RELATIONAL COMMUNICATION: COMPLEMENTARITY AND SYMMETRY AND THEIR RELATION TO . . DOMINANCE- SUBMISSION Thesis for the Degree of Ph. D. MICHIGAN STATE UNIVERSITY PHILIP MARTIN ERICSON. 1972 IIIIIIIIIIIIIIIIIIII This is to certify that the thesis entitled RELATIONAL COMMUNICATION: ' COMPLEMENTARITY AND SYMMETRY AND ‘THEIR RELATION TO OOMINANCE-SUBMISSION presented by Philip M. Ericson has been accepted towards fulfillment of the requirements for Ph.D. degree in COMUI’ITQQIIOM £111: 5 (51121111, [1/ Major professor Date July 21, 1972 0-7839 . ~— ‘5! BINDING .Y HOAG & SONS' .BflOK BINDERY INC LIBRARY BINDEHS III mmaroan. :m : ABSTRACT RELATIONAL COMMUNICATION: COMPLEMENTARITY 'AND SYMMETRY AND THEIR RELATION TO DOMINANCE-SUBMISSION BY Philip Martin Bricson This study examined a proposed theoretic relation- ship between the relational communication concepts of complementarity and symmetry, and Robert C. Carson's interaction theory of personality. On the basis of his theory and previous research on relational communication, predictions were made about the relationship of dominance scores, dominance difference scores, and social class to complementary and symmetrical transactions. Data for the study were obtained from a random sample of 56 husband-wife dyads in their homes in Dayton, Ohio under an Office of Civil Defense grant, Contract #DAHC—20-7l-C-0297.’ Couples filled out a questionnaire that included Edwards' dominance scale and socio-exonomic questions, and then discussed four topics, one dealing with how they happened to meet and marry, one dealing with preparations they would make in the event of a nuclear Philip Martin Ericson attack, one dealing with the independence of the wife from the home, and one dealing with preparations they would make in the event of a tornado warning. These discussions were tape recorded on casette recorders, were transcribed, and were coded by a coding scheme modified from Mark (1970). Only the last three topics were analyzed in this study. A 2x2 Analysis of Variance design was used to test four hypotheses that included the variables of dominance difference and social class. None of these hypotheses were supported and several significant reversals to what was predicted were found. Rather than upper-class couples having a higher proportion of symmetrical transactions, lower-class couples had the higher proportion; rather than lower-class couples having a higher proportion of comple- mentary transactions, upper-class couples had a higher proportion; rather than upper-class couples with a low dominance difference having a higher proportion of symmet- rical transactions than lower-class couples with a low dominance difference, the reverse was the case. Additional hypotheses concerning a predicted higher proportion of complementary transactions than symmetrical transactions, a predicted linear relationship between dominance scores and one-up messages, and a predicted higher husbands' dominance score/one-up message correlation than wives' dominance score/one-up message correlation were also not supported. Philip Martin Ericson The lack of significant predicted findings is dis- cussed in terms of the difficulty of predicting from a non-interaction phenomenon--fi11ing out a dominance scale-- to an interaction phenomenon--discussing a series of topics, and is also discussed in terms of a lack of social class variables that may predict relational communication behaviors. Descriptive statistics were used to indicate per- centages of individual messages, control dimensions of those messages, and percentages of transaction types. The most significant finding to come from the study con- cerns the stability in the use of transaction types, as measured by a deviation from an expected random distribu- tion score, and the fact that the non-Civil Defense topic involved a more rigid use of certain transactions than did the other topics. It is suggested that this difference may be due to the content, structure or "public" nature of the topic. Suggestions for further research relate to a re- definition of relational communication complementarity, using a set of predictions based on actual relational and transaction data, and measuring the reward aspect of com- munication behaviors that might be enacted for certain initiating behaviors. Accepted by the faculty of the Department of Communication, College of Communication Arts, Michigan State University, in partial fulfillment of the require- ments for the Doctor of Philosophy degree. Erwin P. Bettin haus Director of The51s Guidance Committee: Erwin P. Bettinghaus T’Chairman Richard V. Farace John T. Gullahorn Verling C. Troldahl RELATIONAL-COMMUNICATION: COMPLEMENTARITY AND SYMMETRY AND THEIR RELATION TO DOMINANCE-SUBMISSION By Philip Martin Ericson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1972 To Susan, Karl and Pam ii ACKNOWLEDGMENTS I wish to acknowledge the help provided me by my guidance committee members--Dr. Erwin Bettinghaus, Dr. Vincent Farace, Dr. John Gullahorn, and Dr. Verling Troldahl. Dr. Bettinghaus, my guidance committee chair- man and thesis director, gave me freedom to explore and use various alternative analyses and provided clarity and direction when it was most needed. Dr. Farace helped spur my interest in family communication research through his persistent questioning and allowed me the opportunity to utilize his project's husband-wife interaction data. Dr. Gullahorm increased my awareness of the area of small group behavior through the use of computer simulation and served on my committee despite severe time pressures. Dr. Troldahl provided me with methodological knoWledge of and insights to many of the analyses used in this study. I have learned much and benefited greatly from their advice, interest, and friendship. I also wish to acknowledge the advice and ideas of a good friend and colleague, Edna Rogers, who contri- buted greatly to the modification of the relational communication coding scheme. Although our contributions to that modification may have been equal, her tendency iii not to accept all the ideas I suggested and her intellec- tual probing were of immense help in thinking through many of the conceptual and operational problems we faced. I also wish to thank Frank Millar--who served ably as a field administrator for the data collection and with whom I shared some of the frustrations, disappointments, and satis- faction resulting from that data collection, and Cay Bettinghaus--who gave advice and clues for the data collec- tion procedures and served as a professional interviewer. Ken Villard, Joanne Helfrich, and Jeff Tully also deserve my thanks for helping, respectively, to collect the data, write a computer listing program for the data analyses, and provide suggestions on the use of several computer analyses. Finally, I wish to thank my wife and children who withstood many of my frustrating moments and helped share in many of the satisfying moments of the last three years. They made the effort worthwhile and it is to them that I dedicate this thesis. iv TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . . . xi INTRODUCTION . . . . . . . . . . . . . . . . . . . . l 1A THEORY OF PERSONALITY AND INTERACTION . . . . . . 5 Summary . . . . . . . . . . . . . . . . . . . . 22 A REVIEW OF THE LITERATURE CONCERNING NEED AND BEHAVIOR COMPLEMENTARITY . . . . . . . . . . . 25 Summary and Conclusion . . . . . . . . . . . . . 40 A REVIEW OF COMPLEMENTARITY BEHAVIORS AND LITERATURE CONCERNING SYMMETRICAL AND COMPLEMENTARITY RELATIONAL COMMUNICATION . . . . . . 41 Complementary Behavior Characteristics . . . . . 41 Managerial-Autocratic Behavior . . . . . . . 42 Responsible-Hypernormal Behavior . . . . . . 43 Cooperative- -Overconventional Behavior . . . 43 Docile- -Dependent Behavior . . . . . . 44 Self- -effacing- -Masochistic Behavior . . . . . 44 Rebellious- Distrustful Behavior . . . . . . 44 Aggressive- Sadistic Behavior . . . . . . . 4S Competitive- Narcississtic Behavior . . . . . 45 Relational Communication Concepts . . . . . . . 46 Personality and Relational Communication Complementarity . . . . . . . . . . . . . . . . 65 DYADIC DIMENSION AND SITUATIONAL VARIABLES, OPERATIONALIZATIONS AND SOME HYPOTHESES . . . . . . 68 Some Situational and Dyadic Dimension Variables . . . . . . . . . . . . . . . . . . . 69 History . . . . . . . . . . TaSk O O O O O O O O O O : O O O O Socio- Economic C1ass . . . . . . . . . . Hypotheses . . . . . . . . . . . . . . . . . Summary and Conclusion RESEARCH METHODOLOGY . . . . . . . . . . Sample . . . . . . . . . Method of Data Collection . . . . . . Operationalization of the Variables Dominance . . . . . . . . Socio- Economic Class . . Complementary and Symmetrical Transactions RESULTS . . . . . . . . . . . . . . . . . . . . Coder Reliability Checks . . . . . . . . . Transformations Applied to Dependent Variables Statistical Analyses . . . . . . . . . . . . Hypothesis 1 . . . . . . . . Hypotheses 2 and 3 . . . . . . . Hypotheses 4 and 5 . . . . . . . . . . . Hypotheses 6 and 7 . . . . . . . . . . . Summary . Descriptive Statistics . . . . Times and Probes across Topics . . . . Message Types by Individuals . Message Types by Dominance Difference, Social Class, Sex, and Topics . . . . . Control Aspects of Messages Transaction Types . . . . . . . . Summary . . . . . DISCUSSION OF RESULTS Complementarity . Dominance Scores . . . . . . . . . . . Social Class Findings . . . Stability of Transaction Types . . . . . . vi Page 69 71 72 74 77 79 79 81 84 84 89 91 111 111 118 120 121 123 126 127 130 133 134 138 142 145 147 163 167 167 171 175 177 Page Effects of Topics . . . . . . . . . . . . . . . 178 Advantages and Disadvantages of Coding Scheme . . . . . . . . . . . . . . . . . . . 180 Suggestions for Further Research . . . . . . . 183 Conclusion . . . . . . . . . . . . . . . . . . 190 REFERENCES . . . . . . . . . . . . . . . . . . . . 191 APPENDICES . . . . . . . . . . . . . . . . . . . . 197 vii Table 1. 10. 11. 12. 13. LIST OF TABLES Mutual rewards and costs based upon "inherent" complementarity and anticomplementarity . . . . Means and standard deviations of the dominance variable for the EPPS normative sample and the Phase II samples . . . . . . . . . . . . . Dominance difference scores . . . . . . . . Education level of husbands . . . . . . . . First stage percentage agreements between coders on use of three digit codes . . . . . Second stage percentage agreements between coders on use of three digit codes . . . . . . t test between mean proportions of complev mentary and symmetrical transactions . . . . . Proportion of complementary transactions by social class and dominance difference . . . . . Comparison of upper- and lower-class high dominance difference dyads' mean proportions of complementary transactions . . . . . . . . . Proportion of symmetrical transactions by social class and dominance difference . . Comparison of upper- and lower-class low dominance difference dyads' mean proportion of symmetrical transaction . . . . . . . . Correlation coefficients between dominance scores and proportions of one-up messages . Z test for husbands' and wives' dominance scores and one-up proportions correlation coeffiCients O O O O O I O O O O I O O O 0 viii Page 16 86 88 90 113 116 122 125 125 128 128 129 130 Table Page 14. Times for topics by social class . . . . . . . 134 15. Times by social class, dominance difference and correlated topics . . . . . . . . . . . . . 135 16. t test of probes by social class . . . . . . . 136 17. Frequencies and percentages of message types across sample and across 3 topics . . . . . . 139 18. Comparison of message types from Mark (1970) and present study . . . . . . . . . . . . . . . 141 19. Ranges of husbands' dominance scores and dyad dominance difference scores . . . . . . . . . . 143 20. Frequencies and percentages for control dimensions of messages . . . . . . . . . . . . 146 21. Frequencies and percentages of transaction types by topic . . . . . . . . . . . . . . . . 147 22. R deviation scores by dominance difference, social class, and correlated topics . . . . . . 154 23. Proportion of complementary transactions by dominance difference, social class, and correlated topics . . . . . . . . . . . . . . . 156 24. Proportion of message type 11 (assertion as support) by dominance difference, social class, sex and correlated topics . . . . . . . . . . . 228 25. Proportion of message type 13 (assertion as extension) by dominance difference, social class, sex and correlated topics . . . . . . . 229 26. Proportion of message type 14 (assertion as answer) by dominance difference, social class, sex and correlated topics . . . . . . . 230 27. Proportion of message type 19 (assertion as initiation-termination) by dominance dif- ference, social class, sex and correlated topics . . . . . . . . . . . . . . . . . . . . 231 ix Table Page 28. Proportion of message type 23 (question as extension) by dominance difference, social class, sex and correlated topics . . . . . . 232 29. Proportion of message type 31U (unsuccessful talkover as support) by dominance difference, social class, sex and correlated topics . . . 233 30. Proportion of message type 318 (successful talkover as support) by dominance difference, social class, sex and correlated topics . . . 234 31. Proportion of message type 33U (unsuccessful talkover as extension) by dominance differ- ence, social class, sex and correlated topics . . . . . . . . . . . . . . . . . . . 235 32. Proportion of message type 338 (successful talkover as extension) by dominance differ- ence, social class, sex and correlated topics . . . . . . . . . . . . . . . . . . . 236 33. Silences (code 000) by dominance difference, social class and correlated topics . . . . . 237 34. Proportion of competitive symmetrical trans- actions by dominance difference, social class and correlated topics . . . . . . . . . 238 35. Proportion of submissive symmetrical trans- actions by dominance difference, social class and correlated topics . . . . . . . . . 239 36. Proportion of neutralized symmetrical trans- actions by dominance difference, social class and correlated topics . . . . . . . . . 240 37. Proportion of transitory transactions by dominance difference, social class and correlated topics . . . . . . . . . . . . . . 241 LIST OF FIGURES Figure 1. The interpersonal behavior circle 2. Control dimensions of message types 3. Transactional types 4. R deviation score means across three topics 5. R deviation scores by topics xi Page 14 102 107 151 153 CHAPTER I INTRODUCTION A popular trend in the field of communication today is the study of communication as a process that requires interdependent participants. Instead of analyzing communi- cation as a source-to-receiver phenomenon, the focus has shifted to a general systems perspective where participants are viewed as interdependent and are connected by relation- ship rules. From the systems point of view these rules function to maintain the communication system by defining the roles of the participants and how they are to interact with one another. With the advantage of focussing on the interacting system rather than on an individual and with the view that systems possess similar characteristics across levels, the general systems framework has been utilized in psychotherapy to study ”normal" and non—normal families. These studies have looked at the relational patterns that husbands and wives use to define their respective roles. The general systems model on which much of the analysis of communication in family-therapy situations has been based is most thoroughly developed by Watzlawick, Beavin and Jackson (1967). According to them, communica- tion is an interaction process which is characterized by "two or more communicants in the process of, or at the level of, defining the nature of their relationship." (Watzlawick, g£_al., 1967, p. 121.) From this systems perspective they develop five axioms of communication, one of which specifies the types of roles being defined. Formally stated the axiom is: "All communicational inter- changes are either symmetrical or complementary, depending on whether they are based on equality or difference." (WatzTawick, g£_21., p. 70.) This axiom is based on concepts originally developed by Gregory Bateson in 1936--name1y, symmetry and comple-I mentarity. These interaction concepts pertain to the roles that participants will hold in a given relationship. Symmetrical interaction is where the participants tend to mirror each other's behavior; complementary interaction exists when one participant's behavior complements the other's. The former is characterized by "equality and the minimization of difference" while the latter is based on "maximization of difference." (Watzlawick, g£_§1., pp. 68-9.) This maximization of difference exists when one participant is in a superior or "one-up" position and the other participant is in an inferior or "one-down" position. Neither participant can adopt one of these positions with- out the other nor can a participant impose the relationship upon the other. As Watzlawick, g£_a1., state, "each be- haves in a manner which presupposes, while at the same time providing reasons for, the behavior of the other: their definitions of the relationship fit." (p. 69.) Although the previous discussion has indicated the, two major relational constructs of the model, there are two important weaknesses to this paradigm. One is that no operationalizations are provided for complementarity and symmetry. The other is that the model has very little explanatory power and no predictive power to indicate when a certain relationship will emerge. The former problem has been attacked and partially solved in the writings of ‘ Sluzki and Beavin (1965) and Mark (1970). Whereas Sluzki and Beavin indicate a typology of dyads and a categoriza- tion of symmetrical and complementary transactions, Mark develops a coding scheme that indexes statements made by dyad members and classifies them as symmetrical or comple- mentary depending on the statements immediately preceding them. The latter problem concerning a lack of explanation and predictability has not received much attention. For one thing, Watzawick, Beavin and Jackson themselves are not that concerned with these issues. As they state, the questions are tangential to their exposition. Since our aim is intensive rather than extensive, 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 on how and not why the interactional system operates. (WaEETawick, et al., pp. 130-1.) For another, finding a theory that integrates these rela- tional concepts, provides predictions of certain relation- ships, and does not, itself, suffer from major theoretical faults is not an easy task. Nevertheless, in an attempt to answer some of these problems, this present study sought to integrate the relational concepts of symmetry and com- plementarity with a theoretical framework in order to better explain why a symmetrical or complementary relation- ship emerges, and predict when either might occur. The focus of this study was on how husbands and wives communicate on a relational level while discussing a set of three topics on which they had been asked to reach a joint decision or conclusion. It was hoped that the integration of the relational communication concepts with a theoretic framework of interaction would lead to a better understanding of relational communication, and that the interaction or discussion data would provide additional information concerning the patterns that husband and wife dyads use to communicate. CHAPTER II A THEORY OF PERSONALITY AND INTERACTION A theoretic framework which seems to have close ties to the relational concepts explored by Watzlawick, et a1. is one developed by Robert C. Carson in his book, Interaction Concepts of Personality (1969). Carson's theory of interpersonal behavior as manifested by dyadic interaction is eclectic. He draws on the personality theories of Sullivan and Leary, the exchange theory of Thibaut and Kelley, and the information theory of Miller, Galanter, and Pribram, and attempts to create a framework that integrates aspects from each. Emphasizing the im- portance of personality in interaction, Carson argues that The behavior of the two persons engaged in a typical dyadic interaction is determined by the dispositional tendencies inherent in each of them at the time and by their perceptions of their own and of each others' behavior, as well as their perceptions of other aspects of the situation--perceptions biased in turn by their dispositional tendencies. (Carson, 1969, p. 12.) It is the purpose of this chapter to discuss Carson's framework in order to later derive testable hypotheses concerning what types of relational communica- tion will emerge when certain personality types interact. Carson's conception of personality is heavily based on ideas articulated by Harry Stack Sullivan. It is Carson's interpretation that, for Sullivan, personality was . . .nothing more (or less) than the patterned regu- larities that may be observed in an individual's relations with other persons, who may be real in the sense of actually being present, [or] real but absent and hence 'personified' or illusory. (Carson, p. 26.) These regularities emerge in interpersonal interactions in which needs are met through a reciprocal process of integra- tion. In other words, through interpersonal interaction, (1) complementary needs are resolved, or aggravated; (2) reciprocal patterns of activity are developed, or disintegrated; and (3) foresight of satisfaction, or rebuff, or similar needs is facilitated. (Carson, p. 41.) The types of interaction in which people find them~ selves engaged may be quite numerous. Nevertheless, Carson suggests that there are certain basic dimensions of the types of interpersonal relationship defined in any given interaction. The isolation of these dimensions has been the purpose of several studies, some of which Carson re- views. Among those studies mentioned are: Roger Brown's (1965), in which he argued the two dimensions of status and solidarity are common to most, if not all, human inter- action; Borgatta and others' (1958, 1960), where factor analyses pointed to the factors of assertiveness, like- ability, emotionality, intelligence and responsibility; and Becker and Krug's (1964) in which a two-centroid factor analysis of teacher/parental ratings of children's behaviors yielded an Introversion-Extraversion factor and an Emotional Stability-Instability factor. On the basis of this review of this research, Carson concludes that major portions of the domain of interpersonal behavior can profitably and reasonably accurately be conceived as involving variations on two independent, bi-polar dimensions. One of these may be called a dominance- submission dimension; it includes dominant, assertive, ascendint, leading, controlling (etc.) behaviors on the one hand, and submissive, retiring, obsequious, unas- sertive, following (etc.) behaviors on the other. The poles of the second principle dimension are perhaps best approximated by the terms hate versus love; the former includes hateful, aggre551ve, rejecting, punish- ing, attacking, disaffiliative (etc.) behaviors, while the latter includes accepting, loving, affectionate, affiliative, friendly (etc.) social actions. (Carson, p. 102.) Focussing on studies that looked at a circumplex ordering of interpersonal factors, Carson singles out the work of Leary (1957) and his associates. Their basic model is a two factor (Dominance vs. Submission and Hate vs. Love) circumplex that has 16 distinctive, behaviorally defined segments ordered into 8 general categories. Each of these categories has ”a moderate (adaptive) and an ex- treme (pathological) intensity.” (Leary, 1957, p. 65.) The 16 behaviors, ordered into the octant categories by the moderate and then the extreme form of the behavior, are: Managerial-Autocratic, Responsible-Hypernormal, Cooperative-Overconventional, Docile-Dependent, Self- effacing-Masochistic, Rebellious-Distrustful, Aggressive- Sadistic, and Competitive-Narcissistic. Each of these behaviors has a complementary behavior that can and is very often elicited. This idea is based on Leary's belief that interpersonal behaviors are, . . .in part, security operations (a Sullivan term) employed by persons to maintain relative comfort, security, and freedom from anxiety in their interactions with others. The ur ose of interpersonal behavior, in terms of its securlty-maintenance functions, is to induce from the other person behavior that is comple- mentary to the behavior proffered. (Carson, p. 112.) Such complementary behaviors are considered rewarding for the participants; i.e., each has performed a service for the other. The rationale for viewing complementary behaviors as rewarding is based primarily and most heavily on the theoretical analysis originally developed by Thibaut and Kelley (1957). Their analysis, like Homans' (1961), is based upon an exchange view of human interaction where "the hedonic outcome attained by a person is calculated in terms of the rewards he receives and the $3535 he incurs in that interaction . . ." (Carson, p. 123.) For them, rewards are any positively valued consequences that accrue to a person from an interaction, such as the ability to engage in preferred behavior, need-satisfaction, consensual validation of one's at- titudes, experiencing of personal acceptance by the other, and so forth. (Carson, p. 123.) Costs, on the other hand, are negatively valued conse- quences such as the effort or energy expenditure required to make particular responses; the arousal of anxiety, embarrassment, or other unpleasant feelings; the simul- taneous instigation of competing or conflicting re- sponses; noncomplementary behavior in the other person, and so forth. (Carson, p. 123.) Although one implication from these definitions is that one can identify rewarding and costly acts from a basis of generally accepted behavior, Carson argues that the general social desirability of various kinds of be- havior is correlated very imperfectly with the reward or cost value of the behavior as it is actually per- formed in given circumstances. (Carson, p. 125.) The unit of analysis set forth by Thibaut and Kelley is a behavior sequence or set that refers to an organized pattern of verbal and/or major acts that may be instrumental in gaining or arriving at, or consummatory in appreciating, some goal. For Carson, the set is very similar to a Plan as articulated by Miller, Galanter, and Pribram (1960); the behavior sequence is like the tactical- level TOTE units. Since each person in a dyad brings with him a repertoire of sets, the interaction in the dyad will involve behavior sequences that involve the least cost and highest reward for each. These reward and cost outcomes will determine whether or not the interaction progresses and the relationship between the participants remains stable. 10 Each person in the dyad has, according to Thibaut and Kelley, two reference points from which evaluation of interaction outcome is made. One is known as the Compari- son Level (CL), "a kind of adaptation level that represents the neutral point for the person on a scale of goodness versus badness of outcome." (Carson, p. 129.) This level probably varies from person to person and also within the same person from time to time. When net outcomes (rewards minus costs) fall below an individual's CL, they are per- ceived as unsatisfactory; when they fall above the person's CL, they are experienced as satisfactory and pleasant. Depending on the levels of the CL, the outcomes of a rela- tionship may be high for one individual and low for another or vice versa. ”The magnitude of the hedonic effect will in each case be proportional to the 'distance' between outcomes and the CL." (p. 130.) Whereas the first reference point, the CL, is im- portant for determining the attractiveness of the relation- ship, a second reference point is influential in determining the quality (in terms of outcomes) of alternative relation- ships. This point, known as the Comparison Level for Alternatives (CL is that against which relationship alt)’ outcomes are evaluated. As interpreted by Carson, the CLalt represents the lowest outcome level a member of a dyad will accept in the light of the outcomes avail- able to him in alternative relationships. If outcomes in a relationship drop below CLalt: the person will leave the relationship in order to engage his current 11 next-best alternative. The height of the CLalt there— fore depends mainly upon the reward-cost p051tions experienced or believed by the person to exist in the most satisfactory of the available alternatives to the present relationship, where, as in the case of the CL, these anticipated outcomes are weighted by their current salience for the person. While the height of the CL in relation to outcomes determines a person's satisfaction with a relationship, the hiehgt of his CLalt may be said to determine his dependence upon the relationship. (pp. 130-1.) If the outcomes for each member exceed his CLalt’ the dyadic interaction will likely produce rewarding outcomes for both members and the relationship will undoubtably continue. The conditions under which a given relationship will produce better outcomes than alternative relationships is when, and if, each member can produce rewards for the other without incurring corresponding excessive costs. For a rela* tionship to persist it must produce 'ointl experienced outcomes above each member s CLalt' a condition that is tantamount to interdependence. (p. 132.) Based on these considerations, Carson argues that personal styles (tendencies to enact sets that fall pre- dominantly within a range of the interpersonal behavior circle) may become habitual due to reinforcement of certain behavior sequences by others. The "personality" of an individual, then, is basically a generalized stylistic variation he enacts with others. For example, if person A enacts a given form of behavior, it may func- tion both as a reinforcement (positive or negative, depending upon its degree of complementarity) for B's immediately preceeding behavior, and as a prompt for B 12 to respond in a complementary way. We may infer, then, that one of A's "purposes” in enacting a given form of behavior is that of producing in B a particular form of behavior in response. If’A consistently enacts that same type of behavior over time, and with other persons as well, we are justified in assuming that the receipt of behavior which is complementary to it has generalized optimal outcome value for A, relative to other kinds of behavior he might receive. (p. 143.) As indicated earlier, this rewarding outcome has been ex- plained by Leary as a function of a person's security-~ maintaining equipment. By prompting complementary behavior (for example, behavior in the other person that confirms the initiator's self-concept), the initiator may feel some reward; by failing to prompt complementary behavior in the other (for example, anticomplementary behavior by him that disconfirms the initiator's self concept), the initiator may incur some cost. Behaviors that are complementary tend to occur on ”the basis of reciprocity in respect to dominance-submission, and on the basis of correspondence [symmetry] in respect to hostility-affection." (p. 145.) These characteristics are drawn from the interpersonal behavior circle developed by Leary. It should be emphasized, however, that Leary's description of interpersonal behaviors is not explicit 1 regarding the symmetry of the behaviors along the hostility- affection dimension. He holds that ”interpersonal reflexes [behaviors] tend (with a probability significantly greater than chance) to initiate or invite reciprocal interpersonal reflexes from the 'other' person in the interaction that 13 lead to a repetition of the original reflex" (Leary, 1957, p. 123) and does admit that this probability often does not hold for individual interactions as, for example, in the case where "aggression usually breeds counteraggression." (p. 125.) However, because his concern is with the recip~ rocal behaviors described in the interpersonal behavior circle, he never emphasizes specific symmetrical behaviors. The notion of symmetrical behaviors does appear, nevertheless, when he discusses certain octant categories of the behavior circle. In his discussions, he states that Aggressive/ Sadistic behaviors tend to pull Rebellious/Distrustful and Self-effacing behaviors while Cooperative/Overconventional behaviors tend to pull Overconventional and Hypernormal behaviors (pp. 272, 305). As can be seen in Figure l, f the Aggressive/Sadistic and Cooperative/Overconventional categories are at the hate and love axes, respectively, and the categories of those behaviors being pulled lie next to the initiating behaviors, a relationship better described as symmetrical than as complementary. Carson accepts Leary's view concerning the reci— procity of behaviors along the dominance-submission dimension and the correspondence of behaviors along the hostility-affection dimension, and he is somewhat more explicit in indicating what behaviors may be expected to be symmetrical. He agrees with Leary that both Cooperative 14 i3 g§ 5% Figure 1. The interpersonal behavior circle From: Robert C. Carson, Interaction Concepts of Personality. Chicago: Aldine Publishing Company, 1969, p. 108. and Aggressive behaviors will tend to initiate or pull symmetrical behaviors, although his major focus throughout his book is on complementary behaviors. For describing interaction outcomes, Carson collapses the octant categories of the circle into four quadrants. These are: hostile dominance (H-D, encompassing Aggressive- Sadistic and Competitive-Narcissistic behaviors); friendly submission (F-S, encompassing Docile-Dependent and 15 Cooperative-Overconventional behaviors); friendly dominance (F-D, encompassing Managerial-Autocratic and Responsible- Hypernormal behaviors); and hostile submission (H-S, encompassing Rebellious—Distrustful and Selfueffacing— Masochistic behaviors). Complementary behaviors that are rewarding for both participants are evidenced by the following combina- tions: Aggressive/Competitive behaviors (H-D) are enacted by one individual and Self-effacing/Rebellious behaviors (H-S) are enacted by the other; Managerial/Responsible behaviors (F-D) are enacted by one individual and Cooperative/Docile behaviors (F-S) are enacted by the other; and the reverse of these combinations, Cooperative/ Docile behaviors (F-S) enacted by one are met with Managerial/Responsible behaviors (F-F) by the other and Self-effacing/Rebellious behaviors (H-D) by the other. These complementary behaviors, expressed in terms of rewarding or costly outcomes, may be seen in Table l, p. 16. As this interaction-outcome matrix shows, the diagonal running from the bottom left corner to the top right corner expresses mutual rewards (positive signs) for complementarity. It should also be noted from this matrix that Carson includes the symmetry of hostility- affection behaviors within his notion of complementarity by combining quandrants. (For example, friendliness (F) 16 Table 1. Mutual rewards and costs based upon "inherent" complementarity and anticomplementarity. Ill .1 reason we omwms § \ ‘ 33 E 3 Egg 59m :m ggw 8 o ems ES k°5 Peasoua's ”E; 02; ms .13.. OPTIONS c a. A ( g m < I? d > 3339 315.6 see gear (on; mok omh mm- AGGRESSIVE] coupennve - + amuwmm Mb) _ 3,3 , MANAGERIAL/ RESPONSIBLE - + eemwuoas (F-D) '- + cmwmumwymmue aamwmm * — (F3) '0' .- semerncmo/ BEBELLIOUS + '- aewwuons (H-e) + _ From Robert C. Carson, Interaction Concepts of Personality. Chicago: Aldine Publishing company, 1969, p. 146. initiates friendliness; when it is combined with dominance, friendliness still pulls friendliness while the dominance pulls submission.) When behaviors from the quadrand H-D are enacted with those from F-D and behaviors from the HsS quadrant are combined with F—S behaviors, Carson calls the resulting interactions anti-complementary and costly to both 17 participants as indicated by the four cells containing mutual costs (negative signs). Four other cells of the matrix--H-D vs. F-S, F-D vs. H-S, F-S vs. H-D and H-8 vs. F-D may or may not result in complementarity depending on whether or not the individ« uals accept the non-symmetrical component of the message; i.e., the hostility-affection factor. If one member of the dyad views the hostile behaviors of the other as "kidding" in spite of the unfriendly messages present in the inter- action, there may be room for negotiation regarding the affective quality that is to obtain in the relationship. A Docile, admiring response [F-S] to a Competitive, exhibitionistic display [H-D] represents an intersec- tion of behaviors that is not wholly lacking as a basis of relationship. (Carson, p. 147.) However, Carson also notes that such interaction patterns often degenerate into a struggle for the superior position. The remaining four cells of the matrix—-those con- tained in the diagonal running from the top left corner to the bottom right corner-—might best be described as having pure correspondence or symmetry. Carson does not discuss these outcomes, but his discussion of all the other combi- nations implies that they would probably be unlikely to occur or, if they did occur, they would be of low mutual reward. Using these four quadrants, Carson focusses on the concepts of interpersonal power and dependence. Drawing 18 on two notions of power advanced by Thibaut and Kelley and on one set forth by Jones and Gerard (1967), he defines power as "the ability to produce variability in another person's outcomes. Theoretically, any need in the other person can be used as a basis of power over him, provided only that one has the resources to affect the extent to which the other's need is met.” (pp. 167-8.) This is illustrated by Jones and Gerard's contact control and by Thibaut and Kelley's fate control and behavior control. Under the first type of control, one person gains high outcomes just from the other person's availability for interaction; e.g., the other is the only one who will con- sistently reward the first person's particular behavior. Under the second type of control, one person is uncondi« tionally dependent upon the other's behavior; i.e., the first person's "fate" is dependent upon the other person's whim in choosing to enact or not enact complementary rewarding behavior. Under the third type of control, one person's outcomes are determined, not by the independent behavior of each, but rather by particular conjunctions of the behaviors of each; i.e., by varying his own behavior, the first person can motivate the other person to make corresponding changes in his behavior. When the distribution of power in a dyad is unequal, the high-power person is less dependent upon the other than the other is upon him, since he has more to give or to take 19 away. When power is equally distributed, each member wields high power over the other and each has counterpower to blunt the other's demands--creating an interdependent dyad that is, for Thibaut and Kelley, a highly cohesive one. Since the low-power person in an unequal dyad lacks strong counterpower, he has to follow one or more avenues to alter the unfavorable balance of power in the relation— ship. He may comply with the high-power person's desires, thereby giving him (the low-power person) more predictable, if only very slightly changed, outcomes. He may also raise his CI by proposing to cultivate more attractive alter- alt native relationships, thus possibly and probably forcing the high-power person to eliminate outcomes that are least favorable to the low-power person. Other avenues available to the low-power person are a distortion of the value of outcomes received, development of his own resources for counter power, and propagandistic enhancement of his own behavior products. For example, he may distort the value of outcomes received by concealing the true effects of the high power person's behavior; i.e., outcomes that are highly positive may be treated as routine or may even be devalued, while the effects of more negative outcomes may be minimized or exaggerated by the low-power person. By distorting the values, the low-power person can reduce the power of the high-power person by "reducing the range of outcomes through which the latter believes he 20 can move the person of lesser power.” (Carson, p. 166.) The low-power person can also develop his own resources; he can emphasize those of his behaviors which are valued by the high-power person and therefore provide himself with additional counterpower. Finally, another strategy available to the low-power person for altering an unfavor- able balance of power £3 to propagandize his own behavior products. Through appropriate and effective communication, he may sell himself and his behaviors and persuade the high-power person that the behavioral outcomes in the interaction are higher or more rewarding than the facts warrant. 