittant changes in the character of relationship, the theorists made few other predictions based on their analysis. They have made no claims, as yet, for the explanatory usefullness of their system. Watzlawick, Beavin, and Jackson (1967) as serted that: [since] our aim is intensive rather than exten sive, it is necessary to explore the interactional explanations first, before the integration of premises from other frames of reference. Thus, we will hold with an answer which is descriptive rather than explanatory, that is focused on how and not why the interactional system operates. (pp. 130-131) In the exploration of how communications systems operate, the control dimension of communication has been 79 deemed of central importance. Several types of transac tions have been elucidated, and one fundamental classifi cation has emerged. Watzlawick et. al_. (1967) stated as an axiom that: All communicational interchanges are either symmetrical or complementary, depending on whether they are based on equality or difference.(p. 70) Styles of influencing the definition or relation ships may be studied through the use of these and other related categories. The theoretical classifications have been developed and defined in a manner relying on clinical and observational data and on logical considerations. Bateson (1936) began this process. Jackson, Riskin, and Satir (1961), Haley (1963), Watzlawick, Jackson and Beavin (1966), and others, have extended the process. The follow ing discussion draws from these sources, which contain parallel typologies of interaction. In reading the section which follows, it is impor tant to bear in mind the transactional bases of the defin itions. As Bateson and Jackson (1964) pointed out, there is, strictly speaking, no such thing as a complementary piece of 'behavior.' To drop a brick may be either complementary or symmetrical; and whatever it is depends upon how this piece of behavior is related to preceding and subsequent behaviors of the vis-a-vis. (p. 273) 80 Complementary Interactions Complementary interactions form a category in which the control dimension is obvious. They are based on behaviors which are different in nature and fit together. One person takes a "one-up" position, implicitly deciding that this is an interaction in which he is to be in con trol. The respondent, in a "one-down" position, accepts the other's definition. Bateson and Jackson (1964) gave the examples of dominance and submission, exhibitionism and spectatorship, and succoring and dependence. Watzla wick et. al. (1967) stressed that "it is important to empha size the interlocking nature of the relationship, in which dissimilar but fitted behaviors evoke each other." (p. 69) In a complementary relationship, the person who is "one- up" is in no way necessarily more powerful or better than the person who is "one-down." The person in the "one- down" position consents to the other's being "one-up." The "one-up" person is only apparently in overt control of the relationship. Bateson (1958) claimed that adherence to comple mentary interaction would lead to schism. Submission and assertion, he believed, would accelerate each other, lead ing to schism unless checked. For the relationship to continue, some symmetrical interaction would have to occur, 81 at least temporarily. Symmetrical Interactions Symmetrical interactions, by contrast, are based upon equality. Each person is struggling with the other for control of the relationship, and "the partners tend to mirror each other's behavior." (Watzlawick, et. al. , 1966, p. 68) Symmetrical behavior tends to take on a competi tive cast. Both partners are doing or attempting to do the same thing. If two people are boasting to one another, each responding with a boast to the previous boast, their communication is symmetrical. Each is trying to dominate the other with the greatest boast. Bateson and Jackson (1964) stated that: symmetrical transactions characterize rivalries and other relationships, where A is stimulated to do something because B has done this same thing; and where B does more of this because A did some of it; and A does more of it because B did some, and so on.(p. 270) Bateson (1958) saw schism as the likely outcome of symmetrical, as well as complementary, interactions if they continue unchecked. Escalating competition, accord ing to him, would lead to a breakdown of the relationship. The desire to maintain the relationship would produce some temporary change from symmetrical behavior. 82 Metacommunicative Interactions Communication about communication which is occur ring is termed metacommunication. Although there is a metacommunicative aspect to the relational component of any message, this label is reserved here for overt commu nication about communication in progress. This category includes comments on the contradictions in paradoxical messages. It is the tool we have for trying to clarify confusion, for bringing covert messages to the foreground, and for responding to others' communicative styles. An example would be, "You are smiling as you say that, but your teeth are clenched." Extensive use of metacommunica tion is advocated for marital and family therapists by Satir (19 64) and other communication theorists. Because metacommunication is one form of an at tempt to wrest control of the flow of discourse away from another person, it is regarded as a subclass of symmetri cal responses in this study. The subclass is identified because of the interest this type of communication carries for family and marital psychotherapists. Metacomplementary Interactions If one looks beyond the first, or obvious, level of communication, other strategies may come to light. 83 Metacomplementary relationships are those in which one person (A) permits or forces the other (B) to take charge of defining the relationship. On one level, B is in charge; B may define the relationship as complementary or symmetrical. However, since A specifically allows or per mits B to take charge, A is in the one-up position on a superordinate level. The level at which the decision is made about who is to decide who is to take charge of a relationship is seen as more powerful than the derivative level of determining who is to take charge. As Haley (1963) pointed out: The person who establishes a metacomplementary relationship with another is controlling the maneuvers of the other and so controlling how the other will define the relationship, (p. 12) For instance, the wife who indicates to her hus band that he is to make the decisions relating to their social life is controlling the definition of their rela tionship in this interchange. If he accepts this deci sion, he may decide on the content of their social life, whereas she has determined the process by which the con tent will be decided. She has decided who will decide. While graciously allowing him to decide, she has asserted her power by taking charge of determining who is in charge. Metacomplementary communication introduces the 84 idea that control may be gained at different levels in a relationship. The person in apparent control may have been appointed to that job by the other. One could imagine even more complex strategies involving metacomplementary behav ior. If, for example, the person who authorizes another to assume control of some aspect of their relationship feels coerced into the authorization, another level of control would be exercised in the relationship. However, strat egies this complex have not been explicated and will not be considered further in this study. Because the person defining a relationship as meta complementary is in a one-up position on a superordinate level, metacomplementary responses will be viewed as a sub class of one-up responses in this study. Paradoxical Interactions None of the relati.onships described thus far has involved the concept of denial in communication. When one attempts to control a relationship while denying tha,t at tempt, paradoxical communication is involved. The Pa,lo Alto authors dwell on paradoxical communication as the es sential ingredient for symptomatic behavior. As Haley (1963) sees the issue, A relationship becomes psychopathological when one of the two people will maneuver to circumscribe the other's behavior while indicating he is not. (p. 17) 85 A person can do this with paradoxical communication. A paradox is posed whenever "a person is offered two direc tives which conflict with each other and demand a response." (p. 17) One directive qualifies the other in such a way as to make compliance with both impossible. For instance, "when one person directs another not to follow his direc tives, the paradox is obvious. The receiver cannot obey the directive nor disobey it." (p. 17) In order to pose a paradox, one must deny some as pect of the message to be communicated. Watzlawick et. al. (1967) gave a formal definition. Accordingly, a paradox occurs when: ...a message is given which is so structured that (a) it asserts something, (b) it asserts something about its own assertion, and (c) these two asser tions are mutually exclusive.(p. 212) Paradoxical messages, then, involve an attempt to deny that one is defining a relationship, and they require a response. Gill (1971) argued that paradoxes can be established only because: specific syntactic statements can be uttered in a [semantic] way in which they do not accomplish the task for which they are (syntactically) designed.... (p. 6) This fact makes possible the use of paradoxical communica tion, even when one is not aware of the paradoxical nature of his messages. According to Haley (1963), there are three possible 86 types of response to a paradoxical message. One may leave the relationship, comment on the impossible situation he is in, or respond while indicating that he is not responding to the other person. If the receiver chooses the third way, a paradoxical relationship is said to exist. One paradox has been answered by another. An example of paradoxical communication is present ed by the husband who swears at his wife when drunk, blam ing this behavior on alcohol. He denies that it was truly himself who swore at his wife, holding the alcohol respon sible. While it is obvious that he is denying that it is really he who swore at her, the conflicting directives are not so obvious. On the one hand he is saying, "Respond to me as someone who has sworn at you." On the other hand, he is saying, "Don't respond to me in that way since I did not swear at you; the alcohol is responsible." The husband has put his messages across in such a way that he can deny re sponsibility for it. The paradoxical message can generate a paradoxical response. In the case of the drunk husband, the wife might respond with anger at her husband, while denying that she is angry at him for swearing at her. This would be likely, since, if she became angry at him for cussing her out, he could claim he didn't mean what he said since he was drink ing. She may berate him for drinking, but not hold him responsible for the tantrum which made her angry. If he 87 drinks frequently, he can influence his wife to feel sorry for him— and not be angry— since he is a man who cannot control his drinking problem. She would be more at the mercy of the paradox, feeling confused and angry, with no place to put her feelings. If she were to leave him, she would feel guilty for leaving a poor, sick man with an al cohol problem. In this situation in a paradox is posed and answered with paradoxical behavior, and a paradoxical rela tionship exists. She can only identify the paradox or leave the field to escape. Haley (1963) pointed out that the person posing the paradox wins control of the relationship, at the expense of being able to take responsibility for winning that con trol. The initiator of the paradox wins control because the respondent absolutely cannot define the relationship, either by following directives or not following them, since he is being asked to do both at once. In the case of the drinking husband, he is able to define the relationship as one in which it is all right for him to swear at his wife at the same time that he receives no credit for this defin ition. He gets his wife to respond to his anger, but with out acknowledging it as his anger. He cannot be sure of his definition of the relationship as one in which he can swear at his wife, since alcohol is responsible. He cannot be sure of finding out his wife's response to his anger, since the anger is not presented as really his. She can't 88 know what he feels for her, since what he expresses with and without alcohol are divergent. Thus, according to Haley (1963), It seems to be a law of life that one must take the responsibility for one's behavior in a rela tionship if one is ever to receive credit for the results, (p. 19) Beier (1966), writing of communication in which the covert and overt aspects of a message conflict, stressed the lack of awareness usually present in paradoxical mes sages. He wrote that: People give [paradoxical] information of which they themselves are not aware. They construct their own and the respondent's awareness with great skill and are then totally surprised when they obtain re sponses for which they cannot account, though they unknowingly worked hard to elicit such response activity in the first place.(p. 24 6) If this is true, an important part of our communication must be paradoxical in nature, and we must not be aware of it. Further, we cannot gain certitude about our relation ships when our messages contain contradictions. Gill (1971) carried Beier's contention to a logical conclusion. He noted that children being taught— by adults who themselves use paradoxical behavior— to distrust their own perceptions. "Since everyone else feels certain things and denies it, the children learn that it is 'good' to fol low suit." (p. 18) Indeed, Beier (1966) claimed, exposing someone else's paradoxical behavior typically is punished 89 in our society. With these things in mind, the task of psychotherapy could be construed as maneuvering people into facing their paradoxical behavior. When paradoxical messages are sent repeatedly in contexts in which there would be heavy sanctions for leav ing the relationship and for commenting on the conflictual nature of the message, the famous "double bind" situation exists. Bateson, Jackson, Haley, and Weakland (1956) ar gued that living in a network of such communication pat terns is a major contributing factor in the development of schizophrenia. They cited the following example of double bind behavior: A young man who had fairly well recovered from an acute schizophrenic episode was visited in the hospital by his mother. He was glad to see her and impulsively put his arm around her shoulders, whereupon she stiffened. He withdrew his arm and asked, 'Don't you love me any more?' He then blushed, and she said, 'Dear, you must not be so easily embarrassed and afraid of your feelings.' The patient was able to stay with her only a few minutes more and following her departure he as saulted an aide and was put in the tubs. (p. 259) The son could not interpret his mother's conflictual mes sages accurately and maintain his relationship with her. The double bind required the son to deny his own percep tions and accept his mother's interpretation of the situa tion and definition of his feelings. He became enraged. The double bind experience is expected by the theorists 90 to create feelings of helplessness, fear, and rage within the victim. This exemplifies the power attributed to para doxical communication by the Palo Alto communication ana lysts . Considerable speculation and some research has been done on the double bind. The difficulties with theory and research in this area are pointed out in Schuham's (1967) and Olson's (1972) reviews. They noted that the definition of double bind has varied over time and between theorists. At the operational level, the definitions have been con strued in several different and generally inadequate ways. Ringuette and Kennedy (1966) found that even the major theorists responsible for the development of the theory were unable to agree on whether or not the same messages represented double binds. Thus, the standing of the double bind theory, and the claims made for it, have not yet gained substantial empirical support. In forms not specifically labeled double bind, paradoxical communication is seen as the "universal symp tom" of neurosis by Kaiser (1965). It is believed to be almost omnipresent in our society (Gill, 1971) , and it is seen to occur most often without the perpetrator or the victim realizing that paradoxical messages are being sent. Any variable attributed such power and such a central place in understanding dystunctional family systems, as well as 91 disturbed individuals, is of interest to psychotherapists. Transactional Disqualifications Because they tend to invalidate preceding communi cations, transactional disqualifications form a subset of paradoxical interactions having particular theoretical in terest. Ransom (1970) defined them as occurring when "the response is incongruous with the frame created by the pre vious statement, and metacommunicative indicators labelling the incongruity are absent." (p. 115) This type of inter action exists, for example, when a person answers, "I had an apple for lunch." when asked what time it is. The first statement is disqualified by the incongruity of the second. However, each meaning is congruent in itself. The incongruence is apparent only when the context is con sidered. The sender can claim that either the contextual or non-contextual meaning is the "true," or intended, meaning. Sluzki, Beavin, Tarnoplosky, and Veron (1967) listed several types of transactional disqualifications: (1) evasion, an unlabelled change of subject following a statement which did not end a topic of discussion; (2) sleight of hand, a change of subject labelled as an answer; (3) status disqualification, a change of subject from con tent to speaker accompanied by the invocation of relative 92 status to denegrate the previous message; and (4) redun dant question, a repetition of part of the previous message in the form of a question. Mark (1970) pointed out that such messages are like one person's saying to another, "You don't exist, given that definition of our relationship." For this reason, these messages are of particular interest to psychother apists and theorists dealing with families and couples. These responses are included as a subclass of paradoxical interactions because they represent one person's attempt to control the relationship by disqualifying the context of his response. Little empirical work has been done using the com munication strategies described above. The apparent sub tlety of the task, and the extent to which context and non-verbal behavior qualify the meaning of verbal communi cation, may well have discouraged potential investigators. The importance of this field of study is suggested by the voluminous theoretical work dealing with these strategies of communication, and by the extensive refer ences to them in discussions of marital and family therapy. Further, some steps toward operationalizing the relevant concepts have been taken. Interest in this field also has been arroused among students of relative power in mar 93 riage. Olson and Rabunsky (1972) and Sprey (1972) repre sent an emerging concensus that such power is a subtle variable, and that the strategies of exerting control over the relationship should be studied to further our under standing of the process of powering. In the next section, previous attempts to opera tionalize the strategies of communication will be presented and evaluated. Empirical Background The primary focus of this section is on exploring and comparing the different approaches used in studying the strategies of communication presented in the previous section. First, two studies which did not use formal in teraction coding will be presented. Then the coding pro cedures used in other studies will be presented and com pared in detail. Studies without Formal Interaction Coding Gorad (1971) developed a method of operationalizing one-up behavior for which one disclaims responsibility 94 (i.e., paradoxical behavior) in a game situation. He rea soned that alcoholics hold alcohol, or their "involuntary" addiction to it, responsible for much of their assertive behavior. To see if alcoholic men would do this with their wives more than would normal husbands with normal wives, he used an interaction game in which monetary rewards were contingent on joint performance. Subjects chose responses of "share," "win," or "secret win" in each of the 50 rounds of the game. Only the "secret win" choice resulted in some monetary loss to the partner. Since subjects were told the experimenter could substitute this choice for either spouse's response in any round of the game, whether or not a person had made this choice was supposed to be hidden from the partner. Since the experimenter, in fact, never substituted, it seems plausible that subjects could figure this out. Gorad found significant differences in the pre dicted direction between alcoholics and the other partici pants in the number of "secret win" choices made. His other hypotheses were not supported. In spite of the in genuity of the game, the extent to which game behavior re flects communication strategies in other contexts remains open to question. Using a somewhat more naturalistic approach, Jack 95 son, Riskin, and Satir (1961) presented and analyzed a five minute segment of a family interview. The analysis was essentially "blind" and done by the authors them selves. As part of the analysis, marginal comments iden tifying the communicational stance of the speaker were presented for each speech. The categories used were "com plementary one-up," "complementary one-down," and "sym metrical." Instances of disqualification also were noted. Brief, general definitions of the strategies of communica tion were used as guidelines in this clinical presenta tion. The authors argued that: It is conceivable, after making due allowances for context, that the patterning of complementary and symmetrical responses will give a rough and simple index to varying types of communications in families, and that these patterns in turn will be correlated with specific manifestations of mental illness.