1 Given these varieties of power distribution in dyads and the postulate that people seek to maximize their outcomes and minimize their anxieties, Carson reviews the importance of norm development in the dyad by presenting Thibaut and Kelley's formulations. In their judgment, according to Carson, norms or behavioral rules, "whose acceptance is shared in some degree by both members of the dyad," (Carson, p. 174) have advantages for severaltypes of dyads. For dyads in which power is distributed un- equally, the low-power person may invoke normative regula- tions tx> blunt the use of power by the high-power person; the high-power person, by the same token, may invoke norms to achieve desired outcomes from the low-power person in order to reduce the costs of using many of his own power 21 options. For dyads in which power is distributed equally, procedural rules may be adopted to regulate the use of power by each person and minimize confrontations. The development of norms, however, is based primarily on the interdependence of the members of the dyad and the pattern of outcomes to which such interdependence leads. In other words, norms are most likely to be formed under conditions in which the outcomes in an interdependent dyad are im- perfectly correlated over the various cells of the interaction-outcome matrix. Such a matrix will contain some cells in which the outcomes to the two parties are relatively correspondent, and some in which they are relatively discrepant. A relationship of this type contains structural elements of both "cooperation" and "competition," and is therefore sometimes called a mixed-motive relationship. (p. 178.) When a class of these norms and/or other norms applies "to a person's behavior with regard to some speci- fic external problem or in relation to a special class of other persons," we have a role. (Thibaut and Kelley, 1959, p. 143.) Since a role or set of behavioral rules can be coded in terms of the categories of the interpersonal be- havior circle, peoples' personalities may be explained as a function of the roles they enact with others. As Carson notes, when a person is acting in role. . .his role behavior is supported by normative rules, and corresponding normative rules usually specify also the appropriate reciprocal or counter-role behavior. This ”appropriate" counter-role behavior, perhaps not surprisingly, is 22 almost always very closely related to what we have been calling complementary behavior. (Carson, p. 182.) If formal role assignments are absent or unclear, implicit contracts may be formed to regulate the dyadic members' interaction. These contracts are often found in situations that are regarded as "sensitive” or "embarrassing" for the participants. They are seldom, if ever, discussed and yet the behaviors of the individuals are quite regulated by the implicit rules. Other implicit contracts may be found in dyads that have just formed. Here, the rules emerge from the initial, tentative behaviors enacted by the members. If the participants find these initial behaviors rewarding, they will probably form "a more or less binding agreement as to how their relationship is to proceed for at least some period of time into the immediate future.” (p. 187.) Summary Personality, as conceived by Carson, is an inter- action phenomenon that varies with the individuals inter- acting and the rewards available to each from the types of behavior enacted. The major postulate of this theory is that rewards will be the highest and mutual when the behaviors of the dyadic members are complementary. By collapsing the octant categories of the interaction be- havior circle into four, Carson argues that certain quadrant 23 combinations will tend to occur more often than others. For example, Aggressive/Sadistic and Competitive/ Narcississtic behaviors (categorized as the H-D quadrant) will tend to initiate or pull Rebellious/Distrustful and Self-effacing/Masochistic behaviors (those categorized as the H-S quadrant) because these combinations are viewed as rewarding for both individuals in the dyad. If, however, H-D behaviors were to pull F-D behaviors (those which may be Managerial/Autocratic or Responsible/Hypernormal), the lack of complementarity would be viewed as costly for the individuals because the interaction would most likely pro- duce anxiety. Any of these combinations, of course, will be dependent not only on the individuals' predispositions to respond to others in certain ways, but will also be governed by role regulations or by norms established by the members of the dyad. Because this theory explains why certain behaviors in a dyad might be enacted and attempts to predict what behaviors will be initiated by other behaviors, it is felt that it will increase the predictability of relational communication patterns. To that end, the following chap- ters will seek to: (1) review the literature that concerns complementarity; (2) examine the literature that deals with relational communication and attempt to integrate its concepts into Carson's theory; (3) examine the importance 24 of situational factors on the interaction phenomena of personality and relational communication; and (4) develop a series of testable hypotheses. CHAPTER III A REVIEW OF THE LITERATURE CONCERNING NEED AND BEHAVIOR COMPLEMENTARITY The complementarity of behaviors discussed in the previous chapter concerns those personality traits that are manifested in interaction. This complementarity comes from a theory whose most basic postulate is hedonism; i.e., rewarding or costly interaction outcomes. A variant of this complementarity is that pertaining to general needs. This latter type is based on a need schema articulated by Murray (1938). Expressed in terms of needs, this comple- mentarity has been used to explain why one person chooses another with whom to interact. Mbre specifically, it has been used to explain interpersonal attraction, mate- selectiOn, and marital happiness.~ Since the two types of complementarity deal, in some way, with personality, this chapter will review studies and literature concerning both. Probably the most outspoken proponent of complemen- tarity in selective association has been Robert F. Winch. In his examination of 25 young married couples (1955a, 1955b, Winch, Ktsanes and Ktsanes, 1954), he found qualified support for the prediction that mates would select each other on the basis of need-complementarity. Using data 25 26 obtained from a structured interview, case history inter- view, and a TAT by which raters assigned respondents' answers to need areas, Winch and his associates examined interspousal correlations across the need variables of abasement, achievement, approach, autonomy, deference, dominance, hostility, nurturance, recognition, status aspiration, status striving, succorance, anxiety, emotion- ality, and vicariousness. The number of correlations which were significant in the direction predicted by Winch's need theory exceeded the number expected to occur by chance, a finding which led the authors to conclude that "persons like our subjects tend to select mates whose needs are complementary rather than similar to their own." (Winch, Ktsanes and Ktsanes, 1954, pp. 247-8.) The hypotheses tested by Winch and his associates and by later researchers are formulations of two types of complementarity. Type I complementarity involves need gratification where the needs are the same for one person as they are for another, but one's needs differ in inten- sity. For example, one individual may be high in dominance and finds telling others what to do a gratifying experience. When he interacts with a person low in dominance who enjoys being told what to do, both individuals' needs are grati- fied (or, stated in terms of Carson's framework, the rewards to each are higher than their costs). Type II involves need gratification of different types; 27 If A is highly ascendant, we should expect A to be more attracted maritally to B who is submissive than to C who, like A, is ascendant. If A is somewhat sadistic, we should expect A to be more attracted maritally to B who is somewhat masochistic than to C who is sadistic. (Winch, Ktsanes and Ktsanes, p. 242.) Attempts to replicate and/or further investigate the complementarity of needs among engaged and married couples have led to inconsistent and negative findings. Bowerman and Day (1956) examined engaged couples for need» complementarity and found no support for either a similarity or complementarity hypothesis. Schellenberg and Bee (1960), like Bowerman and Day, measured needs by the Edwards Per- sonal Preference Schedule and investigated the complemen— tarity of these needs among both married and engaged couples. Like the Bowerman and Day study, too, they found no support for the theory of complementary needs. Murstein's (1961) study of newlyweds and middle«aged married couples produced only one relationship that was significant in support of the complementarity hypothesis. Among the nonnewlywed interspousal correlations of EPPS scores, dominance was found to be complementary (r r -.42). Unlike the previous studies which tested both types of complementarity, Murstein investigated Type I need com- plementarity. Contrasted to this non-supportive evidence are studies by Kerckhoff and Davis (1962) and Rychlak (1965). The former study examined college couples who had indicated 28 they were seriously considering marriage. The authors predicted that (l) the degree of value consensus is posi* tively related to progress toward a permanent union and (2) the degree of need complementarity is positively re- lated to progress toward a permanent union. (Kerckhoff and Davis, p. 297.) Breaking couples into "short-term" (those who had gone togehter less than 18 months) and "long-term" groups (those who had gone together for 18 months or more), they found that value consensus signifi- cantly related to progress toward marriage for the short- term group only and need complementarity was significantly related to progress for only the long-term group. Their interpretation of these results is that "filtering factors” operate in mate selection at different stages of the selection process. At the early stages of the relation- ship, similarity of social status variables like class, religion, etc. is of major importance. At later stages, value consensus is of most importance. At still later stages, need complementarity becomes important. Rychlak's study of males participating in a manage- ment training session investigated need similarity, need compatibility (an example of Winch's Type II complemen- tarity), and need incompatibility. Using personality traits from Edwards PPS, he compared similar needs for dominance, nurturance, exhibition, order, and autonomy, compatible needs expressed in the combinations of 29 dominance/deference, nurturance/succorance, and exhibition/ affiliation, and incompatible needs of order/change, dominance/aggression, and autonomy/abasement. Subjects worked in groups of six on a manufacturing and then a dis“ cussion problem. They then indicated sociometric choices for individuals they would most and least like to have as a supra-ordinate (boss), a subordinate (employee), and a peer (neighbor). Results showed no proof for similarity of needs, but significance was found for nurturance/ succorance under need compatibility and for order/change under need incompatibility. Both these latter findings, however, varied depending on the role relationship. Speci- fically, it was found that highly nurturant subjects chose highly succorant individuals as potential neighbors; the same compatibility was not found for other role relation— ships. The incompatibility of need for order versus need for change was found for only the subject-neighbor rela- tionship; i.e., subjects high in need for order were most likely to prefer a neighbor high in need for change. In the subject-boss relationship, subjects high in need for order preferred a boss who would be low in need for change. As indicated by the studies cited above, experimen- tal evidence for need complementarity in mate selection and interpersonal attraction as postulated by Winch is inconsistent and weak. The major support for Winch's theory 30 of complementary needs comes from his own study. Results non-supportive of it come from a variety of individuals and settings. The reasons for such inconsistency have been ex- plained from three different perspectives by three differ- ent writers. The first perspective related to different experimental designs and instruments. As articulated by Winch in reference to Bowerman and Day's study which was nonsupportive of the need-complementarity hypothesis, the issue of different research designs between the two studies is important. As he states: the design of Bowerman and Day differs very appreciably from mine. The major difference concern: sub'ects (they used dating couples rather than married coupIes); variables (more than half of their correlations in- volved variables not used in my study); data-gathering procedure (they used the Edwards PPS, which--though ingeniously conceived--has no known validity for mea— suring needs) and even concept of complementariness (they hypothesized that every variable ought to cor- relate interspousally in a positive direction with every other variable). (Winch, 1958, p. 108.) It should be noted that the differences in design that Winch refers to (particularly the use of the Edwards PPS) are also characteristic of the other nonsupportive Studies described earlier. A different explanation for the lack of support is stated by Levinger (1964). His perspective is based more on conceptual rather than methodological issues and his assessment of Winch's hypotheses and the research 31 nonsupportive of it concerns the distinction between com- plementarity and similarity, instances of hypothesized complementarity, and internal versus external sources of need gratification. The first issue Levinger deals with concerns the way Winch has defined complementarity. As indicated earlier, Type I complementarity refers to similar needs that are different in intensity; Type II complementarity concerns needs that are different in kind. According to Levinger, these definitions present sufficient conditions for a complementary relationship but "it leads one errone- ously to the conclusion that if A's and B's needs differ neither in kind no; in intensity, then the relationship is not complementary." (Levinger, p. 154.) He argues that when A and B have a need that is the same, is equal in in- tensity, and is possessed by both in moderate quantity, complementarity can still be said to exist. It is possible, he further states, that the negative findings of the Bower- man and Day (1956) and the Schellenberg and Bee (1960) studies are a function of this particular situation. For any given need, if in eneral both partners should indicate it in moderate inten51ty, then the correlation between the partners' amounts would be reduced artifi- cially to zero, and yet there would be complementarity, indeed. (Levinger, p. 155.) The second issue with which Levinger deals concerns the domain of complementarity. It is his argument that the 32 identification of various needs for either Type I or Type II complementarity is not based on a clear theoretic rationale. For example, Type I complementarity of the need for dominance is sensible; one partner seeks to in- fluence and control the behavior of the other and the other seeks to be influenced and controleed. However, complementarity of the need for achievement is more diffi- cult to understand. As he states, it is obscure why A and B would be better off when one is high and the other low on need achievement, as op- posed to both being either high or low. For a maximal coordination of their efforts, maximum similarity of achievement motivation might well be most advisable. (p. 155.) The third issue concerning the need complementarity hypotheses refers to the source of gratification of any person's needs. A marriage partner may find that a parti- cular need or series of needs is not being met by his spouse but is being satisfied by some of his interactions with others. In short, there is a substitutability of sources of need gratification. Because this condition exists in a marital relationship, studies dealing with need complementarity should specify the distinction between need gratification within the marriage and need gratifica- tion in all other situations. Levinger argues that al- though Winch, e£_gl. mentioned the different ways in which marriage partners could satisfy their needs, the analyses of the data (1954, 1955a, 1955b) did not seem to distinguish 33 between "within" and "outside the marriage" need gratifica- tion. Furthermore, he states that the studies by Bowerman and Day (1956) and Schellenberg and Bee (1960) that utilized Edwards PPS virtually ignored this need gratification dis- tinction. In addition, since the EPPS was designed to measure needs of a student in a general college environ- ment, the negative findings of these researchers merely demonstrate that marriage partners do not show any consistent cor- relation between such eneral needs. These studies can hardly claim to test the complementarity of those needs in the specific marriage situation. (Levinger, p. 156.) In support of his argument, Levinger refers to a study by Katz, Glucksberg and Krauss (1960) in which they examined complementarity of needs among married couples and the relationship between total satisfaction and scores on a modified version of the EPPS. They found that when items describing general need for nurturance and need for succorance had been rewritten to pertain specifically to spouse relations, they found their most significant find- ings of complementarity. It is Levinger's conclusiOn that until investigators take account of this issue more directly in their measurement of specific needs, there will be no further evidence to support the original hypothesles] by Winch and his colleagues. (p. 156.) The final explanation of why results regarding the need complementarity hypotheses have been inconsistent has 34 been set forth by Rosow (1957). Rather than attacking methodological weaknesses or operationalization procedures of the Winch studies and the study by Bowerman and Day, he focusses on the problem of conceptual clarity concerning need syndromes. Like Levinger, he is concerned that any prediction from Winch's ideas of what needs will be com- plementary is based on shaky theoretical ground. What is necessary, he feels, is a methodology of global personality types. As he states, constructed personality types offer clearer theoretical guides in specifying the complement of a need in any given case. Otherwise, without some idea of the need's function in a personality organization, the chance of correctly predicting the complement of a need rating is seriously reduced. We might inadvertently (and with no check) observe only that portion of a whole need system which happens to be tapped by the situation under examination. . . Clearly, such a broader context can enrich the item-analysis of complementary needs. With some conception of the overall personality picture, we are better able to assess the meanin of observed need fits in the marriages of various personality types. (Rosow, pp. 227-8.) According to Berscheid and Walster (1969), studies by Ktsanes (1955) and R005 (1956) that have used such a holistic approach in analyzing Winch's original data have supported the complementarity of needs. As indicated by the previous review of the litera- ture concerning the need complementarity hypotheses, there is not overall support for all the variables tested. How- ever, there is support for several of the individual hypotheses. For example, Type I complementarity for . 35 dominance was found by Murstein (1961) and Type II comple- mentarity for nurturance/succorance was found by Rychlak (1965) and by Katz, Glucksberg and Krauss (1960). Given these findings, one can readily accept Ktsanes' (1955) comment that "the complementarity need hypothesis is a mOre complicated principle than the mere principle of 'opposites attract.'" (p. 551.) If the previous literature review were to be the sole basis for rejecting or accepting the basic hypothesis of complementarity, Carson's theory would have little sup- port. However, one should note that the research mentioned earlier focussed on complementarity of needs as measured only by pencil and paper tests. Although Winch's and Carson's notions of complementarity have some similarity-- e.g., certain personality variables tend to be the most gratifying or rewarding when those of one individual com- plement those of another-~Carson's focus is on interaction behaviors and he is better able to predict what behaviors will be most often complementary and, therefore, most re- warding. As discussed in Chapter II, the framework from which he predicts is based on concepts presented by Leary (1957) and Thibaut and Kelley (1959). Predictions based most heavily on Leary's interpersonal behavior circle have led to research that has had, as its focus, the complemen- tarity of specific types of behaviors in given situations. 36 Research conducted by Leary and his associates at the Kaiser Foundation (Leary, 1957) supports the notion that behavior offered by one individual will complement that of the other individual when that behavior falls along the dominance—submission axis of the behavioral circle. Also, there is support for the view that behaviors along the hostility~affection axis will be more similar than complementary. Although the evidence on which this support is based is not drawn from a specific study, evi- dence from various psychotherapy groups led Leary to con- clude that Interpersonal reflexes [behaviors] tend (with a prob- ability significantly greater than chance) to initiate or invite reciprocal interpersonal responses from the 'other' person in the interaction that lead to a repe- tition of the original reflex. (Leary, p. 123.) In addition, Leary's note that exceptions to this prob— ability can be found in individual interactions, particu- larly those where aggression leads to counter-aggression, lend some support to the notion of symmetry for behaviors that fall along the hostility-affection axis. A more specific test of the Leary notion of com— lementarity was done by Heller, Myers and Kline (1963) in a client/interviewer setting. Using the two axes of the behavior circle, these authors tested five hypotheses. For the dominance or control dimension they hypothesized that (1) "Dominant client behavior will evoke dependent 37 interviewer behavior” and (2) ”Dependent client behavior will evoke dominant interviewer behavior.” (p. 117.) For the hostility-affection or affect dimesion they hypo- thesized that (l) "Hostile client behavior will evoke hostile interviewer behavior," (2) "Friendly client behavior will evoke friendly interviewer behavior,” and (3) Hostile client behavior will evoke interviewer anxiety. Subjects were 34 interviewers-in-training; clients were actors who role played either a dominant-friendly, dominant-hostile, dependent-friendly, or dependent-hostile client. Dependent measures of interviewer affect and control were obtained from observers using a modified form of the Leary Inter« personal Check List, the items of which were made interview specific. Interviewer anxiety was measured by an observer using an investigator-constructed behavioral anxiety check list. Results from a treatment by treatment by subjects ANOVA supported the control and affect hypotheses, leading the authors to conclude that clients may evoke reciprocal behaviors from their therapists even though this influence may not be per— ceived . . . [and that] evoked behavior should also be considered as a function of the real stimulus qualities of the therapeutic interaction. (p. 121.) One further test of Leary's complementarity is that of Altrocchi (1959). His study examined the relationship between the degree of dominance in an individual, the degree of dominance in people to whom he is attracted and the 38 degree of dominance he attributes to others. Two specific hypotheses were tested: (1) People choose to interact with people complementary to themselves in dominance and (2) Dominant people perceive other people as less dominant than submissive_ people perceive them to be. (p. 303.) Ninety-six male subjects viewed a movie of an interaction among four male social objects, each of whom was selected because he was different in dominance but similar in lik- ability. Subjects rated these social objects on a five item choice questionnaire (indicating with whom they might wish to interact at a later time), rated themselves and the objects on Leary's Interpersonal Check List and completed Gough's Dominance Scale. Results supported neither of the hypotheses. Altrocchi's conclusion is that people differenti- ally influence others and affect interpersonal choice and perception. Since influence results from interaction, a lack of it between subjects and objects, as was the case in this study, leads to a lack of perceptual cues regard- ing the types of behavior that may be expected. When interaction does occur, a person may influence others to behave in ways complementary to him and, furthermore, he is justified in perceiving them as complementary to himself, his deviant perceptions of them not necessarily being misperceptions. (p. 307.) A more recent study of complementarity is that by Peabody (1970). Fifty subjects were asked to assume that 39 each of 12 different persons (P) had a characteristic that had resulted from a reaction to the characteristic of some other person (0) acting toward him. The 12 adjectives used to describe these 12 persons had been drawn from a list of needs described by Murray, g£_21. (1938) and included such terms as dominating, submissive, aggressive, and friendly. Subjects were asked to select, from a list of 14 similar or dissimilar adjective characteristics, those which would most reasonably describe person 0. Results support the Leary and Carson predictions concerning complementarity of dominant-submission behaviors and symmetry of hostility-affection behaviors. For example, when the characteristic of P was aggressive, subjects chose to describe O's characteristics as aggressive, incone siderate or unfriendly. If P was friendly, subjects viewed 0 as friendly, relaxed and flirtatious. When P was domi- nating, O was considered submissive and dependent. When P was submissive, O was described as dominating or aggres- sive. For the symmetrical characteristics along the hostility-affection dimension, 88 to 100% of the subjects chose the same or a similar characteristic to describe 0's possible reaction to P. For the complementary characteris- tics along the dominance-submission dimension, only 4 to 30% of the subjects chose characteristics for O that were the same or similar to that prescribed for P. In short, there is support for the proposal that certain 40 characteristics in interpersonal relations are complementary and others symmetrical. Summary and Conclusion The review of literature dealing with the Winch and Leary notions of complementarity indicates that the latter formulation has slightly more conclusive support than does the former. More specifically, research has es- tablished that, although there is not overall support for the basic hypotheses of complementarity formulated by Winch and his associates, there is support for Type I complementarity for dominance and there is support for Type II complementarity for nurturance/succorance. For Leary's notion of complementarity, studies have shown that behaviors categorized close to the hate-love axis of his interpersonal behavior circle tend to invite similar or correspondent behaviors, while behaviors categorized close to the dominance-submission axis tend to initiate recipro- cal or complementary behaviors. BecauSe the complementarity that Leary deals with is more Specific in terms of what behaviors one would ex- pect to complement others, the argument that Carson sets forth concerning complementary behaviors seems reasonable in the light of both experimental support and the afore- mentioned weaknesses of the Winch complementarity theory. CHAPTER IV A REVIEW OF COMPLEMENTARY BEHAVIORS AND LITERATURE CONCERNING SYMMETRICAL AND COMPLEMENTARY RELATIONAL COMMUNICATION Carson's notion of the complementarity of behaviors as categorized by the quadrants of the interpersonal be- havioral circle has been discussed as being more rewarding than costly along the dominance-submission factor for in- dividuals interacting. It was also pointed out that Carson's personality framework considers behaviors along the hostility-affection dimension to be correspondent. Both complementary and correspondent behaviors have been referred to only in general terms; e.g., a person who exhibits friendly-dominant behavior will find it more rewarding to interact with a person who is submissive and friendly than he would find interacting with another dominant individual. Little has been said to specify what these behaviors might be, either in terms of general be- havioral characteristics or in terms of communication characteristics. Because the purpose of this study is to increase the predictability of the relational communica- tion model through Carson's theoretical framework, the specific types of behaviors one should expect and the 41 If u .3. U i 42 relationship between complementarity of these behaviors and relational communication have to be specified. In order to isolate these behaviors and explore this relationship, this chapter will re-examine Leary's interpersonal behavioral circle and the relational communication concepts of symmetry and complementarity through a review of the relevant litera- ture . Complementary Behavior Characteristics The descriptions of the behaviors in the octant categories of the interpersonal behavior circle as found in Carson (1969) parallel those found in Leary (1957), although the latter places more emphasis on describing clinical psychiatric cases. Because Carson's descriptions are framed more in terms of normal interpersonal behaviors, his descriptions will be cited. Mana erial-Autocratic Behav1or This behavior is characterized by strong, assertive and confident elements which imply leadership, power or expertise. When enacted by an individual, such behavior has a friendly component, communicates the message, "I am a strong, competent, knowledgeable person on whom you may rely for effective guidance and leadership,'" (Carson, 1969, p. 107) and tends to invite Self-effacing and Docile 43 behavior. In other words, dominance pulls submission. When extreme, this behavior is suggestive of a dogmatic or pedantic individual. Responsible-Hypernormal Behavior This behavior is one of positive affect that com- municates the message, "I am a strong, competent, empathic person on whom you may count for understanding and emotional support." (Carson, p. 109.) There is an element of domi- nance within the friendliness and when initiated, it tends to invite Docile and Cooperative behavior. Extreme forms of this behavior are characterized by pity, soft-hearted- ness, and self-sacrifice. Cooperative-Overconventional Behavior Affection, cooperation and compromise characterize this behavior. When enacted, it suggests positive and affiliative tones and invites others to enact Responsible and Cooperative behavior. The message communicated is, "'I am an exceedingly friendly, agreeable, unchallenging person who would like you to like me.'” (Carson, p. 109.) The extreme form of this behavior involves conventionality and effusiveness. 44 Docile-Dependent Behavior Submissiveness characterizes this behavior although there is an element of friendliness. It conveys, "'I am a weak and helpless person in need of your aid and your sup- port,'" (Carson, p. 109) and invites Managerial and Respon- sible behavior. When extreme, it is characterized by ingratiating dependency. Self-Effacinngasochistic Behavior Submissiveness also characterizes this form of be- havior but, unlike the positive effect which is found in Docile-Dependent behavior, this form has elements of aggression and hostility. It communicates the message, "'I am a weak, deficient, unworthy person justly deserving of your domination, rejection, and contempt.'" (Carson, p. 110) and invites others to be Aggressive and Competitive. Extreme forms of this behavior involve groveling weakness and self-condemnation. Rebellious-Distrustful Behavior Hostility, disaffiliation, passivity and powerless- ness are elements that comprise this behavior.‘ It communi- cates the message, "'I reject and mistrust you, for you are, or are certain to become, unworthy of my affection and esteem,'" (Carson, p. 110) and invites Aggressive and 45 Competitive behavior. Cynicism, bitterness and wariness are also found along this range of behaviors, as are marked suspicion and rejection of authority when the behaviors are of an extreme form. ggfiressive-Sadistic e avior This behavior involves hostility, such as sarcasm and criticism, expressed with a dominant tone and invites others to engage in Self-effacing and Rebellious behavior. The message communicated to others is,"'I am a threatening and dangerous person, and you are a suitable target for my wrath.'" (Carson, p. 110.) When enacted, this behavior tends to inspire anxiety and fear in others. When in its extreme form, it involves cruelty and sadism. Competitive-Narcississtic Behavior Dominance is one element of this behavior, as it is an element of Managerial-Autocratic behavior. However, unlike the latter which is characterized by friendliness, this behavior has an aggressive component. It communicates the message, "'I am superior to you, and you, being a lesser person, are hardly worthy of my serious considera- tion,'” (Carson, p. 111) and invites others to engage in Rebellious and Self-effacing behavior. Its extreme form is characterized by boasting and exhibitionism. 46 As noted in Chapter II, Carson collapses these eight categories into four, using as his descriptors, the dimension names. The resulting quadrants of hostile dominance, friendly dominance, friendly submission and hostile submission lead to outcomes that reflect both symmetry along the hate-love dimension and complementarity along the dominance-submission dimension. However, this symmetry and complementarity are relationships which occur between personality traits or reflexes, reflexes that en— compass gross elements of human behavior. Because the elements referred to be the previous descriptions of cate« gory behaviors are not described in terms of communication, the remaining part of this chapter will be devoted to an examination of symmetry and complementarity as conceived for and found in the literature dealing with relational communication. Relational Communication Concepts In general, the notion of relational communication pertains to the messages that individuals in a dyad use to define their relationship. These definitions concern the respective role each individual is to play at a given moment in the interaction and are an index of control each individual has and/or is given by the other individual. Depending upon whether or not both individuals are compet- ing for control of the relationship, or one seeks and the 47 other gives him control, the resulting relationship may be either symmetrical or complementary. Probably the first reference to symmetry and com- plementarity as it pertains to relational communication is found in Gregory Bateson's Nayen, first published in 1936 and revised in 1958. Working among the Iatmul natives of New Guinea, Bateson noted the process of role differentia- tion, a process he termed schismogenesis. Formally defined, schismogenesis is ”a process of differentiation in the norms of individual behavior resulting from cumulative interaction between individuals." (Bateson, 1958, p. 175.) This differentiation may be complementary (for example, assertiveness on one individual's part is replied to with submission by another individual), it may be symmetrical (for example, boasting by one individual is met with boasting by another individual), or it may be a mixture of both. When the latter occurs, the change from a comple- mentary to a symmetrical relationship or the change from a symmetrical to a complementary relationship for even a short time will help to stabilize the original relationship. [This phenomenon is discussed by later writers as homeo- stasis, a view attributable to these psychotherapists' emphasis on a systems approach to family and dyadic inter- action.] The particular roles that individuals play in either a symmetrical or complementary relationship are (‘03 48 established and maintained through each individual's reac- tions to the other. Although this implies that communica- tion is one of the processes by which these relationships are defined, Bateson says little in Néyen to specify why and how such role differentiation occurs. In 1951, Bateson articulated the importance that communication has for defining relationships by postulating that every message has two sorts of meaning, a report about events at a previous moment and a command or stimulus for events at a later moment. (Ruesch and Bateson, 1951, p. 179.) In discussing this in terms of defining a relation- ship between individuals, Bateson notes that when A speaks to B, whatever words he uses will have these two aSpects: they will tell B about A, convey- ing information about some perception or knowledge which A has [report]; and they will be a cause or basis for B's later action [command]. In the case of language, however, the presence of these two meanings AVE may be observed by syntax. A's words may have the syntax of command, which will partly obscure their re- port aspects. 