(p. 323) Certainly, more explicit coding guidelines will be needed if such progress is to be made. Studies using Formal Interaction Coding Sluzki and Beavin (1965) made the first attempt to operationalize definitions of the relational aspects of transactions. They listed examples of symmetrical and 96 complementary transactions, as follows: giving/taking instruction = complementary (giving = one-up, taking = one-down) asking/answering = complementary (asking = one-down, answering = one-up) asserting/agreeing = complementary (asserting = one- up, agreeing = one-down referential statement/referential statement = symmet rical agreeing/agreeing = symmetrical giving instructions/countering with instructions = symmetrical This paper was theoretical in nature. Its major thrust was to suggest the possibility of operationally defining the strategies of communication. Obviously, such a paper did not take into account many particular problems which come up in interaction coding, such as that posed by over lapping conversation. Further, relational nuances provid ed by voice tone and other non-verbal components were ig nored. Mark (1970, 1971) used the Sluzki and Beavin ap proach as a starting point for further development of such a system. He reported that his raters found his Codes to be "considerably easier to handle, more precise and re- 97 taining more information than the Sluzki-Beavin method which was originally tried." (1971, p. 230) Mark's coding system represents a considerable expansion of the earlier one. He increased the categories to include "talk-over," meaning speech while another per son is speaking, and to differentiate successful from un successful interruptions on the basis of whether or not the interrupting speaker gains the floor by his maneuver. Two categories were added to recognize the posing and an swering of a subject's own question and the following of one's assertion by a question. His "disconfirmation" code was defined to be the equivalent of transactional dis qualification, which was defined in the preceding section. For Mark, the unit of interaction to be coded is the "speech," "all that one speaker says before another gains the floor." (1971, p. 224) However, he introduced ambiguity into this definition. He specified that a speaker could follow himself in the sequence of speeches by successfully interrupting his partner or by making "a major shift" in his speech. Examples of the latter would be following an assertion or question with an instruction or order. Mark's second digit codes six and seven, "as sertion and question" and "question and assertion," by contrast, each label two grammatical events, but are coded as one speech. These facts bring to light some inconsis- 98 tency in Mark's definition of a "speech" as the unit of interaction to be coded. Mark's approach calls first for labeling each speech with a three digit code. The first digit repre sents the speaker; the second represents one of the "modi fied grammatical format" categories; and the third labels "what it is in terms of a response to the preceding speech. The second and third digit codes are presented in Table 1, along with the corresponding codes from Ericson and Rogers (1973) . The three digit codes are then classified into the categories of symmetrical, one-up or one-down. The rules for this procedure are extremely complex. When they are translated into words, his definitions are those found in Table 2, where they are presented beside the control di mension codes used by Ericson and Rogers. The derivative classes of one-up, one-down, and symmetrical are then combined once more in Mark's scheme. His system produces nine codes at the transactional level; each of the three control dimension codes is joined with each other code in a two speech transaction. The trans actions then are labelled with such titles as "competi tion towards one-up," "symmetrical-symmetrical" and "com petition towards one-down." Upon examination, it becomes clear that Mark moves 99 TABLE 1 A COMPARISON OF SECOND AND THIRD DIGIT CODES: MARK (1971) VS. ERICSON AND ROGERS (1973) MARK ERICSON AND ROGERS Second Digit Codes: Third Digit Codes: 1. 2. 3. 4. 5. 8. 9. 0. Question Assertion Instruction Order Talking Over Assertion & Question Question & Assertion Other (filled hesitation) Laughter (re sponse to own speech) Silence (over 3 seconds) Assertion Question Talk-Over Non-Complete Other Silence EJLLOrities: Talk-Over, question, assertion, non-complete & other 1. Agreement 2. Disagreement 3. Extension 4. Answer 5. Disconfirma- tion 6. Topic change 7. Agreement & Extention 8. Disagreement & Extension 9. Other (coughs & vocal pauses) 0. Laughter (re sponse to partner) 1. Support (seeking & giving) 2. Non-Support 3. Extension 4. Answer (if commit ment is implied) 5. Instruction 6. Order 7. Disconfirmation 8. Topic Change 9. Initiation-Termina- tion 10.Other 0. Silence Priorities: Initia- tion-Termination, Answer, Disconfirma tion, TopicrChange, Order, Instruction, Support, Non-Support, Extension, Other. 100 TABLE 2 A COMPARISON OF CONTROL DIMENSION CODES: MARK (1971) VERSUS ERICSON AND ROGERS (1973) MARK 1>2 ERICSON AND ROGERS One-Up a) ,Any instruction or order. b) Question in disagreement, disconfirmation, or disa greement and extension. c) Assertion in disagreement, disconfirmation, or topic change. d) Assertion in disagreement and extension, unless a second coding preceded or followed by assertion in extension or answer. e) Assertion in answer or ex tension, unless preceded or followed by either of these. f) Second coding of assertion in agreement and extension. g) Talk-over, if the interrup tion is successful. h) Talk-over in disagreement, if followed by talk-over in agreement. i) First speech of interac tion, unless followed by assertion in extension or a) Any form of speech used in non-support, answer, instruction, order, disconfirma tion or topic change. b) Any form of speech, except non-complete, used in initiation or termination. c) Any talk-over used in extension. answer. One-Down a) Assertion in agreement, un less followed or preceded by talk-over, other, or as sertion in agreement. b) First coding of assertion in agreement and extension. c) Assertion in other. d) Question, unless in disa greement, disconfirmation, or disagreement and extensi' a) Any form of speech used in seeking or giving support. b) Any talk-over or gues tions, used in other. c) Any question used in extension. 101 TABLE 2 (Continued) MARK if2 ERICSON AND ROGERS One-Down e) Talk-over, unless a success ful interruption. f) Grammatical code of "other." Symmetrical One Across a) Pair in which second and third digit codes of the two speeches are identi cal. b) Talk-over, other, or as sertion in agreement, followed or preceded by any of these. c) Second coding of disagree ment and extension, fol lowed or preceded by assertion in extension or answer. d) Assertion in answer or ex tension, followed or pre ceded by either of these. e) First speech in interac tion, if an assertion in other followed by asser tion in extension or in answer. Speeches in the "assertion and question" category are coded as assertions the first time and as questions the second time. The opposite rule applies to speeches in the "question and assertion" category. 2in Mark's scheme, "first coding" refers to the coding of a speech as a response to the preceding message, while "second coding" refers to the coding of the same speech as a stimulus for the following response. a) Any assertion or other, used in extension or other. b) Any non-complete used in extension, other, initia tion or termination. 102 away from the theoretical literature in some of his defi nitions. For instance, his operational definition of "symmetrical" appears to be more restricted than the theo retical definition. He calls symmetrical only those transactions composed of identical or nearly identical syntactic and relationship behavior of the two partners. The focus for him becomes parallel behavior rather than the attempt to demonstrate equality with one's partner through competition for control of the discourse or by outdoing the other. Mark's definition of one-up seems to includes all argumentative communication. Such behavior would seem to be competitive and would more appropriately fall in the symmetrical category. The person who is argu ing is struggling for control of the discourse; he is not in control. Similar reasoning applies to his classifica tion of successful interruptions as one-up. With all of his complex scoring procedures, it seems that Mark lost sight of the critical nature of the control of the rela tionship which must be the very essence of such strategies of communication. Mark used this coding system to analyze audio tape recordings of three discussions between each of fifteen upper class and fifteen lower-class Caucasian couples with children under ten years of age. The couples were asked to decide together what to do if these situations occurred: 103 (1) a civil defense emergency, (2) hearing your child dis cuss sex with incorrect information, and (3) having your child ask to stay up late to watch an educational program on television. Four raters were trained for about two hours, then worked in independent pairs to code three tapes (without transcripts) for a reliability study. The agreement be tween pairs of raters for unitizing was virtually 100%, while the agreement for over all categorization was 92%. The total number of speeches used in this comparison was only 275. Mark found that all but one of his couples tended to use consistent patterns of communication across all topics. In other words, patterns of transactional codes recurred a great deal in most of his couples. Several significant differences between data from upper- and lower- class couples also emerged. Generally, the data supported the view of upper class interaction as "supportive and ex tending" but with "strong covert control" indicated by the large number of disconfirmations (1970, p. 77) . The lower- class conversations appeared to be series of assertions interspersed with disagreements, silence, and "rapid-fire" interruptions (1970, p. 75). There was a tendency for upper-class couples to use more symmetrical behavior, while lower-class couples demonstrated more"competition 104 toward one-up" behavior. All but very few of the inter actions studied showed consistency of relational communi cation across topics, and Mark identified several recurrent patterns. Ericson (1972) and Rogers (1972) developed their coding scheme as a refinement of Mark's (1971) system. (See also Ericson and Rogers 1973). Their system is based on a similar three digit approach, but many changes were made. (See Tables 1 and 2 for a comparison.) At this first level of coding, the following modifications were introduced: (1) the four overlapping code categories were deleted; (2) a separate category for initiations and termi nations was established; (3) "order" and "instruction" were shifted from grammatical (second digit) categories to rela tionship (third digit) categories; and (4) "agreement" and "disagreement" were replaced by "support" and "non support." Support was defined to include "the giving and seeking of agreement, assistance, acceptance and approval," while non-support was "used to denote disagreement, rejec tion, demands and challenges." (Rogers, 1972, p. 83). The last change is particularly unfortunate. The seeking and giving of support are two behaviors of a clearly comple mentary nature when joined together in a transaction. A transaction such as "I want your support" followed by "You have it" would be coded as symmetricalr of the competition- 105 toward-one-down variety, in the Ericson and Rogers system. At the second stage of classification, where the grammatical and relationship codes are grouped into codes of control dimensions, major change also was evident. Ericson and Rogers added "one-across" to this group of codes, defining such speeches as attempts to neutralize control. The theoretical literature has no space for such a code, since the theorists insist that there is a non neutral control component in all communication. Ericson and Rogers tried to deal with this problem by saying one- across statements have less intense implications for con trol than one-up or one-down statements. However, the state of the art in this field has not advanced far enough to permit intensity scaling. If it were that far advanced, it would still be inaccurate to create a sepa rate category for low intensity complementary communica tions, omitting information about whether they were one- up or one-down. In removing "symmetrical" from the control dimen sion and reserving it for the transactional dimension, Rogers and Ericson deviated further from Mark and from the theoretical literature. However, they did recognize the extreme narrowness of Mark's definition of "symmetrical." In Ericson and Rogers' work, nine transactional 106 codes were generated from all possible combinations of two message codes. These codes included three sub-types of symmetry, in which the control codes of both speeches in a transaction were the same. Two sub-types of complementari ty, one-up followed by one-down and one-down followed by one-up, also were generated. The remaining combinations of control codes included one-across codes and were con sidered "transitory." Ericson and Rogers, like Mark, chose the "speech" as the unit of interaction to be coded. They, too, found the need for double coding speeches in which two sets of codes were needed to describe the syntactic and relation ship qualities of the speech. They had transcribing typ ists unitize the interaction for coding and, therefore, had no interrater reliability data to report on this proc ess . Interrater agreement on coding reached an average of 86% and ranged from 68% to 100% on 24 discussions selected at random and rated after two training sessions were held. These writers used their coding scheme to investi gate transactional patterns among 66 married couples with children under age twelve. The subjects volunteered after being selected at random from a telephone directory. The subjects discussed the following topics involving their relationship and emergency situations: (1) how they met and married; (2) sharing home basements in the event of a 107 nuclear emergency; (3) career versus family interests; and (4) family plans for protection in the case of a natural emergency. Ericson (1972) analyzed the data from this study in relationship to measures of dominance and social class. None of his hypotheses were supported by the data. He did find that lower-class couples engaged in a higher propor tion of symmetrical transactions than did upper-class cou ples. The opposite relationship held for complementary transactions. This finding was not in agreement with Mark's (1970) results. The coding system he used ma,y have been responsible for this. 47% of the speeches were coded as "one-across," and 60% of the transactions were coded as "transitory." With this large proportion of the data falling in newly created categories, with no place in his hypotheses or the relevant theoretical literature, it is not surprising that his hypotheses were not supported. It is surprising that he analyzed the data at all. Rogers (1972) related the same communication data to an index of role strain. She found more symmetrical transactions among couples with high than with low role strain scores. In looking at her data on control level classification, only the number of "wife one-up control movements" differentiated significantly between groups. 108 At the primary, three digit, level of coding, she found significant differences between groups on the proportions of support messages and unsuccessful interruptions. Cou ples with higher role strain scores showed less of these behaviors than did the other couples. Rogers' findings relating to communication strategies are subject to the same criticisms applicable to Ericson's results. There are serious shortcomings in the coding sys tems used in all three of these studies. Some of the con ceptual problems present in Mark's (1970, 1971) work were overcome by Ericson (1972) and Rogers (1972) . However, new problems were created with the creation of new control dimension and transactional codes, with the inclusion of giving and seeking support in one category, and with the removal of "symmetrical" from the control dimension coding. Nevertheless, the Ericson and Rogers work does indicate a strong concern with the implications for control of commu nication present as one speaks, and with the need for simplifying and rationalizing coding procedures. It is questionable whether grammatical codes should be used at all when the relationship aspect of com munication is to be coded. Logically, it can be argued that any form of speech can be used to accomplish any semantic purpose. Gill (1971) discusses this point in the 109 introduction to his study, which will be considered next. Gill (1971) developed his approach to codification working directly from the theoretical material and without the mixed blessings of exposure to Sluzki and Beavin's (1965) article or Mark's (1970) dissertation. Ericson and Rogers (1972) worked without reading Gill's study. If they had done so, several conceptual pitfalls might have been avoided. Gill made a crucial differentiation between syntax and semantics. He pointed out that: syntactics may be considered to be the elements and combination rules of language. These are found in dictionaries, rhetorics, etc. Seman tics, on the other hand, is what people use syntactical elements to do. (p. 4) Given this understanding, it is important to note that a variety of syntactical elements may be used to ac complish approximately the same semantic tasks. Similarly, a variety of semantic tasks may be accomplished with the use of the same syntactical elements. A parallel situa tion will help to explain what this means. The same tools and construction materials may be used to build several different houses. Virtually the same house may be built with a variety of different tools and materials. The tools and materials are like syntactical elements, while the houses are like semantic meanings. 