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 of re- port, and B may fail to notice that this report has influenced him in a certain direction. (Ruesch and Bateson, p. 180.) The importance that this postulate had for the study of interaction is demonstrated by the emphasis that certain psychotherapists have placed on role relationships. Most of the writers dealing with the relationship concepts of symmetry and complementarity have been associated, in 49 some way, with Gregory Bateson and the staff of the Mental Research Institue, a behavioral science research foundation organized in the San Francisco Bay area in 1958. One of these, Jay Haley, discussed these concepts in terms that are very similar to those used by Bateson. For Haley, a symmetrical relationship is one characterized by exchanges of similar behavior. "This type of relationship tends to be competitive; if one person mentions that he has succeeded in some endeavor, the other person mentions that he_has succeeded in some equally important endeavor. The people in such a relationship constantly emphasize their equality to, or symmetry with, the other person." (Haley, 1958, p. 44.) A complementary relationship involves one person giving and the other receiving. In such a relationship, one individual is in a superior position and the other is not. A 'superior' position means that the person initiates action and the other follows that action; he offers criticism and the other accepts it, he offers advice and the other assumes he should, and so on. In such a relationship the two people tend to fit together or complement each other. (Haley, p. 44.) [Haley, also points out that when one person lets or forces another person to define the relationship, the former indi- vidual is at a higher level and is defining the relation- ship as complementary. Such a relationship is termed meta-complementary.] 50 These relationships are defined by what indivi- duals say and how they qualify what they say. The former refers to Bateson's notion that every message communicated between two people will define the relationship "if only by expressing the idea 'this is the kind of relationship where this sort of thing is said . . .'" (Haley, p. 44.) The latter, the qualification of what is said, refers to maneuvers or messages about the relationship. As Haley states: Maneuvers consist of 1) requests, commands, or sugges- tions, that another person do, say, think, or feel something, and 2) comments on the other person's communicative behavior. Should A ask B to do some- thing, then B is immediately posed the problem of whether this is the sort of relationship where A has the right to make the request. B is also affected by whether the request was made tentatively or apologeti- cally, or whether it was a rude command. Since the relationship is in question, B must either do what A says and accepts [sic] A's definition of the relation- ship, or refuse to do it and thereby counter with a maneuver to define the relationship differently. He may as a third possibility, do what A says but qualify his doing it with a statement that he is "permitting" A to get by with this and therefore he is doing it but not agreeing with A's definition of the relationship. (Haley, p. 45.) These relationships and the tactics and techniques used in defining them are characteristic of all two—person systems. No relationship in a dyad will always be symmet- rical or complementary; shifts from one type of relation- ship to the other will occur at various points in time and may occur rapidly or progress slowly in one direction. Members of, what might be called, mature dyads often define 51 certain areas of their relationship in which each behaves either symmetrically orcomplementary. Members of, what might be termed, non-normal or pathological dyads often cannot or refuse to either define the relationship or accept the other member's definition. It is this latter type of dyad on which most of the later literature con- cerning these relationships has focussed. In particular, the major emphasis has been on dyads composed of members of families undergoing psychotherapy. Among those later writers dealing with these rela- tionship concepts and the importance they have for the analysis of family interaction are Jackson (1959), Jackson, Riskin and Satir (1961), Bateson and Jackson (1964), Watzlawick (1964), and Watzlawick, Beavin and Jackson (1967). For the most part, they conceptualize symmetry and complementarity the way Haley did and like him, they do not emphasize the operationalizations of these terms. Jackson, Riskin and Satir (1961) suggest that symmetry and complementarity may be identified by examin- ing the statements made by dyadic participants. More specifically, a symmetrical statement is a "comment on the equality of some aspect of a relationship." It is charac- terized by competitive statements. For instance, symmetri- cal statements that might occur in an adolescent peer relationship are "I can run faster than you," stated by one youth and ”I can run further than you," stated by the 52/ ,. ” other. A complementary statement, on the other hand, is "one that 'asks' or 'offers.' Arbitrarily, we call 'asking' the complementary one-down position and 'offering' the com- plementary one-up position. Complementarity may have to be judged in terms of content and in terms of following state- ments." (p. 323.) Aside from this definition, however, no specification is made as to the types or classes of state- ments that would fit the asking and offering categories. As defined by Bateson and Jackson (1964), symmetri- cal relationships are those characteristic of "rivalries and other relationships, where A is stimulated to do some- thing 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." Complementary inter- action includes, for example, dominance and submission, exhi- bitionism and spectatorship, succoring and dependence, and so forth--a series of patterns where there is a mutual fitting between A's behavior and that of B. (p. 270). Like the Jackson, Riskin and Satir definition, however, this does not specify the classes of patterns and, in addi- tion, does not identify what the patterns are. The authors do, however, provide a general focus for the observation of these relationships. Because the definition of the rela- tionship between the two individuals involves the behaviors of both, 53 there is, strictly speaking, no such thing as a comple- mentary 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. (Bateson and Jackson, p. 273.) 1! Aside from these general references to what units should be observed, most of the literature, as indicated earlier, concerns itself with definitions on a conceptual level and does not concern itself with operationalizing the relationship concepts. At present, there seem to be only two references that focus on ways to operationalize symmetry and complementarity: one is an article by Sluzki and Beavin (1965) entitled "Simetria y complementaridad: una definicion operacional y una tipologia de parejas" ["Symmetry and complementarity" An operational definition and a typology of dyads”]; the other is a dissertation by Mark (1970) entitled ”Parameters of normal family communi- cation in the dyad.” Sluzki and Beavin approach the operationalization of symmetry and complementarity by indicating their agree- ment with Jackson's (1959) and Watzlawick's (1964) concep- tual definitions of these terms. They also utilize a unit of analysis suggested, in part, by Bateson and Jackson (l964)--namely, two consecutive transactions, one by one participant and one by the second participant. Based on this unit, they provide the following general examples of symmetrical and complementary transactions: 54 Giving/Taking instructions: complementary (giving 1 up, taking 1 down) Asking/Answering: complementary (asking 1 down, answer- ing 1 up) Affirming/Accepting: complementary (affirming 1 up, accepting 1 down) Referential information/Referential information: symmetrical Accepting/Accepting: symmetrical Giving instructions/Countering with instructions: symmetrical (Sluzki and Beavin, 1965, p. 326). Taking every consecutive pair of message, each transaction can be labeled by giving the first and second messages a one-up, one-down or symmetrical value. For example, the conversation below would be categorized by the transaction values following it. Al What do you think? (Question) Bl I think that X . . . (Referential Statement) A2 Yes, I agree . . . (Acceptance) B2 Furthermore, I believe that X . . . (Referential Statement) A3 Now let's talk about X and Y. (Instruction) B3 No, let's talk about 2. (Negation and Instruction) A4 Okay, that Z is . . . (Acceptance and Referential Statement) B4 2 also is . . . (Referential Statement Transaction Values A1 Question C++ Bl Referential Statement (Answer) Bl Referential Statement (Answer) C++ A2 Acceptance A2 Acceptance C++ B2 Referential Statement (Amplification) 55 Transaction Values B2 Referential Statement (Amplification) S A3 Instruction A3 Instruction S B3 Negation and Instruction B3 Negation and Instruction CIT A4 Acceptance and Referential Statement A4 Acceptance and Referential Statement S B4 Referential Statement (Sluzki and Beavin, p. 326) The transactional values following each unit of analysis indicate what relationship was defined by the two consecutive messages from both members of the dyad. In the conversation outlined above, individual A begins the interaction by taking a one-down position through his question to individual B. B takes the one-up position by answering A's question, thus defining the first trans- action as complementary. B continues as one-up until A replies to B's referential statement with an instruction. A's instruction thus makes the relationship symmetrical. B, however, does not relinquish control and counters A's instruction with a negation and an instruction of his own. At this point, A defers to B's negation and instruction, takes the one-down position again by accepting the nega- tion, but tries to gain control by concluding his message with a referential statement. Finally, B responds to the 56 referential statement by matching it, thus re-establishing symmetry. As shown by this example, transactions may be labeled complementary or symmetrical depending on the con- trol defining aspects of a message and what a message is in relation to one preceding it. By isolating message content in terms of questions, referential statements, instructions or orders, denials or negations, agreements or affirmations, the authors operationalize complementarity and symmetry in terms of the amount of control these gram- matical forms have. The coding scheme presents a graphic representation of these relationships and is an index of how each member defines them for any given transaction. One weakness of this scheme is that it is not clear in its specification of the conditions under which grammati- cal forms are one-up or one-down. The authors do indicate that acceptance and asking are considered one-down and referential statements, answers and giving instructions are considered one-up. However, it is unclear as to whether a referential statement is always one-up or whether a ques- tion or asking is always one-down. Other weaknesses are that the coding scheme allows for only sequential messages from both individuals of the dyad, does not include ways to code sequential messages that are spoken by the same indi- vidual, and has no way of coding messages from both partici- pants that represent talking-over the other. 57 Mark's approach to operationalizing these concepts is based on Sluzki and Beavin's scheme and attempts to better deal with its weaknesses. Like them, Mark utilizes a transactional unit of analysis that includes consecutive messages from both members of the dyad. Unlike them, he is somewhat more specific as to what these messages are and how they are to be labeled. In addition, he provides more information on how this coding scheme indexes the relationship being defined by a particular transactional unit. For him, the individual message is "a 'talk' by an individual of any length, beginning with his first word, and continuing until the other party in the dyad speaks." (Mark, 1970, p. 42.) Each talk is coded as a three digit number, the first digit indicating the sex of the speaker, the second digit indentifying the kind of speech it is in terms of a modified grammatical format, and the third digit referring to what the speech is in relation to the speech that precedes it. The modified grammatical codes for the second digit are: question, assertion, instruction, orders, talking over, assertion and question, question and asser- tion, other (referring to non-fluencies) and laughter. The third digit includes the following: agreement, disagree- ment, extension, answer, disconfirmation, topic change, agreement and extension, disagreement and extension, other (coughs and vocal pauses) and laughter. (Mark, p. 42.) The coding categories he developed are as follows: 58 First digit code: Male = 1 Female = 2 Second digit code: question = l assertion = 2 instruction = 3 orders = 4 talking over = 5 assertion and question = question and assertion = other = 8 laughter \lO‘ 9 Third digit code: agreement = l disagreement = 2 extension = 3 answer = 4 disconfirmation = 5 topic change = 6 agreement and extension = 7 disagreement and extension - 8 other = 9 laughter = 0 (Mark, p. 42.) Operationally, Mark defines a complementary rela- tionship as that which involves the interchange of state- ments which put one participant in a one-up position to the other or in a one-down position to the other. His codes indicate that a complementary relationship may be defined by such types of preceding and following state- ments as: Question as extension (one-down), assertion in disagreement (one-up); assertion in agreement (one-down), order as topic change (one-up); and talking over in agree- ment (one-down), assertion in extension (one-up). Each statement is given an appropriate code from the second and third digit code scheme and the transactional value (whether it is one-up or one-down) for each code is dictated 59 by a series of rules. Symmetry is indicated by exchanges which are identical. For example, if one individual gives an assertion in extension and the other individual follows with an assertion in extension, this unit will be defined as symmetrical. His coding scheme enabled him to take a unit of speech from one participant of his husband-wife dyads and categorize it according to what it was in relation to the previous statement. In terms of the code, a statement by the wife that is a question in extension would be coded as 213; a reply by her husband that is an order in disagree- ment would be coded 142. The wife's statement would be coded as oneedown and the husband's statement would be coded as one-up. For that particular exchange then, the relationship would be complementary with the wife one-down and the husband one-up. As the coding scheme indicates, Mark enlarges the two basic relational categories--symmetry and complemen- tarity--to include several others. The nine possibilities he utilizes are these: one-up, one-down; one-up, one-up; one-down, one-down; one-up, symmetrical; one-down, sym- metrical; symmetrical, one-up; one-down, one-up; symmetri- cal, one-down; and symmetrical, symmetrical. Of these, Mark views only the one-up, one-down; one-down, one-up; and symmetrical, symmetrical as fairly stable patterns. The others tend to lead to an escalation of differing 6O viewpoints regarding who is to have control of the rela- tionship, patterns that are broken only by physical conflict, or by one individual leaving the field, or by one person taking a one-down position, in effect saying "I'll let you control me." The similarities and differences between Mark's coding scheme and that suggested by Sluzki and Beavin may be seen from the sample conversation below that is taken from their article. (1965, p. 326.) Al What do you think? Bl I think that X . A2 Yes, I agree . . B2 Furthermore, I believe that X . A3 Now let's talk about X and Y. B3 No, let's talk about 2. A4 Okay, that Z is B4 2 also is Message Descriptions Transactional Values S 6 B Mark S 6 B Mark A1 Question Question as other Bl Referential . . . CII II Statement Assert1on 1n extens1on Bl Referential . . . Statement Assert1on 1n exten51on C++ +1 A2 Acceptance Assertion in agreement A2 Acceptance Assertion in agreement CII 11 B2 Referential . . . Statement Assertion 1n exten51on B2 Referential . . . Statement Assert1on 1n extens1on S 1+ A3 Instruction Instruction as topic change A3 Instruction Instruction as topic change B3 Negation S +1 and In- Instruction in disagreement struction 61 . . Transactional Message Descr1pt1ons Values 8 8 B Mark S G B Mark B3 Negation and Instruction in Instruction disagreement A4 Acceptance and C++ II Referential Assertion in agreement Statement A4 Acceptance and Referential Assertion in agreement Statement S ++ B4 Referential . . . Statement Assertlon 1n extens1on As this comparison shows, the two schemes are similar only along general characteristics: both have grammatical forms that correspond to asserting or request- ing relationship control and both utilize sequential messages as the unit of analysis. Differences, however, exist not only in terms of the description of statements but also in terms of how two statements combine to form a relational or transactional value. The first difference, as evidenced by the message descriptions, is that Mark's descriptors more clearly in- dicate the grammatical form of the message and what it is in relation to the previous speaker's message. The second difference concerns message units that Sluzki and Beavin view as symmetrical. Whereas they consider the message units B2/A3 and A3/B3 to be examples of symmetry, Mark's scheme leads to the conclusion that these units are examples of competition where each member is trying to be one-up 62 over the other. As noted earlier, Mark views message units as symmetrical when the statements of the two participants are identical. This situation did not exist in the sample conversation above as shown by the message descriptions from his scheme. This is also evident when the numerical codes for this conversation are examined. Al Question as other = 119 B1 Assertion-extension = 223 A2 Assertion-agreement = 121 B2 Assertion'extension = 223 A3 Instruction-topic change = 136 B3 Instruction-disagreement = 232 A4 Assertion'agreement = 121 B4 Assertion-extension = 223 Only if the last two digits of each consecutive message had been identical--for example, 123 rather than 119 for the message Al from member A--would Mark's rules have led to a transactional value of SS, symmetrical. The narrowness of this operationalization is evi- dent when one examines the conceptual definitions of symmetry offered by Haley (1958), Jackson, Riskin and Satir (1961) and Bateson and Jackson (1964), to name a few. For the most part, they and other writers concerned with relational communication conceive of symmetry as similar and not necessarily identical behavior. By includ- ing only identical messages in his operationalization, Mark restricts his code and his chances for more correctly identifying, what most writers consider, symmetrical behavior. 63 Although it is not the purpose of this chapter to present the relational communication coding scheme developed for this proposed study, it should be noted that there is one additional weakness in Mark's scheme that should be corrected, if for no other reason than to clarify the operationalizations of the relational communication con- cepts. As discussed earlier, each digit of the three digit code refers to something different: digit one identifies the dyadic member, digit two refers to what kind of a speech or message it was in terms of a modified grammatical format, and digit three indicates what the speech was relative to the statement that came before it. The second digit, since it is more descriptive as to what a statement is in terms of a grammatical form, should be internally consistent. Likewise, the third digit code, since it refers to what a statement is in relation to a previous statement and involves a higher level of infer“ ence, should also be internally consistent. An examina- tion of the second and third digit codes, however, shows that the second digit is not internally consistent. Whereas the categories of questions, assertion, talking over, assertion and question, question and assertion, and laughter are easily identifiable at a descriptive level, the categories of instruction and orders are at a higher level and require inferences on the part of the observer. In order to maintain the internal consistencies of the 64 second and third deigits, the categories of instruction and orders should be placed in the third digit code. Such a modification will make each digit code more consistent in terms of what each is intended to index and, furthermore, will more clearly separate the levels of each digit. In addition, these changes may make the scheme more accurate in reflecting the control dimensions of messages, a factor of validity that would further increase the scheme's potential utility. Despite these weaknesses, Mark's development of the three digit code is important as the basis of an in- strument that indexes relational communication messages and patterns and does so efficiently and fairly reliably. [Mark checked inter-coder reliability over nine dialogues for three different types of reliability--a unitizing check to determine if coders agreed on the number of alternating speeches, a categorizing check to see if coders were using identical categories, and a total reliability check based on a summing of category reliability values over all nine dialogues. For the first reliability check, the value approached 100% with only two in two hundred and seventy five speeches not agreed upon. The second reliability check across dialogues had values ranging from 0.85 to 1.00. The total reliability check yielded a value of 0.9163. (Makr, 1970, pp. 55-6)]. Using his scheme to code lower and upper class husband and wife interactions, he found 65 different patterns being emphasized by each class. For example, upper class dyads displayed more symmetrical behaviors and tended to have more covert disagreements (disconfirmations) than lower class dyads. Lower class couples, on the other hand, tended to use more overt dis- agreements, more competition to speak (talking over in disagreement) and more silence. He also found that lower class pairs had a greater tendency to use complementary patterns than did the upper class dyads. Personality and Relational Communication Complementarity Having suggested the types of complementary and symmetrical behavior expected from Carson's personality framework, and having pointed to the operationalization of these concepts in terms of the relational communication framework, the similarity between the two approaches can now be examined. As discussed earlier, the complementarity of be- haviors in terms of personality varies depending on whether the initiating behavior of one individual falls along a hostility-affection dimension or a dominance—submission dimension. For example, when the initiating behavior has components of affection and dominance, the theory predicts that the initiated behavior will have components of affec- tion and submission. In more specific terms, 66 complementarity would be said to exist when a person who enacts strong, assertive and confident behaviors pulls help-inducing weakness and conforming behaviors from the other person. If this same complementarity of dominant«submissive behaviors is discussed in relational communication terms, a dominant individual could be viewed as the one who has the most control as expressed by his one-up messages; a submissive individual could be viewed as the one who has the most one-down messagesf This relationship between dominance and one-up messages has not, however, been es- tablished by previous research. Stating that a dominant individual is one who has the most control as expressed by his one-up messages is tautalogical, but whether or not a person's score on a dominance scale is related to his one-up messages remains a question to be answered by ex- perimental research. If this relationship holds, the dominant individual would probably have more assertions as directions and assertions as orders than would the sub- missive individual. This is not to say that all messages exchanged between such personality types will be comple- mentary, but there may be a significant tendency for the dominant person to express his dominance through one-up messages, and correspondingly, the submissive person will probably express his submissiveness through one-down messages. 67 This relationship between dominance-submission complementarity and relational communication is probably the strongest and most logical one that can be predicted from Carson's framework. As noted in Chapter II, Leary predicts six possible complementary relationships and Carson predicts four after collapsing the octants into quadrants. Since the notion of dominance is most consis- tent with the concept of control in relational communica- tion, the complementarity of dominance-submission may have utility as a predictor of relational or transactional patterns between dyads and one-up/one-down messages of individuals. It will be the purpose of the next chapter to specify these relationships and the predictions that may be derived from them, after discussing the importance of other interaction related variables. CHAPTER V DYADIC DIMENSION AND SITUATIONAL VARIABLES, OPERATIONALIZATIONS AND SOME HYPOTHESES As the previous chapters have indicated, Carson's formulation of personality is an interaction one, which incorporates concepts of hedonism and behavior complemen- tarity. Considered alone, his concept of complementarity has power to predict certain types of behavioral sequences, but the theory he develops needs to be supplemented with other factors to increase its predictability concerning relational communication messages and patterns or trans- actions. One general factor that would seemingly increase the predictive power of Carson's theoretic framework is that of the situation in which individuals find themselves interacting. Some support for this position is provided by Lewin in his formula of behavior, B = f(P,E) [behavior is a function of personality and environment]. As stated by Carson, the idea is that a person's behavior in any situation is jointly deter- mined by the characteristics of that situation [E], as he perceives them, and by the particular behavioral dispositions of which he is possessed at the time [P]. (Carson, p. 9.) 68 69 Within this general situational factor are usually encompassed such variables as the task on which individuals are working and the location in which they may find them- selves. Although these two variables are important for the description and prediction of one's general behaviors, since the former includes the rules by which the individuals will be governed and the latter includes such factors as space, setting and geographical environment, there are other variables that are probably as related to communica- tion behaviors and just as important in predicting them. As Rogers (1971) suggests in her paper on dyadic systems and interpersonal communication, some of these involve a temporal dimension: for example, history of the relation- ship and time allocated for certain interactions. Others involve an activity or task dimension: that is, functional integration and normative regulation. Since most of the research on relational communication has been done in either abnormal or normal family settings, variables appro- priate to husband-wife pairs will be discussed in order to derive predictions concerning relational messages and transactions in those dyads. Some Situational and Dyadic Dimension Variables History The variable of history of a relationship is useful in defining dyads about which predictions may be made 70 concerning communication behaviors. When individuals who have never met interact for the first time, much of their communication involves a mutual searching for a definition of the relationship. In addition, considerable attention is given, says Carson, to determining the position each will take vis-a-vis the other in respect to dominance and affection. If the relationship promises to be a workable one from the standpoint of the level of complementarity of proffered stances, and from the standpoint of the apparent intensity or flexibility with which they are held, then the parties may go on to explore further the outcomes that may become mutually available to them through further interaction. (Carson, p. 152.) It is presumed that if and when the relationship becomes more intimate over time, relationship patterns and comple- mentarity of behaviors will be stabilized through what Jackson (1957, 1959, 1965) and Watzlawick, Beavin and Jackson (1967) term homeostatic mechanisms. History thus becomes important in terms of the narrowing of the ranges. of behavior found in a dyad that will be judged as reward- ing by both participants. It would be expected that dyads with a history would have established rules regarding both their content and their role relationship communication and would have stable and consistent patterns of inter- action. Because husband-wife dyads are characterized by intimacy and history, predictions about relational com- munication will be formulated around this type of dyad. 71 Task Another variable that is important for predicting communication is the task in which individuals are involved. In other words, the purpose for which the two individuals are communicating is expected to have some effect on what they might say and how they may say it. Expressed in terms of the functions of communication, the task may be one of production, innovation, or maintenance. (Berlo, 1969.) According to Berlo, the former function involves getting a job done in terms of compliance with a predetermined posi- tion; it is ”the use of communication for control, for maximally efficient repetitive output of the system. It involves persuasion, the giving and taking of instructions or orders, etc." (1969, III, p. 10.) The second function, innovation, also involves both information exchange and control but for the purpose of exploring new ways of doing things. The third function, maintenance, is, to a large extent, a homeostatic function; "it is communication de- signed to maintain the human components of that system." (Berlo, 1969, III, p. 10.) It would be characterized by messages designed to reinforce system members' self— concepts. In any given system, communication may, and probably will, serve more than one of these functions at any given time although one may receive more emphasis than another. As noted above, a production emphasis would be characterized 72 by control messages and, as Rogers (1971) says, by a ”high proportion of task related informational messages." (p. 64.) As operationalized in most of the research, this production function involves the dyad discussing a series of topics that are relevant to family decisions. In Mark's (1970) study, dyads were asked to discuss and decide how they would deal with the following situations: (1) one in which "their child was overheard discussing sex with a friend using misinformation;" (2) one in which ”night- time television programming began to offer programs for which their children wanted to stay up late;" and (3) one in which they had only a short time to seek shelter after a Civil Defense alert was broadcast. (Mark, 1970, pp. 35-6.) Since the notion of dominance~submission comple- mentarity implies control on the part of one individual, and because a production task implies a need for direction or control, this study used similar discussion topics that asked couples to reach a joint decision. It was hoped that this type of task would generate interaction and allow a dominant individual to exert his influence. Socio-Economic-Class One additional variable that has been examined in the study of communication of different groups is that of social or socio-economic class. For example, Schatzman and Strauss (1955) found that respondents reporting their 73 experiences in a natural disaster differed in their communi' cation behavior depending on their social class. The two groups they examined--lower class and middle c1ass--differed, not only in their language characteristics, i.e., grammar and vocabulary, but also in (a) the number of kinds of perspective utilized in com- munication; (b) the ability to take the listener's role; (c) the handling of classifications; and (d) the frame- works and stylistic devices which order and implement the communication. (p. 330.) Research by Strauss (1968) showed that lower or working class and middle class families also differ in their com- munication behavior. He found that lower class family groups, composed of husband, wife, and a child who was at least 12 years of age, from three different societies-- Bombay, India; Minneapolis, Minnesota; and San Juan, Puerto Rico--differed significantly from similar middle class family groups in the volume of intrafamily communication during a problem solving task. As indicated earlier, Mark (1970) also found differ- ences in relational communication between husband-wife dyads classified as lower-middle to lower class and similar dyads identified as upper-middle to upper class. As pointed out in the previous chapter, he found that complementary be— haviors tended to occur more often among lower class pairs and symmetrical behaviors tended to occur more often among upper class dyads. 74 Given these differences between different socio- economic or social class groups along a series of communi- cation dimensions, it was felt that the SEC variable might be useful as a predictor of certain relational communication transactions. Hypotheses On the basis of the previous discussion and the concept of dominance-submission complementarity, a set of hypotheses were derived that integrated the dyadic dimen- sion and situational variables discussed in this chapter with the relational communication concepts. At a very general level and from the theoretic framework developed by Carson, the following hypothesis was formulated: H1: Husband-wife dyads will have a significantly higher proportion of complementary transactions than symmetrical transactions. [It should be noted that the stating of this and all other hypotheses in terms of proportions was based on an examina- tion of the interaction times for the topics. These times ranged from 11/2 to 12 minutes per topic for different dyads and to utilize frequencies as a measure seemed untenable. In order to use parametric tests, it was decided to frame the predictions in terms of proportions.) This first hypothesis sought to test the basic as- sumption set forth by Carson that people will interact in 75 a complementary fashion for a whole series of behaviors. Although the trait of dominance is of prime interest in predicting relational communication patterns, Carson's major tenet also incorporates such traits as competitive- ness, self-effacement, aggressiveness, and rebelliousness. If his assumption is correct, and if dominance is highly correlated with one-up messages, it would be expected that complementary transactions--those in which one person is in control of the relationship and one accepts that control-- will occur more often than will symmetrical transactions'- those in which both individuals seek control. Other hypotheses that were tested that incorporated the notions of dominance-submission complementarity, social class, and complementary or symmetrical transactions are as follows: H2: Husband-wife dyads who have different dominance scores will have a significantly higher proportion of complementary transactions than will husband- wife dyads who have similar dominance scores, regardless of social class when the dyad is in- volved in a decision-making discussion. H3: Husband-wife dyads from a lower-class environment who have different dominance scores will have a significantly higher proportion of complementary transactions than will similar dominant type dyads from an upper—class environment when the dyad is involved in a decision-making discussion. H4: Husband-wife dyads who have similar dominance scores will have a significantly higher proportion of symmetrical transactions than will husband-wife dyads who have different dominance scores, regard- less of social class when the dyad is involved in a decision-making discussion. 76 H5: Husband-wife dyads from an upper—class environment who have similar dominance scores will have a significantly higher proportion of symmetrical transactions than will similar dominant type' dyads from a lower-class environment when the dyad is involved in a decision-making discussion. Whereas Hypothesis 1 is important in terms of Carson's assumption of the complementarity of behaviors, these latter four hypotheses are important in determining the efficacy of dominance and social class as predictors of relational communication. However, even if these hypo- theses were confirmed, there would still exist the possi- bility that the relationship between one's dominance and his one-up messages could be negative rather than positive. Since these hypotheses deal with dyadic information, the degree to which one's dominance is a predictor of one's controlling messages can be tested only with information about the individual. Therefore, to test the assumption that dominance is, indeed, a predictor of one's controlling or one-up messages, the following hypotheses were formulated: H6: There will be a significant positive correlation between an individual's score on a dominance scale and his proportion of one-up messages. H7: Men will have a significantly higher correlation between their dominance scores and their propor- tions of one-up messages than will women. It should be emphasized that the dominance— submission complementarity suggested in the hypotheses is that based on a friendly-dominance quadrant discussed by Carson. Similar predictions could have been made on the 77 basis of hostile—dominant behaviors, but it seemed more logical to assume that in an observational setting designed to gather data to test the aforementioned hypotheses, friendly-dominant behaviors would be those enacted by dyad members. Summary and Conclusion The hypotheses developed in this chapter have at- tempted to predict what types of relational messages and transactions may be expected from a notion of complemen- tarity of one type of personality behaviors. Although other predictions could be made from Carson's framework that might have some applicability to relational communi- cation, it was felt that the concept of dominance has the strongest ties with the concepts of symmetrical and complementary messages and transactions. Since the latter deal with the control of the relationship and dominance is a generalized trait of control, the relationship between the two seems particularly valid. If we accept Haley's (1959) view that "everyone is constantly involved in de- fining his relationship or countering the other person's definition," (p. 335) then the theoretic framework dis- cussed in this paper would seemingly strengthen and in- crease the predictive power of the relational communication 78 framework and allow us to better describe the circum‘ stances under which we would expect certain types of messages and transactions to occur. CHAPTER VI RESEARCH METHODOLOGY The purpose of this chapter is to set forth infor- mation concerning the sample, methods of data collection, operationalizations of the variables, and the methods of statistical analyses. On the basis of information presented in the previous chapter, it is obvious that this study con- cerned itself with relational communication found in husband-wife interactions that were concerned with decision-making for a set of predetermined topics. Sample The husband-wife dyads for this study were a selected sub-sample from a randomly selected sample of homes for an Office of Civil Defense research project, Communication Processes in Civil Defense Program, OCD Contract #DAHC-20-7l-C-0297. These homes, drawn from a telephone listing of metropolitan and suburban Dayton, Ohio residences, were the data base for Phase I of the project, a telephone interview. The sub-sample of dyads for Phase II of the project and the data base for this study was selected on the basis of two criteria: (1) 79 80 whether the couple interviewed in Phase I was married; and (2) whether the couple had children under 12 years of age. Using these criteria, a list of 142 families was drawn up from the list of 399 interviews completed in Phase I and attempts were made during the spring of 1972 to con- tact each family. Of the 142 families drawn from Phase I, 65 were interviewed, 44 refused, and 33 could not be located or were not available during an interview period of three weeks. In addition, one couple was interviewed who was at an address for a Phase I family but who had not participated in the telephone interview. This brought to 66 the total number of completed interviews for Phase II. A letter given to each couple explained the purpose of the study and indicated they would receive ten dollars for their time. (See Appendix I) No attempt was made, unlike Mark's (1970) study, to determine if either the husband or wife had ever sought or been given psychiatric help. The average age of the wives in this Phase II sub- sample was 32.6 years; that of the husbands was 35.9 years. Wives had an average educational level of 12.6 years of school while husbands had an average of 14.0 years. Couples were married an average of 11.6 years. Four of the couples were Black, representing 6¢ of the sub-sample and the re- maining 62 couples were White. [This percentage of Blacks is slightly less than half of the 13.2% of Blacks and other 81 minorities reported for the Dayton, Ohio urbanized area by the 1970 census. (1970 Census of Population: Ohio.)] Method of Data Collection The purpose of the home interviews was to obtain data on: (1) how husbands and wives spend their time together and their time with others (See Appendix II, Part I), (2) husband and wife's processing of information-- including Civil Defense material (Appendix II, Part II), (3) their answers regarding who does certain tasks in the home and their views on who should do these tasks (Appendix II, Part III), (4) their scores on the trait of dominance (Appendix II, Part IV), and (5) their socio-economic back- ground and their satisfaction with their marriage (Appendix II, Part V). In addition to these questions, contained in questionnaires that the husband and wife completed sepa- rately, the home interview was intended to collect data on how husband-wife dyads communicate on a relational level when presented with a set of topics on which they are asked to reach a decision. For this study, the topics dis- cussed fell under two headings--family behavior in civil defense situations and family behavior in a non-emergency situation. More specifically, one topic dealt with the sharing of home basements, one topic dealt with family actions under threat of a tornado, and one dealt with the couple's reactions to whether a wife should have an 82 independent career. (See Appendix III.) Since these discussions were the data base for the relational communi- cation analysis, they were tape recorded on casette tapes and transcribed. (See Appendix IV for the set of rules given to the typists/transcribers.) Couples were given each topic separately and were asked to discuss it; i.e., husbands and wives were asked to talk to each other about each topic. They were told they had up to 10 minutes per topic. Although only three topics were utilized in the analysis for this study, one other topic was included to set the couple at ease and possibly insure at least a five minute discussion. This topic was discussed first and dealt with the couple's courtship; specifically, couples were asked to discuss with each other how they happened to meet and marry. After this topic, both husband and wife were given a written copy of each topic and were given suggestions on what they could discuss. These suggestions included (1) reactions to, (2) feelings about, (3) expectations of, (4) problems with, and (5) solutions for the situations posed by each topic. To further insure at least a minimum five minute discussion, these suggestions were written up as probes and were to be used by the interviewers when the following con- ditions arose: (1) the couple did not talk for at least 5 minutes, (2) neither husband or wife spoke for a period of 15 seconds, and (3) the couple spent most of their time 83 discussing only one of the suggestions. [It should be noted that these instructions were not finalized until four couples had been interviewed. As a result, the inter- viewer for those couples did not utilize probes and the couples' discussions were not as long as those collected later. It should also be pointed out that the probes were not always utilized by interviewers, especially when a couple indicated they no longer wished to discuss a par- ticular topic, and were also not always useful in producing a discussion of at least 5 minutes.] Problems with one interviewer, one noisy tape re- corder, and couples who either did not understand the topics, the nature of the task, or just could not talk in the presence of the tape recorder initially reduced the number of interviews for analysis to 64. Further examina— tion of the sample and transcripts showed that (1) one black couple had no children, and (2) six couples had too few transactions for analysis on one topic. Since it was intended that the sample be homogeneous in terms of the couples' family situation--i.e., at least one child under the age of 12,-- the childless couple was not included in the analysis. Although it might have been possible to utilize only white couples to establish some comparability with most of the samples studied in the previous research on relational communication, a decision was made to include 84 two black couples. Since using only white couples would have resulted in a sample of 54, the inclusion of the two black couples was done primarily to obtain equal size cells for a 2x2 analysis of variance design. With the exclusions mentioned previously, the final sample had 56 couples. Qperationalizations of the Variables Dominance As noted from the hypotheses, the factor variables for this study were dominance scores and socio-economic class information, and the dependent variables were com- plementary and symmetrical transactions, and one-up mes- sages. The dominance variable was measured by twenty eight items from Edwards Personal Preference Schedule (Edwards, 1959). [This scale was used despite Winch's (1958) criti- cism concerning the unknown validity of the EPPS scales, because of (1) Edwards' (1959) report that several studies have shown high correlations between self-ratings and scores on the EPPS scales, a factor that may be indicative of construct validity, and (2) Becker's (1964) use of the .BPPS dominance scale for the construction of dominance difference scores that he suggests provide a superior operationalization of complementarity than one based on correlated individual dominance scores.] These items con- taiJied a pair of statements, one of which was defined as chaaracteristic of the dominant personality. Subjects were 85 asked to pick one of the two statements, and the dominance score was determined by the total number of dominant items picked with a possible range of from 0 to 28. (See Appen- dix II, Part IV.) One item, number 13, was changed from the original pairing in Edwards schedule. Rather than using the Heterosexual statement, "I like to participate in discussions about sex and sexual activities," another Heterosexual statement, "I like to be regarded as physically attractive by those of the opposite sex,” was substituted. This was done in the belief that the former statement might cause negative reactions among the respondents. Ten of the 112 subjects did not complete the total scale and skipped or missed one or more of the items. Specifically, nine subjects (6 wives and 3 husbands) missed one item, and one subject (wife) missed four items. Follow- ing a suggestion by Edwards (1959), those items were scored by flipping a coin with heads indicating choice A and tails indicating choice B. Using this procedure for the thirteen missed items, five dominant choices resulted. Although it was not known what the effect of using just the dominant items would be, it is interesting to note (See Table 2) that the means and standard deviations for the men's and women's dominant scores in this study are very'similar to those reported by Edwards (1959). A t test between men's and women's mean dominant scores for each groupIshowed significant differences. Apparently, using 86 5.:0503 How came on» cusp Ho>oH was“ an powpmfi zapcmuflmwamflm mw :woa macaw Hoo.o v m«« n.coEoz how came on» menu Ho>oH wasp um Howpma xaunmowmficmfim ma :moE machv Ho.o v m« ANIO HNIN NNIo HNIN momemm momcmm om om co co Nmme Hmoe z z z z z z on.¢ mo.m mh.m «ewe.ea en.e mm.< mo.m «¢N0.¢H mm.v Am.m v~.oa «om.eH cage: :02 c0503 no: case: so: dose; :02 :oEoz so: coaoz no: m.mm memo: m.mm mcwoz m.am mcmoz mangem Hagen HH amaze oHQEwm ouonEou HH ommcm mHQEmm pfise< Haemeou mmam mmmm 03p HOW OHQMHHN> OUfimfifiEOU 06:. MO meMHwH>OU thfiflwum mug wfimmz .moHQEmm HH omega ecu was onEmm o>wumspoc .N oHan 87 the dominant scale alone without the other EPPS variables did not adversely affect the normality of the distributions of dominance scores for this sample. Since the hypotheses concerned dyadic information, each wife's dominance score was subtracted from her hus- band's dominance score to give a dominance difference score for each dyad. The use of the difference scores for dominance is based on Levinger's (1964) criticism of the negative correlation coefficient as an indicator of complementarity of needs that may be the same in kind and equal in intensity, and on Becker's (1964) use of dominance scores in examining the relationship between dominance complementarity and Authoritarianism. In addition to hav- ing an advantage over the correlation coefficient as an indicator of complementary needs that are similar in scale value, this type of index also provides a measure of dyadic information concerning the dominance trait. As shown in Table 3, p. 88, these difference scores ranged from 1 to 19 with the median at 6.50. Splitting at this point, 28 dyads with scores ranging from 1 to 6 were placed in a group containing those referred to in the hypotheses as having similar dominant scores, and 28 dyads with scores ranging from 6 to 19 were placed in a group corresponding to those dyads identified in the hypo- theses as having different dominant scores. To check Vfllether these groups did, indeed, differ in terms of their 88 Table 3. Dominance difference scores. Scores V Frequencies 1 4 2 3 3 4 Mean = 3.85 n = 28 4 3 5 10 6 8 ---Median = 6.50 7 4 8 5 9 3 10 3 ll 4 Mean = 9.46* n = 28 12 2 13 1 l6 1 19 _;1 56 *p < 0.001 (This mean is significantly larger at this level than the mean for the low dominance difference group.) dominance difference scores, the low dominant difference group mean and the high dominant difference group mean were compared in a t test. The resulting t value, 8.33, was significant at p < 0.001. 89 Socio-Economic Class The socio-economic class factor variable was based on Mark's (1970) study and his operationalization. Using Bergel's (1962) argument that upper-class people are college graduates and lower-class people have not finished high school, Mark collected information on in- come, education level, and job title and description. His cutting points for placing people in the upper—class category were income above $8,000 per year and some col- lege level education. Similar cutting points had been proposed for this study to identify dyads as upper and lower class; namely that upper-class dyads would be those in which the income was over $8,000 and the major wage earner had at least two years of college, and lower-class dyads would be those in which the income was less than $8,000 and the major wage earner had less than two years of college. The income information from this sample, however, did not allow splitting by these cutting points. Of the 66 dyads, only 3 indicated their family's income was less than $8,000, whereas 37 stated their income was between $8,000 and 14,999, and 26 indicated their income was over $15,000. Instead, relying on part of Bergel's argument, dyads were split into classes by education level of the husband. As shown in Table 4, the range runs from grade eight to six years of graduate school. A median split on 9O memos can swap Ho>oH mNAH um Nowhma zfipconmNcmNm ohm memos omonhv Hoo.o v m« A.mm50Nm mmmHu Nozoa No Ho>oH :oNpmosco Nazca ecu com om co MI flNNV Nooeum openemcu co meme» me an H mcmv Hoonum mumscwhu mo muse» Nsom H wN u G H mmHV finger—um oumaéwho Mo whmm> $0.53.. H .NN.oNuM m mNNV Nooeum opeseaco mo memo» 039 m mm u a N NNNV Nooeum mumseeco mo New» «no N «No.0Nuw NH AONU onNNou mo meme» Neon NN m NmNV onNNou mo meme» emcee o :NNemx, N Neau ImmoNNou mo whee» 039 N N NNNV uwoNNoo mo New» one 4 emNeoz NN u = NH NNNV meacu eNmNoze ON mm u : NN.NNIM N NNNV ovate noeo>mNm N om.NNnm e NONV memcu apnea m N NNV oeeeu epemNm N mowucoscoum mowocozcohm onEmm Hmcfim Ho>oq oHQEmm ouonEou .mvcmnmsz mo Ho>oa coaumoswm .v oanmh 91 the total sample places 33 subjects in a category of those having one or less years of college and the remaining 33 in a category of those having at least two years of col- lege. Split this way, the respective means for the two groups are 11.90 and 16.67 years of education, a differ- ence which is significant at p < 0.001. Also listed in this table are the categorizations of subjects after the bad tapes, and poor discussions were removed. Using a median split at 14.5, the 28 subjects in the lower group had an average education level of 11.71 years while the 28 subjects in the higher group had an average education level of 16.79 years, a difference that is, again, significant. Operationalized this way, the two groups were considered lower and upper class. Although they have been and will continue to be referred to in this study as lower and upper class groups, they probably-- on the basis of respondents' reporting of income--more accurately represent lower-middle and upper-middle class populations. Despite the possible inaccuracies of these labels, the groups did differ significantly on education and were, therefore, used to check the assumption that they also differed in communication behaviors. Complementary and Symmetrical Transactions The measurement of these dependent variables was based on a three digit coding scheme that was very similar 92 to Mark's (1970), but contained slight modifications. These modifications and the rationale for them were developed jointly by this writer and a colleague, Edna Rogers. The changes made in the scheme were based on the weaknesses of Mark's scheme discussed in Chapter Four, and concerned the internal consistencies of the second and third digit code categories, a renaming of some of the code categories, the addition of two new categories and the elimination of one code category, a separation of Mark's combined code categories in digits two and three, the specification of different types of symmetry, and the creation of a neutralizing or leveling of control trans- action. The categories used in the modified scheme are as follows: lst Digit 2nd Digit 3rd Digit 1. Wife 1. Assertion 1. Support 2. Husband 2. Question 2. Non-support 3. Talk-over 3. Extension 4. Non-complete 4. Answer 5. Other 5. Instruction 6. Order 7. Disconfirma- tion 8. Topic change 9. Initiation- termination 0. Other Like the original coding scheme developed by Mark, this one utilized the first digit to refer to the two speakers, the second digit to refer to the grammatical format of the 93 speech, and the third digit to refer to the response mode of the speeche-i.e., what the speech was in reference to the previous one. Following Mark, this writer defined a speech as any talk by an individual that was of any length, beginning with the first word and continuing until the other dyadic member spoke. As noted in Chapter Four, it was argued that the second digit categories of Mark's scheme were not intern- ally consistent since his categories of instruction and orders seemed to be quite different from those of asser- tion, question, talking over, and laughter. Because it was felt that the categories of this digit should be gram- matical formats, these categories were placed under the third digit. The resulting categories were assertion, question, talk-over, non-complete, and other. An "assertion” was used to refer to any completed referential statement that was either in the declarative or imperative form. A "question" referred to a speech which took an interrogative form. Although the remaining three categories are not grammatical forms, pg: 33, they are descriptive of the form a message takes. A "talk«over," therefore, was used to refer to the way a speaker entered the interaction, or what Mark calls a style of mobiliza- tion (1970, p. 41). It was considered successful if the first speaker relinquished the floor when the second speaker started speaking, and it was considered 94 unsuccessful when the first speaker continued what he was saying despite the second speaker's attempt to interrupt or talk over. Since it was felt that both types of talk- overs indicate similar attempts at control, no special relational codes were developed to differentiate the two in terms of relational messages. Instead, two codes were developed for use on the transcribed copies of the taped discussions. Successful talk-overs were denoted by the use of brackets [ ], and unsuccessful talk-overs were indicated by the use of parentheses ( ). In order to check whether this technique of isolating both types of talk-overs was reliable, 5 of the original 66 transcripts were randomly selected and were coded by this writer and independently by another re— searcher. Reliability checks were done for total number of talk-overs, and agreement on both successful and un- successful talk-overs. For the total number of talk-overs-- i.e., whether or not a message was a talk-over, either successful or unsuccessfu1--the percentage agreement was 0.71. For successful talk-overs, the percentage agreement was 0.63, and for unsuccessful talk-overs, the percentage agreement was only 0.36. Given these rather low figures, it was decided to abandon the independent coding and instead utilize a method of double checking and resolving disagreements. For this procedure, 95 transcripts were read and coded by this writer, read and checked by two other researchers, and any disagreements were resolved. The "non-complete" category was used to indicate any utterances--other than those categorized as talk- overs--that were initiated but were not completed. It included such messages as "well,” "uh," "but," "I, uh, guess...uh," etc. The final category under this digit, "other," referred to verbal utterances that were indis- tinguishable or were not classifiable to form. The third digit code categories were used to refer to the response mode of the speech. The "support” category indicated both the giving and seeking of accept- ance, agreement and/or approval. The "non-support” category was used to refer to disagreements, rejections, demands, and challenges. The "extension" code was used to indicate a speech that continued the flow or theme of the preceding speech. Included under this category was a non-committal response to a question. The category "answer" was reserved for a response to a question that was definitive or at least had substance and/or commit- ment. Whereas a non-committal response to a question such as "I don't know" was coded as an extension, a defini- tive response such as "It was thirty degrees” was coded as an answer since the two types of responses have differ- ent control defining natures. 96 "Instruction" and "order" both denote a regulative response but of different intensities. The former referred to a suggestion or evaluative statement which was often qualified, while the latter referred to an unqualified command with little or no explanation. For example, a message like "I think it's time for you to go to bed now because you have school tomorrow" was coded as an instruc- tion, while a message like "Go to bed" was coded as an order. The categories of ”disconfirmation" and "topic change" were taken directly from Mark's scheme. Both refer to a response switch or non-continuance, but a ". . . disconfirmation occurs after a statement has been made which demands [Italics in original] a response to it by the other individual and he does not respond to the demand . . . ," while a topic change occurs with the ”. . . introduction of a new idea after discussion of the [original topic] . . . ." (Mark, 1970, p. 44.) Thus, a disconfirmation was used to indicate a message or response by one individual that ignored or bypassed the request of the other individual. For example, a response like "Hey, look, it's snowing!" to the question, "Can we talk about what we're going to do?" was considered a disconfirmation. A topic change referred to a response that had little con- tinuity with previous messages and no response continuity was requested. For example, if one person said "I think 97 we should get the kids' hair cut" and the other said "We should all go over to your mother's on Saturday," the latter would have been considered a topic change. "Initiation-termination" was used to denote a message that either began or ended an interaction. Its use signified starting and ending points for particular discussions or conversations, and it also was used for any message by the husband or wife following a probe by the interviewer. "Other" was a residual category for any response made that was undistinguishable or unclear. Since silence in an interaction may be of signifi- cance in describing the way certain dyads interact in decision-making discussions, this scheme followed Mark's use of a 000 code to indicate periods of silence that were longer than 3 seconds. Every additional 5 second period of silence was also coded as 000. To illustrate how this scheme was used to cate- gorize messages, a sample from a lower class couple's discussion of topic 3 is shown below. Wife: Well, what happens when it is over and 123 you come up and everything is gone, then what? There is no food, there is no place to sleep, there's nothing, what do you do then? Husband: Well, contact your red cross and insur- 214 ance, something, try to get a place to stay, you know, temporary, until you know you could, uh start to clean up, fix up. That's about all you could do. 98 Wife: Well, I guess you could uh, if everything 113 you had was just ruined, if you could contact one of your nearest relatives, one Husband: (Right.) 231 Wife: closer to you, by phone, you could 113 probably walk there and at least make arrangements for . . Husband: Right. 211 Wife: . . . someplace to stay. Because you 113 wouldn't even have clothes, or anything, maybe everything would be gone. The wife's first message was coded 123 to indicate a question in extension that sought information about the topic being discussed [tornadoes]. The husband's message was coded as 214 to show that it was an assertion that served as an answer by providing information pertinent to the wife's question. The wife's second message was an assertion in extension, 113, that was not successfully interrupted by the husband's talk-over, coded 231. The wife's third message was also coded 113 to indicate an extension of what she had said prior to her husband's talk-over. The husband agreed with her statement, 211, and she followed his message with another assertion in extension. Coding decisions were guided by a set of priority considerations to facilitate the categorization of any given message. (See Appendix V for the priority scheme, definitions of categories, and examples used to train the coders.) These priorities were based on the predominant function of the message. Thus, for the second digit codes, 99 the message was first considered as a talk-over (#3). A talk-over could have been a question, assertion, or non- complete but an interruptive speech was coded as a talk- over, independent of form. If the speech was not a talk-over, it was then considered as a question (#2), an assertion (#1), or a non-complete (#4). If it could not be categorized as any of these, i.e., if it was an INDISTINGUISHABLE, it was coded as "other" (#5). Similar priorities governed the third digit codes and the use of them in categorizing the response mode of the message. In terms of what a message was in relation to a previous, it was first considered an an initiation or termination (#9). Although messages coded in this category might have also been coded in other categories-~e.g., as an answer, a non-support, or a topic change--, the first consideration was whether the message initiated or terminated. If the message was not an initiation-termination, it was considered as an answer. If the main response function of the message was to pro- vide information, it was coded as an answer (#4). If, however, the message was a response that gave support (#1) or non-support (#2) to a question that sought approval or acceptance, it was coded as one of these rather than as an answer. For response switching, the message was first considered a disconfirmation (#7). If it was not codable here, it was then considered a topic change (#8). For 100 messages that had a regulative function, the first con- sideration was whether the message was an order (#6). If it was not, it was coded as an instruction (#5). If the message was not codable in any of these categories, it was then considered an extension (#3), a message of support (#1), or a message of non-support (#2). It was considered as an other (#0) if it fit neither of these previous cate- gories or if it was indistinguishable. Thus far, the discussion of the coding scheme has outlined only the first step in measuring complementary and symmetrical transactions. The second step required a translation of the message codes to a control dimension; that is, each message was given a control direction in terms of whether the message denoted a demand for control of the relationship, whether it referred to an acceptance of control by another individual, or whether it indicated a somewhat neutralized or non—demanding, non-accepting control. A message that referred to a demand for control was designated as one-up (t); a message that referred to an acceptance of control was designated as one—down (I); a message that referred to a neutralized control or level- ing effect was designated as one-across (+). For example, if a message was an assertion which was an extension of the dialogue, it was coded as a _13 and represented a one-across (+) movement. If a message was an assertion expressing support for a previous message, it was coded 101 as an _11, which was a one-down (I) movement. A non- supportive assertion was coded as a _12 and designated a one-up (+) movement. Expressed in terms of message forms and response modes, the following code categories were viewed as con- trol maneuvers toward one-up; non-support responses, questions demanding an answer, answers with substance, instructions, orders, initiation«termination, disconfirma« tion, topic change, and talk-overs that were anything but supportive or unclassifiable. Whereas the first five of these codes referred more to the control of the dyadic relationship, the latter four categories denote "floor" control maneuvers. One-down code category combinations were: ques— tions that sought information or support, continued the dialogue (extension) or had an uncodable third digit (other), support responses, non-complete phrases that, in effect, demanded that the other take control, and talk- overs that supported the other. Neutralizing or leveling of control maneuvers were seen as extensional or continuation messages that carried a dialogue along with little attempt at controlling the dialogue. Those category combinations that were seen as one-across maneuvers were: assertions, non-complete phrases, and "other" (unclassifiable) message forms that were extensions. These included statements of continuance, 102 "filler" phrases, and non-committal responses to questions. Other one-across codes were non-completes that initiated or terminated an interaction, and assertions, non-completes and "other" that had no clear response mode; i.e., that contained a third digit code of zero or "other." These control aspects and the category combina- tions that may have been either one-up, one-down, or one- across are presented in the matrix of combined form and response dimensions (digits two and three) in Figure 2. c: o w-I a o 9 c m no H o E g s m 0 s: "-1 be cu m a) G. O H 0H F: 0) p p g, .H U '«H U H (U H :3 m In :3 c: cu c: o m c o H H o u whH h a. ' 0 3 H m u -H +IE o o. c u m m 'u m a. wIH .: s o x a a H -H o ::o u m I: Q) (U or-I O "U 4..) .H 4..) o l 2 3 4 5 6 7 8 9 O assertion l I I + + + + + + + + question 2 I I I I I I I I I I talk-over 3 I I I I I I I + + + non-complete 4 I I + + + + 1 1 1 1 other 5 I 4‘ + I + f 1. .f 1. + Figure 2. Control dimensions of message types. [It can be noted from this matrix that the control defining nature of a message is based on both the form it takes and what it is in relation to a previous message.] 103 The rules set forth earlier for determining whether a message was classified as one-up, one-down, or one-across may be more clearly seen in the list of second and third digit combinations which follows. There are a total of 50 combinations, consisting of 35 one-up messages, 8 one-down messages, and 7 one-across messages. Digit Two Type of Code Number Control 1 one-up I = 12, 14, 15, l6, l7, 18, 19 one-down I = 11 one-across + = 13, 10 2 one-up I = 22, 24, 25, 26, 27, 28, 29 one-down I = 21, 23, 20 3 one-up I = 32, 33, 34, 35, 36, 37, 38, 39 one-down I = 31, 30 4 one-up I = 42, 44, 45, 46, 47, 48 one-down I = 41 one-across + = 43, 49, 40 5 one-up I = 52, 54, 55, 56, 57, 58, 59 one-down I = 51 one-across + = 53, 50 From these rules, various message types were coded in terms of their control dimension. For most interactions, each message was coded by one three digit number and there- fore had one control direction. However, since a message functions both as a response to a previous message and as a stimulus for a following message, some messages in this study combined several control directions that may or may not have been similar. For example, if the response 104 interpretation was categorically different from the stimu- lus interpretation, two sets of three digit numbers were used to describe the message and two control dimensions were eventually identified on the basis of the preceding rules. The use of this double coding may be illustrated by a sample interaction from a lower class couple's dis- cussion of tornado preparations. Wife: Well, what if, what if the whole top 123 I caved in the part of the basement, then what do you do, just get in the nearest corner where it isn't? Husband: Right. Uh, what if the phone rang 211-223 I-I and the tornado was coming over? Would you, uh, want to answer the phone, do you think somebody would be in trouble or something? What would you do? Wife: [1 think I would probably stay in 134 I the basement.] In this interaction, the husband's message illustrates a situation in which his message is an agreement or support and also a question seeking information, both of which are considered one-down (I). In the next sample interaction, also taken from the same couple's discussion of tornado Irreparations, the wife's message is both an extension-- “filich is coded one-across (+)--and a question seeking illformation--which is coded one-down (I). Husband: If I could get home? 223 I Wife: I don't know if you could. What if 113-123 +-I you couldn't get home, then what would you do? 105 The preceding explanation of message codes indi- cates the basis for obtaining transactional or relational information concerning the way dyadic members defined their roles through various control maneuvers. The de- pendent variables for this study-—complementary and symmetrical transactions-—were thus operationalized by combining the directional control dimension of one individual's message with the directional control dimen- sion of the other individual's message. Each transaction unit therefore represented dyadic information and pro- vided a measure of the role relationships between the participatns for a particular set of messages. For ex- ample, the sample interaction by a lower class couple presented earlier can be used to illustrate both indi- vidual message codes, individual control dimensions, and dyadic or transactional values. Message Control Transactional Code Code Code Wife: Well, what happens 123 I when it is over and ...do you do then? Husband: Well, contact your 214 I red cross and insur- It ance...you could do. 1? Wife: Well, I guess you 113 + could, uh, if every— thing...relatives, +I one Husband: (Right) 231 I 1+ Wife: closer to you, by 113 + phone, you could ++ probably...for Husband: Right 211 I 106 For a message from one individual that had two control codes, both directions were utilized in producing the transactional measure. For instance, in the following interaction from a lower class couple's discussion of tornado preparations, the wife's message leads to the transactional codes of I+ and II. Message Control Transactional Code Code Code Husband: If I could get 223 I home? Wife: I don't know if 113-123 +-I you could. What if you couldn't +1 get home, then what would you do? Husband: I'd find, I would 214 I try to find the nearest shelter... 