110 Within semantics, Haley (1963), Watzlawick, Beavin and Jackson (1967) and others, have distinguished between the grammatical and relationship aspects of meaning. The argument just applied to the distinction between syntax and semantics also fits these two aspects. For example, "What the hell do you mean by doing that?" is a question, but may serve to denegrate the listener. On the other hand, if spoken gently with a smile, it may be used to indicate appreciation for a favor. This argument seems to be a sufficient basis for disregarding grammatical format in coding the relationship component of communication. With this consideration in mind, Gill (1971) de veloped his approach to coding. The coding stipulations were specific to his particular experimental situation, in which the experimenters were trained to adopt a one-up stance. The guidelines consisted of the modal examples presented in Table 3. He contended that this approach was better than a system of rigid definitions because of the necessity of paying attention to the implications for con trol of communication present in the discourse. Rigid guidelines for coding aspects of speech which are supposed ly components adding up to the control dimension could make coding of this dimension less accurate than guidelines for coding it directly. This is because the former coding is one step removed from the latter, and because of the dif- Ill TABLE 3 GILL'S (1971) RELATIONSHIP CODING GUIDELINES Experimenter One-up; One-down: Symmetrical: Paradoxical: A) a) a) b) a) b) Asked questions or agreed or disagreed with subject, without becoming involved in a heated exchange. Answered subject's one-up questions seeking information. Beeame involved in heated exchange. Responded to subject's symmetrical maneuvers by a series of responses which escalated into conflict. Disqualified his response. Made statements which could not be re sponded to unambiguously. One-up: One-down: Subject a) Symmetrical: Paradoxical: a) b) a) b) c) d) e) f) a) b) Asked a question designed to gain in formation so that subject could better respond in a one-down fashion. Following experimenter agreement, con tinued the same description, line of reasoning, or answer which he had pre sented prior to agreement. Following experimenter disagreement, changed his position or made a retrac tion (to comply). Unable to go on. Asked a question which disrupted the previous set. Silence. Comments not related to discussion topics. Following experimenter agreement, changed his position (logically)• Following experimenter disagreement, defended his position or changed his position (as a maneuver to gain more advantage). Disqualified his response. Made statements yhich aould not be re sponded to unambxguously. 112 ference between semantics and syntax explicated earlier. Gill's guidelines would produce different codes for many responses than would Mark's (1970) or Ericson and Rogers' (1973) systems. For example, Gill places ques tions on the part of subjects generally in the one-up cat- etory in his experimental context. This decision must be saluted since a question which is primarily seeking in formation contains a very strong directive for the re spondent to answer. The nature of the next statement is greatly constrained by the question. Gill included "para doxical" as a category in his study, and he created no new categories, differing in these ways, too, from the others. Gill used the "remark" as the unit of conversation to be coded. Generally, a remark was the same as a "speech," the unit used in the above studies. However, if a speech was made of two clearly defined segments sepa rated by a pause, it was coded as two remarks. This strategy overcame the difficulty imposed by the need for dual coding of some responses when the speech is the unit coded. Since Gill's raters coded tapes while listening to them together, and without transcripts, joint decisions were made about unitizing the data. Therefore, no reli ability data on this process are available. In a check of interrater agreement using three randomly selected ses sions, the average interrater agreement on both semantic 113 classification and sequence of response was 87%. Agree ment for the different pairs of raters among the three raters used ranged from 74% to 98% for the different strategies coded. Gill did not discuss the procedures used in training raters, but did state that they were trained. Gill monitored the interaction of 32 college stu dents, each talking with an experimenter for ten minutes about six subjects of social relevance. One such topic was "What should be done about the poverty problem?" Ex perimenters agreed with the subjects half of the time and disagreed the other half of the time. The experimenters were assumed to be in a one-up position and were in structed to use one-up responses as much as possible. Ex perimenters posed the questions for subjects to discuss. The interaction, then, was somewhat standardized and con trolled by the experimental design. The results indicated that the relationship re sponses during periods of agreement and disagreement were significantly different. In particular, complementary one-down responses occurred more frequently in the agree ment condition and symmetrical responses were more common when the experimenter disagreed with the subject. Gill concluded that subjects were more comfortable with their one-down positions when the experimenter agreed with them 114 than when the experimenters disagreed. The discomfort in the disagreement condition, he reasoned, led them to try to maneuver out of the one-down position by making sym metrical responses. Because of the superior logic and faithfulness to theoretical roots in Gill's study, his approach to coding was modified for use in the present research. Rationale The present study was designed with two major goals in mind. The first was to develop an approach to coding the relationship dimension of communication between mates which would be consistent with the theoretical defi nitions of the strategies of communication. Considera tions related to this purpose are explored first in this section. The second goal was to apply this coding system to discussions of topics found to be important to marital satisfaction, and on which agreement and disagreement were maximized. The reasons for choosing this task and the subjects used are presented next, along with a general outline of the experimental design. Gill's (1971) guidelines for coding the strategies of communication were modified for use in the present re search. His approach of using "modal examples" as "rough 115 guidelines" for classifying responses was chosen as a starting point because of its consistency with the defi nitions supplied by the major theorists and presented in the first part of this chapter. His guidelines allow the coders to focus on the relationship, or control, dimension of communication. The other coding procedures examined were found to be inadequate in this regard and to deviate in important ways from the theoretical substructure of the communication strategies. Gill's guidelines took into ac count the context of the communication examined, and were created specifically for his experimental task. Gill's (1971) guidelines were modified to suit the situation in which spouses discuss topics of agreement and disagreement relating to their relationship with each other. This situation is potentially more varied and com plex than was Gill's. Several differences stand out: 1. The subjects are not strangers meeting for the first time. Husbands and wives are engaged in long-term, intimate, interdependent relationships, with a history of communication rules behind them. 2. The topics for discussion were areas of agreement and disagreement relating to the mates' perceptions of how they feel about particular aspects of their relationship. Such topics were expected to be of higher valence for sub 116 jects in this study than were the topics Gill used. They seemed likely to provoke stronger emotional, responses in the subjects. 3. Both people in the dyads studied were subjects. Neither was briefed ahead of time or instructed to take a particular role in the conversations. These differences were expected to make communica tion more complex, and scoring more difficult, in the pre sent study than in Gill's. Nevertheless, the investiga tion of communication strategies between mates, with the implications for control in defining relationships, would seem to be possible, as shown in the studies reviewed in the previous section. Gill's guidelines for coding one-up, one-down, symmetrical, and paradoxical responses were revised to fit this less structured experimental context. In addition, guidelines were developed for coding subclasses of these strategies which have particular interest for students of psychopathology and communication. These subclasses are: metacommunicative and metacomplementary responses and transactional disqualifications. It was hoped to deter mine how frequently, and under what conditions, such re sponses occurred. The task of having couples discuss important areas 117 of their marital relationship about which they agreed and disagreed strongly was chosen for several reasons. First, the variable of agreement and disagreement was used in Gill's study, and has been used in many studies of marital interaction. Second, it was felt that one could assume there was greater stress on the relationship in the dis agreement condition than in the agreement condition. Such stress would be expected to have an influence on relation al communication. Third, it was decided to maximize this presumed difference in stress by choosing topics for dis cussion on which agreement and disagreement were maximal on issues highly correlated with marital satisfaction. Since the relational aspects of communication were being studied, it was deemed preferable to maximize this differ ence rather than to standardize the topics. Precedent for this decision can be found in the Rodnick-Goldstein re search group's decision to have members of disturbed fami lies interact over problems specific to their particular families (see, e.g., Goldstein, M. J., Judd, L. L. Rod- nich, E. H., Alkire, A. A., and Gould, E., 1968). A con flict resolution paradigm was rejected for this study since tasks of this type carry an expectation that dis agreement should be resolved and because the open-ended topics used in this study do not fit easily into such a format. 118 The decision to use married couples as subjects was made because the marital relationship seems to provide the opportunity to explore communication in the context of prolonged, chosen, intimate contact between man and woman. It was decided to use as subjects couples in marital ther apy in comparison with couples considered "normal" and not in psychotherapy. This made possible the examination of the question of whether couples whose marriages are in trouble enough for them to seek help differ from the other groups in the strategies of communication they employ un der conditions of agreement and disagreement. Since data on difference between husbands and wives on the dependent measures were easily available in this study, it was de cided to evaluate them for any difference which might ap pear. Hypotheses This research was designed to permit the testing of several hypotheses relating the strategies of communi cation to the experimental conditions and subject vari ables discussed above. The rationales for hypotheses re lated to each communication strategy are set forth first in this section. Then the hypotheses are presented in the form of a list. 119 Predictions about Paradoxical Responses Paradoxical response rates were presumed to be re lated to stress and to familiarity with paradoxical com munication. The disagreement condition was assumed to be more stressful than the agreement condition. It is rea sonable that more stressful circumstances are more likely to elicit symptomatic behavior than less stressful ones. Since it is logical that there is less to hide when agree ing with somebody than when disagreeing, there would seem to be little need for paradoxical communication when under relatively mild stress. Thus, it was predicted that more paradoxical communication would occur in the disagreement condition than in the agreement condition. There is some empirical support for such a prediction. Gill (1971) found his subjects made no paradoxical responses in his agreement condition, but made an average of 1.6 such re sponses in the disagreement condition. The difference in stressfulness of the two condi tions was considered to be greater for people who have relatively greater difficulty accepting themselves or each other. Disagreement over important aspects of their rela tionship was expected to be more stressful for the mates in marital therapy as opposed to those not in therapy. The clinic subjects were expected to be more likely than 120 others to feel a need to avoid accepting responsibility for some aspect of their communication, particularly in a high stress situation. Further, if paradoxical communica tion is related to symptomatic behavior, and if symptomat ic behavior is related to clinic status, then members of this group would have learned the methods of paradoxical communication better than the non-clinic subjects. Therefore, the proportion of paradoxical communi cations was anticipated to be smallest among the non-clinic subjects discussing areas of their relationship on which they agreed. The proportion of paradoxical communications was expected to be greatest among the clinic group dis cussing topics on which they disagreed. Overall, the non clinic group was expected to use a smaller proportion of paradoxical communications than the clinic group. It was predicted that all subjects would use a greater proportion of paradoxical statements when in the disagreement condi tion than in the agreement condition. Further, it was anticipated that the difference in paradoxical response rates between the two conditions would be greater for clin ic subjects than for non-clinic subjects, since the former were deemed likely to experience the difference between agreement and disagreement as more marked than the latter. 121 Predictions about Transactional Disqualifications Transactional disqualifications, messages which are inconguous with the set created by the preceding re sponses, were coded as a subset of paradoxical responses. The rationale and predictions for this group of responses were the same as those presented for paradoxical responses. Predictions about Symmetrical Responses Differences in the rates of symmetrical responses also were expected. Gill (1971) found a mean of 1.1 sym metrical responses among his subjects while in the agree ment condition and a mean of 45.5 symmetrical responses while in the disagreement condition. This difference was significant at at least the .05 level. Gill reasoned that: The fact that Ss did not remain in a compatible [i.e., complementary] one-down position when they were disagreed with may have been largely due to the previous cultural conditioning of the Ss. Thus, it may be argued that being disagreed with served as an aversive stimulus of sorts, and when presented with this stimulus, Ss made responses to terminate this aversive stimulus, i.e., ration alization, further explanations in an attempt to convince Es, etc. On the other hand, Ss were virtually never symmetrical when they were agreed with (no S asked "Why do you agree with everything I say?") (p. 68). 122 While it may be true that having someone disagree with one is aversive, Gill's findings must be limited in generality to the situation in which a person in a one-up position disagrees with another, since his experimenters maintained a one-up position throughout. It may be that disagreement from a person in a subordinate or symmetrical relationship would function differently. It was expected that the disagreement condition in this study would be more stressful than the agreement con dition, and differences in communication strategies simi lar to those found by Gill were predicted. The state of being in disagreement over an important aspect of one's marriage was regarded as aversive. Responses aimed at terminating such a state by bringing about change in the other person's opinion, or the topic, or engaging in a competitive interchange, were classified as symmetrical. Gill's findings suggested that attempts to change the mind of one's mate would be most common in the disagreement con dition. Common sense suggests the same outcome. There fore, it was predicted that both partners would show a higher proportion of symmetrical responses in the disagree ment condition than in the agreement condition. The assumption was made that non-clinic subjects would be able to change from one-down to symmetrical re sponses more easily than clinic subjects. This followed from the belief that clinic subjects would be less able to terminate interpersonal discomfort in an assertive manner, and would be more likely to accept passively any definition of their relationship offered by their mates, than would non-clinic subjects. Following this reasoning, it was predicted that non-clinic subjects would use more symmetri cal responses than clinic subjects. Since non-clinic subjects were presumed to be able to change from one-down to symmetrical responses more easi ly than clinic subjects, and since the desirability of making such a shift was demonstrated above, a significant interaction also was predicted. It was hypothesized that the difference in the proportion of symmetrical responses between conditions would be greater for non-clinic than for clinic subjects. Predictions about Metacommunicative Responses Metacommunicative responses were expected to be too rare to make testing hypotheses based on them reason able. Theoretical considerations, however, suggested some basis for making two mutually exclusive predictions about them. These hypotheses were to be tested if enough meta communicative responses occurred to make such evaluation 124 worthwhile. The prediction that these responses would be more common among non-clinic than clinic couples was made on the basis of the following argument. One must be able to refrain from conforming to the response expectations of one's partner in order to talk about the nature of commu nication taking place. It was thought that this ability would be more common among non-clinic than clinic subjects, because of the presumed greater "ego strength" of the former. The other prediction, that clinic couples would make more metacommunicative responses than non-clinic cou ples, was suggested by Broderick (1973). He pointed out that clinic couples could receive training in metacommuni cation as part of their therapy. If this were the case, then it would be likely that they would use this type of response more than non-clinic couples. Metacommunicative responses were grouped as a sub class of symmetrical responses. The first prediction made above was consistent with the other hypotheses about sym metrical responses. The second was not. If the second hypothesis was to be supported by the data, data analyses utilizing symmetrical responses were to be done excluding metacommunicative responses from this category. Such a finding, then, would have been considered as an artifact of the treatment process. 