45> In both cases, the transactional codes indicate the pat- terns of control similarity or dissimilarity between the dyadic members. There were three transactional or relational types that resulted from the combination of the three possible control dimensions of paired messages. These were comple- mentarity or a complementary relationship, symmetry or a symmetrical relationship, and a transitory relationship. The particular transactional or relational types are shown in the matrix in Figure 3. Complementary trans- actions, those having both a one-up and one-down dimension, are represented by cells 2 and 4 of the matrix. Symmetrical 107 One-Up One-Down One—Across I I + One-Up . (1) 11 (47* 11 (77* 11 One-Down I (2) 1+ (5) 1+ (8) 1+ One-Across + (37++ (fiji +1 (9) ++ Figure 3. Transactional types. transactions are shown in cells 1, 5, and 9. As these cells suggest, there are three types of symmetry rather than the one referred to by Mark. These types were based on the belief that messages of various types may be matched according to their control aspects rather than their gram- matical aspects, and that the former matching is more logical than is the latter in terms of the concept of symmetry. Cell 1 (II) represents competitive symmetry, a transaction type that is similar to Watzlawick, Beavin, and Jackson's (1967) symmetrical escalation (p. 107) and refers to a "struggle" for control. Cell 5 (II) refers to submissive symmetry, a transaction type which is char- acterized by acquiescence of both members to control by the other. Cell 9 (++) is termed neutralized symmetry and refers to mutual behaviors that represent neither a seeking of nor a submission to control. The cells 3, 6, 7, and 8 represent a new type of transaction. Previous investigators have conceptualized the control defining nature of a message as being either 108 complementary or symmetrical. The addition of the one- across (+) direction, which was an attempt to increase the sensitivity of the control measure, produced the neutralized symmetry described above, and a third type of transaction termed transitory. Where cells 3 and 7 refer, respectively, to movements from neutralizing con- trol toward one-up and from one-up to neutralizing, cells 6 and 8 refer, respectively, to movements from neutraliz- ing toward one-down and from one-down to neutralizing. It is from these transactional types that the data for this study were obtained. Although the transi- tory transaction is of some conceptual importance, only the data on complementary and symmetrical interactions or transactions were of major concern. In addition, al- though individual message data were generated, will be described extensively in the next chapter, and were used to test two of the hypotheses, the data for testing the majority of the hypotheses were of a dyadic nature and were therefore those of the transactional type. Earlier in this chapter, it was stated that several dyads were excluded because of bad tapes resultf ing from a noisy tape recorder, lack of children, and poor discussions. Having discussed the operationalization of the dependent variables--complementary transactions, symmetrical transactions, and one-up messages,-- it is now possible to specify the criterion which resulted in a dyad 109 being included in the analysis or being excluded because their discussion was poor. This criterion was that each topic for a dyad had to have a minimum of 5 transactions or else the dyad would be excluded. The reason for this number of transactions is tied to the way the first and last messages of an interaction were coded. Since both the first and last messages were arbitrarily defined as being one-up with a third digit code of initiation- termination (#9), it was felt that six messages were necessary in order to allow all three types of trans- actions to occur. Using 5 transactions as the minimum cutting point, 6 couples were excluded. Once the final sample was selected, transcripts were coded by three coders who had been trained over 3 separate two hour periods. Each coder was given 20 dif- ferent transcripts and each was also given 6 transcripts that were coded by all three for reliability checks. [See the next chapter for information on reliability figures.] When the coding of the transcripts was com- pleted, the codes were punched on data cards for use in a listing procedure developed by this writer and program- med by Mrs. Joanne Helfrich. This computer program printed out the following information: individual mes- sage categories expressed in terms of who said what and what that message was in relation to the previous message, control directions for individual messages, transactional 110 types that occur between pairs of messages from dyadic members, sums of various types of messages spoken by each dyad member, sums of messages expressed in terms of their control directions, and sums of the transaction types. Given the operationalizations presented in this chapter, the data from the coded husbanduwife interactions were subjected to both statistical and descriptive analy- ses. The specific methods used to examine the data and the results of these analyses will be presented in the next chapter. In more detail, it will be the purpose of the next chapter to: (l) discuss the reliability figures for the coding scheme as implemented by three coders; (2) describe the transformation procedures applied to the interaction proportion data; (3) show the results of the statistical analyses applied to the data to test the hypotheses; (4) present information concerning average times of and use of probes per topic; (5) explore the relationships among the data; and (6) examine the data in order to suggest steps for further research. CHAPTER VII RESULTS This chapter contains the reliability figures across coders for the relational coding scheme and also reports the results of both the statistical and descrip- ‘tive analyses applied to the husband-wife interactions on ‘three topics. The statistical analyses pertain to the ‘testing of the seven hypotheses discussed in Chapter V 311d concern the results of a t test, a 2x2 Factorial Anialysis of Variance, the Pearson Product Moment Correla- tion, and a 2 test. The descriptive analyses concern the tixmes and probes used in the topics, the sums and percen- tanges of message types for all persons across all topics, t}1e sums and percentages of the control dimensions for tllese message types, the sums and percentages of all five t)rjpes of transactions across all dyads for all topics, 811d a comparison of transaction types with a random dis- lrribution to check for stable patterns. Coder Reliability Checks As discussed in the previous chapter, each of 'the three coders was given 26 transacripts, 6 of which Vvere the same across coders and were to be used for 111 112 checking inter«coder reliability. These 6 transcripts had been randomly selected, based on a procedure suggested by Scott (1955), and represented 11% of the 54 transcripts used in the analyses. The actual reliability checking was done in two stages. In the first stage, each coder was given 12 transcripts, 3 of which were reliability checks. As soon as these 3 transcripts were coded, they were collected and were checked for agreement across coders for use of the 3 digit codes. The reliability figures that were calculated were percentages of agreement, the formula for which may be found in Holsti (1963, p. 49) and is 2“31,23 C1 + C2 R: where C1,2 refers to the number of agreements between two coders and C1 + C2 indicates the sum of all category assignments by the two coders. Since three coders were used in this study, three reliability figures were calcu- lated, each of which indicates the percentage of agreement between two coders. The first figures reported in Table 5, p. 113, are those indicating coder agreement percentages across three transcripts and are based on a comparison of the entire 3 digit code. This was done as a checking pro- cedure to see if and where problems might be occurring 113 Table 5. First stage percentage agreements between coders on use of three digit codes. Average . Coders Coders Coders . . . Top1c Reliab1l1ty A/B A/C B/C A/B/C Dyad 132 50 _ 54 _ 54 = l 69 - 72 33 - .82 65 .83 .79 2 %%= .91 3;:— = .88 %%= .84 .87 66 = 58 _ 58 = 3 §§ 75 gg - 65 g3 .67 69 84 82 78 4 = .82 = .82 = .82 .82 E m 9? m Dyad 159 84 _ 86 __ 88 = l 121'- .69 120" .72 T23 .72 .71 68 _ 76 = 80 = 2 TT7 - 58 1T7 64 TTT 70 64 36 _ 40 _ 42 = 3 -§3 - .67 §§ - .75 E? .81 .74 4 17166: .69 181%” .76 120%: .74 .73 0.71 Dyad 163 230 _ 212 _ 218 = l 777 — .84 276 — .76 771- .79 .79 88 = 92 = 90 = 2 109' .80 108 .85 i09 .82 .82 94 = 96 = 92 = 3 T07 92 102 .94 T02 .90 .92 102 104 98 4 ——— = .80 = .82 ——— = .78 .80 127 126 125 0783 Average Reliability across 12 topics equals 0.78. 114 in the use of the coding scheme. As shown in the table, the percentages range from 0.58 to 0.94 across coder com- parisons and across topics with an average percentage agreement across the three coders and across all 12 topics of 0.78. [Although this study analyzed only the last three topics discussed by the dyads, it was felt that since all four topics had been coded, it was necessary to include reliability figures for each.) It had been hoped that a figure of at least 0.80 could be obtained across all 6 transcripts to suggest the utility and efficacy of the coding scheme. Although the figure of 0.78 is close to what had been hoped for, it represents only a partial reliability figure and, as noted above, was calculated primarily to check the first results from the coders. An analysis of the three transcripts showed some errors in the use of the first digit speaker code. Out of 1980 messages across the 12 topics, 7 errors in coding the speaker were detected. Other problems in not correctly coding talk-overs (which had been identified on the transcripts by brackets or parentheses), and dis- agreements over what should be considered an assertion or non-complete helped lower the reliability figures. Given these problems, it was decided to hold another training session in order to point out the types of errors being made and make recommendations for minimizing them. These recommendations included: (1) reading transcripts twice 115 to check on categorizations for all three digits; (2) paying attention to brackets and parentheses used to in- dicate talk-overs; and (3) specifying in more detail, rules for coding certain audibly filled hesitation pauses such as "Mmm" and "Oh." Using these recommendations, the remaining 9 transcripts were checked and/or coded, the three trans- cripts described in Table 5 were recoded by randomly assigning a coder to each, and the coders were given another 16 transcripts to code, 6 of which had been ran- domly selected, were common to each coder, and served as the reliability checks in stage two of the checking pro- cedure. This second check was viewed as the criterion stage for determining whether the coded data needed re- coding or whether they could be submitted to various analyses. Again, using percentage of agreement figures, reliability was determined for 24 topics across 6 dyads. As shown in Table 6, p. 116, the pair comparison percent- ages range from 0.68 to 1.00 and the average reliability across all 72 pair comparisons is 0.86. A comparison of these second stage figures with those calculated in the first stage shows that the former are higher, both in terms of the minimum and maximum points of the range, and in terms of the overall mean. In addition, an examination of the types of errors and disagreements in the second 116 .ow.o m_maco mo_ao+ em mmoLom >+___nm__oc ommco>< “- q- Ln V Ln v [\ M <- C) O) O) O\ N O) 3.104.: E.uM_._u neufl 3. "mm. mw.umum NRHmw em. “mg. mm. "M9: 8.1% e 8.1mm irmlmm 5.qu aim NNJW gum. mm. uwmlm mm. "mm 3. "mm. m 2.qu mN.nmlw". 2.14M. gum Ermaum mm. "mm". a. "MIMI" .0. "mm. N E. "mm E. umwm E. ubmm NN. "If: prawn. 8.1%. mm. "mm“. 3.1%.": mm. "11%. _ 8m 02 m? 08 02 m} 08 02 N} 2%: muonoo mcouoo mcouoo mcocoo mcouoo muonoo mconoo muonoo mcouoo . mm. 83 En 8.5 So 8.6 3.1%?" $.umwu. mo.ul_%ol_ em. "bum. NN.H.NIN$ em.ub_.m_. No. "mm No. "mm oo._n.wlm e E.ummmm.u.wlmmm.unmwmm."WORHWWom.u.wmoo._um.moo._ummoo._nm.wm _w.uu_mNm_. 3."me mm.u._o|NI" mm. umfl NEIWIMI" 8.1%.": mm. "mm mm. um oo._umm N S. u mum. om." m E." .mlm em. "Mm“: mm. "MINI" E. "mm mm. "Pm. mm. umfi 8.1m": _ 08 o}. m? 08 02 mm? 8m 02 m} o . QB mcouoo mcouoo mconoo mcovoo mcovoo mcouoo muonoo mcoooo mcocoo . NNN 85 So 85 Nm. 83 .mouoo +_m_u mocc+ +0 mm: :0 mcouoo coo3+oa m+coeoocmm omm+coocoa omm+m ncooom .o o_nmk 117 stage coding showed fewer speaker code errors (only 1 error in some 4720 messages compared to 7 errors in 1980 messages in stage one), fewer errors in coding talk-overs, and fewer disagreements over the categorization of words such as ”Mnn" and "Oh." Apparently, the training session held after the first reliability check was useful in minimizing these types of errors. In order to determine the possible source of dis- agreements between coders, comparisons were also made between both 2nd and 3rd digit codes. For the 2nd digit codes, the average reliability across the 72 comparisons was 0.93; for the 3rd digit codes, the average reliability was 0.91. These two figures do not average out to the total reliability figure of 0.86 because the latter figure is based on the total number of categorizations by two coders. Since coders differed in their use or non-use of double codes, a major source of non-agreements is these varying number of codes. Most of the disagreements over the 2nd digit categorizations seemed to be over whether a statement should be coded as an assertion or as a non- complete. Most of the disagreements over the 3rd digit categorizations seemed to be over whether a message should be considered an extension or whether it should be coded as an answer, a support, or a non-support. Although the 2nd digit disagreements over catego- rizing a message as an assertion or a non-complete were a 118 prime source of unreliability, it should be emphasized that when the 3rd digit digits of the code were the same and the 2nd digits had this aforementioned difference, the control aspect of the message would still be the same except for one instance; (a -49 and a -19 would have dif- ferent control dimensions.) Overall, therefore, the re- liability figures for both the 2nd and 3rd digits and for the entire 3 digit code indicate quite acceptable levels of disagreement or error when the control dimensions of messages are considered. Given these reliability figures, the coded data were submitted to the computer listing program to obtain information for both the statistical and descriptive analyses. The 6 transcripts used in the reliability check were included in the analyses by choosing the high- est inter-coder agreement and randomly selecting one of the transcripts coded by those two coders. Since the inter-coder reliability between coders A and B was the highest, 2 transcripts coded by coder A and 4 transcripts coded by coder B were selected by using this procedure. Transformations Applied to Dependent Variables As was discussed in Chapter V, the hypotheses tested in this study were framed in terms of proportions in order to use parametric statistics. Nevertheless, in 119 order to satisfy the assumptions of several parametric statistical tests, transformations of proportions are recommended by Hays (1963, p. 380), Winer (1962, p. 221), Edwards (1950, p. 203), Snedecor (1946, p. 316), and Cochran (1943). Referring to the use of proportions or percentages in the analysis of variance, Snedecor comments that such data may sometimes be subjected to analysis without transforming them. Percentages which express frequency of occurrence per hundred units when calculated from counts of 100 or more in the numerator and when ranging between 20% and 80%, may be expected to yield valid tests in analysis of variance . . . On the other hand, if percentages result from less than 100 affected individuals, or if the event enumerated is infrequent, some transformation of the variable may be necessary before analysis of variance is carried out. (1946, p. 316.) In addition to this transformation for the analysis of variance, many of these same writers also suggest that a similar transformation be applied to the data for any t tests or correlation analyses. Given these suggestions and the fact that two of the dependent variables--complementary and symmetrical transactions--ranged from 5 to over 150, and the other dependent variable—-one-up messages--also had a wide range, it was decided to use a transformation suggested by Winer (1962). He states that for proportions, the variances and means have a relationship indicated by the following equation: 120 Oij = Uij(l ' Uij) He then suggests the following transformation where Xijk is a proportion and the result is an angle measured in radians. For proportions between .001 and ijk = 2 arcs1n YXijk .999, Xijk assumes values between 0.0633 and 3.0783. All statistical tests applied to the data for hypotheses test- ing used this transformation, although, for descriptive purposes, means will be expressed in both radians and percentages. For the analyses used to test the hypotheses, the dependent variables--expressed in proportions--were trans— formed by a computer subroutine utilizing the arcsin library function. Since the transformed variable could only assume values for proportions greater than 0 and less than 1, any proportions which were found to be these latter values were changed to 0.001 and 0.999, respec- tively, for all statistical tests. Statistical Anaiyses The statistics utilized to test the hypotheses in this study were of four types: a t test to examine the difference between the correlated means of the proportion of complementary and symmetrical transactions; a 2x2 121 Factorial Analysis of Variance to check the predictive ‘power of the social class and dominance difference score 'variables; the Pearson Product Moment Correlation Coeffi- cient to test the relationship between dominance scores and.the proportion of one-up messages; and a 2 test to (explore the difference between two Pearson Product Moment (Sorrelation Coefficients--the r for husbands' dominance sscores and their proportion of one-up messages and the r :Eor wives' dominance scores and their proportion of one- tip messages. An a level of 0.05 was established for re- ;jection of the null hypotheses, and statistical tests Ivere two-tailed. PQKpothesis l This hypothesis was based on Carson's theoretic Ixrediction that complementarity is a preferable and more :Erequent interaction phenomenon than is anti«comp1ementar- ity (symmetry). Stated in terms of relational communica- ‘tion, it predicts that husband-wife dyads will have a isignificantly higher proportion of complementary trans- iictions than symmetrical transactions. [Since there was 110 a priori reason to assume that couples would differ in their use of complementary or symmetrical transactions across topics, the proportions used to test this hypothe- Sis were based on the total number of transactions across three topics. In addition, symmetrical transactions 122 represent the sum of competitive, neutralized and submis- sive symmetrical transactions.] Rephrased in the null form the hypothesis predicts that husband-wife dyads will not differ in regard to their proportion of complementary and symmetrical transactions. Because the dependent variables--complementary and symmetrical transactions“- were measured on the same subjects, a t test for correlated means was applied to the arcsin means. As shown in Table 7, the mean proportion of com- plementary transactions as expressed in percentage form is 12% whereas the mean proportion of symmetrical transactions as a percentage is 29%. The difference between the two is significant but is obviously in the wrong direction predicted by H1. Symmetrical transactions made up a higher proportion of all the transactions that occurred than did those designated as complementary. As a result, H10 is rejected, but in this case, H1 is not the true alternative hypothesis. Table 7. t test between mean proportions of complementary and symmetrical transactions. Complementary Symmetrical Transactions Transactions t df p Mean Mean (Expressed in radians and percentages) 0.69/12% 1.14/29% 14.92 55 <0.0001 123 Hypotheses 2 and 3 These are two of four hypotheses drawn from Carson's ideas and previous research on relational com- munication. H2 was an attempt to predict, from two indi- vidual's dominance scores and their social class, what proportions of complementary communication behaviors would occur in a decision-making situation. Specifically, it predicted that husband-wife dyads who have different dominance scores will have a significantly higher propor- tion of complementary transactions than will husband-wife dyads who have similar dominance scores, regardless of social class when the dyad is involved in a decision- making discussion. Stated in the null form, the hypothe- sis predicts that husband-wife dyads who have different dominance scores [i.e., ”high dominance difference." (See Chapter VI)] will not differ in their proportion of complementary transactions from husband-wife dyads who have similar dominance scores [i.e., "low dominance difference,"] regardless of social class when the dyad is involved in a decision making discussion. H3 was based heavily on Mark's (1970) research that showed that lower-class couples had used complemen- tary transactions more than did upper-class couples. Like H2, it predicts that high dominance difference dyads will have a significantly higher proportion of complementary 124 transactions, but more specifically, it predicts that lower-class couples who have different dominance scores [i.e., high dominance difference] will have a significantly higher proportion of complementary transactions than will upper-class couples who also have high dominance differ- ence scores. In null form, this hypothesis predicts that husband-wife dyads from a lower—class environment who have different dominance scores will not differ in their pro- portion of complementary transactions from similar dominant type dyads from an upper-class environment when the dyad is involved in a decision-making discussion. These hypotheses were tested with a 2x2 Factorial Analysis of Variance with factors of Social Class (high and low) and Dominance Difference (high and low). As in the case of H1, the proportions of complementary trans- actions were based on sums across three topics. The re- sults of this analysis of variance and the results of a special test applied to the data for H3 are shown in Table 8 and Table 9, respectively. As the results in Table 8 show, the significant main effect is accounted for by Social Class and not be Dominance Difference as predicted. In addition, an examination of the cell means (Table 8) and the results of an F statistic (Table 9) suggested by Lindquist (1953, p. 213) shows that rather than lower-class high dominance difference dyads having a higher proportion of complementary transactions, the reverse is the case and 125 Table 8. Proportion of complementary transactions by social class and dominance difference. Cell Means Dominance Difference Social . Class ngh Low (expressedin radians and percentages) Upper 0.73/13% 0.72/12% Lower 0.61/9% 0.68/11% Analysis of Variance Table Source of Variance SS df MS F p Social Class 0.11 l 0.11 5.09 <0.05 Dominance Difference 0.01 l 0.01 * ns Social Class by Dominance Difference 0'02 1 0°02 1'02 ns Error 1.08 52 0.02 Total 1.22 55 *less than 1. Table 9. Comparison of upper- and lower-class high domi- nance difference dyads' mean proportions of complementary transactions. _- r 4.7 -____1 4.— i ——‘ _: F = msAwBi/msw’ df = (a - 1)/(N - ab); F = 0.11/0.02 = 5.32* df = 1, 52 *p <0.05 126 significantly so. Therefore, neither H2O nor H30 are rejected in favor of H2 and H3. Hypotheses 4 and 5 As was the case with H2 and H3, the bases for these hypotheses concerning symmetrical transactions were Carson's theoretic framework and previous relational com- munication research. H4 predicts that husband-wife dyads who have similar dominance scores, i.e., low dominance difference, will have a significantly higher proportion of symmetrical transactions than will husband-wife dyads who have different dominance scores, regardless of social class when the dyad is involved in a decision-making discussion. H5 predicts that husband‘wife dyads from an upper class environment who have similar dominance scores will have a significantly higher proportion of symmetrical transactions than will similar dominant type dyads from a lowereclass environment when the dyad is involved in a decision-making discussion. The null forms of these hypotheses are as follows: H40: Husband-wife dyads who have similar dominance scores will not differ in their proportion of symmetrical transactions from husband-wife dyads who have different dominance scores, regardless of social class when the dyad is involved in a decision-making discussion. H50: Husband-wife dyads from an upper class environ- ment who have similar dominance scores will not differ in their proportion of symmetrical trans- actions from similar dominant type dyads from a 127 lower-class environment when the dyad is involved 1n a dec1s1on-making discussion. Like H2 and H3, these hypotheses were tested with a 2x2 Factorial Analysis of Variance split by Social Class and Dominance Difference. The results of this test and the test of the Social Class factor at the low dominance dif« ference level of the 2x2 design are shown in Table 10 and Table 11, respectively. These results show significant effects for both Social Class and Dominance Difference, but in the opposite directions predicted by the two hypotheses. Rather than low dominance difference cells having a higher proportion of symmetrical transactions, the high dominance difference cells have the higher proportion. Furthermore, the main effect by Social Class indicates that lower-class couples and not upper-class couples have a higher mean proportion of symmetrical transactions. Given these reversals and the insignificant results of the F statistic applied to the upper- and lower-class levels of the low dominance difference level (see Table 11), neither H4O nor H5O are rejected in favor of H4 and HS. Hypotheses 6 and 7 These hypotheses were included in this study to check whether one's dominance score is a predictor of one's controlling or one-up messages. H6 predicts that there 128 Table 10. Proportion of symmetrical transactions by social class and dominance difference. Cell Means Dominance Difference Social Class High Low Texpressed’in redian§_and proportions) Upper 1.10/27% 1.08/26% Lower 1.26/35% 1.12/28% Analysis of Variance Table Source of Variance SS df MS F p Social Class 0.14 l 0.14 8.72 <0.005 Dominance Difference 0.09 l 0.09 5.99 <0.05 Social Class by Dominance Difference 0’04 1 0‘04 2'81 ns Error 0.81 52 0.02 Total 1.08 55 Table 11. Comparison of upper- and lower-class low domi- nance difference dyads' mean proportion of symmetrical transactions. \ v F = df = (a - 1)/(N - ab); msAwBi/msw’ F = 0.01/0.02 = 0.81 df = 1, 52 129 will be significant positive correlation between an in- dividual's score on a dominance scale and his proportion of one-up messages. H7 predicts that men will have a significantly higher correlation between their dominance scores and their proportion of one-up messages than will women. Rephrased in null form, H6O predicts that there will be a zero correlation between an individual's score on a dominance scale and his proportion of one-up messages. H7O predicts that men and women will not differ in the respective correlations between their dominance scores and their proportion of one-up messages. The correlations between dominance scores and proportions of one-up messages are shown in Table 12. Table 12. Correlation coefficients between dominance scores and proportions of one-up messages. r n p Husband 0.100 56 ns Wife 0.116 56 pg_ Total 0.143 112 ns As indicated, none are significant; given the non- significance of the total r, H6O is not rejected. The correlation for the wives' dominance scores and their proportions of one-up messages is 0.116, and that for husbands' dominance scores and their proportions 130 of one-up messages is 0.100. A 2 test was used to test the difference between these two correlation coefficients and the results are shown in Table 13. Although the two Table 13. 2 test for husbands' and wives' dominance score and one-up proportion correlation coefficients. = 0.100; r = 0.116; 1‘H w 2 = 0.100 - 0.117/O.194 Zr r = 0.0876 H W coefficients are in a ratio that is Opposite to that pre- dicted by H7, the 2 test for the difference between the two is not significant and H70 is therefore not rejected. Summary The hypotheses tested in this study sought to predict complementary and symmetrical communication be- haviors from Carson's general notion that complementarity is most probable because it is most rewarding, and from social class findings that have shown lower class groups tend to have fewer messages than do upper class groups. H1 was drawn directly from Carson's framework and pre- dicted a significantly higher proportion of complementary transactions than symmetrical transactions. As was noted in Chapter VI, complementarity was defined in terms of 131 relational communication as a transaction involving a one-up and one-down message. Symmetrical transactions were defined as those involving control dimensions that are similar and the three types were summed to test this first hypothesis. The results of a t test showed signifi- cance, but in the opposite direction; i.e., symmetrical transactions made up a significantly higher proportion of the total than did complementary transactions. H2 predicted that dyads having different dominance scores would have a significantly higher proportion of complementary transactions than would dyads having similar dominance scores. H3 predicted that lower-class couples having different dominance scores would have a signifi— cantly higher proportion of complementary transactions than would upper-class couples having different dominance scores. The results of a 2x2 ANOVA split by class and dominance difference scores showed no effect by dominance difference, although a main effect by social class was found. An F test of the upper- and lower-class high dominance difference cell means showed that the effect by social class was in the opposite direction predicted by H3. H4 and H5 were concerned with symmetrical trans- actions; H4 predicted that dyads having similar dominance scores would have a significantly higher proportion of symmetrical transactions than would couples having 132 different dominance scores; and H5 predicted that upper- class dyads having similar dominance scores would have a significantly higher proportion of symmetrical trans- actions than would lower-class dyads having similar dominance scores. A 2x2 ANOVA showed significance for both social class and dominance difference, but both were in the opposite direction predicted. H6 and H7 were based on a predicted relationship between one's dominance score and his or her proportion of one-up messages. H6 predicted that the correlation between the mens' dominance scores and their proportion of one-up messages would be significantly higher than the correlation between the women's dominance scores and their proportions of one-up messages. None of these correlations were significant. The most consistent finding that emerges from these statistical tests is that social class seems to account for both complementary and symmetrical transactions, although this finding is not What was predicted on the basis of research by Mark (1970). Rather than *~ lower-class dyads showing a higher proportion of comple- mentary transactions, upper-class dyads showed a higher proportion; rather than upper-class couples having a higher proportion of symmetrical behaviors, lower-class couples had a higher proportion. 133 The findings concerning dominance difference scores are also in a direction opposite to that predicted. In the case of H4, high dominance difference dyads had a higher proportion of symmetrical transactions than did low dominance difference dyads. The findings concerning dominance scores may also said to be consistent since no relationships were found between them and one-up message proportions. In short, none of the research hypotheses in this study that were generated from Carson's theoretic frame- work and from previous relational communication research were supported; i.e., predicting from dominance scores, dominance difference scores, and social class led to both non-significant results and to results that were not con- sistent with either theoretical nor previous research results as evidenced by significant but reversed direc- tions on Hypotheses l, 3, 4, and 5. The possible reasons for these findings and a more complete explanation of them will be found in the next chapter. Descriptive Statistics The information in this section concerns the times and probes across topics, totals for both various message and transaction data, and the equal or non-equal use of various transactions by the dyads. In addition to this descriptive information, this section also includes 134 statistical tests done to explore some of the relation- ships found in and suggested by the data. Times and Probes across T6pics One of the findings from previous research on communication research among various social classes is that lower-class groups tend to talk less than upper- class groups. The times by topics and social class for dyads in this study are reported in Table 14. As the Table 14. Times for topics by social class. _—‘ 1 w - .—__ I“ !—— 4— Social . . . Class Top1c l Top1c 2 Top1c 3 Upper 7 min 44 sec 7 min 50 sec 6 min 20 sec Lower 7 min 30 sec 7 min 22 sec 6 min 2 sec means suggest, times do not appear to differ significantly by class although a difference may exist by topic. To check whether times might differ by dominance difference groups, a mixed 2x2x3 Analysis of Variance split by Social Class, Dominance Difference, and correlated Topics was completed. The results of this analysis are shown in Table 15, and confirm the implications drawn from the means in Table 14. In short, times do not differ by class but differ over topics. 135 Table 15. Times by social class, dominance difference and correlated topics. Iwr Analysis of Variance Table Source of Variance SS df MS F p Social Class 17568.59 1 17568.59 * ns Dominance * Difference 26400.21 1 26400.21 ns Social Class by Dominance 11533.71 1 11533.71 * ns Difference Error 2359735.09 52 45379.52 * Topics 276195.08 2 138097.54 15.09 <0.0005 Topics by Social 1117.51 2 558.76 * ns Class Topics by Dominance 19837.75 2 9918.88 1.08 ns Difference Topics by Social Class * and Dominance 3848.89 2 1924.45 ns Difference Error 951798.76 104 9151.91 Total 366803S.59 167 *less than 1. Part of the similarity in times by class may be due to the use of probes. As noted in the previous chap- ter, probes were to be used by the interviewers if, for 136 example, the dyad did not talk for a minimum of 5 minutes. In all, 138 probes were used across the three topics and all couples. However, 51 of these were used for the upper-class dyads with a mean of 1.82 and 87 were used for the lower-class dyads with a mean of 3.10. The results of a t test on these means are shown in Table 16. As indicated by these results, probes do not differ across the two groups and therefore may be assumed not to be reason for the similarities of times across the two class groups. Table 16. t test of probes by social class. ?;:w Social Class Mean SD n t p Upper 1.82 1.89 28 1.61 ns Lower 3.10 3.70 28 The differences in times across topics may have been due to the topics discussed. As was pointed out in Chapter VI, Topic 1 in this study concerned preparations for a nuclear attack. More specifically, couples were asked what they would do if, after turning on the radio, they learned there had been a nuclear attack. In addition, they were asked whether or not they would share their cellar or basement with their neighbors, or, if they had no basement, would they go to someone else's house for protection. Topic 2 dealt with husbands' and wives' 137 feelings about whether both should be able to have inde- pendent careers, jobs, and interests. Topic 3 asked couples to describe what they would do if they learned of a tornado warning and were told they had 15 minutes to take shelter. An examination of the content of the three topics suggests that in Topic 1, couples were dealing with a hypothetical situation which involved many alternatives. In other words, the fact that the situation had never hap- pened before seemed to lead individuals to ask each other questions and speculate on what might happen. Many couples discussed whether or not they would go to their basement, and what they would bring with them for supplies. In Topic 2, the feelings each individual had regarding a "working wife" seemed to occupy much of what was said. This topic, therefore, was one which provided both the husband and wife a chance to air their attitudes. In Topic 3, couples seemed to view the purpose of the dis- cussion as repeating what they had done before in previous tornado warnings. Since most couples had lived in an area which had frequent tornado watches and some tornado warnings, the discussions evoked information on what the couples had done before. This involved going to the base- ment with preparations, going into a centrally located room for shelter, or ignoring the warning. 138 One tentative conclusion that may be drawn from this overview of the content level is that Topic 3 was dismissed by many couples in the sense that they had been through a similar situation before, had or had not made a series of plans which they did or not implement. It seems probable, therefore, that this particular topic was treated the same way as is a question that asks for specific in- formation. Because there was little to be said about pre- parations once the couple had discussed what they were going to do in this "hypothetical" situation, the amount of time spent on this topic was less. Messa e Types by Indiv1duals In an attempt to see what types of verbal mes- sages constituted the various transaction types, messages were summed across individuals and across all three topics. [For totals across 4 topics, see Rogers (1972).] These frequencies and the percentage of the total each repre- sents are shown in Table 17, pp. 103*104. As these results indicate, Assertions as Extensions (code 13) account for 37%, Assertions as Support account for 13%, Successful Talk-overs as Extensions account for 11%, Non-completes as Extensions account for 8%, and Questions as Extensions account for 7% of the total verbal messages. [As dis- cussed in Chapter VI, Successful Talk-overs were defined as those which, when initiated, resulted in the other Table 17. 