125 Predictions about One-down Responses Relationships opposite those predicted for sym metrical responses were predicted for one-down responses. The reasoning behind these predictions was just the oppo site from the reasoning presented in discussing the pre dictions about symmetrical responses, presented above. The rationale will not be discussed again, except to cite Gill's (1971) finding which lent support to the first prediction. Gill reported a mean of 43.0 one-down re sponses in the agreement periods, compared with a mean of 20.4 one-down responses in the disagreement periods. Three hypotheses relating to one-down responses were advanced. First, it was predicted that one-down re sponses would be more frequent in the agreement condition than in the disagreement condition. Second, it was an ticipated that these responses would be more common among clinic than non-clinic subjects. A significant interac tion also was predicted, suggesting that the difference in the proportions of one-down responses between the two con ditions would be greater for the non-clinic than the clin ic subjects. 126 Predictions about One-up Responses A basis for one prediction about rates of one-up responses was found in the literature. Gorad (1971) ad vanced this position most succinctly. He suggested that alcoholics, like other people with psychiatric symptoms, would have more difficulty assuming responsibility for one-up behavior than would normals. Since a one-up re sponse requires that such responsibility be accepted, it was predicted that one-up responses would be more frequent among non-clinic than among clinic subjects. No other predictions were made regarding the frequency of one-up responses. Predictions about Metacomplementary Responses Metacomplementary responses were regarded as a subset of one-up responses. They were expected to be too infrequent to justify statistical analysis. It was de cided to code them because of their theoretical interest, and to report the findings in anecdotal form. Predictions about Husband-Wife Differences Due to the sexist nature of our society, one could 127 presume that women experience themselves as less powerful than men. However, no such assumption was made in this study. This assumption would generate predictions that wives would use more frequent paradoxical and one-down responses than would husbands. It was decided to examine non-directional hypotheses relating to husband-wife main effects and interactions, for the sake of exploring these data. The specific hypotheses elucidated above compose a test of the more general hypothesis that strategies of com munication between mates are influenced in a predictable manner by the variables of clinic versus non-clinic status and by the content variable, discussion of areas of agree ment versus disagreement. The particular hypotheses which composed this test are reformulated in this list. Paradoxical Responses 1. The percent paradoxical responses will be greater in the disagreement condition than in the agree ment condition. 2. The percent paradoxical responses will be greater among clinic subjects than among non-clinic sub jects. 3. The difference in the percent of paradoxical re- 128 sponses between the two conditions will be greater for clinic subjects than for non-clinic subjects. Transactional Disqualifications 4. The percent transactional disqualifications will be greater in the disagreement condition than in the agreement condition. 5. The percent transactional disqualifications will be greater among clinic subjects than among non clinic subjects. 6. The difference in the percent of transactional disqualifications between the two conditions will be greater for clinic subjects than for non-clinic subjects. Symmetrical Responses 7. The percent symmetrical responses will be greater in the disagreement condition than in the agree ment condition. 8. The percent symmetrical responses will be greater among non-clinic subjects than among clinic sub jects. i | ! 129 9. The difference in the percent of symmetrical re sponses between the two conditions will be greater for non-clinic subjects than for clinic subjects. Metacommunicative Responses 10. The percent metacommunicative responses will be greater among non-clinic subjects than among clinic subjects. The opposite prediction also will be tested. One-Down Complementary Responses 11. The percent one-down responses will be greater in the agreement condition than in the disagreement condition. 12. The percent one-down responses will be greater among clinic subjects than among non-clinic sub jects . 13. The difference in the percent of one-down responses between the two conditions will be greater for non clinic subjects than for clinic subjects. One-Up Complementary Responses 14. The percent metacommunicative responses will be 130 greater among non-clinic subjects than among clinic subjects. Husband-Wife Differences 15. Main effects and interactions involving husband- wife differences, generally, and in combination with clinic versus non-clinic status and agreement versus disagreement, also will be evaluated. They will be tested against the null hypothesis, which suggests no differences. CHAPTER IV METHOD Having presented the background, rationale, and hypotheses for this study in the preceding chapters, the research design and procedures will now be articulated. The design of this study calls for the comparison of com munication strategies used by husbands and wives partici pating in, and not participating in, marital therapy. The strategies used are compared under two experimental condi tions, each requiring the discussion of important aspects of the subjects' marital relationship. The conditions are: discussion of an area of agreement, and discussion of an area of disagreement. Subjects The subjects were ten married couples currently in marital therapy and ten couples not in marital therapy. The couples in therapy (clinic group) were clients in three pastoral counseling centers in the area of Clare mont, California, Whereas therapists were asked to refer couples with particularly disturbed marriages, the willing ness of couples to participate was the final selection 131 132 criterion. Many of the couples reported to be most highly conflicted were not willing to be subjects. The subjects not in therapy (non-clinic group) were referred by a local minister and a graduate student in counseling psychology. They believed these couples to be "normal" and to be representative of the same general com munity from which the clinic couples were selected. Topic Selection Procedure The Marital Agreement Questionnaire (MAQ) was de signed for this study. Questions one through eight were taken from the Marital Adjustment Scale (Locke and William son, 1958). The six point scale of agreement used for all questions also was taken from this scale. Subjects were asked to rate and rank for agreement ten areas considered important to marital satisfaction. The topics included such items as family finances, time spent together, and sex relations. A copy of the MAQ appears in Appendix A. In this study, the MAQ was used for selecting an extreme example of agreement and of disagreement. The two topics having the highest and lowest average ranking, when both partners' rankings were considered together, were selected. These topics were used as the basis for the discussions from which the ratings of communication strate gies were made. 133 Procedure After potential subjects were referred for partici pation in the program, they were telephoned by the experi menter and informed of several things: (1) that participa tion in the couples' discussion program requires that both partners spend about one hour together with the experiment er; (2) that the session would involve discussion of impor tant aspects of their relationship with each other; (3) that, since the program was being carried out as a research project, the discussions would be tape-recorded and partic ipants would be asked to fill out some questionnaires; (4) that all information would be confidential and would be used for research purposes only. The findings reported would not be identified with the participants as individ uals in any way. If the couple was interested in partici pating, an appointment was arranged at their convenience. The experimental sessions took place in a room used for therapy by the Claremont Pastoral Counseling Center. The subject couples were seen separately. To assure inde pendent responses, they were seated separately at a table six feet long to fill out the questionnaires. They were at an easy conversational distance from each other for the discussions. The research assistant remained with the sub jects, at the distance of a few feet from them. It was believed that her presence might facilitate their focussing 134 on the discussion topics. A tape recorder with a built-in wide-range microphone was used as inconspicuously as pos sible during the discussions. The research assistant1 was a woman graduate student in developmental psychology at Claremont Graduate School. She was not familiar with the strategies of communication or with the clinic or non clinic status of the subjects. In each session, the research assistant introduced herself and indicated that the subjects would be asked to complete some questionnaires and to discuss together two areas of their relationship. She then asked the subjects to fill out three forms: (1) Consent for Participation in a Research Project (See Appendix B), (2) the MAQ, and (3) The Personal Information Form (See Appendix C). The forms were completed in the order listed. As soon as the MAQ was completed by both mates, the topics for discussion were selected. They were chosen for maximal concensus that the partners agreed on one topic and disagreed on the other. The Experimenter Check Sheet (See Appendix D) was used in this process. The subjects were presented in an order counterbalanced for agreement and disagreement. Half the clinic subjects and half the non clinic subjects discussed areas of agreement first. Each couple was instructed approximately as fol~ ^rs. Genie Hartmann. 135 lows: I have just looked through the questionnaires you completed and have selected two items for you to discuss. I shall give you the first topic, and you will have ten minutes to discuss your feelings about it with each other. You are to share your thoughts and feelings about the topic as it re lates to your own relationship. When your ten minutes are up, I'll signal time to stop and get ready for the next topic. Do you have any ques tions about the procedure? If not, I'll turn on the tape recorder and we can get started. Your first topic is, e.g., your feelings about the amount of time you spend together. The assistant then turned on the recorder and said, "Go ahead." After the first discussion, the experimenter indi cated that time was up and introduced the second topic, indicating when to begin and stop. When both topics were discussed, she indicated that "That is all." She thanked the couples for their participation and spoke informally with them for a few minutes to help relieve any anxiety which might have been generated. The tape recordings were transcribed by the Clare mont Secretarial Service, and the transcriptions were used in conjunction with the recordings as the material to be coded. The transcripts were checked for accuracy and cor rected by the experimenter. One-fourth of the discussions selected at random were unitized (i.e., divided into seg ments for coding) and coded by two independent raters who were not aware of which couples were from the clinic sample and which were not. This made it possible to obtain rates of interrater agreement on unitizing and coding. 136 Coding Procedure The coding procedure involved correcting the tran scripts, identifying each segment of interaction to be coded, and assigning each message to one of the communica tion strategies. Gill's (1971) approach to defining the strategies formed the basis for establishing the categories used in this study. The codes used were the following: one-up (including metacomplementary); one-down; symmetrical (including metacommunicative); and paradoxical (including transactional disqualifications). Raters were trained with sample recordings and transcripts obtained in a pilot study. They were directed to focus on the relationship aspect of the message, rather than on the content. Each message was to be considered as a response to the preceding message. In the recognition that context modifies meaning, the di rections for coding were written to include types of re sponses relevant to the situation in which the couple were speaking. Modal examples, based on possible responses in the situation, were used as guidelines for rating. For purposes of this study, a message was defined as any segment of interaction which seemed to the raters to be a single statement, serving a single semantic pur pose. A response with two separate segments requiring different codes was coded as two messages, while several 137 uninterrupted sentences serving one communicative purpose were designated as one message. The coding guidelines are presented below: 1. Complementary one-up, where a subject actively defines the nature of the interchange by, e.g., analyzing, persuading, directing, questioning, or introducing a new topic of discussion in a non-competitive manner; or agrees with the partner when the partner is taking a position in itiated by the subject; or disagrees with the partner with out becoming caught up in a defensive or competitive inter change; or moves to close a discussion without interrupting or aggressively "shutting down" the partner. la. Metacomplementary, where a subject states or implies that he gives permission for, or asks, or encour ages, the partner to define the nature of the following interchanges. 2. Complementary one-down, where a subject re sponds in a compliant manner to a directive, question, expository statement, or persuasive comment; or agrees with the partner, following or building on the partner's line of reasoning; without competing; or changes his own logical position to be closer to that of the partner; or indicates briefly that he understands what the partner is saying. 138 3. Symmetrical, where a subject changes his posi tion to be further from that of the partner; or defensively maintains his own position which is in conflict with that of his partner; or becomes involved in a heated or competi tive interchange; or asks a question or makes a statement disrupting the previous set without disqualification. 3a. Metacommunicative, where a subject comments on the communication which is occurring. 4. Paradoxical, where a subject disqualifies one or more aspects of his response (e.g., who is responsible for the message; to whom the message is directed; that a message is being sent; or that it is being sent in response to the preceding message, if it is), or makes a statement containing conflicting directives. 4a. Transactional Disqualification, where a sub ject makes a response which is incongruous with the frame created by the previous response, and does not label the incongruity. There are four subtypes:'*' (1) evasion (un labelled change of subject when the prior message does not end a topic of discussion; (2) sleight-of-hand (change of subject labelled as an answer); (3) status disqualification (change of subject from content to speaker, where relative status is invoked to disqualify the previous message; and ■*"The subtypes were identified by Sluzki, Beavin, Tarnoplosky, and Veron (1967). 139 (4) redundant question (repetition of part of the previous message in the form of a question when intent appears to be critical). If a message seemed to fit more than one of these categories, it was coded for the major apparent strategy. For example, if an interruption was paradoxical in nature, it was coded as symmetrical because the fact of interrupt ing was seen as overriding the importance of any other particular category into which the message might fall. For the statistical analysis of the discussions, one rater's coding was used. Generally, the tone of voice and context of a message were considered to be very impor tant determinants of which code was the most appropriate. Multivariate analysis was used. There were one between- couples variable (couples), two between subjects variables (sex and clinic versus non-clinic status), and one within- subject variable (agreement versus disagreement). Since the data to be analyzed were proportions, the analyses were performed on arc-sine transformations of the raw data. CHAPTER V RESULTS This chapter reports the results of this study. The first four sections provide the background for the fifth, which includes evaluations of the hypotheses of this research. Section one compares the clinic and non clinic groups of subjects on demographic characteristics. The topics discussed by the two groups are presented in the second section, while the third contains the results of the investigation of interrater reliability. Section four compares the two groups on the numbers of messages used in the discussions. The last section is made up of two parts: (a) a comparison of clinic and non-clinic sub jects and husbands and wives on the use of the strategies of communication during periods of agreement and disagree ment; and (b) a comparison of the same groups on certain sequences of strategies of particular relevance to this study. Demographic Characteristics Data comparing the clinic and non-clinic groups on several demographic variables are presented in Table 4 . 140 141 TABLE 4 DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLES Clinic Group Non-Clinic Group Signifi cance Age of Males mean s.d. Age of Females mean s .d. Education of Males s.d. Education of Females mean s.d. Years Married mean s.d. Family Income mean s .d. 35.7 7.51 32.7 15.65 2.10 14.80 1.94 10.65 7.26 $ 16,050 $ 5,169 Number of children mean 2.20 s.d. 1.99 27.9 2.7869 3.75 26.4 2.8873 p <.02 p <. 01 17.40 1,9783 n.s. 1.62 16.30 1,7045 n.s. 1.79 6.00 1,6101 n.s. 4.57 $13,250 2,2029 p<.05 $ 6,424 0.55 2,3645 p<.05 0.65 142 No significant differences emerged for the educational levels of males or females, or for the number of years married. The clinic group had significantly more family income (p.<.05), more children (p.<05), and was older (p.<.02), than the non-clinic group. In general the subjects were middle-class, tended to have at least two years of college, and had been married for an average of slightly over eight years. One of the clinic couples had seven children , but most of the subjects had zero to two children. The subjects tended to range from the early twenties to the late thirties in age. Two clinic husbands and one clinic wife had each been married one time before the present marriage. No other subjects had prior mar riages. Discussion Topics The ranks from the MAQ were used for selecting discussion topics. A rank of one indicated maximal per ceived agreement, and a rank of ten indicated maximal per ceived disagreement. Table five presents a comparison of the average ranks assigned the selected topics by the clinic and non-clinic groups of subjects. The average rank assigned by all subjects considered together to the topics of the agreement discussions was 1.80; and the mean rank for the disagreement topics was 9.2 The means for the 143 TABLE 5 DISCUSSION TOPIC RANKS FOR CLINIC AND NON-CLINIC GROUPS1 Group Average Rank for Husbands and Wives Agreement Disagreement Clinic mean = 1.7 mean = 9.2 (n = 10 couples) s.d. = 0.64 s.d. = 0.75 Non-Clinic mean = 1.9 mean = 9.2 (n = 10 couples) s.d. = 0.90 s.d. = 0.81 Total mean = 1.8 mean = 9.2 (n = 20 couples) s.d. = 0.77 s.d. = 0.78 1Ranks were taken from the Marital Agreement Questionnaires. 144 TABLE 6 TOPICS DISCUSSED BY CLINIC AND NON-CLINIC COUPLES Topic-*- Number of Discussions Agreement Disagreement Total Clinic Non- Clinic Clinic Non- Clinic Aims and Goals in Life 4 6 0 0 10 Finances 2 1 1 1 5 Responsibilities around the House 1 0 1 3 5 Good and Proper Behavior 0 1 1 1 3 Sex Relations 1 1 1 0 3 In-Laws 1 0 1 1 3 Demonstration of Affection 0 0 2 1 3 Time Spent Together 0 0 1 2 3 Angry or Hurt Feelings 0 0 2 1 3 Friends 1 1 0 0 2 TOTAL 10 10 10 10 40 ^The topics were taken from the Marital Agreement Questionnaire. 