139 across sample and across 3 topi CS. Frequencies and percentages of message types Message Code (2nd and 3rd digits) f Percent 11 Assertion as support 1291 12.96 12 Assertion as non-support 143 1.44 13 Assertion as extension 3719 37.35 14 Assertion as answer 214 2.15 15 Assertion as instruction 17 0.17 16 Assertion as order 8 0.08 17 Assertion as disconfirmation 8 0.08 18 Assertion as topic change 5 0.05 19 Assertion as initiation-termination 354 3.55 21 Question as support 7 0.07 22 Question as non-support 1 0.01 23 Question as extension 702 7.04 29 Question as initiation-termination 46 0.46 31 Unsuccessful talkover as support 493 4.95 31 Successful talkover as support 204 2.05 32 Unsuccessful talkover as non-support .18 0.18 32 Successful talkover as non-support 38 0.38 33 Unsuccessful talkover as extension 529 5.31 33 Successful talkover as extension 1067 10.72 34 Unsuccessful talkover as answer 5 0.05 34 Successful talkover as answer , 31 0.31 35 Successful talkover as instruction 4 0.04 36 Unsuccessful talkover as order 2 0.02 36 Successful talkover as order 4 0.04 37 Successful talkover as disconfirmation l 0.01 38 Successful talkover as topic change 3 0.03 39 Unsuccessful talkover as initiation- 5 0 05 termination ' 39 Successful talkover as initiation- 5 0 05 termination ' 30 Unsuccessful talkover as other 69 0.69 30 Successful talkover as other 1 0.01 41 Non-complete as support 14 0.14 42 Non-complete as non-support 4 0.04 43 Non-complete as extension 822 8.25 44 Non-complete as answer 12 0.12 49 Non-complete as initiation-termination 29 0.29 40 Non-complete as other 17 0.17 53 Other as extension 5 0.05 59 Other as initiation-termination 4 0.04 50 Other as other 60 0.60 9961 100.00 140 person stopping talking. Talk-overs were considered un- successful when the first speaker continued what he or she was saying despite the second speaker's attempt to interrupt or talk over.] An examination of message types by their format, i.e., digit two, indicates that 57 1/2% of the messages are assertions, 7 1/2% are questions, 25% are talk-overs (unsuccessful 11%, successful 14%), 9% are non-completes, and 1% are other. When the third digit codes are examined, the following percentages are found: Extensions account for 68%, supportive messages account for 20%, non-supportive messages account for 2%, answers account for 2 1/2%, initiation-termination messages account for 4 1/2%, and; the remaining categories account for 3% of the total verbal messages. At a descriptive level, these findings suggest that couples in this sample tended to utilize messages which kept the discussion going, asked and answered ques- tions, and provided support. Disruptive messages, i.e., non-supportive and disconfirmatory messages, did not appear often and one might conclude that, at the individual level, the messages used were what might be expected from a dis- cussion in which couples had been asked to reach a deci- sion. This conclusion, however, is only tentative since no comparisons were able to be made with non-decision-making 141 discussions. The only comparisons possible, therefore, are with what one might logically expect and with results found by Mark (1970). As shown in Table 18, these com- parisons suggest a similarity for assertions as extensions Table 18. Comparison of message types from Mark (1970) and present study. Mark (1970) % Persent Study % Assertion as extension 31 37 Assertion as support 9 13 Talkover as extension 3 16 Question as extension 1 7 and assertions as supportive messages, although the last two categories in the table are not similar across the two studies. Again, these comparisons can only suggest that for discussions in which couples are asked to arrive at a decision, there is a tendency to utilize a large number of extension or continuation messages and suppor- tive messages. One other message type that was coded but was not included in the previous totals is that of silence. Since the frequencies in Table 17 reflect verbal messages, the number of silences were added to the total number of verbal messages to calculate the percent of all coded messages accounted for by silences. In all, over 3 topics, 142 there were 472 silences which account for 4 1/2% of all coded messages. Message Types by Dominance, Socia Class, Sex, and Topics Given the overall lack of support for any of the hypotheses, and, particularly, the one's based on the dominance difference scores, it was decided to further investigate the relationship of both dominance scores and social class to both the individual and dyadic data. Since there was some effect by social class for H2 and H4, couples were split into the two classes and then were separated by a median split in terms of whether or not their dominance difference scores were based on the hus- band's high dominance score or the husband's low dominance score. Then, the difference scores in these high and low husband dominance cells were split at the median to form 2 additional cells. In other words, with each class grouping, 28 couples were split into 4 cells on the basis of both husbands' dominance scores and on the difference scores with their wives. The bases of these splits are shown in Table 19. With the social class splits, each cell of the 4x2 matrix contained 7 couples. Since previous research by both Mark (1970) and Kenkel (1957) had shown differ- ences by sex for certain types of messages, another split 143 Table 19. Ranges of husbands' dominance scores and dyad dominance difference scores. f Husbands Upper C1355 Lower Class Dominance Range of R f Scores/ Husbands' Dominance ange O , Dominance . . . Husbands . Dom1nance Dom1nance Difference D . Difference Difference Scores og1nance with Wife C°res High/Low 16-27 1-6 13-21 1-6 High/High 18-24 9-19 14-24 8-13 Low/Low 7-15 2-6 4-14 1-5 Low/High 0-15 7-16 6-13 5-12 was made by sex. Finally, since there was a possibility that differences may have existed by topics, this was added as a correlated fourth factor to give a mixed 4x2x2x3 Analysis of Variance of Design by which to test the various message types. This design was run on each of the 39 verbal mes- sage types and on silence. Of these 40 tests, 17 proved to be significant, although 7 will not be reported because the data on which they were based represented less than 2% of the total messages. The Analysis of Variance tables for these results will be found in Appendix VI rather than in the text. Grouping the findings by similar effects, the following results concern the effects over topics: (1) Assertions as supportive messages (code 11) were used 144 more often in Topic 2 than in Topics 1 and 3; (2) Asser- tions as extensions (code 13) were used more in Topic 2 than in Topics 1 and 3; (3) Assertions as answers (code 14) were used least in Topic 2 and more in Topics 1 and 3, were used most by low dominant upper class husbands whose dominance scores were quite different from their wives', and were used least by lower class wives whose dominance scores were quite similar to their high dominant husbands'; (4) Questions as extensions (code 23) were used least in Topic 2 and more in Topics 1 and 3, were used most by upper class wives whose dominance scores were quite dif- ferent from their low dominant husbands', and were used least by high dominant upper class males whose dominance scores were quite similar to their wives'; (5) Unsuccess- ful talkovers as supportive messages (code 31) were used most often by upper class subjects on Topic 2; (6) Unsuc- cessful talkovers as extensions (code 33) were used most often by high dominant subjects who had similar dominance scores to their spouses, and were used most often in Topic 3; and (7) Successful talkovers as extensions (code 33) were used most often in Topic 3 by high dominant sub- jects who had similar dominance scores with their spouses and were used least in Topic 2 by low dominant subjects who had different dominance scores from their spouses. Other effects involving social class are: (1) Assertions as initiations-terminations (code 19) were used 145 most often by lower class subjects; and (2) Successful talkovers as supportive messages (code 31) were used most often by upper class subjects. Finally, silence was found to be used most by lower class couples in Topic 1 and was used least by upper class couples in Topic 2. Control Aspects of Messages Another analysis that was conducted at the indi- vidual rather than at the dyadic level was an examination of message types in terms of their control dimensions. On the basis of the definitions of control aspects provided in Chapter VI, message types were summed to provide totals of the three control dimensions. For one-up messages, message types 12, 14, 15, l6, 17, 18, 19, 22, 29, 32, 33, 34, 35, 36, 37, 38, 39, 42, 44, and 59 were summed; for one-down messages, message types 11, 21, 23, 31, 30, and 41 were summed; and for oneeacross messages, message types 13, 43, 49, 40, 53, and 50 were summed. The totals across these frequencies are shown in Table 20, and pro« vide a clue as to why Hypotheses 6 and 7 were not supported. Since these particular hypotheses predicted a positive cor- relation, it seemingly would have been necessary that the proportion of oneeup messages be higher than 25%. This is particularly evident when one examines both the dominance score and the proportion of one-up message distributions 146 Table 20. Frequencies and percentages for control dimen- sions of messages. - At Frequencies Percentage One-up + 2528 25% One-down + 2781 28% One-across + 4652 47% Total 9961 100% in terms of their skewness coefficients. For the former distribution, the coefficient of skewness is 0.2894; for the latter, the coefficient is 0.2408. Because these coefficients do not really differ, the correlation of 0.143 found between dominance scores and proportions of one-up messages is most probably due to the small propor- tion of one-up messages and not due to skewness in the one-up message distribution. In other words, the non- significant correlation apparently reflects a "true" non- linear relationship between one's score on the EPPS dominance scale and his or her proportion of one-up messages. The other message totals in Table 20 also show that with the high percentage of one-across messages, transactions involving one-across messages should have predominated. Given these data and this probability, it was decided to examine all transaction types, even though 147 the original focus of this study was on complementary and symmetrical transactions only. Transaction Types The following information concerns dyadic data, some of which was used to test 5 of the 7 hypotheses. These data represent the five types of transactions defined in Chapter VI, namely, complementary, competitive symmetry, submissive symmetry, neutralized symmetry, and transitory. In order to check the totals for each, the transaction types were summed both within and over topics and are reported in Table 21. Table 21. Frequencies and percentages of transaction types by topic. Topic 1 Topic 2 Topic 3 Total Complementary 13% 11% 12% 12% Competitive 9 Symmetry 7o 6% 7% 7% Submissive 9 9 9 Symmetry 40 4o 40 4% Neutralized 9 9 9 Symmetry 180 17. 17% 17. Transitory 58% 62% 60% 60% Total 100% 100% 100% 100% Message Totals n = 3369 n = 2371 n = 2762 n = 8502 148 These results indicate why H1 was not supported, mainly because complementary transactions represent only about 12% (xf the totals for each topic while symmetrical transactions account for some 27% of the totals for each topic. Of more interest, perhaps, is the possible consis- tency both within and across the three topics. One of Mark's (1970) findings was that transaction types did not differ greatly across topics, a finding that is predicted by many psychotherapists for groups with-a-history. As is evident from the table, none of the transaction types varies more than 4% from Topic 1 to Topic 3, a finding that suggests consistency for types across topics. In short, even though the topics differed in content, the overall method of discussing them at the relational level may not have differed across topics. Such consistency or stability, if found in this study, would coincide with the findings from several studies that have examined various types of communication patterns. For example, research by Dreschler and Shapiro (1963), Haley (1964), and Murrell and Stachowiak (1967) has shown that inter- action patterns of both normal and abnormal families are fairly stable, although Haley found that family triads with a schizophrenic, delinquent, neurotic or otherwise abnormal member had more rigid interaction patterns than did normal family triads, while Murrell and Stachowiak 149 found the opposite. In order to ascertain whether the husband-wife dyads in this study had stability within topics as well as across topics in their use of trans- action types, a method suggested by Haley (1964) was used to examine the relational patterns. Haley's concern was in identifying patterns of the order in which family members speak. With a triad composed of a father (F), mother (M) and child (C), there are six possible categories of speaking order--FM, FC, MF, MC, CF, or CM. To check whether these patterns were repetive, Haley hypothesized that "if the order in which family members speak differs from that order which would occur with random behavior then the family is following patterns which repeat." (1964, p. 52.) Since random responses would occur equally on an infinite run if family members were behaving randomly, each of the speaker order categories would occur 16.66 per cent of the time and large deviations from these values would give an indica- tion of non-random patterns. By subtracting 16.66 from each family's actual use of a particular category, ignor- ing the sign and summing the differences between actual and expected percentage of category use, Haley calculated an R Deviation or deviation from random index. From this index, he could tell whether a family had a rigid pattern of speaking order, indicated by a large R Deviation, or whether the patterns tended to occur equally, indicated 150 by a small R Deviation or one close to zero. The results with 80 families--40 normal and 40 abnorma1--showed that the normal family R Deviation mean, 19.16, was significantly less than the abnormal family R Deviation mean, 29.46. This procedure was modified by this writer to check whether dyads differed from a random distribution in their use of transaction types within topics, and to check whether differences existed among groups categorized by their dominance difference scores and their social class. With five types of transactions, each would occur 20.00 per cent of the time if transactions occurred ran- domly. Using Haley's technique, an R Deviation was calcu- lated for each dyad by subtracting 20.00 from each category's occurrence and summing the absolute differences. Pure randomness would be indicated by an R Deviation of zero and rigid use of one transaction type would be indicated by a deviation score of 160. These R Deviations were first calculated for the total sample across the three topics, were plotted, and are shown in Figure 4. As the figure shows, the range of R Deviation Means across three topics is from 65 to 120 with an over- all mean of 89.03 and a standard deviation of 12.21. Although somewhat leptokurtic, the distribution approaches normality and indicates that 56 dyads differed over three topics in their use of various transactional patterns. Even more interesting is the fact that the distribution shows non-randomness in the use of transaction types. 151 .mofimou oopnp mmouom mauve opoom :ofiumfl>om m .w madman Anamoev monoum :oflpmfi>on m OOH OmH OOH omH ONH OHH OOH OO OO O5 O0 Om OO om ON OH O _ . a . . _ q . . la 4 l q . 4 a O .N .v HN.NH u Om m0.00 u M Hmuoe x .O .O .OH speAp go leqmnN 152 This non-randomness or stability coincides with the find- ings mentioned earlier and confirms the notion that there are stable patterns of interaction between husband-wife pairs. However, in order to further check this stability across topics, R Deviation Scores were computed for each dyad on each topic, were plotted, and are shown in Figure 5. These distributions indicate that couples tended to have similar variations in their use of transactions for Topics 1 and 3 while the variability in use of patterns for Topic 2 was somewhat less. Since a lower R Deviation mean score indicates a more equal use of each transaction type, the Topic 3 R Deviation mean of 95.75 and the 7 scores that are equal to or greater than 120 show a some- what rigid use of or two transaction types. To check whether or not the R Deviation means were significantly different and to see whether or not dominance difference scores, social class, and topics might account for the differences among the means, a 4x2x3 Analysis of Variance--split by the four levels of dominance difference described earlier, by the two levels of social class, and by the 3 correlated topics-—was run. The results of this ANOVA are shown in Table 22. The only significant finding is by topics, a finding that indicates that dyads used fewer transaction types and maintained a narrower range of communication behavior for 153 OOH OHH .monop kn mohoum :oHumH>on m .m ohsmHm monoum coHumH>oo m OOH OO OO On OO Om ow om ON OH O omH OVH omH ONH l I 11 d u d N n .N w. 3 l L? O 1.. £1: ... mm .o m. 3.3 "win; m. H onoH .O s OH\O N n m 1N w l .v 0 3.3 u mm 3 mimmuwima .c m. N UHOOH e w 3o : m m 0. I? M” O 3.2 u Om .o J 3.3 "mica P m que .m m w 154 Table 22. R Deviation scores by dominance difference, social class, and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 2652.18 3 884.06 1.98 ns Social Class 17.76 1 17.76 * ns Dominance Difference * by Social 495.02 3 165.01 ns Class Error 21419.68 48 446 24 Topics 4121.88 2 2060.94 9.58 <0.0005 Dominance Difference 2694.39 6 449.07 2.09 ns by Topics Social Class * by Topics 7.86 2 3.93 ns Dominance Difference by Social 1677.64 6 279.61 1.30 ns Class by Topics Error 20644.62 96 215.05 Total 53731.03 167 *less than 1. Topic 2. This finding also indicates that the assumption drawn from the percentages of transaction types in Table 21 concerning stability of patterns was incorrect. Rather 155 than discussing the three topics in a similar fashion, a factor that had been an a priori assumption, dyads treated Topic 2 differently in terms of their relational patterns. Further evidence of this is provided by an Analy- sis of Variance that was run on complementary transactions with factors of dominance difference (4 levels), social class, and correlated topics. As the result in Table 23 show, the main effect by social class still exists (and in the same direction as found in the ANOVA run for H20 and H30) and a main effect by topics is also evident. With a mean of 0.66 for Topic 1, a mean of 0.56 for Topic 2, and a mean of 0.69 for Topic 3, it seems evident that Topic 2 was a "different" topic. [Symmetrical transac- tions were also run in a 4x2x3 ANOVA and the only significant finding was by social class, a finding that coincides with the ANOVA run to test H4 and H5 0 0'] To illustrate the various types of interaction patterns found in Topic 2, samples from dyads who had R Deviation Scores of 40, 95, and 160 are shown below. The first sample is taken from an upper class dyad in which the husband's dominance score was low and the wife's dominance score was even lower, giving a high dominance difference. The interactions below are taken from the last third of the discussion and begin with a question by the wife. 156 Table 23. Proportion of complementary transactions by dominance difference, social class, and correlated topics. Analysis of Variance Table 3233.2” 88 df MS F P gggigggfige 0.02 3 0.01 * ns Social Class 0.38 l 0.38 4.84 <0.03 Dominance g;f§::§:§e 0.04 3 0.01 * ns Class Error 3.76 48 0.08 Topics 0.52 2 0.26 3.45 <0.04 Dominance Difference 0.74 6 0.12 1.61 ns by Topics ggciggigiass 0.10 2 0.05 * ns Dominance 2332:6252? . * us by Topics Error 7.31 _96 0.08 13.32 167 *less than 1. 211': E II: 2:13 zm 2 157 What do you do in the evenings with our family? What do we do together? In the summertime we have it made. We all go swimming; but what do we do in the winter time, which is much longer than the summer? Are we talking about disciplining the children or are we talking about family activities? Family activities is what I'd like to discuss because we completely (0h) disagree on disciplining the children. Don't you feel like I do the (If. you) major portion of that? And I have to agree that I do it poorly. If you consider the full day, yes. But the time that we're here together, no. Well, just for the record, I have to dis- agree with that. Well, just for the record, you're wrong. Okay. What do we do as a family? Okay, what about family activities. As the control dimensions show, 4 of the 10 trans- actions are transitory, 3 are complementary, l is submissive symmetry, and 2 are competitive symmetry. This somewhat random pattern in this sample is characteristic of the transaction sequences found in this couple's discussion of of of in consistency of patterns as they dealt with it. cates the volatility of the topic for them.] independent careers for a wife, and reflect the lack [It is interest to note that 22% of this couple's statements this topic were non-supportive, a finding that indi- Another type of transaction sequence is illustrated by a lower class couple, whose R Deviation Score is 95, and whose members both had high dominance scores resulting in a low dominance difference. 158 The following sample occurs near the beginning of the discussion and begins with a statement by the wife. W 2:: 22122 2:13 2332 3 Some nights, one of us is gone every night of the week but others we're home to- gether. Yesterday was... [It's really hard to say because I believe in being in community groups and helping where I can. If somebody knows something of importance, I think I should know that and be able to help people, cause you're only put on this earth for one thing, and that's to help each other. Anyway, every* body doesn't know everything. As long as it doesn't put too much strain on a rela- tionship, I think it is a good thing to belong to different organizations.] Yeah, but a lot of... [That are beneficial] Well, I go along with a lot of what you said but of course, when you first get married you don't have any children, your wife might as well work, because you're furnishing a home and everything you want and I think (Right) to rely on the husband, well, you're young and a lot of times you're not making that much and you need the money. As long as it doesn't interfere with the normal routine. Oh, no, but [Like some people that we know.] [Well the husband works one shift, the wife works the other, you] (The wife works another.) wave as you're going in and out the door. 1133 113+ 233+ 113‘ \ 4L»; f I I 231+ 113+\ '3 w++ 11 I 213% 3++ 143+’ 233+ :: 133+\ :++ ++ ++ +9- 233! 143+ In this particular sample, 8 of the 12 transactions are trasitory, 2 are competitive symmetry, and 2 are neu- tralized symmetry. As the previous sample of interaction illustrated a somewhat random use of transaction types, this sample shows more use of the transitory type. 159 The third type of transaction sequences is that from a couple whose R Deviation Score is 160, who are lower class, and who both have low dominance difference scores . The sample below represents the entire discussion of Topic 2, although in abstracted form since in 8 minutes, the couple had only 7 transactions. H ESE Now, now my opinion on this now. I figure 219+ when two people first get married, I be- \ lieve, I don't, I actually believe in \ divorces that people don't get along ‘ perfect. . . . But, like I say, I don't y believe a woman should work, except for + the first year, now she could work but 1 after that I believe her place is at home, ' unless they decide not to have a family. : l l . . . I don't know what your opinion is of it now. Well, I'd say the same. Well I know I 113+ worked from 65, we got married in 65, I worked up until the year 68. I had no ‘ children. . . . But like I say, he didn't I want no children the first year and al- ++ though I wanted children right away, but 1 uh, he never prevented from having child- I ren at any time. , That's true 211+ But uh, after a year we were wanting 113+ more, both of us did. . . . I'm not the \ type of person that has to go all the \ time. . . . I feel that before I let my & home go or my furniture or anything I would + feel that I should go and try to keep it 9 til he got in, but other than that I don't I feel a woman should work, if she can make I a home. And that's about all I could say I on the subject. I I agree on that last part. That like in 211+ a emergency now. I never thought about 213+ that til you said it, but if a man's laid \ off or disabled or something like that, ‘ like you say, most of the time you've got ‘ insurance to cover it, but your insurance \ don't last all your lifetime. . . . That ‘ walkout one time I worked shop it was, I \ was treated like a king, we had no kids 160 then, but like she said we didn't try to / prevent any. Matter of fact, I think you , got pregnant the same summer. I W Yeah 111+ H The same as our walkout, when you got 213+ pregnant, yeah. And us, stuff like that. \ Now that's about the only time I could \ see it, if it's an emergency, but like I \ say, I believe a woman's place is in the \ house after their first year of married \ life, because in that first year, like )9 you said, you work, there's so many things f you can buy, your furniture, and television, I and like a car. . . . There's so many / things to do. Keeping the house clean, I stuff like that. That's about all I can I think of. ’ W That's all I can say, too. 119+ As is obvious from the control dimensions, all 7 transactions are of the transitory type, a finding that seemed to occur in discussions where there were a series of, in effect, monologues. These particular samples indicate the different types of sequences used to discuss the topic. As noted earlier in the discussion of times, Topic 2 provided a chance for each spouse to voice his or her opinion con- cerning the independence of the wife; for some couples the topic appeared to be controversial, for others it seemed an area about which little had been said, and for still others it appeared to be something upon which both spouses had previously agreed. Given these various modes of airing opinions and the fact that the topic seemed to generate a different type of content--i.e., deeply held opinions of something rather than information about 161 something~-it is not too surprising that couples inter- acted differently on this topic than on Topics 1 and 3 in terms of their relational communication. In order to see whether the percentages of other transaction types reported in Table 21 differed across dominance difference, social class and/or topics, and in order to provide further insight as to what contributed to the high R Deviation Mean for Topic 2, a 4x2x3 ANOVA was run for competitive symmetry, submissive symmetry, neutralized symmetry and transitory types of transactions. The results of these tests are reported in Tables 34-37 in Appendix VII. These show that competitive symmetry differed significantly across topics with the lowest pro- portion appearing in Topic 2 and the higher proportions in Topics 1 and 3. Submissive symmetry differed over dominance difference groups and topics: the highest pro- portion of submissive symmetrical transactions was used in Topic 3 by dyads in which the male was of high dominance and wife was of low dominance; the lowest proportion of this type of transaction was found in Topic 2 for dyads in which the husband was of low dominance and the wife was also of low dominance. Neutralized symmetrical transac- tions differed significantly by class, with the lower class having the higher proportion. No significant dif- ferences were found for transitory transactions by domi- nance difference groups, social class or topics. 162 These findings indicate that, although competitive symmetry was not a major pattern, it contributed to the higher R Deviation Mean of Topic 2. This was obviously due to the lower proportion of this type of transaction in Topic 2 as compared to the proportions in Topics 1 and 3. On the basis of an examination of the content of the topics, it was suggested earlier that Topic 2 was differ- ent from Topics 1 and 3 because it dealt with attitudes regarding the independence of the wife, and, therefore, might have been expected to have more competitive symmet- rical transactions. However, the opposite to what might have been predicted was found. It may well be that competitive symmetry was used more in Topics 1 and 3 because they were least likely to generate severe arguments. Given the "public” nature of the discussions, couples may have been less likely to let their transactions in Topic 2 lead into a struggle for control that often is characterized by major disagreements or severe arguments. It may also be that the findings concerning neutralized symmetry--namely, that lower class couples used a significantly higher proportion than did upper class couples--reflect the tendency of lower class couples to disclose less about themselves and their be- haviors in situations where they are, in a real sense, communicating to an interviewer and an unknown group of higher class researchers. Since Topic 2 dealt with an 163 area that might have been a source of disagreement for many couples, and since a competitive symmetrical rela- tionship would have highlighted any disagreements, the relationships that did emerge may have been partly the result of the dyads' refusal to become engaged in public arguments over the husband's and wife's respective atti- tudes. [This speculation seems similar to a notion raised by Goffman (1959) in his discussion of the pre- sentation of self by members of a team of performers who interact within a social establishment. As he says, concerning the members of the team, we find that familiarity prevails, solidarity is likely to develop, and that secrets that could give the show away are shared and kept. A tacit agreement is maintained between performers and audience to act as if a given degree of opposition and of accord existed between them. Typically, but not always, agreement is stressed and opposition is underplayed. (p. 238.)] Summar The descriptive statistics in this study were con- cerned with both individual and dyadic data. It was found that topic times did not differ across social class or dominance difference, but did differ significantly over topics. Whereas the average times for Topics 1 and 2 were approximately 7 minutes, 30 seconds, the average time for Topic 3 was 6 minutes 11 seconds. The lower time for the last topic was probably due to dyads treating 164 it very matter-of'factly; since it involved preparations for a tornado warning and since it was something most couples had gone through before, the discussion involved few speculations and took less time. Another finding at the individual level of analysis was that Assertions as extensions accounted for 37% of all message types, Assertions as support accounted for 13%, Successful talkovers as extensions accounted for 11%, Noncompletes as extensions accounted for 8%, and Questions as extensions accounted for 7% of the total verbal mes- sages. These suggest that dyadic members tended to use messages which were functional--they kept the discussion going, provided support for the other, asked questions, and/or provided answers. Few disruptive messages were found in the total sample. One other interesting finding concerns the per- centage of messages that were talkovers. Of the total messages, 25% were talkovers, with 14% successful and 11% unsuccessful. This suggests that most dyads did not seem to have a "rule" for allowing one person to finish what he was saying before he was interrupted or talked over. Instead, dyads seemed to operate as if there was a rule that said, "we may interrupt or talk over at any time." In any case, talkovers were used either to support the other or extend what the other was saying. 165 When message types were examined by dominance dif- ference, social class, sex, and topics, the following major findings emerged: Assertions as support and Asser« tions as extensions were used more in Topic 2 than in Topics 1 and 3, while Questions as extensions and Asser- tions as answers were used more in Topics 1 and 3 than in Topic 2. Also, upper class subjects used more Successful talkovers as support than did lower class subjects, while lower class subjects had a higher proportion of Assertions as initiations-terminations. At the message level, 47% of all messages had the one-across (+) control dimension, 28% had the one-down (+) control dimension, and 25% had the one-up (+) control dimension. When these are broken into transaction types over the three topics, 60% (If the transactions were transitory, 17% were neutralized symmetry, 4% were submissive symmetry, 7% were competitive symmetry, and 12% were complementary. Finally, a check of the actual occurrence of all transaction types against a random distribution of expected occurrence showed that the variation in stability of pat- tern or transaction usage was somewhat normal, and that the overall R Deviation Mean of 89.03 indicated non- randomness in the use of transaction types. Further analy- sis by dominance difference, social class, and topics indicated that dyads treated Topic 2 differently in terms 166 of their relational communication--i.e., there was more rigidity in the use of certain transaction types on this topic than on Topics 1 and 3. Such rigidity was found to exist due to the major use of transitory transactions. CHAPTER VIII DISCUSSION OF RESULTS The factors to be discussed in this chapter concern: (1) the general issue of complementarity; (2) the use of dominance scale scores to predict relational communication transactions; (3) the significant social class findings that were opposite to what was predicted; (4) the general stability of transaction types; (5) the effects of the topics on the stability of transaction types; (6) the advantages and disadvantages of the relational communica~ tion coding scheme; and (7) suggestions for further research. Complementarity The information presented in Chapters II, III, and IV concerned the theoretical frameworks set forth by Leary, Thibaut and Kelley, and Carson and the general notion that complementarity is the expected form of in- teraction behavior because it is the most rewarding. One would expect, therefore, to find behaviors in dyadic in- teractions that complement each other. Or, stated and defined in terms of relational communication concepts, 167 168 one would expect to find a one-up message from one person followed by a one-down message from the other, and vice versa. With this operationalization, it was predicted that there would be a higher proportion of complementary trans- actions than symmetrical transactions. As shown by the results of the t test in the previous chapter, the reverse was found to be the case, and significantly so. Further- more, an examination of the various types of transactions indicated that the complementary type accounted for no more than 12% of the total transactions, while the three types of symmetry accounted for 28% and transitory transactions accounted for 60%. One explanation of these results is that the coding scheme may have been heavily biased in favor of one-across messages such as Assertions as Extensions. That is, the rules for coding messages indicated that the third digit 3 code or extension category be used when there was ambiguity as to what response mode a message should receive. Rather than serving as a residual category in cases where coders were unsure of the response mode, the 3 code may have served as the "easy" category because it was neutral. Therefore, fewer messages may have been coded as one-up or one-down which could have led to fewer complementary transactions. 169 Another, perhaps, more useful explanation of these results is that the conceptual definition of complemen- tarity in relational communication terms was incorrect for the coding scheme and should be changed. The definition used in this study was based on the prevailing conceptual definitions of relational communication transactions set forth by writers like Haley (1958), Jackson (1959), Bateson and Jackson (1964), Watzlawick (1964), and Watzlawick, Beavin and Jackson (1967). Defined as a transaction having a one-up and one-down message, complementarity was found to account for a small percentage of the total. Since this study was exploratory in the sense that it was the first time the modified coding scheme was used, no clear expectations of the transitory transactions had been formed. As a result, given the high percentage of one- across messages (47%), the expectation that complemen- tarity would account for a large proportion of the transactions was in error. From this, however, stems a potentially important and useful suggestion if this coding scheme is to be used to further test some of Carson's predictions. Rather than defining a complementary transaction as being only one-up/one-down messages, the definition could include those transactions defined in this study as transitory-- i.e., one-up/one-across and one-down/one-across message combinations. This definition would then reflect three 170 types of complementarity and would still allow for three types of symmetry. Such a definition would also meet Levinger's criticism of the tendency to measure comple- mentarity in terms of differences great enough to give a significant negative correlation coefficient. Although Levinger's (1964) concern is in terms of the complemen- tarity of needs examined by Winch (1955a, 1955b) and others, his argument has merit at the conceptual level too for the relational communication concept of comple— mentarity. For one thing, although the idea of complementary transactions implies polar ends of a continuum of con- trolling behavior and has been defined as such, comple- mentarity can still be said to exist even though behaviors are not really opposite. For example, if one person asks a question, the other answers, and the first person extends the answer, the second transaction would be coded as one- up/one-across. This type of transaction, termed transitory in this study, is, at second glance, an example of a rela- tional communication transaction in which one person complements the message of another without using a diametrically opposed control dimension. For another, extending the definition of comple- mentary transactions to include the transitory transactions increases the importance of one-across messages. Instead of being considered simply a neutralized type of control, 171 such messages could be viewed as a mid-point on a dominance-submission continuum, thereby making the opera- tionalization of relational communication complementarity more logically reflect the continuity implied by the wider conceptual definition. Dominance Scores Six of the seven hypotheses in this study were con- cerned with the relationship between dominance scores and relational transactions or one-up messages. The general notion that scores on a dominance scale would reflect complementarity was based on research by Winch (1955a, 1955b), Winch, Ktsanes and Ktsanes (1954), Murstein (1961), Rychlak (1965), and Katz, Glucksberg and Krauss (1960). As the review of the literature indicated, the concept has been most often operationalized in terms of a negative correlation coefficient, a methodological procedure that both Levinger (1964) and Becker (1964) have criticized. Both argue that a negative correlation coefficient pro— vides a test of only one aspect of complementarity, since a significant correlation coefficient may indicate the relationship between two end points of a continuum. In order to minimize this problem in this study, dominance- submission complementarity was operationalized, following Becker (1964), in terms of dominance difference scores split at the median. 