145 two groups did not differ appreciably from the total group averages. The topics discussed by the couples in each group are presented in Table 6. "Aims and Goals in Life" was most frequently selected. It was discussed by six non clinic and four clinic couples as an area of agreement. "Friends" was discussed least. It was selected for only two couples, both perceiving agreement about this area of their relationship. Three topics were discussed only as areas of disagreement: "Demonstration of Affection," "Time Spent Together," and "Angry or Hurt Feelings." All of the topics were used in at least two discussions, and most of the topics were discussed by both clinic and non clinic couples. Interrater Reliability For purposes of assessing interrater reliability, recordings of ten discussions were coded independently by two raters. The recordings included five discussions be tween clinic couples (two topics of agreement and three topics of disagreement) and five discussions between non clinic couples (three topics of agreement and two topics of disagreement). The agreement and disagreement dis cussions from two couples in each group were used, and one other discussion from each group of couples was included. 146 The discussions were selected at random, with the restric tion that the above criteria be met. The raters were not informed of which discussions were taken from which group. Interrater reliabilities were assessed for two processes, unitization and codification. The results are presented in Table 7. Interrater agreement on unitization, the designa tion of segments of conversation as messages, was 95.3%. Thus, about 95% of the 793 units identified by the rater-*- who coded all of the recordings also were identified as messages by the other rater.2 Agreement between raters on assigning codes to messages was calculated for all messages taken together and for each category of messages considered separately. Over-all agreement was 91.3%, while agreement on coding the separate categories ranged from 75.0% for metacommuni- cative, to 100% for metacomplementary. The mean interrater agreement for the different categories was 88.0%. Agree ment between raters on coding particular conversations ranged from 87.1% for a discussion of a topic of agreement by a non-clinic couple, to 95.2% for a discussion of a topic of agreement by a clinic couple. These levels of -*-Lewise Busch, the present investigator. 2Richard Busch, graduate student in pastoral care and counseling. 147 TABLE 7 INTER-RATER AGREEMENT! Process Coded % Agreement between two Coders Unitization (n = 793) 95.3 Codification Categories total (n = 793) - 91.3 One-Up (n=225) 89.3 One-Down (n=217) 94.0 Symmetrical (n=308) 92.2 Paradoxical (n= 21) 85.7 Transactional Disqualification(n=15) 80.0 Metacommunicative (n= 8) 75.0 Metacomplementary (n= 4) 100.0 mean 88.0 1Inter-Rater agreement was calculated for five dis cussions between clinic couples and five discussions be tween non-clinic couples. Half the discussions were about topics of agreement. Half were about topics of disagree ment. 148 interrater agreement were considered to be sufficiently high for the present purposes. Numbers of Messages Used Data comparing the number of messages coded per discussion for the variables of clinic status, spouse, and discussion class are presented in Tables 8 and 9. Table 8 presents means and standard deviations of the number of messages per discussion, while Table 9 summarizes the analysis of variance for this variable. Only the contrast between the clinic and non-clinic groups was significant (p.<.05). The non-clinic group used more messages per discussion than the clinic group. The difference in the number of messages between the agreement and disagreement conversations approached significance (p.c.10), with more messages occurring during disagreement than agreement. Strategies of Communication Relative Frequencies Multivariate analysis of variance was used to com pare the relative frequencies with which the different strategies of communication were used. The factors were clinic status, spouse, and discussion class. 149 TABLE 8 NUMBER OF MESSAGES USED Variable Mean Standard Deviation Total1 (n = 80) 44.0 17.83 Clinic Status Clinic (n = 40) 38.9 17.37 Non-Clinic(n = 40) 49.1 16.82 Spouse Husband (n = 40) 43.4 17.84 Wife (n = 40) 44.6 17.81 Discussion Class Agree (n = 40) 41.2 19.34 Disagree (n = 40) 46.8 15.68 n = 80 discussions = 40 subjects x 2 discussions per subject. 150 TABLE 9 SUMMARY TABLE OF ANALYSIS OF VARIANCE FOR NUMBER OF MESSAGES PER DISCUSSION Source df SS MS F Total 79 25444.0 Between Subjects 3!9 18978.0 A (clinic status) 1 2060.4 2060.5 4.584* B (spouse) 1 28.8 28.8 0.064 AB 1 76.1 76 .1 0.169 S/AB 36 16182.7 449.5 Within Subjects 40 6466.0 C (dis cussion class) 1 638.5 638.5 3.971** AC 1 9.8 9.8 0.061 BC 1 0.5 0.5 0.003 ABC 1 28.8 28.8 0.179 SC/AB 36 5788.5 160.8 *p.<.05 **p.<.10 151 The following communication strategies were used as variables: one-up, one-down, symmetrical, paradoxical, transactional disqualification, and metacommunicative. One strategy, metacomplementary, was not included for separate investigation because it was used rarely and by only ten subjects. As specified in the hypotheses, metacommunica tive responses were investigated separately and were in cluded in symmetrical responses, of which they form a sub class. Similarly, transactional disqualifications were examined separately and as they contributed to the larger class of paradoxical responses. Metacomplementary mes sages were included in one-up responses. The relative frequencies with which the different response classes occurred comprised the raw data for this analysis. This approach to the data was utilized because of the large differences between subjects in the total number of messages used. Since the data were in this form, and since relative frequencies are not normally distributed, arc-sine transformations of the proportions were used in the actual analysis. This type of transformation adjusts the distribution of relative frequencies to meet the cri terion of homogeneity of variances, required to meet the assumptions underlying the statistical model. To provide an overview of the distribution of the raw data, Table 10 presents the means and standard devia 152 tions of the relative frequencies (percents of total mes sages) for the different strategies in the cells used in the analysis. The results of the multivariate analysis are sum marized in Table 11. Only significant results of the uni variate analyses made within significant multivariate con trasts can be considered as truly significant univariate findings. Only the discussion-class main effect was found to be significant (p.<.009) for all six variables consid ered together. However, the difference between husbands and wives approached significance (p.<.08). The clinic- status factor clearly was not significant (p.<.16). The only interaction approaching significance was between the clinic-status and spouse factors (p.<.07). From these findings alone, it is clear that all the hypotheses invol ving differences between clinic and non-clinic subjects or husbands and wives were not supported. The univariate analyses performed within the larger analysis indicate how much the different variables contrib uted to the differences reflected in the over-all F ratios. Those strategies showing the most difference on the dis cussion-class variable were symmetrical (p.c.OOl), one-up (p.<.05), and transactional disqualification (p.c.09). A significantly higher rate of symmetrical messages was used in the disagreement than in the agreement discussions, TABLE 10 MEANS AND STANDARD DEVIATIONS OF COMMUNICATION STRATEGIES Agree Disagree T o t a 1 M F Total M F Total M F M+F One-Up % Clinic mean 33.0 37.7 35.4 20.8 25.5 23.2 26.9 31.6 29.8 s.d. 22.0 11.0 20.3 16.9 Non-Clinic mean 35.0 29.6 32.3 36.2 25.4 30.8 35.6 27.5 31.6 s .d. 13.9 10.5 11.8 7.8 Total mean 34.0 33.7 33.8 28.5 25.5 27.0 31.3 29.6 30.4 One-Down% Clinic mean 32.2 36.2 34.2 24.8 31.7 28.3 28.5 34.0 31.2 s.d. 19.0 16 .7 15.5 23.5 Non-Clinic mean 38.2 31.2 34.7 33.6 29.3 31.2 35.9 30.3 33.1 s.d. 13.1 15.0 10.3 9.8 Total mean 35.2 33.7 34.5 29.2 30.5 30.0 32.2 32.1 32.2 Symmetrical % Clinic mean 31.3 27.3 29.3 55.0 41.3 48.2 43.2 34.3 38.7 s.d. 16.0 17.4 18.5 19.9 Non-Clinic mean 23.5 36.2 29.9 26.5 45.9 36.2 25.0 41.1 33.0 s.d. 14.2 18.6 8.8 15.0 Total mean 27.4 31.8 29.6 40.8 43.6 42.2 34.1 37.7 35.9 153 TABLE 10 MEANS AND STANDARD DEVIATIONS OF COMMUNICATION STRATEGIES (Continued) Agree Disagree Total M F Total M F Total M F M+F Paradoxical % Clinic mean 6.9 1.9 4.4 4.1 5.6 4.9 5.5 3.8 4.6 s.d. 4.1 2.3 4.2 5.6 Non-Clinic mean 6.0 2.6 4.3 4.9 2.4 3.7 5.5 2.5 4.0 s .d. 5.6 3.6 3.8 1.5 Total mean 6.5 2.3 4.4 4.5 4.0 4.3 5.5 3.1 4.3 Transactional Disqualification% Clinic mean s.d. 3.2 2.6 0.8 1.4 2.0 1.4 2.0 1.0 2.5 1.2 2.3 0.9 1.6 Non-Clinic mean s .d. 3.1 4.1 0.6 1.2 1.9 1.3 1.9 0.4 0.9 0.9 2.2 0.5 1.4 Total mean 3.2 0.7 2.0 1.4 0.7 1.0 2.3 0.7 1.5 Metacommunicative % Clinic mean s.d. 0.6 1.4 1.6 2.5 1.1 2.3 2.3 2.0 2.0 2.2 1.5 1.8 1.6 Non-Clinic mean s.d. 0.8 1.2 1.2 1.4 1.0 1.4 1.6 1.4 1.6 1.2 1.1 1.3 1.1 Total mean 1.3 1.4 1.1 1.9 1.7 1.8 1.6 1.6 1.4 154 TABLE 10 MEANS AND STANDARD DEVIATIONS OF COMMUNICATION STRATEGIES _________________________(Continued)_________________________ Agree D i s agree Total M F Total M F Total M F M+F Metacomplementary % Clinic mean 0.7 0.6 0.7 0 1.1 0.5 0.4 0.8 0.6 s.d. Non-Clinic mean 1.5 1.8 0.5 0.6 0.6 0 2.4 0.8 0 0.4 0.7 0.3 0.5 s.d. Total mean 1.1 1.4 0.6 0.6 0.6 1.6 0 0.4 0.5 0.5 0.5 0.6 0.5 t-* Ul Ul 156 TABLE 11 SUMMARY TABLE OF MULTIVARIATE ANALYSIS OF COMMUNICATION STRATEGIES df P less than Factor A (Clinic Status) 6,67 Variable: One-Up One-Down Symmetrical Paradoxical Trans. Disq, Metacommuni- cative 1,72 1.616 0.464 0.272 2.434 1.045 0.536 0.257 0.16 0.50 0.60 0.12 0.31 0.47 0.61 Factor B (Spouse) Variable; One-Up One-Down Symmetrical Paradoxical Trans. Disq. Metacommuni- cative 6,67 1,72 0.246 0.001 0.959 6.659 9.518 0.431 0.62 0.98 0.33 0.01 0.003 0.51 Factor C (Discussion Class)6,67 Variable: One-Up 1,72 One-Down Symmetrical Paradoxical Trans. Disq. Metacommuni- cative 3.163 4.147 1.695 11.766 0.016 2.983 2.584 0.01 0.05 0.20 0.001 0.90 0.09 0.11 157 TABLE 11 SUMMARY TABLE OF MULTIVARIATE ANALYSIS OF COMMUNICATION STRATEGIES (Continued) df P less than Factor AB (Clinic x Spouse) Variable: One-Up One-Down Symmetrical Paradoxical Trans. Disq. Metacommuni- cative 6,67 1,72 2.026 3.595 2.396 11.581 0.414 0.068 0.034 0.07 0.06 0.13 0.001 0.52 0.80 0.85 Factor AC (Clinic x Discusssion)6,67 Variable: One-Up One-Down Symmetrical Paradoxical Trans. Disq. Metacommuni- cative 1,72 0.731 2.518 0.146 2.892 0.381, 0.014 0.610 0.627 0.12 0.70 0.09 0.54 0.91 0.44 Factor BC (Spouse x Discussion) 6,67 Variable: One-Up One-Down Symmetrical Paradoxical Trans. Disq. Metacommuni- cative 1,72 1.186 0.158 0.153 0.042 4.178 3.036 1.103 0.32 0.69 0.70 0.84 0.05 0.09 0.30 158 TABLE 11 SUMMARY TABLE OF MULTIVARIATE ANALYSIS OF COMMUNICATION STRATEGIES (Continued) df F p less than Factor ABC (Clinic x Spouse x Discuss- ion) 6,67 1.295 0.27 One-Up 1,72 0.162 0.69 One-Down 0.000 0.99 Symmetrical 1.260 0.27 Paradoxical 2.537 0.12 Trans. Disq. Metacommuni- 0.026 0.87 cative 0.283 0.60 159 while the opposite relationship held for one-up messages. Although not a significant difference, there were more transactional disqualifications in the discussions of areas of agreement than of disagreement. The significant difference found for symmetrical responses between periods of agreement and disagreement was predicted. This is the only hypothesis (number seven) proposed in this research which was clearly supported by the findings. The other multivariate F ratios with low probabili ties can be interpreted as providing direction for further study. The main effect for the spouse factor and the interaction of clinic status with spouse both approached significance. The variables showing the greatest differen ces between husbands and wives were paradoxical (p.<.01) and transactional disqualification (p.<.003). They were used more frequently by husbands than by wives. None of the other strategies approached the five percent level of probability on the spouse factor. Considering the clinic- status and spouse interaction, an interaction of low proba bility (p.c.OOl) was found for the proportion of symmetri cal messages used. Clinic husbands used symmetrical mes sages more frequently than non-clinic husbands, and the opposite relationship emerged for wives. The interaction between clinic status and spouse on the percentage of one- 160 up messages approached the five percent level of probabil ity (p.<.06), and the directions of the differences between groups were exactly the opposite of those characterizing symmetrical messages. None of these findings were predic ted, though spouse differences and interactions involving them were simply to be explored. Less confidence may be placed on the findings of univariate F ratios with low probabilities which appeared in multivariate factors which themselves had higher proba bilities of occurring by chance. The multivariate F ratio for the interaction of spouse with discussion class (p.<.32) showed strong dif ferences on two communication strategies, (p.< .045) and transactional disqualification (p.<.09). Higher propor tions of each were used by husbands in the agreement con dition than in the disagreement condition, while wives used paradoxical messages more frequently in the disagree ment than in the agreement condition and did not differ be tween discussions in their usage of transactional disquali fications. Only the difference found for wives' use of paradoxical responses was predicted (hypothesis one), and this difference was predicted for all subjects. Several interactions between clinic status and dis cussion class were predicted. However, the multivariate F ratio for this factor indicates no over-all differences 161 (p.<.6). With the caution due this finding, it is noted that symmetrical messages were used at a higher rate in the disagreement condition than in the agreement condition and that this difference was greater for clinic than non-clinic subjects (p.<.09). The direction of this interaction is opposite that predicted in hypothesis nine. Since the dependent variables in the hypotheses were specified as relative frequencies of messages falling in the particular categories used, only the results of the multivariate analysis could be used in deciding whether or not the null hypotheses suggesting no differences were re jected. Only one of the eleven experimental hypotheses proposed was clearly supported by these data. This was the first of the univariate analyses examined, number seven, which predicted higher proportions of symmetrical responses in the disagreement condition than in the agreement condi- i | tion. | Sequential Analyses In order to further explore the data with respect to the rationales presented for the hypotheses, several sequential analyses were used. In the rationales presented for the hypotheses relating to symmetrical and one-down responses, numbers seven to nine and eleven to thirteen respectively, it was noted that the predictions were 162 founded on the belief that subjects would respond dif ferently in the conditions of agreement to messages which involved active definition, or attempts to control defini tion, of the relationship. Such stimulus messages include one-up and symmetrical strategies. The pattern of responses postdicted here involved differences between the rates of using symmetrical and one-down strategies as responses to one-up and symmetrical stimuli. To investigate these differences, and any re lated interactions of the independent variables of this study, modified sequential analyses were used. The proportions of the one-down and symmetrical messages among all responses occurring immediately after one-up or after symmetrical messages sent by a subject's mate were compared across discussion classes, between spouses, and between clinic and non-clinic groups. Cell means for these data are presented in Table 12. Two analy ses of variance, one for each stimulus class, were per formed on arc-sine transformations of the proportions. This approach is considered to be a modified sequential analysis because messages which did not immediately follow or precede messages from the spouse were excluded. The hypotheses used in this investigation parallel exactly the predictive hypotheses concerning symmetrical and one-down responses advanced in the major part of this TABLE 12 CELL MEANS FOR ONE-DOWN AND SYMMETRICAL RESPONSES TO ONE-UP AND SYMMETRICAL STIMULI Percent One-Down Responses To One-Up Stimuli (UP) Agree Disagree Total M F Total M F Total M F M+F Clinic 47.6 59.3 53.5 44.1 62.0 53.0 45.9 60.7 53.3 Non-Clinic 62.0 55.3 58.6 46.5 45.9 46.2 54.3 50.6 52.4 Total 54.8 57.3 56.1 45.3 53.9 49.6 50.1 55.6 52.8 Percent Symmetrical Responses To One-Up Stimuli (US) Clinic 24.9 28.6 26.7 35.2 23.2 29.2 30.0 25.9 28.0 Non-Clinic 21.1 33.2 27.2 29.3 43.3 36.3 25.2 38.3 31.7 Total 23.0 30.9 27.0 32.3 33.3 32.8 27.6 32.1 20.0 163 TABLE 12 CELL MEANS FOR ONE-DOWN AND SYMMETRICAL RESPONSES TO ONE-UP AND SYMMETRICAL STIMULI (Continued) Percent One-Down Responses To Symmetrical Stimuli (SD) A g r e e D i s a g r e e T o t a 1 M F Total M F Total M F M+F Clinic 25.5 33.5 29.5 24.2 33.7 28.9 24.9 33.6 29.2 Non-Clinic 37.7 35.4 36.5 43.8 29.0 36.4 40.7 32.2 36.4 Total 31.6 34.4 33.0 34.0 31.3 32.7 32.8 32.9 32.8 Percent Symmetrical Responses To Symmetrical Stimuli (SS) Clinic 43.4 40.8 42.1 58.2 46.4 52.3 50.8 43.6 47.2 Non-Clinic 39.0 56.6 47.8 32.8 55.9 44.4 35.9 56.3 46.1 Total 41.2 48.7 45.0 45.5 51.2 48.3 43.3 50.0 46.7 165 study. Thus, the proportions of symmetrical responses to one-up and to symmetrical stimuli were expected to be greater in the disagreement than in the agreement dis cussions and to be greater among non-clinic than clinic subjects. Further, the proportion of symmetrical responses relative to one-down responses was expected to be greatest in the disagreement condition for the non-clinic subjects. The spouse variable was included for exploration, as in dicated by the original hypotheses. The results of the analysis of variance for one- down versus symmetrical responses to one-up stimuli are summarized in Table 13. Only the interactions containing U (response class) were of interest in this analysis. Only the interaction of clinic status, discussion and response class (ACU) was found to be significant (P.<.05). The contrasts of interest within this interaction were evaluated using Scheffe's test for significant con trasts, which was performed on non-transformed data. The results are presented in Table 14. Results of contrasts involving US combinations will be explored first. The proportion of symmetrical responses to one-up stimuli was higher in the disagreement condition than in the agreement condition among both clinic and non-clinic couples. The proportion of US in the agreement condition did not differ between clinic and non-clinic couples, but was greater for 166 TABLE 13 SUMMARY TABLE FOR THE ANALYSIS OF VARIANCE COMPARING ONE-DOWN AND SYMMETRICAL RESPONSES FOLLOWING ONE-UP STIMULI Source d.f. SS MS F E A (clinic status) 1 0.521 2.864 n.s. B (spouse) 1 0.579 5.220 <.05 C (discussion) 1 0.057 0.455 n.s. U (response- 1 9.293 12.609 <.01 class E (couple)/A 18 3.274 0.182 AB 1 0.163 1.472 n.s. AC 1 0.015 0.117 n.s. BC 1 0.228 1.466 n.s. AU 1 0.392 0.532 n.s. BU 1 0.300 0.249 n.s. CU 1 0.858 3.512 < .10(ri.£ BE/A 18 1.996 0.112 CE/A 18 2.274 0.126 UE/A 18 13.266 0.737 ABC 1 0.026 0.167 n.s. ABU 1 1.546 1.286 n.s. ACU 1 1.633 6.678 <.05 BCU 1 0.653 3.516 < . 10 (n. s, BCE/A 18 2.814 0.156 BUE/A 18 21.627 1.202 CUE/A 18 4.400 0.244 ABCU 1 0.011 0.060 n.s. BCUE/A 18 3.345 0.186 167 TABLE 14 RESULTS OF SCHEFFE'S TEST FOR SIGNIFICANT CONTRASTS IN THE CLINIC STATUS X DISCUSSION X RESPONSE CLASS INTERACTION, UD VS. US1 Cells Difference Conclusion2 CAUD-CDUD +0.4 n. s. CAUS-CDUS -2.7 Significant NAUD-NDUD +12.4 Significant NAUS-NDUS -9.1 Significant CAUD-NAUD -5.2 Significant CDUD-NDUD +6.8 Significant CAUS-NAUS -0.4 n. s. CDUS-NDUS -7.1 Significant [CDUD-CDUS]-[NDUD-NDUS] +13.9 Significant |[CAUS-CDUS]|-|[NAUS-NDUS]|| -6.4 Significant |[CAUD-CDUD]|-|[NAUD-NDUD] 1-12.0 Significant 1Abbreviations used in this table signify: A = Agree, D = Disagree, C = Clinic, N = Non-Clinic, UD = One-down response following one-up stimulus, US = symmetri cal response following one-up stimulus. 2The 95% confidence interval was used in assessing these contrasts. 168 non-clinic couples in the disagreement condition than for clinic couples. Finally, the proportion of symmetrical re sponses to one-up stimuli was greatest, absolutely and in comparison with one-down responses, for the non-clinic couples in the disagreement condition. These results, ex cept for the finding of no difference between groups in the proportion of US in the agreement condition, were pre dicted. Many of the contrasts involving proportions of UD also were found to be reliable. The proportion of one- down responses to one-up stimuli was greater in the agree ment than in the disagreement condition only for non clinic subjects. The proportion of UD was greater for clinic couples than for non-clinic couples in the disagree ment condition. These results, except for the finding of no difference in the proportions of UD between discussions for the clinic group, are consistent with the post-hoc hy potheses set forth earlier. Table 13 reflects two more findings of interest in exploring the hypotheses. The discussion by response class interaction approached significance (p.c.10). UD combina tions had higher relative frequencies in the agreement than in the disagreement discussions, and the opposite relation ship held for US pairs. The interaction of spouse, dis cussion, and response class also approached significance 169 (p.<.10). Wives tended to use higher rates of one-down responses in both conditions than did husbands; the dif ference was greatest in the disagreement condition. The difference in proportions of one-down and symmetrical re sponses was greater for males than females in the agree ment condition and greater for females in the disagree ment condition. The results for the analysis of variance for one- down and symmetrical responses to symmetrical stimuli are summarized in Table 15. No significant results related to the present hypotheses emerged. The results of the sequential analyses indicate some support for the post-hoc reformulations of the ori ginal hypotheses relating to one-down and symmetrical re sponses only when preceded by one-up stimuli. To investigate the differences between one-up and symmetrical stimuli preceding these responses, Pearson product moment correlations between the two classes of re sponse to each stimulus were calculated. Again, arc-sine transformations of the proportional data were used. The correlation of UD with US was -0.64, which accounted for 41 percent of the variance; and the correlation of SD with SS was -0.55, accounting for only 30 percent of the variance. Thus, the relationship between the two response classes was found to be stronger when one-up messages were the 170 TABLE 15 SUMMARY TABLE FOR THE ANALYSIS OF VARIANCE COMPARING ONE-DOWN AND SYMMETRICAL RESPONSES FOLLOWING SYMMETRICAL STIMULI Source d.f. SS MS A (clinic status) 1 0.664 2.144 n. s. B (spouse) 1 0.404 2.015 n. s. C (discussion) 1 0.215 1.413 n. s. S (response-class) 1 4.060 5.077 .05 E (couple)/A 18 5.571 0.310 AB 1 0.029 0.144 n.s. AC 1 0.409 2.686 n. s. BC 1 0.176 1.307 n.s. AS 1 0.319 0.399 n.s. BS 1 0.098 0.085 n.s. CS 1 0.007 0.014 n.s. BE/A 18 3.613 0.201 CE/A 18 2.741 0.152 SE/A 18 14.392 0.800 ABC 1 0.00001 0.0001 n.s. ABS 1 2.569 2.226 n.s. ACS 1 0.096 0.183 n.s. BCS 1 0.043 0.111 n.s. BCE/A 18 2.424 0.135 BSE/A 18 20.774 1.151 CSE/A 18 9.460 0.526 ABCS 1 0.332 0.857 n.s. BCSE/A 18 6.965 0.387 171 stimuli than when symmetrical messages were. This conclu ded the sequential analysis. CHAPTER VI DISCUSSION This chapter is divided into five major sections. The first deals with an evaluation of the method used to code the discussions. The method of subject selection is discussed in section two, and the method of generating dis cussions is evaluated in section three. Section four is devoted to a discussion of the results of the study and their implications for research and theory using these variables. The final section deals with this study in re lation to the field of marital and family interaction re search. Codification The approach to codification used here was designed to be as true as possible to the thrust of the theoretical definitions set forth by the theorists from the Mental Re search Institute (MRI). Accordingly, the emphasis was placed on the semantic task attempted with the relationship component of each message, and each message was coded as a response to the preceding speech of the speaker's partner. With the emphasis on the relationship component of communi- 172 173 cation, very high interrater agreement (91 percent) was ob tained. This finding suggests that it is possible for trained individuals to discern the different communication strategies without a great deal of difficulty. It was ob served that attending to voice tone was particularly impor tant in coding and that the use of messages instead of speeches as the units to be coded simplified the procedure. It seems reasonable that research attempting to tap any relationship messages could benefit by using such a codifi cation unit because relationship messages frequently change within speeches. In the present study, at least, agreement on unitization was high (95 percent), and it was felt that the gains in interrater agreement afforded by this procedure were well worth the loss in mechanical objectivity which would have been afforded by using the speech as the coding unit. A few refinements were suggested by experience with this system. The first concerns paradoxical messages, which include transactional disqualifications. Some para doxical messages seemed to be recognized by the speaker as such; others did not. The former type were often humorous and/or sarcastic. The latter seemed to occur without the speaker's being aware of the incongruity in what he or she was doing. It seems quite possible that the two types of paradoxical messages might have different effects on the 174 feelings inspired in the recipient and on the communication. Although inferences about another person's awareness may be difficult to validate, such apparent differences seemed readily discernable in listening to the tapes. Such dif ferences could be built into a coding system like the one used here. The second difficulty noted with the present system relates to one-down responses. Most often, comments such as "uh-huh" and "yeah," which seemed to be uttered as ha bitual responses at high rates by some subjects and low rates by others, were classified as one-down messages. The meaning conveyed often seemed to be "I understand what you are saying," "I want you to know I am listening," or "I agree." The rates of one-down responses seemed to be in flated by high numbers of these responses. Messages of this kind were used frequently by some subjects during the speech of their partners. Such unobtrusive interruptions, then, also inflated the number of messages attributed to the partner since each message was restricted to no more than one speech in length. The use of a separate category for such statements, joined with the introduction of the possibility of coding messages greater than one speech in length, might increase the accuracy with which coding se quences would reflect the course of the conversation. Finally, the use of more extensive sequential or 175 time-sequential analyses of communication strategies (or any other process variable) could prove useful in under standing the nature of communication. Sackett1 is current ly refining appropriate statistical procedures for between- and within- subject comparisons of this nature. The introduction of these changes into the codifi cation and analytic procedures would be likely to increase the accuracy with which the data reflect the conversations. Subject Selection Procedures It is considered possible that the couples making up the clinic and non-clinic groups in this study were not samples of two populations differing on the dimension of marital disturbance. No measures of such disturbance were used because the behavioral difference of seeking help for marital difficulty versus not seeking such help originally was thought to be sufficient. However, the therapists re ferring couples for the study indicated that the most dis turbed couples they were seeking generally were those not willing to participate. Apparently, the prospect of dis- 1For preliminary work, see Sackett, G. A nonpara- metric lag sequential analysis for studying dependency among responses in observational scoring systems. Mimeo graphed paper. Regional Primate Research Center, Univer sity of Washington. 176 cussing their relationship in the context of research was aversive for the more seriously disturbed couples. In lis tening to the tape recordings of the conversations, the clinical impression conveyed by most, but not all, of the clinic couples was one of relative disturbance. The clini cal impression formed of the non-clinic couples was pre dominately, but not entirely, "normal." Thus, the extent to which the two groups of subjects differed on the amount of marital disturbance may not have been great enough to justify the assumption that they came from populations dif fering enough on this dimension to justify using such small samples. Some other approach to sample selection, and the use of larger samples, would seem advisable. Further, bet ter control of demographic differences between samples to be contrasted would seem advisable. Procedures for Generating Interaction The present research was designed so that couples would be free to move as they wished in discussing topics of agreement and disagreement. The topics were not stan dardized across couples because the perceived amount of agreement or disagreement was to be maximized. However, the freedom of the subjects to do as they wished in re sponse to the instruction to discuss a particular area of 177 their relationships, seemed to have introduced a great deal of variance. Whereas all of the conversations did seem to relate to the topics presented, some of the discussions were much more specific in focus than others, and a few of the conversations on general topics of agreement focussed on specific issues of disagreement within the larger topic. The areas of disagreement generally seemed easier for the subjects to discuss, with the conversations moving more rapidly than in the agreement discussions. A similar find ing was reported by Gill (1971). Generally, it seems that larger sample sizes or more focussed interaction would be necessary to reduce the effects or size of the very large within-cell variance found in this study. Further, less self-conscious interaction might result with less overtly- relevant discussion material. Results Numbers of Messages The number of messages used was found to be greater for non-clinic than for clinic couples. This could reflect faster-paced conversations, more interruptions, and/or more statements such as "yeah" in the non-clinic than in the clinic groups. At any rate, this difference confirmed the necessity of using proportional data in the analyses. 178 Experimental Results Relative Frequencies Multivariate analysis was used in examing the ex tent to which the distribution of messages between: clinic and non-clinic subjects, husbands and wives, and agreement and disagreement over topics considered important to mari tal satisfaction. Of the three factors and their interac tions considered, only the multivariate F ratio for agree ment versus disagreement was significant (p.<.009). This finding was highly reliable and indicated that the distri bution of the communication strategies was different for the two classes of discussion examined. This result was ob tained inspite of the open-ended nature of the topics used for discussion. This finding in itself has some implica tions for research and theory. Jackson, Riskin, and Satir (1961) have argued that the patterning of symmetrical and complementary responses could yield a rough index of family communication styles, which would be "correlated with spe cific manifestations of mental illness." (p. 323). The finding of significant differences in the distribution of communication strategies undier conditions of agreement and disagreement underscores the necessity, at least, of taking the meaning of discussion topics and context into account before establishing a schema of family types built on these 179 strategies. The ways in which the communication strategies var ied between discussion classes was revealed by the univari ate analyses for the different strategies within this fac tor. One-up messages were found to have higher relative frequencies in the agreement than in the disagreement con dition, while the opposite relationship held for symmetri cal messages. This means that, when discussing areas of disagreement in the marital relationship, couples tended to vie for control of the communication more than when dis cussing areas of agreement. In agreement discussions, there was a higher proportion of one-up statements, which overtly direct and control the definition of the relation ship. Since there was some (non-significant) tendency for one-down responses also to be more frequent in the agree ment than in the disagreement discussions, it can be said that the agreement discussions tended toward complementarity while the disagreement discussions tended to be more sym metrical in nature. These were the only significant find ings in the multivariate analysis. A few other findings had low probabilities of oc curring by chance. Most important are the multivariate F ratios for the spouse factor (p.<.08) and the clinic status by spouse interaction (p.<.074). These findings suggest fairly strong trends in the data indicating the likelihood 180 that a design producing reducfed error variance or having a larger number of subjects could have detected reliable dif ferences in the over-all distributions of the communication strategies along the relevant dimensions. The strongest univariate F ratios for spouse differ ences indicate that husbands used higher proportions of paradoxical messages and transactional disqualifications than did wives. Possible explanations of this finding would be based on the general notion that men are less ex pressive than women in this society, and therefore could be prone to express themselves in less congruent way than their wives, and/or on the influence of experimental con text. The experimenter in this study was a woman, and it could be that the husbands experienced themselves as being more on the defensive than did the wives. This could have resulted in the use of higher rates of paradoxical re sponses and transactional disqualifications than feeling that one had an ally in the experimental situation. The interpretation of this finding, thus, should be postponed and could be used as grist for further research. Within the spouse by clinic status interaction, the strongest univariate F ratios were found for the propor tions of one-up and symmetrical messages. Clinic husbands tended to use more symmetrical messages than their wives and more symmetrical messages than their non-clinic counter1 - f 181 parts. The same relationship held for non-clinic wives with their husbands and their clinic counterparts. Thus, clinic husbands and non-clinic wives tended to use the most symmetrical responses. This finding tends to be compatible with the notion that elements of role reversal characterize disturbed, as compared with normal, couples, and that such a tendency reaches into the styles they use in communica tion within marriage. Further research would be necessary to investigate this hypothesis. The findings for one-up responses were essentially the exact opposite of the find ings for symmetrical responses. The above results were based on the frequencies of particular categories of messages as proportion of all mes sages used by the subjects. Only these data could be used in evaluating the hypotheses of this study as they were originally presented. The only a priori prediction gaining support was that the proportion of symmetrical messages would be greater in the disagreement discussions than in the discussions of areas of agreement. Sequential Analyses The sequential analysis relating the predictions derived from the rationales for the original hypotheses to the rates of one-down and symmetrical responses to one-up 182 and to symmetrical stimuli, yielded relevant results. The discussion by response class interaction was in the pre dicted direction, and yielded an F ratio approaching sig nificance (p.<.10). US pairs had higher relative frequen cies in the disagreement condition than in the agreement condition, and the opposite relationship held for UD com binations generally. The interaction of clinic status and response class was not significant, though it had been pre dicted. However, the interaction of clinic status, discus sion, and response class yielded a significant F ratio (p.