172 The results of the tests of the hypotheses indicate however, that none of the predictions based on dominance were supported. Furthermore, in several instances, the reverse of what was predicted was found to be significant. In other words, what had been operationalized by Becker (1964) and this writer as dominance-submission complemen- tarity in terms of scaling behavior did not predict com- plementarity in terms of relational communication behavior. What was complementarity in terms of the EPPS scale values of the husbands and wives predicted symmetrical communica- tion behaviors, while what was symmetry in terms of the dominance scale values predicted complementary communica— tion behaviors. These results are partially clarified by an examination of the correlation between husbands' and wives' dominance scores, the statistic used by Winch. For this sample, the r between husbands' and wives' dominance scores is 0.273,}><0.01, a significant reversal to what would be expected if there were complementarity. Given this finding, it seems obvious that dominance scores do not predict relational communication behavior. Further evidence that dominance scores are also not good predictors of individual communication behavior comes from the 4x2x2x3 ANOVA's run on message types. Whereas the logical prediction would be that high dominant indi- viduals would have more answers and fewer questions than low dominant individuals, it was found that low dominant 173 upper class husbands whose dominance scores were different from their wives' used the highest proportion of Assertions as answers, while lower class wives whose dominance scores were similar to their high dominant husbands' used the lowest proportion of Assertions as answers. In addition, it was found that upper class wives whose dominance scores were quite different from their low dominant husbands' used the highest proportion of Questions as extensions, while high dominant upper class males whose dominance scores were quite similar to their wives' used the lowest proportion. The only findings which might coincide with predic- tions based on a dominance scale were that Unsuccessful talk-overs as extensions were used most often by high dominant subjects who had similar dominance scores with their spouses, and Successful talk-overs as extensions were used most often in Topic 3 by high dominant subjects who had similar dominance scores with their spouses, and were used least in Topic 2 by low dominant subjects who had different dominance scores from their spouses. Even though these latter findings seem reasonable in terms of what might be expected from a dominant indi- vidual, the overall inconsistency of findings offers little support for the notion that scores on the EPPS dominance scale can accurately predict relational patterns or relational messages. One reason for this may relate to 174 how subjects perceived the scale itself. If they treated the scale as a novelty, then the scores would probably not reflect those feelings that indicate a degree of dominance. However, since no other scale was administered at the time to check this validity, this conclusion is extremely speculative. Probably a more important reason for the lack of predicted relationship relates to the two types of behav- ior involved. On the one hand, husbands and wives indi- vidually filled out the dominance scales; on the other hand, husbands and wives then interacted over three topics. Although this writer accepts the argument that filling out the dominance scale and interacting with one's spouse are both forms of behavior, he does not agree that they are in the same domain. Scaling behavior or filling out the dominance scale represents a non-interaction phenomenon; it is something done independently without the vocal support or cues from one's spouse. Discussion of topics, however, represents a systems' phenomenon that epitomizes the interdependency of the husband and wife. In short, predicting from the former phenomenon to the latter is not very useful, and leads to the conclusion that the predictions from Carson's framework should have bypassed the dominance scaling and gone directly to the interaction behaviors. That is, rather than predicting from the theory to the dominance scale scores and from the scores 175 to the complementary and symmetrical transactions, a more useful approach would have been to predict from the theory directly to the relational transactions. Social Class Findings The predictions regarding social class were felt to be consistent with previous research. The results, however, were the opposite to what had been predicted. Where lower class pairs had been predicted to have a higher proportion of complementary transactions, upper class couples actually had a significantly higher propor- tion. Where upper class pairs had been predicted to have a higher proportion of symmetrical transactions, lower class couples had a higher proportion. Although these predictions were based primarily on Mark's (1970) study, these results contradict his findings. One reason for this may relate to the two coding schemes. As discussed earlier, the coding scheme in this study was a modification of Mark's and increased the con- trol dimensions of a message from two to three, and changed the number of transactions from the basic two described by Haley (1958) and Bateson and Jackson (1964) to five. Since the categorizations were slightly differ- ent, the proportions of complementary and symmetrical transactions in the two samples may also have been quite different. A definitive "reason" as to why these 176 relationships emerge is not available at this time, how- ever, because dominance was found to be a non-predictor and there is no research that relates relational communi- cation among social class groups to any personality or ethnological variables. It may well be that no single predictor variable of relational communication among social groups can be isolated; rather, a series of vari- ables may have to be found that can accurately predict the occurrence of transaction types. Until then, one is faced with describing what was found. At this descriptive level, the two social class groups differed, as pointed out earlier, in their propor- tions of both complementary and symmetrical transactions. They did not differ in the overall stability in their choice of various transaction types as shown by the R Deviation Scores. They did differ in the use of silence over topics, with lower class couples having the most silences in Topic 1 and upper class couples having the least silences in Topic 2. At the individual level, upper class subjects used more Successful talkovers as suppor- tive messages, and lower class subjects used more Asser- tions as initiations-terminations. [This finding is related to the use of probes. When a period of silence longer than 15 seconds occurred in a discussion, inter- viewers were told to use a probe. Since more probes were 177 used for lower class couples, messages following a probe were coded as an initiation.] Stabilipy of Transaction Types One of the consistent findings in research on in- teraction patterns is the general stability of patterns over tasks and topics. These findings illustrate that families are organized systems, behaving in an organized fashion with patterns that repeat. In order to see whether or not such stability of patterns was a charac- teristic of this sample's relational communication behav- ior during their discussions of three topics, an R Deviation or deviation from a random distribution score was calculated for each dyad. Tested with a 4x2x3 Analy- sis of Variance split by 4 levels of dominance difference, 2 levels of social class, and 3 correlated topics, the results indicated only a significant effect for topics. The findings show that there was stability across all types of dyads regardless of their dominance scores and their social class and across Topics 1 and 3, although the rigidity or use of fewer transaction types increased for Topic 2. Even with these differences across topics, however, the means are more than half-way from 0, a score indicat- ing that patterns are total random, and slightly less than half-way to 160, a score indicating total rigidity in the 178 use of one pattern. In other words, couples in this dample did have stability in their use of transaction types and did not behave in a random manner. Effects of Topics As noted above, stability in the use of relational transactions or patterns differed only across topics, where, in Topic 2, there was slightly more rigidity in the use of transaction types. This finding is of interest, not only in terms of the stability notion, but also in terms of the slightly different way in which the dyads treated this topic. The topic itself dealt with an independent career, job, or interest for the wife and the results of the 4x2x3 ANOVA show a significant difference in terms of how the dyads dealt with it. Obviously, the belief that the three topics would be treated the same as decision-making dis- cussions was not borne out. As will be recalled, Topic 1 dealt with preparation for a nuclear attack and sharing of the home basement, while Topic 3 dealt with preparations for a tornado warning. Topic 2 was obviously different in content and generated different information. Where TOpic l and Topic 3 described situations that asked for somewhat specific information, Topic 2 provided the dyads a forum in which to voice their, often deeply held, attitudes about the independence of the wife. Topic 2, therefore, 179 seemed to be a somewhat volatile topic for some couples. Some evidence for this is found in the results from a 4x2x3 ANOVA run on complementary transactions. If comple- mentarity--defined as one-up/ond-down messages-~can be said to be an index of non-conflicting or non-competing messages, then the fact that Topics 1 and 3 had a signifi- cantly higher proportion of complementary transactions than did Topic 2 tends to confirm a content difference. These differences at the dyadic level of transac- tion usage were matched somewhat by differences at the individual level that further illustrate the differences in how the topics were treated. For example, results from the series of 4x2x2x3 ANOVA's run on message types show that Questions as extensions and Assertions as answers were used significantly more in Topics 1 and 3 than in Topic 2; and, Assertions as supportive messages and Assertions as extensions were used significantly more in Topic 2 than in Topics 1 and 3. Other individual find- ings that represent interactions between individual, class, sex, and topic also support this difference between the topics and lead to the conclusion that the type of content about which a dyad is asked to reach a decision can affect the way the dyad will communicate relationally. 180 Advantages and Disadvantages of CodingScheme The data analyzed and discussed in this study resulted from the use of coding scheme originally developed by Mark (1970) and modified by this writer and Rogers (1972). The basic reasons for modifying the scheme were to make the scheme more internally consistent, to add another control dimension for messages, to increase the number of symmetrical transactions from one to three, and to create a new neutralizing or leveling of control transaction. The scheme was taught to three coders over 3 separate two hour periods and, except for a 1/2 hour checking session implemented after the coding had begun, no further training was done. Each coder was given 26 transcripts to code, 6 of which were coded by all three and served as reliability checks. As pointed out in Chapter VII, the overall reliability across 3 coders was a respectable 0.86, a figure that suggests coders were not greatly troubled in using similar codes. The more logical advantage of the coding scheme is that it provides an index of control for messages from which both individual and dyadic data may be drawn. The disadvantages of the scheme concern category usage and message sequencing. The former involves cate- gories that were not used to index messages. In this sample, no messages were categorized as Assertion as other 181 (code 10), Question as answer (code 24), Question as in- struction (code 25), Question as order (code 26), Question as disconfirmation (code 27), Question as topic change (code 28), Question as other (code 20), Noncomplete as instruction (code 43), Noncomplete as order (code 46), Noncomplete as disconfirmation (code 47), Noncomplete as topic change (code 48), Other as support (code 51), Other as nonsupport (code 52), Other as answer (code 54), Other as instruction (code 55), Other as order (code 56), Other as disconfirmation (code 57), and Other as topic change (code 58). These categories represent 36% of the total of 50 possible combinations. It is entirely possible that some of these codes might never be found--e.g., Question as answer, Question as disconfirmation, Other as order, Other as disconfirmation--but some of these codes may not have been used due to the emphasis the coders placed on the 2nd digit categories of Assertion and Talkover. The use of only 32 categories still suggests that the scheme may be unwieldy, although the decision to eliminate any of the codes should be done only after further research. The second disadvantage concerns the way the scheme was implemented for the control dimensions. The use of the three digit code to index each message implies that a coder can examine a statement, determine its format (2nd digit), and infer its response mode (3rd digit). For most interactions this coding procedure is probably truly 182 sequential--i.e., one's response to another person's message comes immediately after that initiating or stimulus message. However, in some cases, a message which follows another may not be the response to that message. Instead, it may be the response to a message that came earlier in the interaction. For example, in the following sample interaction, message B3 is actually a response to message Al rather than to A3. A1 We should have gone to your mother's for 213 the holidays. Bl Why? 123 A2 Because she wanted to see us all 214 B2 (But we...) 130 A3 and we haven't been there for a year. 213 B3 What do you mean, "for the holidays?" 123 Although this could be a problem of both validity and re- liability if such message sequences occurred often, there was no indication from an examination of the transcripts that such message sequences were a frequent occurrence. Therefore, until some evidence is found that indicates the 3 digit codes are consistently in error, this dis- advantage may be only minor. In short, given the 0.86 intercoder reliability, statements from the coders that the scheme was easy to use, and disadvantages that may be only minimal over a series of interaction samples, the scheme is functionally useful, both in terms of implementation and reliability. 183 Suggestions for Further Research The major purpose of this study was to examine the relationship between dominance—submission as measured by the EPPS and complementary and symmetrical transactions as indexed by a modified coding scheme originally developed by Mark (1970). Among the findings were: (1) dominance scores did not predict relational transactions; (2) upper class and lower class couples differed in their use of transactions--upper class couples had a higher proportion of complementary transactions than did lower class couples, and lower class couples had a higher proportion of symmet- rical transactions than did upper class couples; (3) dyads had stability in their use of patterns, and (4) dyads communicated similarly at the relational level on the two Civil Defense topics but differently on the "working wife” topic. With these findings, it would seem important to study the way in which dyads communicate relationally about different information and content. Perhaps, the similarity in the stability found for two of the topics was a function of both content and structure; i.e., both topics were concerned with Civil Defense information and both asked couples to discuss possible alternatives and arrive at a joint conclusion. The other topic, however, dealt with the husband's and wife's feelings concerning 184 an independent career, job, and interest for wives and asked the couple to describe their feelings and indicate their joint conclusion. In order to check whether these relational findings are peculiar to these topics, further research is needed in which both structure and content are varied. Such research might provide further clues as to whether or not consistency in the use of transaction types is a function solely of an imposed structure, or whether it results from structure and the dyad's reaction to what is being discussed. This research might also find what types of information lead couples to either random- ness or rigidity in their use of transaction types. Of equal importance is the need to examine the consistency of transaction usage among dyads that have a neurotic or schizophrenic member. The only comparisons that can be made at this point concerning the stability of transaction usage is that it differs from randomness and it differs across topics. The study of ”non-normal" dyads would provide another baseline from which to describe normal husband-wife dyads interacting over a series of topics. Another useful area of exploration is that of paralinguistics. Since the data in this study was, in a sense, abstracted from audio tapes, information concerning intonation and inflection was lost. Even though typists put in question marks to indicate the interrogative form, 185 some messages were difficult to code since the paralin— guistic cues were missing. Research into these cues and any discrepancies between them and the relational coding of messages could provide further insight as to the validity of the coding scheme. Such research might also discover and index, those patterns which seem to further define dyadic relationships or clarify what has been defined in words. Viewed in a larger context, the findings suggest a further examination of the theoretic predictions set forth by Carson, and a further investigation of how such a theory may be useful in the study of family communication. Even though the major theoretic tenet of complementarity set forth by Carson was not supported in this study, the theory seemingly has merit and potential utility when one observes human behavior. For example, expressed in role behaviors, complementarity can be found in supervisor/employee, buyer/ seller, teacher/student, and parent/child dyadic relation- ships. Expressed more in terms of specific communication behaviors, complementarity can be found when one asks a question and the other provides the answer, when one gives an order and the other agrees to do what is ordered, and when one gives an answer to a previous question and the other extends or continues that answer. These communica- tion behaviors are neither uncommon nor random and the implications drawn from Carson's theory attempt to account 186 for such behavioral sequences on the basis that they are rewarding for the individuals who are interacting. If this is, indeed, the case, one would expect to find in most dyadic relationships a tendency or, as Leary argues, a probability that complementarity would be a modal be- havior sequence. Discussed in terms of family communication, the theory implies that spouses will interact in such a fashion as to "complement" each other's communications. This in- dicates that, over a period of time, one would expect to find a significantly higher proportion of message sequences in which one message adds something to another rather than matching it exactly, as in the case where each member is trying to define the relationship by emitting messages that have a similar control dimension. Although the test of the hypothesis that predicted a higher proportion of complementary transactions showed opposite results, it was argued that the operationalization of complementarity in terms of relational communication may have been too narrow. Certainly, the observations of human behavior referred to earlier and the theoretic references to complementarity offered by social psychologists such as Newcomb (1956), Schutz (1958), and Thibaut and Kelley (1959) suggest that Carson's theory should not be dismissed lightly. In order to better test Carson's prediction of com- plementarity, and in order to better examine the proposed 187 relationship between that prediction and communication transactions defined by the relational communication cod- ing scheme, it would seem necessary to make several con- ceptual and methodological suggestions. First, future studies dealing with this concept could use the operationalization of complementarity discussed earlier in this chapter; i.e., defining the transaction as a one-up/one-down, one-up/one-across, or one-down/one-across message sequence. Second, as also discussed earlier, predictions concerning dominance-submission might better be based on actual communication behaviors rather than on scores on a dominance scale. In other words, using data similar to that gathered in this study, predictions could be made from decision-making situations to other decision-making or even non-decision-making situations. For instance, using data coded by the relational coding scheme of this study, one could predict that men will have a significantly higher proportion of one-up messages than will women--that is, they will be more dominant; or, one could predict that women will have a significantly higher proportion of one- down messages than will men--that is, they will be more submissive. Further predictions could be made on the basis of a traditional role structure described by Heiss (1962), namely, that ”the male takes the lead by contributing the major share of ideas, and the female does her part by 188 reacting to his suggestions, by smoothing over the rough spots, etc." (p. 197.) Phrased in terms of relational communication, such a prediction might be: complementary transactions of the one-up/one-down type where the husband is one-up will be a significantly higher proportion of the total transactions than will complementary transactions of the one-up/one-down type where the wife is one-up. Third, research could be based on the reward aspect of behavior. For instance, a scale could be developed that might possibly measure those behaviors deemed reward- ing by each member of the dyad. A subject could be given a description of a specific communication behavior enacted by his spouse in a hypothetical but real situation, and be asked to choose from a list of possible behaviors the be- havior he would be most apt to initiate in response. In addition, he could be asked to rate that behavior in the hypothetical sequence in terms of its value to him. With a series of described behaviors, subjects could indicate those which they felt would be most probably and most re- warding for them. By summing across subjects, lists of probable and rewarding behaviors could be obtained. These lists could be correlated to check--at a non-interaction 1evel--the relationship between proposed behavior and the behavior deemed rewarding. Furthermore, from these lists predictions could be made regarding individual's actual messages and couples' actual transactions. Analyses of 189 the preferred and actual behaviors might give some support to the hedonism element of Carson's prediction. Such research might provide answers to questions like: Do couples engage in behaviors they have rated as most rewarding? Is complementarity a relationship that emerges as a preferred choice? With the addition of variables like "power" or "rules," this research might also provide answers to questions like those raised by Waxler and Mishler, e.g., "Do families have patterned processes for handling power strategies? Does a family respond with a predictable set of actions to disagreements? Are there specific contingency patterns in family inter- action?" (1970, p. 289.) To the extent that these answers reflect the pre- dictions in Carson's theory, to that extent will we increase our understanding of why and/or when people utilize certain control messages and transactions. To the extent that these answers relate to transaction sequences, to that extent will we increase our understand- ing of transaction patterns and the use made of them in certain types of interaction situations. It is hoped that the findings of this study have provided an impetus for seeking both the latter and the former. 190 Conclusion This study attempted to place the relational com- munication framework into a theoretical framework of interaction developed by Carson. As noted in the Intro- duction, "it was hoped that the integration of the rela- tional communication concepts with a theoretic framework of interaction would lead to a better understanding of relational communication, and that the interaction or discussion data would provide additional information concerning the patterns that husband and wife dyads use to communicate." Although the predictions drawn from the theoretic framework were not supported, it is felt that the change in the definition of relational communication complemen- tarity will allow the two frameworks to be merged. The latter hope, concerning information on interaction patterns, was met. In addition to these findings, this study seemed useful in terms of: the development of a computer program that lists the relational transactions and message types for each dyad; the use of R Deviation Scores to check for transaction stability; and the use of proportions and arcsin transformations that allowed higher order statisti— cal tests to be used. REFERENCES REFERENCES Altrocchi, J. Dominance as a factor in interpersonal choice and perception. Journal of Abnormal and Social Psychology, 1959, 59, 303-308. Bateson, G. Naven, 2nd ed. Stanford: Stanford University Press, 1958. Bateson, G. and D. D. Jackson. Some varieties of patho- genic organization. Disorders of Communication. Research Publications,’A}R}N}MfD., 1964, 42, 270- 283. Becker, G. The complementary-need hypothesis: Authoritar- ianism, dominance and other Edwards Personality Preference Schedule Scores, Journal of Personality, 1964, 32, 45-56. Becker, W. C. and R. S. Krug. A circumplex model for social behavior in children. Child Development, 1964, 35, 371-396. Bergel, B. Social Stratification. New York: McGraw-Hill, 1962. Berlo, D. K. The uses humans make of communication, Essay III. Mimeo. Michigan State University, Department of Communication, 1969. Berscheid, E. and E. H. Walster. Interpersonal Attraction. Reading, Mass.: Addison-Wesley, 1969. Borgatta, E. F. Rankings and self-assessments: Some behavioral characteristics replication studies. Journal of Social Psychology, 1960, 52, 297-307. Borgatta, E. F., L. S. Cottrell, Jr. and J. M. Mann. The spectrum of individual interaction characteristics: An interdimensional analysis. Psychological Reports, 1958, 4, 279—319. Bowerman, C. E. and B. R. Day. A test of the theory of complementary needs as applied to couples during courtship. American Sociological Review, 1956, 21, 602-605. 191 192 Brown, R. Social Psychology. New York: Free Press, 1965. Carson, R. C. Interaction concepts ofypersonality. Chicago: Aldine, 1969. Cochran, W. G. Analysis of variance for percentages based on unequal numbers. Journalrofjphe Amepican Statistical Association, 1943, 38, 287F301. Drechsler, R. J. and M. Shapiro. Two methods of analysis of family diagnostic data. Family Process, 1963, 2, 367-370. Edwards, A. L. Experimental desigp in psychological re- search. New York: Rinehart, 1950. Edwards, A. L. Personalypreference schedule, rev. ed. New York: PsyChological Association, 1959. Goffman, E. The presentation of self in everyday life. Garden City: Doubleday, 1959. Haley, J. An interactional explanation of hypnosis. The American Journal of Clinical Hypnosis, 1958, 1, 41157. Haley, J. Research on family patterns: An instrument measurement. Family Process, 1964, 3, 41-65. Hays, W. L. Statistics for psychologists. New York: Holt, Rinehart G Winston, 1963. Heiss, J. S. Degree of intimacy and male-female inter- action. Sociometry. 1962, 25, 197-208. Heller, K., R. A..Myers and L. V. Kline. Interviewer behavior as a function of standardized client roles. Journal of Consulting_Psychology, 1963, 21, 117-122. Holsti, O. The quantitative analysis of content, in Content Analysis, R. C. North, et al. (eds.). Evanston: 'Northwestern Univers1ty Press, 1963. Homans, G. C. Social behavior: Its elementary forms. New York: Harcourt, Brace GEWOrld, 1961. Jackson, D. D. Family interaction, family homeostasis and some implications for conjoint family psycho- therapy, in Individual and familial dynamics, J. Masserman (edi). 'New York: Grune G Stratton, 1959. 193 Jackson, D. D., J. Riskin and V. M. Satir. A method of analysis of a family interview. A. M. A. Archives of General Psychiatry, 1961, 5, 321-339. Jones, E. E. and H. B. Gerard. Foundations of social psychology. New York: Wiley, 1967. Katz, 1., S. Glucksberg and R. Krauss. Need satisfaction and Edwards PPS scores in married couples. Journal of Consulting Psychology, 1960, 24, 203-208. Kenkel, W. F. Influence differentiation in family decision making. Sociology and Social Research, 1957, 42, 18-25. Kerckhoff, A. and K. E. Davis. Value consensus and need complementarity in mate selection. American Sociological Review, 1962, 21, 295-303. Ktsanes, T. Mate selection on the basis of personality type: A study utilizing an empirical typology of personality. American Sociological Review, 1955, 20, 547-551. Leary, T. Interpersonal dia nosis ofypersonality. New York: Ronalleress, 957} Levinger, G. Note on need complementarity in marriage. Psychological Bulletin, 1964, 61, 153-157. Lindquist, E. F. De i n and analysis of experiments in psychology and education. Boston: ‘Houghton Mifflin, 1953. Mark, R. A. Parameters of normal family communication in the dyad. Unpublished Ph.D. dissertation, Michigan State University, 1970. McNemar, Q. Psycholo ical Statistics. 4th ed. New York: John Wiley and Sons,vlnc., 1969. Miller, G. A., E. Galenter and K. H. Pribram. Plans and the structure of behavior. New York: Holt, Rine- hart & Winston, 1960. Murray, H. A. Explorations inypersonality: A clinical and experimental study_of fifty men of collgge age. New York: Oxford, 1938. 194 Murrell, S. A. and J. G. Stachowiak. Consistency, rigid- ity, and power in the interaction of clinic and non-clinic families. Journal of Abnormal Psychology, 1967, 12, 265-272. Murstein, B. I. The complementary need hypothesis in newlyweds and middleaged married couples. Journal of Abnormal and Social Psychology, 1961, 93, I94- 197. Newcomb, T. M. The prediction of interpersonal attraction. American Psychologist, 1956, 11, 575-586. Peabody, D. Symmetry and asymmetry in interpersonal rela- tions--with implications for the concept of projec- tion. Journal of Personality, 1970, 38, 425-433. Rogers, E. Dimensions of dyadic systems and interpersonal communication. Mimeo. Michigan State University, Department of Communication, 1971. Rogers, E. Dyadic systems and transactional communication in a family context. Unpublished Ph.D. disserta- tion, Michigan State University, 1972. R005, D. E. Complementary needs in mate-selection: A study based on R-type factor analysis. Unpublished Ph.D. dissertation, Northwestern University, 1956. Rosow, 1. Issues in the concept of need-complementarity. Sociometry, 1957, 20, 216-233. Ruesch, J. and G. Bateson. Communication: The social matrix of psychiatry. New YOfk: 5W. W. Norton, 1951} Rychlak, J. F. The similarity, compatibility or incom- patibility of needs in interpersonal selection. Journal of Personalipy_and Psychology, 1965, 2, 334-340. Schatzman, L. and A. Strauss. Social class and modes of communication. American Journal of Sociology, 1955, 55, 329-338. Schellenberg, J. A. and L. S. Bee. A re-examination of the theory of complementary needs in mate-selection. Marriage and Family Living, 1960, 22, 227-232. 195 Schutz, W. C. FIRO: A three-dimensional theory of inter- personal behavior. ENew‘York: *Rinehart, 1958. Scott, W. A. Reliability of content analysis: The case of nominal scale coding. Public Opinion Quarterly, 1955, 19, 321-325. Sluzki, C. E. and J. Beavin. Simetria y complementaridad: una definicion operacional y una tipologia de parejas [Symmetry and complementarity: An operational definition and a typology of pairs]. Acta psiquiatrica ypsicologica de America Iatina, 1965, 11, 321- 330. Snedecor, G. W. Statistical Methods, 4th ed. Ames, Iowa: Iowa State College Press, 1946. Strauss, M. A. Communication, creativity, and problem- solving ability of middle- and working-class families in three societies. American Journal of Sociology, 1968, 13, 417-430. Thibaut, J. W. and H. H. Kelley. The socialpsychology of groups. New York: Wiley, 1959? U.S. Bureau of the Census. Census of population: 1970. General social apd economic characteristics. Final Report PC(1) - C37. Chip. EWashington, DlC}: U.S. Government Printing Office, 1972. Watzlawick, P. An anthology of human communication. Palo Alto: Science and"Eehavior Books, 1964. Watzlawick, P., J. Beavin and D. D. Jackson. Pragmatics of human communication. New York: W. W} NOrton, E67. Waxler, N. E. and E. G. Mishler. Experimental studies of families, in Advances in experimental social 5 cholo . V01. 5. ’IZiBerkowitz (ed.). New gork: Academic Press, 1970. Winch, R. F. Mate-selection: A study of complementary needs. New Yofk: Harper 8 Brothers, 1958: Winch, R. F. The modern family. Rev. ed. New York: Holt, Rinehart and Winston, 1964. Winch, R. F. The theory of complementary needs in mate selection: A test of one kind of complementariness. American Sociological Review, 1955a, 20, 52-56. Winch, Winch, Winer, 196 R. F. The theory of complementary needs in mate selection: Final results on the test of the general hypothesis. American Sociological Review, 1955b, 20, 552-555. R. F., T. Ktsanes and V. Ktsanes. The theory of complementary needs in mate selection: An analytic and descriptive study. American Sociological Review, 1954, 19, 241-249. B. J. Statisticalpprinciples in experimental design. New York: McGraw-Hill, 1962. APPENDICES APPENDIX I MICHIGAN STATE UNIVERSITY East Lansing Michigan 48823 College of Communication Arts, Department of Communication Cable: Commdept March 6, 1972 A study is now underway concerning family communica- tion. We are interested in learning how husbands and wives talk about different issues that come up in their daily lives. In order to gain a better understanding of family opinions, we are asking many families to participate in this study. We would like your permission to talk with both you and your spouse, if you have children under 12 years of age. Couples who are willin to participate in this study will receive ten dollars ( 10) for their time. This study is being done by the Department of Com- munication at MSU. The findings will be placed in a report so that no individual's or family's views can be identified. We want to assure you that your opinions will remain totally confidential. In order for the findings to show a true picture of different families' viewpoints, your participation is very important to us. You are part of a random sample and your cooperation is essential to the value of this study. If you have any questions about the study, feel free to ask the person bearing this letter, or contact me directly at my office phone, 517-355-3478. I think you will find this study interesting and you will be making an important contribution to the understanding of families. Sincerely, Richard V. Farace Director of the Family Communication Project 524 South Kedzie Hall RVF:jo 197 APPENDIX II PART I FAMI LY QUE STIONNAI RE like tion your PART PART INSTRUCTIONS This study consists of two parts. First, we would you to fill out a questionnaire on family communica- patterns. In the second part, we would like you and spouse to discuss a few topics. The questionnaire is to be filled out by each of you. Please do not consult your spouse about any of the questions in the questionnaire until you have completed it. After you have completed your questionnaire, please place it into the envelope, seal it, and give it to the interviewer. When both of you have completed the questionnaire, you will be given three topics to discuss. Two deal with emergency situations and the other deals with a family topic. We would like you to take about ten (10) minutes to talk about the possible alternatives for each topic, and decide what you and your family would do in each situa- tion. In case you have any questions about any part of the questionnaire, please feel free to ask the interviewer about them. WE APPRECIATE YOUR COOPERATION IN THIS STUDY AND HOPE THAT YOU WILL FIND IT INTERESTING! 198 The following questions with other people during the course of a day. concern the time you spend We would like you to give a rough estimate of the time you spent, yesterday, with the people listed below. For each ques- tion, please check the response that best estimates this amount of time. 1. How much of your time yesterday_during waking hours was spent... Count the time you were with one another even though others might have been we didn't spend any time together a. with your spouse? present. 1 to 1 to 2 to 4 to 6 to more b. with just your spouse? 30 minutes 30 minutes to 1 hour 2 hours 4 hours 6 hours 8 hours than 8 hours we didn't spend any time alone 1 to 1 to 2 to 4 to 6 to more 30 minutes 30 minutes to 1 hour 2 hours 4 hours 6 hours 8 hours than 8 hours c. with just one or more of your children? we didn't spend any time alone 1 to l to 2 to 4 to 6 to more d. with people other than 30 minutes 30 minutes to 1 hour 2 hours 4 hours 6 hours 8 hours than 8 hours your spouse and children? I wan't with anyone else 1 to 1 to 2 to 4 to 6 to more 19 £0 30 minutes 30 minutes to 1 hour 2 hours 4 hours 6 hours 8 hours than 8 hours 200 a. When you and your spouse were together yesterday, how much time did you spend in conversation? Count the time actually spent talking and listen- ing to your spouse. we didn't spend any time 1 to 30 minutes 30 minutes to 1 hour 1 to 2 hours 2 to 4 hours 4 to 6 hours 6 to 8 hours .__gmore than 8 hours b. How much time yesterday would you have liked to have talked with your spouse? much more somewhat more about the way it was somewhat less ___much less Is the amount of time you spend talking with your spouse yesterday typical of most week days? yes--if yes, go to question 4. no--If no, please estimate the amount of time you spend talking with your spouse on a typical week- day. we don't spend any time 1 to 30 minutes 30 minutes to 1 hour 1 to 2 hours 2 to 4 hours 4 to 6 hours 6 to 8 hours more than 8 hours (PLEASE GO RIGHT ON TO THE NEXT PAGE) 4. 