<.05), and most of the differences predicted were found to be reliable within this interaction. The proportion of symmetrical responses was greatest, absolutely and in rela tion to the proportion of one-down responses, for the non clinic subjects in the disagreement condition. The non clinic subjects also used higher proportions of one-down responses to one-up stimuli in the agreement condition than did clinic subjects. This result indicates that: (1) in the agreement discussions, non-clinic subjects accepted their partners1 control of the communication more than clinic subjects; and (2) that they struggled for control more than clinic subjects in the disagreement condition. This finding suggests more response flexibility on the parts of non-clinic than of clinic couples. In contrast to these findings, no reliable relation 183 ships emerged between the two response classes and the ma jor factors of this study when the responses to symmetrical messages were examined. Apparently, those factors have lit tle influence when the stimulus has the quality of competi tion for control of the communication. Subjects in all cells tended to use rather similar distributions of one- down and symmetrical responses. It could be said, then, that symmetrical stimuli exert more powerful influences than the contextual or individual differences on the dis tribution of these responses, while there is more of a ten dency for the factors to have an influence over the distri bution of the responses considered here, when preceded by one-up stimuli. Investigation of the relative strengths of the dif ferent strategies of communication as stimuli influencing the kinds of responses which follow them could be under taken with more extensive use of sequential analysis. How ever, for such .research to be adequate, one would have to deal with the problem of controlling the intensities of the relationship components of communication. In sum, most of the hypotheses of this research were not supported by the results, though there was a ten dency for the relative frequencies of complementary and symmetrical responses to vary in the predicted way between discussion classes. The finding of no differences between 184 clinic and non-clinic subjects in the rates of using para doxical responses and transactional disqualifications raises doubt about the Palo Alto theorists' conviction that they should be related to the amount of disturbance in primary relationships. As suggested earlier, it may be that there are relatively noxious and non-noxious sub classes of these strategies, and that they may be related to such disturbance. However, the theorists have not de veloped this idea. One final and general observation seems in order. It must be noted that the discussion-class factor, the only one on which repeated measurements were possible, was the only factor exerting significant influence over all the communication strategies considered as a group. Larger numbers of subjects, or more structured interaction sam ples, would be advisable for further research if between- subjects variables are to be utilized. Marital and Family Interaction Research Up to this point, the field of marital and family interaction theory has been greatly influenced by the work of Jackson and his colleagues. The communication strate gies with which they deal have had considerable impact on the development of theory relating particularly to the 185 practice of family psychotherapy. However, the major com munication strategies have been almost totally excluded from the general field of interaction research. This study takes a step toward including them in this realm. In sev eral ways, it raises more questions than it answers, but it does contain some interesting findings which further re search could explore. Some of the more interesting issues raised include recognition that strategies of communication are reliably influenced by the difference between discussing topics of agreement and disagreement deemed important to marital sat isfaction. Thus, the choice of interaction task is likely to influence these strategies; such tasks must be selected with extreme care. Open-ended discussions have advantages in stimulating rather free conversation, but introduce large amounts of error variance. Role-play activities might provide a more adequate compromise between freedom and structure in conducting related research in the future. Obviously, a great deal of work relating such things as theories of marital power or satisfaction to the communication strategies would help relate them to those theoretical bases. The communication strategies could also bring light to those fields. CHAPTER VII SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS Summary Chapter I serves as an introduction to this study, which was designed as a response to a problem frequently encountered in the study of marital and family interaction. The field abounds in theoretical speculation which is poor ly related, if at all, to empirical investigation. One particularly influential way to approach the understanding of families has been developed by theorists from the Mental Research Institute in Palo Alto. They have stressed the importance of the process of defining rela tionships through communication and the interactional na ture of causation within family systems. The purpose of this study was to render important concepts from the Palo Alto theorists amenable to adequate empirical investigation and to use them in a study containing variables relevant to the theory. The uses of the strategies of communication identified by the M.R.I. group were related to the dif ferences between subjects in, and not in, marital therapy, and between husbands and wives as they discussed areas of agreement and disagreement about aspects of their relation- 186 187 ship considered important to marital satisfaction. Several hypotheses relating the communication strategies to clinic status and type of discussion were tested. Chapter II surveys the methods used in family ^in teraction research and includes a series of considerations important in designing or evaluating such research. Some of the major difficulties within the field were highlighted in this process. Special attention was given the problems of adequately relating operational and theoretical defini tions and of choosing methods of generating data which are appropriate to the questions being asked. This discussion provided a broad framework within which to view this study. The third chapter focusses on the theoretical and empirical background specifically dealing with the strate gies of communication, and develops into the rationale and hypotheses for this study. The earlier empirical work done using the communication strategies was found to be inade quate for this study either because of using codification systems which mutilated the theoretical definitions of the different strategies (Mark, 1971 and Ericson and Rogers, 1973) or because they were not amenable to use with married couples interacting (Gill, 1971). Gill's approach was mod ified extensively for use in this study. The specific design of this study is presented in Chapter IV. Generally, the rates of using different commu- 188 nication strategies were compared between subjects on the variables of spouse and clinic status and within subjects on the variable of discussion class. Ten couples in mari tal therapy (clinic subjects) and ten not in such therapy (non-clinic subjects) each participated in ten-minute dis cussions of one topic of agreement and one topic of dis agreement about areas of their relationship considered im portant to marital satisfaction. The strategies of communication used as dependent variables in the analysis of the conversations were: one- up, one-down, symmetrical, metacommunicative, paradoxical, and transactional disqualification. The unit of conversa tion coded was the message, a segment of interaction serv ing a single semantic purpose in defining the relationship. The discussions were coded using transcripts in combina tion with audio-recordings. The results were presented in Chapter V. Multi variate analysis was used to compare the distribution of the relative frequencies of the communication strategies across the factors of clinic status, spouse, and discussion class. Only the factor of discussion class was found to have a significant effect on the distribution of all the strategies considered together. Within this factor, uni variate analyses indicated that symmetrical messages had higher relative frequencies in the disagreement than in the 189 agreement condition, while one-up messages occurred at higher rates in the agreement condition. When the relative frequencies of one-down and sym metrical responses to one-up stimulus messages were com pared across the factors of interest, the interaction of clinic status, discussion class, and type of response was significant. Non-clinic couples tended to vie for control more than clinic couples in the disagreement condition. In the agreement condition, they accepted the control of their partner more than did the clinic couples. The proportion of symmetrical responses to one-up stimuli, absolutely and in comparison with one-down re sponses, was greatest for the non-clinic couples in the disagreement condition. A parallel analysis of the propor tions of these same types of response following symmetrical stimuli yielded no significant results. The correlation between the two types of response to one-up stimuli ac counted for 41 percent of the variance, while the correla tion for the two response classes following symmetrical stimuli accounted for 30 percent of the variance. These findings were discussed in Chapter VI. The significance of the findings for one-up and symmetrical messages indicated that there is a tendency for discussions of important topics of agreement to be more complementary in nature than discussions of areas of disagreement and 190 that the disagreement discussions were most symmetrical in nature. Thus, both clinic and non-clinic subjects tended to struggle for control of their relationship more when in disagreement and to allow control to be less controversial when in agreement with each other. That clinic status has an influence on the extent to which one accepts or rejects a mate's assumption of control was demonstrated by the analysis of variance for symmetrical versus one-down re sponses to one-up stimulus messages. Non-clinic subjects shifted between symmetrical and one-down responses to one- up stimuli between conditions more than did clinic sub jects, demonstrating greater apparent potential to vary their response styles with the context. This suggests greater flexibility in adopting different communication strategies on the parts of non-clinic than of clinic sub jects . Implications for further research dealing with the strategies of communication and relating them to other topics of importance in family interaction research were mentioned throughout this section. Conclusions Several conclusions can be drawn from this study. The most important is that the strategies of communication 191 can be coded reliably and that, therefore, research using them is feasible. Further, no support for the Palo Alto theorists' belief that paradoxical responses should be more frequent in disturbed than non-disturbed relationships was found. However, the belief that communication between equals under relatively stressful conditions would tend to be more symmetrical and less complementary than otherwise was supported. Further, there was support for the belief that members of less disturbed relationships would evidence greater ability than members of disturbed relationships to shift, with changes in conditions, between acceptance and rejection of their partner's control of the relationship. Thus, there was support for part of the theoretical base underlying this research, but no support for that part relating level of disturbance to rates of paradoxical strategies. Implications for Further Research The implications for further research mentioned in the discussion section fall into three major categories. The first relates to methodology. Suggestions were made for reducing error variance, refining the system of codifi cation, and using sequential analysis to reflect better any differences in the distribution of messages between factors 192 of interest. Second, it was noted that several differences between husbands and wives approached significance. Fur ther research investigating these differences could be fruitful. Such research could be related to such variables as experimenter gender and the extent of role reversal. Finally, research relating the strategies of communication to such variables as levels of family disturbance, marital power, and marital satisfaction were suggested. ' • 1 !l Ji X ? \ { I B I B L I O G R A P H Y { I 193 BIBLIOGRAPHY Aldous, Anandam Bales, ] Bales, : Bateson Bateson Bateson Bauman, Baxter, Baxter, J. Strategies for developing family theory. Journal of Marriage and the Family, 1970, 32, 250- K., and Highberger, R. Child compliance and con- gruity between verbal and nonverbal maternal com munication— A methodological note. Family Process, 1972, 11, 219-226. i. F. Interaction process analysis. Cambridge, Mass.: Addison-Wesley, 1950. (a) l . F. A set of categories for the analysis of small group interaction. American Sociological Review, 1950, 15, 257-263. 7b) G. Naven, second edition. Stanford: Stanford University Press, 1958. G., and Jackson, D. D. Some varieties of patho genic organization. In D. Rioch (Ed.), Disorders of communication. Research publication of the As- sociation for Research in Nervous and Mental Dis ease, 42, 1964. G., Jackson, D. D., Haley, J., and Weakland, J. Toward a theory of schizophrenia. Behavioral Science, 1956, 1, 251-264. G. and Roman, M. Interaction testing in the study of marital dominance. Family Process, 1966, 5, 230-242. J. C., and Arthur, S. C. Conflict in families of schizophrenics as a function of premorbid adjust ment and social class. Family Process, 1964, 3, 273-279. J. C., Arthur, S. C., Flood, C. G., and Hedgepeth, B. Conflict patterns in the families of schizo phrenics. Journal of Nervous and Mental Disease, 1962, 135, 419-424. 194 195 Beakel/ N. G. , and Mehrabian, A. Inconsistent communica tion and psychopathology. Journal of Abnormal Psychology, 1969, 74, 126-130. Beavers, W. R., Blumberg, S., Timken, K. R., and Weiner. Communication patterns of mothers of schizophren ics. Family Process, 1965, 4, 95-104. Behrens, M. L., and Goldfarb, W. A study of patterns of interaction of families of schizophrenic children in residential treatment. American Journal of Orthopsychiatry, 1958, 28, 300-312. Beier, E. G. The silent language of psychotherapy. Chicago: Aldine, 1966. Bell, N.W. Extended family relations of disturbed and well families. Family Process, 1962, 1, 175-193. Bing, E. The conjoint family drawing. Family Process, 1970, 9, 173-194. Borgatta, E. F. A note on the consistency of subject be havior in interaction process analysis. Sociometry 1964, 27, 222-229. Borke, H. The communication of intent: A systematic ap proach to the observation of family interaction. Human Relations, 1967, 20, 13-28. Boszormenyi-Nagy, I., and Framo, J. L. (Eds.), Intensive family therapy. New York: Harper and Row, 1965. Bowen, M. Family relations in schizophrenia. In A. Auerback (Ed.), Schizophrenia: An integrated ap proach. New York: Ronald Press, 1959. Bowen, M. A family concept of schizophrenia. In D. D. Jackson (Ed.), The etiology of schizophrenia, New York: Basic Books, 1960. Broderick, C. Verbal communication, 1973. Brodey, W. M. Some family operations and schizophrenia: A study of five hospitalized families, each with a schizophrenic. Archives of General Psychiatry, 1959, 1, 379-402. 196 Caputo, D. V. The parents of the schizophrenic. Family Process, 1963, 2, 339-356. Cheek, F. The schizophrenogenic mother in word and deed. Family Process, 1964, 3, 155-177. (a) Cheek, F. A serendipitous finding: Sex roles and schizo phrenia. Journal of Abnormal and Social Psychology, 1964, 69, 392-400. (b) Cheek, F. Family interaction patterns and convalescent adjustment of schizophrenics. Archives of General Psychiatry, 1965, 13, 138-147. la) Cheek, F. The father of the schizophrenic. Archives of General Psychiatry, 1965, 13, 336-3457 (bT Cleveland, E. J., and Longaker, W. D. Neurotic patterns in the family. In D. Leighton (Ed.), Explorations in social psychiatry. New York: Basic Books, 1957. Drechsler, R. J., and Shapiro, M. I. Two methods of analysis of family diagnostic data. Family Process 1964, 2, 367-379. Ericson, P. M. Relational communication: complementarity and symmetry and their relation to dominance-sub- mission. Unpublished Ph.D. dissertation, Michigan State University, 1972. Ericson, P. M., and Rogers, L. E. New procedures analyzing relational communication. Family Process, 1973, 12, 245-267. Farina, A. Patterns of role dominance and conflict in parents of schizophrenic patients. Journal of Abnormal and Social Psychology, 1960, 61, 31-38. Farina, A., and Dunham, R. M. Measurement of family rela tionships and their effects. Archives of General Psychiatry, 1963, 9, 64-73. Farina, A., and Holzberg, J. Interaction patterns of par ents and hospitalized sons diagnosed as schizo phrenic or nonschizophrenic. Journal of Abnormal Psychology, 1968, 73, 114-118. 197 Ferreira, A. J. Decision-making in normal and pathologic families. Archives of General Psychiatry, 1963, 8, 68-73. Ferreira, A. J., and Winter, W. D. Family interaction and decision-making. Archives of General Psychiatry, 1965, 13, 214-223. Ferreira, A. J., and Winter, W. D. Stability of interac tional variables in family decision-making. Ar chives of General Psychiatry, 1966, 14, 352-355. Ferreira, A. J., and Winter, W. D. Decision-making in normal and abnormal two-child families. Family Process, 1968, 7, 17-36. (a) Ferreira, A. J., and Winter, W. D. Information exchange and silence in normal and abnormal families. Fam ily Process, 1968, 7, 251-276. (b) Ferreira, A. J., Winter, W. E., and Poindexter, E. Some interactional variables in normal and abnormal families. Family Process, 1966, 5, 60-75. Fisher, S., Boyd, I., Walker, D., and Sheer, D. Parents of schizophrenics, neurotics, and normals. Archives of General Psychiatry, 1959, 1, 149-166. Fisher, S., and Mendell, D. The spread of psychotherapeu tic effects from the patient to his family group. Psychiatry, 1958, 21, 133-140. Fontana, A. F. Familial etiology of schizophrenia: Is a scientific methodology possible? Psychological Bulletin, 1966, 66, 214-227. Foudraine, J. Schizophrenia and the family, a survey of the literature, 1956-1960, on the etiology of schizophrenia. Acta Psychotherapeutica, 1961, 9, 82-110. Framo, J. L. Systematic research on family dynamics. In I. Boszormenyi-Nagy and J. L. Framo (Eds.), Inten sive family therapy. New York: Harper and How, 1965. Friedman, C. J., and Friedman, A. S. Characteristics of schizogenic families during a joint story-telling task. Family Process, 1970, 9, 333-353. 198 Gill, J. D. Semantic communication processes. Unpublished Ph.D. dissertation, University of Utah, 1971. Goldstein, M. J., Gould, E., Alkire, A., Rodnick, E. H., and Judd, L. L. Interpersonal themes in the The matic Apperception Test stories of families of disturbed adolescents. Journal of Nervous and Men tal Disease, 1970, 150, 354-365. Goldstein, M. J., Judd, L. L., Rodnick, E. H., Alkire, A. A* and Gould, E. A method for studying social influ ence and coping patterns within families of dis turbed adolescents. Journal of Nervous and Mental Disease, 1968, 147, 233-251. Goodrich, D. W., and Boomer, D. S. Experimental assessment of modes of conflict resolution. Family Process, 1963, 2, 15-24. Gorad, S. L. Communicational styles and interaction of alcoholics and their wives. Family Process, 1971, 10, 475-489. Haley, J. Observation of the family of the schizophrenic. American Journal of Orthopsychiatry, 1960, 30, 460- 467. Haley, J. Family experiments: A new type of experimenta tion. Family Process, 1962, 1, 265-293. Haley, J. Strategies of psychotherapy. New York: Grune and Stratton, 1963. Haley, J. Research on family patterns: An instrument measurement. Family Process, 1964, 3, 41-65. Haley, J. Experiment with abnormal families. Archives of General Psychiatry, 1967, 17, 53-63. Haley, J. Testing parental instructions of schizophrenic and normal children. Journal of Abnormal Psychol ogy, 1968, 73, 559-565. Handel, G. Psychological study of whole families. Psy chological Bulletin, 1965, 63, 19-41. Henry, J. Culture against man. New York: Random House, 1963. 199 Hess, R. D., and Handel, G. Family Worlds: A Psychosocial Approach to Family Life. Chicago: University of Chicago Press, 1959. Hill, R. Payoffs and limitations of contemporary strate gies for family theory systemazation. Unpublished manuscript, 1971. Jackson, D. D., Riskin, J., and Satir, V. A method of an alysis of a family interview. Archives of General Psychiatry, 1961, 5, 321-339. Jackson, D. D., and Yalom, I. Family research on the prob lem of ulcerative colitis. Archives of General Psychiatry, 1966, 15, 410-418. Jackson, P. W. Verbal solutions to parent-child problems. Child Development, 1956, 27, 339-349. Jones, D. M. Binds and unbinds. Family Process, 1964, 3, 323-331. Kaiser, Hellmuth. Effective psychotherapy: The contribu tion of Hellmuth Kaiser (L. B. Fierman, Ed.) . New York: Free Press, 1965. Layman, E. M., and Lourie, R. S. Waiting room “ observation as a technique for analysis of communication be havior in children and their parents. In D. Holch and P. Zubin (Eds.), Psychopathology of communica tion, New York: Grune and Stratton, 1955*1 Leighton, L. A., Stollak, G. E., and Fergusson, L. R. Patterns of communication in normal and clinic families. Journal of Consulting and Clinical Psy chology , 1971, 36, 252-256. Leik, R. Instrumentality and emotionality in family inter action. Sociometry, 1963, 26, 131-145. Lennard, H. L., Beaulieu, M. R., and Embrey, N. G. Inter action in families with a schizophrenic child. Archives of General Psychiatry, 1965, 12, 166-183. Lennard, H. L., and Bernstein, A. Patterns in human inter action. San Francisco: Jossey-Bass, 1969. 200 Lerner, P. M. Resolution of intrafamilial role conflict in families of schizophrenic patients, I: Thought disturbance. Journal of Nervous and Mental Dis ease, 1965, 141, 342-351. Lerner, P. M. Resolution of intrafamilial role conflict in families of schizophrenic patients, II. Journal of Nervous and Mental Disease, 1967, 144, 336-341. Levin, G. Communicator-communicant approach to family interaction research. Family Process, 1966, 5, 105-116. Levinger, G. Supplementary methods in family research. Family Process, 1963, 2, 357-366. Levy, D. M., Goldfarb, W. , and Meyers, D. I. Relational behavior of schizophrenic children and their mothers: A methodologic study. Paper given at 1963 meeting of the American Orthopsychiatric As sociation, Los Angeles. Levy, J., and Epstein, N. B. An application of the Ror schach test in family investigation. Family Pro cess , 1964, 3, 344-376. Locke, H. J., and Williamson, R. C. Marital adjustment: A factor analysis study. American Sociological Review, 1958, 23, 562-569. Loveland, N. T., Wynne, L. C., and Singer, M. T. Family Rorschach: New method for studying family inter action. Family Process, 1963, 2, 187-215. Mark, R. A. Parameters of normal family communication in the dyad. Unpublished Ph.D. dissertation, Michigan State University, 1970. Mark, R. A. Coding communication at the relationship level. Journal of Communication, 1971, 21, 221- 232. McCord, W., Porta, J,, and McCord, J. The familial genesis of psychosis. Psychiatry, 1962, 25, 60-71. McPherson, S. Communication of intents among parents and their disturbed adolescent child. Journal of Ab normal Psychology, 1970, 76, 98-105. 201 Meissner, W. W. Thinking about the family— psychiatric aspects. Family Process, 1964, 3, 1-40. Mendell, D., Cleveland, S. E., and Fisher, S. A five gen eration family theme. Family Process, 1968, 7, 126-132. Mendell, D., and Fisher, S. An approach to neurotic behav ior in terms of a three generation family model. Journal of Nervous and Mental Disease, 12 3, 171- 180, 1956. Meyers, D. F., and Goldfarb, W. Studies of perplexity in mothers of schizophrenic children. American Journal of Orthopsychiatry, 1961, 31, 551-563-! Meyers, J. L. Fundamentals of experimental design. Boston; Allyn and Bacon, 1967. Mills, T. Power relations in three person groups. American Sociological Review, 1953, 18, 351-357. Mischel, W. Personality and assessment. New York: Wylie, 1968. Mishler, E., and Waxier, N. E. Family interaction pro cesses and schizophrenia. International Journal of Psychiatry, 1966, 2, 375-430. Mishler, E. G., and Waxier, N. E. Interaction in families: an experimental study of family processes and schizophrenia. New York: Wiley and Sons, 1968. Morris, G. O., and Wynne, L. C. Schizophrenic offspring and parental styles of communication. Psychiatry, 1965, 28, 19-44. O'Connor, W. A., and Stachowiak, J. Patterns of interac tion in families with low adjusted, high adjusted, and mentally retarded members. Family Process, 1971, 10, 229-241. Olson, D. H. Empirically unbinding the double bind: Re view of research and conceptual reformulations. Family Process, 1972, 11, 69-94. Olson, D. H., and Rabunsky, C. Validity of four measures of family power. Journal of Marriage and the Fam ily, 1972, 34, 224-234. 202 Olson, D. H., and Ryder, R. G. Inventory of Marital Con flicts (IMC); An experimental interaction proce dure, Journal of Marriage and the Family, 1970, 32, 443-448. Olson, D. H., and Ryder, R. G. IMC coding system. Unpub lished manual, 1973. (a) Olson, D. H., and Ryder, R. G. IFC coding system. Unpub lished manual, 1973. (b) O'Rourke, J. F. Field and lab: The decision-making behav ior of family groups in two experimental conditions. Sociometry, 1963, 26, 422-435. Parsons, T., and Bales, R. F. Family socialization and interaction process. Glencoe, 111.: Free Press, 1955, 91-94. Rabkin, L. Y. The patient's family: Research methods. Family Process, 1965, 4, 105-132. Ransom, D. Family communication: A conceptual and method ological analysis. Unpublished Ph.D. dissertation, University of California, Berkeley, 1971. Reiss, D. Individual thinking and family interaction, I. Introduction to an experimental study of problem- solving in families of normals, character disorders, and schizophrenics. Archives of General Psychiatry, 1967, 16, 80-93. (a) Reiss, D. Individual thinking and family interaction, II. A study of pattern recognition and hypothesis test ing in families of normals, character disorders, and schizophrenics. Journal of Psychiatric Re search, 1967, 193-211"! [bj Reiss, D. Individual thinking and family interaction, III. An experimental study of categorization performance in families of normals, those with character dis orders, and schizophrenics. Journal of Nervous and Mental Disease, 1968, 146, 384-403. Reiss, D. Individual thinking and family interaction, IV. A study of information exchange in families of nor mals, those with character disorders, and schizo phrenics. Journal of Nervous and Mental Disease, 1969, 149, 473-490. 203 Reiss, D. Intimacy and problem solving: An automated pro cedure for testing a theory of consensual experience in families. Archives of General Psychiatry, 1971, 25, 442-4S5. fa} Reiss, D. Varieties of consensual experience, I. A theory for relating family interaction to individual think ing. Family Process, 1971, 10, 1-28. (b) Reiss, D. Varieties of consensual experience, II. Dimen sions of a family's experience of its environment. Family Process, 1971, 10, 28-35. (c) Reiss, D. Varieties of consensual experience, III. Con trasts between families of normals, delinquents, and schizophrenics. Journal of Nervous and Mental Disease, 1971, 152/ 73-95. (dj* Ringuette, E. L., and Kennedy, T. An experimental study of the double bind hypothesis. Journal of Abnormal Psychology, 1966, 71, 136-141. Riskin, J. Family interaction scales: A preliminary re port. Archives of General Psychiatry, 1964, 11, 484-494. Riskin, J., and Faunce, E. E. Family interaction scales scoring manual. Published privately, 1968. Riskin, J., and Faunce, E. E. Family interaction scales, I. Theoretical framework and method. Archives of General Psychiatry, 1970, 22, 504-512. (al Riskin, J., and Faunce, E. E. Family interaction scales, III. Discussion of methodology and substantive findings. Archives of General Psychiatry, 1970, 22, 527-537": Tb} Riskin, J., and Faunce, E. E. Interactional family re search: A methodological review. Family Process, 1972, 11, 365-455. Rogers, L. Dyadic systems and transactional communication. Unpublished Ph.D . dissertation, Michigan State University, 1972. Roman, M., and Bauman, G. Interaction testing: A tech nique for the psychological evaluation of small groups. In M. Harrower (Ed.), Creative Variations 204 in projective techniques. Springfield, 111.: Charles C. Thomas, 1960. Roman, M., Bauman, G., Borello, J., and Meltzer, B. Inter action testing in the measurement of marital intel ligence. Journal of Abnormal Psychology, 1967, 72, 489-495. Rosenbaum, C. P. Patient-family similarities in schizo phrenia. Archives of General Psychiatry, 1961, 5, 120-126. Rosenthal, A. J., Behrens, M. I., and Chodoff, P. Communi cation in lower class families of schizophrenics. Archives of General Psychiatry, 1968, 18, 464-470. Ruesch, J. Synopsis of the theory of human communication. Psychiatry, 1953, 16, 215-243. Ruesch, J., and Bateson, G. Communication, the social matrix of psychiatry. New York: Norton, 1951. Ryckoff, I., Day, J., and Wynne, L. Maintenance of stereo typed roles in the families of schizophrenics. Archives of General Psychiatry, 1959, 1, 93-98. Ryder, R. G. Two replications of color matching factors. Family Process, 1966, 5, 43-48. Ryder, R. G. Husband-wife dyads versus married strangers. Family Process, 1968, 7, 233-238. Ryder, R. G. Dimensions of early marriage. Family Pro cess, 1970, 9, 51-68. (a) Ryder, R. G. A topography of early marriage. Family Pro cess, 1970, 9, 385-402. (b) , Ryder, R. G., and Goodrich, D. W. Married couples' re sponses to disagreement. Family Process, 1966, 5, 30-42. Sanua, V. D. Sociocultural factors in families of schizo phrenics: A review of the literature. Psychiatry, 1961, 24, 246-265. Sanua, V. D. The sociocultural aspects of childhood schiz ophrenia. In G. Zuk and I. Boszormenyi-Nagy (Eds.), Family therapy and disturbed families, Palo Alto: Science and Behavior Books, 1967. 205 Satir, 1 Schuham Schuham Schuham Sharan, Singer, Singer, Sluzki, Sluzki, Smith, ] h Conjoint family therapy. Palo Alto: Science and Behavior Books, 1967. A. I. The Double-bind hypothesis a decade later. Psychological Bulletin, 1967, 68, 409-416. A. I. Power relations in emotionally disturbed and normal family triads. Journal of Abnormal Psychology, 1970, 75, 30-37. A. I., and Freshley, H. B. Significance of the nonverbal dimension of family interaction. Pro ceedings of the Annual Convention of the American PsychologTcal Association, 1971, 6, 455-456. S. Family interaction with schizophrenics and their siblings. Journal of Abnormal Psychology, 1966, 71, 345-353. M. T., and Wynne, L. C. Thought disorder and family relations of schizophrenics, III: Methodol ogy using projective techniques. Archives of Gen eral Psychiatry, 1965, 12, 187-200. M., and Wynne, L. C. Communication styles in par ents of normals, neurotics, and schizophrenics: Some findings using a new Rorschach scoring manual. Psychiatric Research Reports, 1966, No. 20, 25-38. C. E., and Beavin, J. Simetria y complementaridad: una definicion operacional y una tipologia de pare- jas [Symmetry and complementarity: An operational definition and a typology of pairs]. Acta psiquia- trica y psicologica de America latina, 1965, 11, 321-330. C. E., Beavin, J., Tarnoplosky, A., and Veron, E. Transactional disqualification. Archives of Gener al Psychiatry, 1967, 16, 494-504. L . C. Verbal discussion versus note passing tasks in the study of family role structure. Journal of Nervous and Mental Disease, 1971, 152, 173-183. Snedecor, G. W. Statistical methods. Ames, Iowa: Iowa State College Press, 1967. 206 Spiegal, J., and Bell, N. The family of the psychiatric patient, In S. Arieti (Ed.), American handbook of psychiatry, New York: Basic Books, 1959, 1, 114- 149. Sprey, J. Family power structure: A critical comment. Journal of Marriage and the Family, 1972, 34, 235- 238. Stabenau, J. R., Tupin, J., Werner, M., and Pollin, W. A. A comparative study of families of schizophrenics, delinquents, and normals. Psychiatry, 1965, 28, 45-59. Straus, M. A. Power and support structure of the family in relation to socialization. Journal of Marriage and the Family, 1964, 26, 318-326. fa) Straus, M. A. Measuring families. In H. Cristensen (Ed.), Handbook of marriage and the family. Chicago: Rand McNally^ 1964. (b) Straus, M. A. Communication, creativity and problem-solv ing ability of middle and working-class families in three societies. American Journal of Sociology, 1968, 73, 417-430. Straus, M. A., and Tallman, I. SIMFAM: A technique for observational measurement and experimental study of families. In J. Aldous (Ed.), Family problem solving. Hinsdale, 111.: Dryden Press, 1971". Strodtbeck, F. L. Husband-wife interaction over revealed differences. American Sociological Review, 1951, 16, 468-473. Strodtbeck, F. L. The family as a three person group. American Sociological Review, 1953, 19, 23-29. Terrill, J. M., and Terrill, R. E. A method for studying family communication.- Family Process, 1965, 4, 259-290. Titchener, J. L., D'zmura, T., Golden, M., and Emerson, R. Family transaction and derivation of identity. Family Process, 1964, 2, 95-120. Titchener, J. L., and Emerson, R. Some methods for the study of family interaction in personality develop- 207 ment. Psychiatric Research Reports, 1958, No. 10, 72-88. Usandivaras, R. J., Grimson, W. R., Hammond, H., Isaharoff, E., and Romanos, D. The marbles test. Archives of General Psychiatry, 1967, 17, 111-118. Watzlawick, P. A review of the double bind theory. Family Process, 1963, 2, 132-153. Watzlawick, P., Beavin, J., and Jackson, D. Pragmatics of human communication, New York: Norton, 1967. Watzlawick, P., Beavin, J., Sikorski, L., and Mecia, B. Protection and scapegoating in pathological fam ilies. Family Process, 1970, 9, 27-39. Waxier, N., and Mishler, M. Scoring and reliability prob lems in interaction process analysis: A methodo logical note. Sociometry, 1966, 29-28-40. Wild, C. Disturbed styles of thinking. Archives of General Psychiatry, 1965, 13, 464-470. Winter, W. D., and Ferreira, A. J. Interaction process analysis and family decision-making. Family Pro cess, 1967, 6, 155-172. Winter, W. D., Ferreira, A. J., and Bowers, N. Decision making in married and unrelated couples. Family Process, 1973, 12, 83-94. Wynne, L,, and Singer, M. Thought disorder and family re lations of schizophrenics, I: Research strategy. Archives of General Psychiatry, 1963, 9, 191-198. JT) Wynne, L., and Singer, M. Thought disorder and family re lations of schizophrenics, II: A classification of forms of thinking. Archives of General Psychia try, 1963, 9, 199-206. ~Tb) Wynne., L., and Singer, M. Thought disorder and family relations of schizophrenics, IV: Results and im plications. Archives of General Psychiatry, 1965, 12, 201-202. 208 Zuk, H. , and Rubenstein, D. A review of concepts in the study and treatment of families of schizophrenics. In I. Boszormenyi-Nagy and J. L. Framo (Eds.), In tensive family therapy, New Yorks Harper and Row, 1965. A P P E N D I C E S 209 APPENDIX A CONSENT FOR PARTICIPATION IN A RESEARCH PROJECT 210 211 CONSENT FOR PARTICIPATION IN A RESEARCH PROJECT I hereby give my consent to participate in a re search project investigating communication between husbands and wives. I understand that our discussions will be tape- recorded, but that any information about me will remain confidential and will be used for research purposes only. The primary investigator in this project is Lewise L. Busch, M.A. Signature Date APPENDIX B MARITAL AGREEMENT QUESTIONNAIRE 212 213 MARITAL AGREEMENT QUESTIONNAIRE The following questions are about how you feel about your marriage. Frank and honest replies are of the highest im portance. There are no "right" or "wrong" answers. The following points are to be observed: (1) Answer all ques tions. Do not leave any blanks to signify no reply. (2) Do not confer with your mate in answering the ques tions or show your answers to him. (3) For each question, place an "x" in front of the most appropriate answer. 1. Do you and your mate agree 4. Do you and your mate on good and proper behav- agree on handling ior? family finances? A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree F. always disagree 2. Do you and your mate agree on aims, goals, and things believed important in life? A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree F. always disagree 5. Do you and your mate agree on amount of time spent together? A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree F. always disagree A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree F. always disagree 3. Do you and your mate agree on friends? _A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree _F. always disagree 6. Do you and your mate agree on demonstration of affection? A. always agree B. almost always agree C. occasionally disagree D. frequently disagree E. almost always disagree F. always disagree qH D n w > t t d D o a > ^ m a n w > 214 MARITAL AGREEMENT QUESTIONNAIRE (Continued) 7. Do you and your mate agree 10. Do you and your mate on sex relations? agree on responsibili ties around home? A. always agree A. always agree B. almost always agree B. almost always agree __C. occasionally disagree C. occasionally disagree D.frequently disagree D. frequently disagree E. almost always disagree E. almost always disagree __F. always disagree F. always disagree Do you and your mate agree on ways of dealing with in-laws? . always agree . almost always agree . occasionally disagree . frequently disagree . almost always disagree . always disagree Do you and your mate agree on handling angry or hurt feelings? . always agree . almost always agree . occasionally disagree . frequently disagree . almost always disagree . always disagree 215 Now, look down the list of topics and your responses. Se lect that topic on which you and your mate are in the greatest agreement. Write the topic next to rank "1" on the rank list and lightly cross it off the topic list. Next, select the topic of greatest agreement from the re maining topics. Write the topic next to rank "2" on the rank list and cross it off the topic list. Proceed in this manner until all the topics are listed in order from greatest agreement to greatest disagreement on the rank list. TOPIC RANK 1. GOOD AND PROPER BEHAVIOR 1._____________ 2. AIMS AND GOLAS IN LIFE 2.____________ 3. FRIENDS 3._____________ 4. FINANCES 4._____________ 5. TIME SPENT TOGETHER 5.___________ _ 6. DEMONSTRATION OF AFFECTION 6.____________ 7. SEX RELATIONS 7.__________ __ 8. IN-LAWS 8._____________ 9. ANGRY OR HURT FEELINGS 9. 10. RESPONSIBILITIES AROUND THE HOUSE 10. APPENDIX C PERSONAL INFORMATION FORM 216 217 PERSONAL INFORMATION FORM NAME_______________________________________PHONE NO._________ ADDRESS____________________________ _________________________ AGE_____________ YEARS MARRIED_______________ RELIGIOUS OR ETHNIC BACKGROUND_____________________________ IF PREVIOUSLY MARRIED, HOW MANY TIMES?____________________ HOW MANY CHILDREN?_________ THEIR AGES___________________ THEIR SEXES__________________________________________________ YOUR OCCUPATION_____________________________________________ APPROXIMATE YEARLY INCOME TO FAMILY________________________ HIGHEST EDUCATIONAL LEVEL REACHED__________________________ HAVE YOU EVER SOUGHT PROFESSIONAL ASSISTANCE FOR PERSONAL OR MARITAL DIFFICULTIES? FOR HOW LONG? APPENDIX D EXPERIMENTER CHECK SHEET 218 219 EXPERIMENTER CHECK SHEET SUBJECTS: MR. & MRS._______________________________________ DATE:________________ TIME:________________ EXPERIMENTER:_______ TOPIC M RANK F RANK AVERAGE AGREEMENT:_________________________ DISAGREEMENT: __________ Comments and work notes:
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Prognostic Expectancy Effects In The Desensitization Of Anxiety Over Invasion Of Body Buffer Zones
PDF
The Effects Of Diphenylhydantoin On The Galvanic Skin Responses Of Psychopathic And Normal Prisoners
PDF
Self-Sacrifice, Cooperation, And Aggression In Women Of Varying Sex-Role Orientations
PDF
Looking Back After Coming Down: Conformity And Commitment In Campus Protest
PDF
Stimulus and response generalization of classes of imitative and nonimitative behavior as a function of reinforcement, task, cues, and number of therapists
PDF
A Behavioral Assessment Of The Reinforcement Contingencies Associated With The Occurrence Of Suicidal Behaviors
PDF
Mental Imagery As A Function Of Muscular Tension And Suggestion
PDF
The Effect Of Anxiety And Frustration On Muscular Tension Related To The Temporomandibular-Joint Syndrome
PDF
Coalition Formation In Conjoint Marriage Counseling
PDF
A Cognitive Dissonance Analysis Of Conformity Behavior, As Applied To The full Denture Patient
PDF
Attention, retention, and incentive processes in observational learning
PDF
Factors Of Adaptation And Rehabilitation In Home Hemodialysis
PDF
Dissonance and Self-Perception Analysis of "Forced Compliance": When Two Theories Make Competing Predictions
PDF
The Development And Evaluation Of Three Therapeutic Group Interventions For Widows
PDF
The Effects Of Repression-Sensitization, Race, And Levels Of Threat On Extensions Of Personal Space
PDF
Death Anxiety In Leukemic Children
PDF
Modification Of Low Self-Confidence In Elementary-School Children By Reinforcement And Modeling
PDF
An Examination Of Positive And Negative Reinforcement In Classical And Operant Conditioning Paradigms In The Primary Psychopath
PDF
Weight Reduction As A Function Of The Timing Of Reinforcement In A Covertaversive Conditioning Paradigm
PDF
The effects of physical proximity and body boundary size on the self-disclosure interview
Asset Metadata
Creator
Busch, Lewise Langston
(author)
Core Title
Strategies Of Marital Communication
Degree
Doctor of Philosophy
Degree Program
Psychology
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
OAI-PMH Harvest,Psychology, clinical
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
Wolpin, Milton (
committee chair
), Broderick, Carlfred Bartholomew (
committee member
), Frankel, Andrew Steven (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c18-738087
Unique identifier
UC11356435
Identifier
7501049.pdf (filename),usctheses-c18-738087 (legacy record id)
Legacy Identifier
7501049.pdf
Dmrecord
738087
Document Type
Dissertation
Rights
Busch, Lewise Langston
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