201 Since weekends may be different from weekdays, please estimate the amount of time you spend last Sunday... a. with your spouse? ___we didn't spend any time together 1 to 30 minutes 30 minutes to 1 hour 1 to 2 hours 2 to 4 hours 4 to 6 hours 6 to 8 hours ___more than 8 hours b. in conversation, talking and listening, to your spouse? we didn't spend any time 1 to 30 minutes 30 minutes to 1 hour 1 to 2 hours 2 to 4 hours 4 to 6 hours 6 to 8 hours more than 8 hours How much time last Sunday, would you have liked to have talked with your spouse? much more :::somewhat more ___about the same ___somewhat less ___much less Was the amount of time you spent talking with your spouse last Sunday typical of most Sundays? yes--If yes, go to question 7. no--If no, please estimate the amount of time you spend talking with your spouse on a typical Sunday. we con't spend any time 1 to 30 minutes 30 minutes to 1 hour 1 to 2 hours 2 to 4 hours 4 to 6 hours 6 to 8 hours more than 8 hours APPENDIX II PART II 202 7. The previous questions dealt with time estimates. Now we are interested in the general kinds of topics you and your spouse talk about. Please indicate how often you and your spouse talk about each of the following topics by checking the appropriate columns. Once a Once or Once or Less Never day or Twice twice than talk more a a month once a about week month it IT a. your job(s) b. your feelings toward each other c. your children d. family leisure time activities e. care of the home f. financial matters g. friends h. relatives i. local community events, e.g., church, P.T.A., elections, com- mittees, etc. j. national and international events k. please specify other topics below (PLEASE GO RIGHT ON TO THE NEXT PAGE) 203 When just you and your spouse talk, who starts most of the conversations between you? My spouse does more than I I do more than my spouse Each of us about the same Don't know, can't say When just you and your spouse talk, how much of the time do you talk in comparison to him/her? Much more Somewhat more About the same Somewhat less Much less When just you and your spouse talk, whose interests or concerns do you talk about? Mainly talk about what my '__—spouse wants to discuss __yMainly talk about what I want to discuss Split about evenly between his/her interests and mine When just you and your spouse talk, who usually ends or stops the conversation? My spouse usually ends them _——1 usually end them :::Split about evenly between us, depends on topic ___Usually ended by outside interference or interrup- tions In general, how satisfied are you with the communica- tion between you and your spouse? ___Very satisfied Somewhat satisfied :::}bout as satisfied as dis- satisfied Somewhat dissatisfied :::Wery dissatisfied (PLEASE GO RIGHT ON TO THE NEXT PAGE) 10. 204 When husbands and wives talk to one another different kinds of feelings may result. Listed below are some of the ways that you may have felt when talking with your husband/wife. Please indicate with the appropriate number (1-5), how often you have had the feeling mentioned in each of the statements. 1 2 3 4 5 Always Often Now and then Seldom Never a. In a conversation with my husband/wife, I am un- comfortable during a period of silence. b. I am satisfied with our ability to talk things out together. c. My husband/wife does not listen to me when I'm talking. d. I find it difficult to express my true feelings to my husband/wife. e. I know my husband's/wife's feelings and emotions from his/her gestures and facial expressions. f. I avoid talking about certain subjects with my husband/wife because it may be unpleasant for us. g. When we're talking, my husband/wife understands me and how I feel. h. My husband/wife encourages me to express my con- cerns. i. I can anticipate what my husband/wife is going to say before he/she says it. j. My husband's/wife's manner of speaking is irritat- ing. R. My husband/wife lets me know how he/she feels about what I'm saying. 1. In our conversations, I don't understand how my husband/wife feels. m. I find other people more interesting to talk to than my husband/wife. 205 n. It's easy to talk to my husband/wife about any problem or complaint. 0. I feel dissatisfied with my husband's/wife's ability to express his/her feelings and emotions in words. 206 11. Each of us receives information from many sources. We are interested in how you learn about different types of events. Listed below are several ways you may have learned about them. Please check all your sources of information for the following typgs of events (e.g., school events, community events, etc.) You may have more than one answer for each event. School Community National 6 Natural & Events Events International Civil Defense Sources Of Events Preparedness Information Co-workers Neighbors My Spouse My Children Newspapers/ magazines TV or radio Delivered’ in mail Never received 12. Now, considering just your own immediate family, who tends to bring the most information about these same events to the attent1on of the other family members. School Community National 6 Natural 8 Events Events International Civil Defense Events Preparedness Who Brings Most Informa- tion My Spouse Myself My children All of us about the same None of us 13. 207 Earlier, we mentioned "delivered in mail" as one way of learning about events. We are now interested in finding out who looks at, and what your family does with the third class mail (like advertisements, pub- lic information announcements) that is delivered to your house. a. In your family, who is usually the first person to look at this type of mail? Advertisements Public Information (store ads, magazine (School events, community ads, special offers) notes, civil defense liter- ature, etc.) My spouse My spouse MYself Myself My children My children 0 one in particular No one in particular b. Who else in the family looks at it? Advertisements Public Information ___My spouse ___My spouse I do I do ___My children ___My children ___No one in particular; ___No one in particular; depends on material depends on material c. What is usually done with this type of mail? Advertisements Public Information Thrown away immediately Thrown away immediately Kept for a short time Kept for a short time Kept for future refer- Kept for future refer- ence ence d. How often is this material discussed with other members of the family? Advertisements Public Information Often Often Seldom Seldom ___Never ___Never APPENDIX II PART III 14. 208 Each family works out its own way of doing things. We would like to ask you about how certain things are done in your family. Please indicate, by using the number of the appropriate response below (1-6), how you and your spouse divide up some of the family jobs. 1 2 3 4 5 6 Husband Husband Both Wife Wife Neither Almost More Than About the More than Almost One Always Wife Same Husband Always a. Who does the grocery shopping? b. Who prepares the meals? c. Who repairs thing (appliances, furniture, toys) around the house? d. Who disciplines the children? e. Who gets up at night, if necessary, with the children? f. Who helps the children with their homework? g. Who decides on the family budget? h. Who makes complaints, if necessary, to salesmen, service repairmen or landlord? i.’ 'Who selects the family car(s)? j. Who plans what to do on a Saturday night? k. Who decides what people you will invite to the house? 1. Who keeps in touch with relatives? m. Who shows affection for the other spouse? n. Who takes the initiative to make up when there's been a disagreement? 0. Who tries to see the other's point of view when there is a difference of opinion? 15. 209 We would now like to ask you how you feel these family jobs should be done. In your family, who do you think should Be responsible for doing the following things, regardless of whether that person actually does them 01‘ not. Please indicate below, by using the number of the appropriate response (1-6), your own preference as to how these jobs should be divided up between your- self and your spouse. 1 2 3 Husband Husband Both AlmoSt More than About the More than Almost Husband Always Always Wife Same a. Who do you think should b. Who do you think should c. Who do you think should furniture, toys) around d. Who do you think should e. Who do you think should necessary, f. Who do you think should their homework? g. Who do you think should h. Who do you think should necessary, to salesmen, landlord? i. Who do you think should j. Who do you think should Saturday night? k. Who do you think should invite to the house? 1. Who do you think should relatives? m. Who do you think should other spouse? n. Who do you think should 4 5 6 Wife Wife Neither One do the grocery shopping? prepare the meals? repair things (appliances, the house? discipline the children? get up at night, if with the children? help the children with decide the family budget? make complaints, if service repairmen or select the family car(s)? plan what to do on a decide what people will keep in touch with show affection for the take the initiative to make up when there's been a disagreement? 0. Who do you think should try to see the other's point of view when there is a difference of opinion? APPENDIX II PART IV 16. 210 Now, we would like to ask you to react to a number of pairs of statements about things that you may or may not like: about ways in which you may or may not feel. This is not a test. It is an attempt to find out how people fEET about themselves. Therefore, there are no right or wrong answers. For each pair of statements, choose the statement you think best describes the way you feel. CIRCLE the letter of the statement. Some choices may be difficult; nevertheless, choose the one that best describes how you feel. (If neither statement accurately describes how you feel, choose the one which you consider to be less inaccurate.) For example, in the sample item below you might choose item (a) if you feel that statement describes you best. Example: (1) a I like to tell amusing stories and jokes at parties. b I would like to write a great novel or play. Please select one statement in each pair by circling the letter of the one you choose. (1) (2) (3) (4) (5) (6) a I feel timid in the presence of other people I re- gard as my superiors. b I like to supervise and to direct the actions of other people whenever I can. a I like to be called upon to settle arguments and disputes between others. b I like to avoid responsibilities and obligations. a I like to tell other people how to do their jobs. b I feel like getting revenge when someone has insulted me. a I like to show a great deal of affection toward my friends. b I like to be regarded by others as a leader. a I like to sympathize with my friends when they are hurt or sick. b I like to be one of the leaders in the organiza- tions and groups to which I belong. a When with a group of people, I like to make the decisions about what we are going to do. b I like to predict how my friends will act in various situations. (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) a 0‘93 0‘93 211 I like to put in long hours of work without being distracted. I like to be regarded by others as a leader. I like to tell other people how to do their jobs. I like to ask questions which I know no one will be able to answer. I get so angry that I feel like throwing and break- ing things. I like to tell other people how to do their jobs. When serving on a committee, I like to be appointed or elected chairman. I would like to write a great novel or play. I like to be one of the leaders in the organiza- tions and groups to which I belong. I like to be able to do things better than other people can. I like to do things in my own way without regard to what others may think. I like to supervise and to direct the actions of other people whenever I can. I like to be called upon to settle arguments and disputes between others. I like to be regarded as physically attractive by those of the opposite sex. When serving on a committee, I like to be appointed or elected chairman. When I am in a group I like to accept the leadership of someone else in deciding what the group is going to do. I like to be able to persuade and influence others to do what I want to do. I like to think about the personalities of my friends and to try to figure out what makes them as they are. I like to keep my letters, bills, and other papers neatly arranged and filed according to some system. I like to be one of the leaders in the organizations and groups to which I belong. I like to write letters to my friends. I like to argue for my point of view when it is attacked. (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28) 0‘93 212 I like to be called upon to settle arguments and disputes between others. I like my friends to do many small favors for me cheerfully. When with a group of people, I like to make the decisions about what we are going to do. I like my friends to sympathize with me and to cheer me up when I am depressed. I like I like neatly system. to be regarded by others as a leader. to keep my letters, bills, and other papers arranged and filed accourding to some I like to do things with my friends rather than by myself. I like to argue for my point of view when it is attacked by others. When I am in a group, I like to accept the leadership of someone else in deciding what the group is go- ing to do. I like to supervise and to direct the actions of other people whenever I can. I like to argue for my point of view when it is attacked by others. I like to experience novelty and change in my daily routine. I like to tell other people how to do their jobs. I like to be the center of attention in a group. I feel depressed by my own inability to handle various situations. I like to be able to persuade and influence others to do what I want. I like to engage in social activities with persons of the opposite sex. When with a group of people, I like to make the decisions about what we are going to do. I like to be able to persuade and influence others to do what I want I like to finish any job or task that I begin. When serving on a committee, I like to be appointed or elected chairman. I like to try new and different jobs--rather than to continue doing the same old things. (PLEASE GO RIGHT ON TO THE NEXT PAGE) APPENDIX II PART V 17. 18. 19. 20. 21. 22. 23. 24. 25. 213 Finally, we would like to ask you just a few more questions about yourself. What is your age? years Please circle the last grade completed in school. 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 1 2 3 4 S 6 Grade School High School College Graduage School What is your present occupation? Approximately, what is your family's yearly income? ___less than $3,000 ____$3,000 to $5,999 $6,000 to $7,999 :$8,000 to $14,999 ___over $15,000 How long have you been married? years Is this your first marriage? yes no How would you describe your satisfaction with your marriage? 1 2 3 Very . ‘. Somewhat Satisfied Satlsfied Satisfied 4 5 6 Somewhat Dissatisfied Very Dissatisfied Dissatisfied How satisfied do you think your husband/wife is with your marriage? 1 2 3 Very . . Somewhat Satisfied Satleled Satisfied 4 5 6 Somewhat . . . Very Dissatisfied D1ssatisf1ed Dissatisfied What about your marriage do you like the least? 214 26. What about your marriage do you like the most? THANK YOU AGAIN FOR YOUR COOPERATION APPENDIX I I I 215 TOPIC 1 Imagine it is a typical weekday and at least one of you is at home. A Civil Defense alert--a loud wailing siren--is sounded. After turning on the radio, you learn that there has been a nuclear attack. You are warned to prepare for the radiation fallout which will reach your area in about two hours. Your neighbors have also heard the warning and need to find shelter. (A) If your house has a cellar or basement, would you be willing to share it with your neighbors during the disaster? (B) If you don't have a basement, would you be willing to go to some- one else's house for protection? Please discuss the pos- sible alternatives and decide on a plan of action. TOPIC 2 Some married couples feel that both the husband and the wife should be able to have independent careers, jobs and interests outside the family. Other couples feel that both should be devoted to the interest of the family, and that the wife in particular should be in the home as a full time homemaker and mother. What are your feelings on this matter and what is your joint conclusion? TOPIC 3 Imagine a tornado watch has been in effect for several hours in your area. You, your spouse, and your children are at home one evening watching television, when 216 a tornado warning is broadcast indicating the sighting of a tornado. You are told that you have about 15 minutes to prepare and take shelter. What would you and your family do in this situation? Please discuss the possible alternatives and decide on a plan of action. APPENDIX IV INFORMATION FOR TYPISTS/TRANSCRIBERS General Procedures Type dyad number from tape and topic number (l,2,3,4) in appropriate places on coding form. Listen to Topic One (1) on tape in order to distin- guish speakers and their style of interaction. After listening to Topic One, rewind tape and, using the coding form, begin typing the interaction when the first member of the dyad speaks. Identify speaker as Husband (H) or Wife (W) by placing the appropriate code letter in the SPEAKER column of the coding form. NOTE: On some of the tapes, the interviewer gives a probe or message designed to extend the discussion. You will be given a sheet which indi- cates those discussions that were probed. Probes will be in the form of suggestions ("You might discuss...") or requests ("Count you discuss...") and will include one or more of the following points: a. Your first impressions or reactions b. Your expectations c. Your feelings d. Any problems e. Any solutions Identify the interviewer giving one of these probes with an (I) by placing this code in the SPEAKER column. When Topic One is transcribed, rewind tape and compare transcription with the actual interaction; correct any errors in the transcribed copy. When Topic One transcription is complete and accurate, proceed to other topics on tape utilizing the same general procedures. Rules for SpecifyingMessages For each person, a message represents everything he or she says until the other person speaks. Include all interviewer probes as messages, but please exclude messages spoken by children. 217 218 Type messages accurately as heard, including grammati- cal errors. For those words that involve mispronunci- ations, determine from context what word was intended and give its correct spelling. Indicate questions by a question mark; pay attention to inflection patterns of a speaker's voice to deter- mine the interrogative form. Pay attention to silence. For any period of silence that is at least three (3) seconds, type the word SILENCE in the middle of the line following the pre- ceding message. If the silence runs an additional five (5) seconds, type another SILENCE below the first. For every additional five second periods of silence, follow the same procedure. For any message that is not distinguishable, identify the speaker and type INDISTINGUISHABLE in place of the message. For a part of a message that is not distin- guishable, type as much of the message that is recog- nizable and where it becomes unclear or garbled, type INDISTINGUISHABLE. For example, the former situation would be typed as follows: H: I suppose I would come from work as soon as I could get out. W: INDISTINGUISHABLE H: Yeh, I guess that's what I'd do. The latter situation would be typed as follows: W: One of us would have to go to the school to check with the teacher and the principal. H: It depends on who it was who accused him of doing INDISTINGUISHABLE. W: We'd sure check up on that teacher. APPENDIX V RELATIONAL COMMUNICATION CODING SCHEME DIGIT ONE DIGIT TWO DIGIT THREE (Speaker) (Format) (Response) 1 = Wife 1 = Assertion 1 = Support 2 = Husband 2 = Question 2 = Non-support 3 = Talk Over 3 = Extension 4 = Non-complete 4 = Answer 5 = Other 5 = Instruction 6 = Order 7 = Disconfirmation 8 = Topic Change 9 = Initiation- Termination 0 = Other Silence is coded by a 000 code for every period of silence longer than 3 seconds. Every additional five second period of silence is also coded as 000. DO NOT CODE ANY MESSAGES BY INTERVIEWER DO NOT CODE LAUGHTER ***************************** To illustrate how this scheme may be used to cate- gorize messages, a sample from an interaction is shown below. Husband: But then I don't remember what'd I 223 say to him? Wife: You explained the wrong ways and the 114 right ways and childbirth and other things. Husband: Oh, yah, I remember that...let's see... 211 Wife: Over in E. Lansing they already have 113 a program in reproduction so he really knew the right way. The husband's first message is coded 223 to indicate a question of extension that seeks information about the topic being discussed. The wife's first message is coded as 114 to show that it is an assertion that serves as an 219 220 answer. The husband's second message is a supportive assertion and the wife's second message is an assertion of extension. Definitions of Code Categories DIGIT TWO "Assertion” "Question" "Talk-Over" "Non-complete" "Other" DIGIT THREE "Support" "Non-support" "Extension" "Answer" any completed referential statement that may be in either the declarative or im- perative form; i.e., a message that has a subject and a verb. Also, words such as "yes," "no," "sure," "Hmm-mmm," "Uh" uh," and "right" which clearly indicate a control function. any statement that takes an interrogative form. any interruption or verbal intervention made while the other person is speaking. any utterances, other than those coded as "talk-overs," that are initiated but are not completed. any utterance that is indistinguishable or is not classifiable as to form. any statement giving or seeking accept- ance, agreement and/or approval. any statement or utterance that is a dis- agreement, rejection, demand, and/or challenge. any statement that continues the flow or theme of the preceding message, or any statement that is a non-committal response to a question. any statement that is a definitive response to a question or at least has substance and/or commitment. "Instruction" "Order" "Disconfirma- tion" "Topic Change" "Initiation- Termination" "Other" Priority DIGIT TWO 221 any statement that is a regulative re- sponse to a in the form of a suggestion; it is often qualified or contains an explanation. any statement that is a regulative re- sponse but is in the form of an unquali- fied command with little or no explanation. any statement that ignores or by-passes the request of the other individual. any response that has little continuity with previous messages and no response continuity was requested. any statement that either begins or ends an interaction; its use signifies start- ing and ending points for particular discussions or conversations. This code is also used for any message by husband or wife that follows a probe by the in- terviewer. any response that is unclear or unclassi- fiable. Considerations for Using Categories 1. First, determine if the message should be considered as a talk-over (#3). A talk-over may be a question, assertion, or non-complete, but an interruptive speech is coded as a talk-over, independent of form. 2. If not a talk-over, the message should be considered as a question (#2). 3. If not a question, it should be considered as an assertion (#1). 4. If not a question, it should be considered as a non- complete (#4). S. If it cannot be categorized as any of these, it is coded as ”other" (#5). For instance, any message or partial message that is, or follows, an "IN-DISTIN- GUISHABLE" or "NON-DISTINGUISHABLE." 222 DIGIT THREE 1. The first consideration should be whether the message initiates or terminates. (Although messages coded in this category can also be coded in other categories-- e.g., as an answer, a non-support, or a topic change--, the major consideration should be whether the message is an initiation or a termination.) If the message is not an initiation-termination, it should be considered as an answer (#4). The main response function of an answer is to provide informa- tion. However, an answer which gives support or non- support should be coded according to its control func- tion. If the message is a response switch, it should first be considered a disconformation (#7). If the message is not a disconfirmation, it should be considered a topic change (#8). If the message has a regulative function, it should first be considered as an order (#6). If the message is not an order, it should be coded as an instruction (#5). If the message is not codable in any of these cate- gories, it should be considered an extension (#3), message of support (#1), or a message of non-support (#2). NOTE: Coding of any of these three categories should follow this priority scheme: i.e., any inde- cision between the three should result in a code of (#3), extension. If a message cannot be categorized as any of these, it is coded as an other (#0). Any message or partial message that is, or follows, an "IN-DISTINGUISHABLE" or "NON-DISTINGUISHABLE" should be coded as an other. Coding of Dual Response Messages As shown on page one, each message by a speaker is coded with a three digit number. In most cases, these messages will have only a three digit code to indicate who was speaking, what form the message was, and what the message was in response to the other speaker's previous message. In some cases, however, a speaker's message may serve more than just one response mode; it may represent 223 support as well as being of a regulative nature. In these situations, such a message will receive at least two three digit numbers to categorize its different response func- tions. For example, in the following sample interaction, the second speaker's message illustrates a situation in which his message is an agreement or support and also a question seeking information. Speaker 1 (Wife): I think we should leave now. 115 Speaker 2 (Husband): Okay, I guess so, but what 211-223 about bringing the food? Speaker 1 (Wife): Well, we'll have to come 114 back for it. In the next sample interaction, the second speaker's mes- sage shows both support and non-support of the first speaker's previous message. Speaker 1 (Husband): I think we should leave now. 215 Speaker 2 (Wife): Okay, I guess so, but I don't 111-112 like you always deciding what time we leave. Speaker 1 (Husband): That's too bad! 212 Sample Interaction The following example represents the form of interac- tion that you will possibly code and illustrates both single and dual response messages. Wife: That's a very interesting situation, 119-123 but I really don't know how I'd handle it, what about you, honey? Husband: Well, I think that in the case of an 214 emergency we would probably first tune in the radio and listen for directions and at the same time begin to gather up some food... Wife: [...and clothing...] 133 Husband: ...to the basement... 243 Wife: [We'd have to take blankets...] 133 Husband: ...where we'd... 243 Wife: [...because we'd need the warmth in 133 the basement and besides the children would need it.] Husband: Right, that sounds about right. 211 In this example, the wife's first message receives two codes--a 119 to indicate an assertion that, arbitrarily, is defined as starting the interaction, and a 123 to show 224 a question that seeks to extend the discussion of the topic. The husband's first message is an assertion that serves as an answer. The next message from the wife is enclosed in brackets to indicate a successful interruption and is coded 133 to indicate a talk-over in extension. The husband's continuation of his original thought is coded 243 to indicate a non-complete that is an extension. The wife again comes in with a talk-over in extension and the husband tries to complete his original statement. The wife again successfully talks over and completes her idea. Finally, the husband makes an assertion that supports his wife's previous message. In some situations, messages which begin as assertions (or talk-overs) may end as questions, or vice versa. For example, note the following interactions: Wife: After we met, you asked asked me to 113-123 come visit you, right? Husband: Yeah, then, uh, we went home to visit 211-213 my folks. The wife's message is double coded to indicate that she kept the discussion going (113) but then sought information agreement for what she had said (123). The husband pro- vided agreement (211) and then continued the flow of the conversation (213). Words that might frequently turn assertions into questions are "okay?", "right?", and "huh?". Husband: And what would you do? I think you 223-215 should get some blankets, flashlight, and water together, open all the win- dows and get the kids to the basement. Wife: Shouldn't I also bring some food? We11,123-1l3 oh wait, I could bring some graham crackers for the kids. Here, the husband asked a question (223) but didn't allow his wife to answer. Instead, he followed his question with an instruction (215). The wife responded with a question (123) and then followed that with an assertion in extension (113), rather than waiting for her husband to answer her question. In some cases, one person may have an unsuccessful talk-over followed by another message. For example: Wife: Husband: 225 I could carry the metal buckets from 113 the garage down to the basement and fill them with water from the water heater. (They'd be too heavy for you. You'd 232-212 need...). One of the boys would have to help you carry them or you'd trip all over yourself. Here, the wife had a message that continued the previous discussion (113), the husband had an unsuccessful talk- over in disagreement or non-support (232), and then fol- lowed with a statement that was also a disagreement with what the wife had said. DIGIT TWO Assertion Question Talk-over Code Categornyxamples ”I think, uh, we should talk about the public shelters." "it's going to depend on several things." "You're the one who has to decide." "I'll get it.” "How old are you?" "Right?” "Did you listen?" "What do you think?" Any message or partial message enclosed within brackets [ ] or parentheses ( ) that is a verbal interruption. These messages may take any grammatical form but will be identified by the brackets or parentheses. NOTE: In some cases, a series of three messages may be coded as follows: H [It would probably be at least two hours} W (What about...) H Ibefore we could come out of the base- ment.] In this sequence, the husband had a success- ful talk-over, indicated by the [ }?, the wife had an unsuccessful talk-over while the husband was talking, indicated by the ( ), and the husband's message was continued, as indicated by the F ]. Any message within a [ }+, like any message enclosed within regular brackets [ ], is considered a 226 successful talk-over. Any message contained within a [ ] is coded as an Assertion or Non- complete in Extension and is pop coded as a talk-over in Extension. Non-complete-"Well, ah...” ”Mmm..." -"But what I really thought..." "Umm..." -"That was in..." "Ah..." Other DIGIT THREE Support Non-support Extension Answer: Instruction "...and besides, uh, we, uh...” All messages that do not fit in the previous four categories. QUESTION FORM - "Are you okay?" - "Could you help me with that?" - "Can I come too?" - "What do you think?" ASSERTION FORM - "Yes, I agree." "Yeah." - "Okay, I'll help." "Right." - "That's a great idea." _ "Hmm_mmm. H "Sure . H QUESTION FORM - "Why would you want to do something stupid like that?” - "Are you crazy?" - "I suppose you're smarter?" ASSERTION FORM - "I don't like it." "Uh-uh." - "We won't do it." - "That's ridiculous." "Furthermore, its the best way to do it." "I don't know." (NOTE: In order to be coded as an extension, a message has to continue the flow or theme of a preceding message, or be a non- committal response to a question.) "Its forty miles to the nearest public shelter." "You have to add water, then oil." (NOTE: An answer is a definitive response to a question.) "I think we should go." "You have school tomorrow and its time you went to bed." "You shouldn't do that because you'll get hurt." "You go get it, okay?” Order 227 "Close the door." "Go on to the next one." "Don't do that." "You do it." Disconfirmation - Topic Change Initiation Termination Other (Are you going?) "It's fourteen above in Kalamazoo." (What should we do tomorrow night?) "Its raining outside." (NOTE: Disconfirmations represent messages that disregard requests or demands of previous messages.) (The baby's learning to walk now.) "Where's tonight's paper?" (I bought a new dress.) "Hey, guess who I saw today at work?" (We're eating late tonight.) "Let's go to the hockey game on Saturday." "Well to start off with, I think we need to consider what the conditions are." "We would probably go to the basement right away.” (NOTE: An initiation is any message that starts a dialogue or discussion.) "That's about it." "There's no more to say." (NOTE: A termination is any message that brings a dialogue to a close.) Any message with an unclassifiable re- sponse mode. I.e., messages or parts of messages labeled as "IN-DISTINGUISH— ABLE" or "NON-DISTINGUISHABLE." APPENDIX VI 228 Table 24. Proportion of message type 11 (assertion as support) by dominance difference, social class, sex and correlated topics. !_ —r Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.09 0.03 * ns Social Class 0.33 0.33 1.94 ns Sex 0.002 1 0.002 * ns Dominance Difference * by Social Class 0'18 3 0'06 ns Dominance Difference * by Sex 0.13 3 0.04 ns Social Class by Sex 0.05 1 0.05 * ns Dominance Difference by Social Class 0.43 3 0.14 * ns by Sex Error 16.29 96 0 17 Topics 0.58 2 0.29 3.81 <0.03 Dominance Difference * by Topics 0.11 6 0.02 ns Social Class * by Topics 0.10 2 0.05 ns Sex by Topics 0.02 2 0.01 * ns Dominance Difference by Social Class 0.22 6 0.04 * ns by Topics Error 8.16 108 0.08 Not accounted for** 7.74 98 Total 34.442 335 *1ess than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 229 Table 25. Proportion of message type 13 (assertion as ex- tension) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.05 3 0.02 * ns Social Class 0.38 l 0.38 2.82 ns Sex 0.002 1 0.002 * ns Dominance Difference * by Social Class 0'03 3 0'01 ns Dominance Difference * by Sex 0.39 3 0.13 ns Social Class by Sex 0.27 l 0.27 1.96 ns Dominance Difference by Social Class 0.62 3 0.21 1.53 us by Sex Error 12.99 96 0.14 Topics 0.74 2 0.37 7.04 <0.001 Dominance Difference * by Topics 0.29 6 0.05 ns Social Class by Topics 0.32 2 0.16 2.99 ns Sex by Topics 0.05 2 0.02 * ns Dominance Difference by Social Class 0.66 6 0.11 2.09 us by Topics Error 5.69 108 0.05 Not accounted for** 10.04 98 Total 32.522 335 *1ess than 1 **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics in- teraction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 230 Table 26. Proportion of message type 14 (assertion as answer) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.26 3 0.09 1.40 ns Social Class 0.02 l 0.02 * ns Sex 0.21 l 0.21 3.36 ns Dominance Difference by Social Class 0.36 3 0.12 1.94 ns Dominance Difference by Sex 0.55 3 0.18 2.98 <0.04 Social Class by Sex 0.06 1 0.06 1.03 ns Dominance Difference by Social Class 0.63 3 0.21 3.39 <0.02 by Sex Error 5.96 96 0 06 Topics 0.74 2 0.37 8.13 <0.001 Dominance Difference * by Topics 0.18 6 0.03 ns Social Class by * Topics 0.02 2 0.01 ns Sex by Topics 0.21 2 0.10 2.28 ns Dominance Difference by Social Class 0.37 6 0.06 1.36 ns by Topics Error 4.88 108 0.05 Not accounted for** 3.47 98 Total 17.92 335 *Less than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics in- teraction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 231 Table 27. Proportion of message type 19 (assertion as initiation-termination) by dominance differ- ence, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.07 3 0.02 * ns Social Class 0.77 l 0.77 4.50 <0.04 Sex 0.19 1 0.19 1.08 ns Dominance Difference * by Social Class 0'40 3 0'13 ns Dominance Difference * by Sex 0.45 3 0.15 ns Social Class by Sex 0.06 1 0.06 * ns Dominance Difference by Social Class 0.05 3 0.02 * ns by Sex Error 16.51 96 0.17 Topics 0.04 2 0.02 * ns Dominance Difference * By Topics 0.13 6 0.02 ns Social Class * by Topics 0.06 2 0.03 ns Sex by Topics 0.11 2 0.06 1.55 ns Dominance Difference by Social Class 0.17 6 0.03 * ns by Topics Error 3.98 108 0.04 Not accounted for** 6.12 98 Total 29.11 335 *less than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 232 Table 28. Proportion of message type 23 (question as ex- tension) by dominance difference, social class, sex and correlated topics. 1* Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.73 0.24 1.42 ns Social Class 0.01 0.01 * ns Sex 1.47 1.47 8.55 <0.004 Dominance Difference * by Social Class 0'32 3 0'11 _ ns Dominance Difference by Sex 1.73 3 0.58 3.37 <0.02 Social Class by Sex 0.02 l 0.02 * ns Dominance Difference by Social Class 1.36 3 0.45 2.65 us by Sex Error 16.46 96 0 17 Topics 1.65 2 0.82 11.03 <0.0005 Dominance Difference * by Topics 0.30 6 0.05 ns Social Class by Topics 0.41 2 0.21 2.76 ns Sex by Topics 0.37 2 0.19 2.49 ns Dominance Difference by Social Class 0.71 6 0.12 1.59 ns by Topics Error 8.06 108 0.07 Not accounted for** 5.84 98 Total 39.44 335 *less than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics in- teraction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 233 Table 29. Proportion of message type 31U (unsuccessful talkover as support) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.87 0.29 2.06 ns Social Class 0.90 0.90 6.43 <0.02 Sex 0.03 0.03 * ns Dominance Difference * by Social Class 0’10 3 0'03 ns Dominance Difference * by Sex 0.26 3 0.09 ns Social Class by Sex 0.11 1 0.11 * ns Dominance Difference by Social Class 0.18 3 0.06 * ns by Sex Error 13 43 96 0 14 Topics 0 04 2 0.02 * ns Dominance Difference by Topics 0.56 6 0.09 2.12 ns Social Class by Topics 0.27 2 0.13 3.08 <0.05 Sex by Topics 0.14 2 0.07 1.60 ns Dominance Difference by Social Class 0.13 6 0.02 * us by Topics Error 4.73 108 0.04 Not accounted for** 4.69 98 Total 26.44 335 *less than 1 **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 234 Table 30. Proportion of message type 318 (successful as support) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.10 0.03 * ns Social Class 0.60 0.60 10.20 <0.002 Sex 0.12 0.12 1.99 ns Dominance Difference * by Social Class 0'06 3 0'02 ns Dominance Difference by Sex 0.29 3 0.10 1.61 ns Social Class by Sex 0.05 l 0.05 * ns Dominance Difference by Social Class 0.03 3 0.01 * ns by Sex Error 5.68 96 0 06 Topics 0.17 2 0.08 2.51 ns Dominance Difference * by Topics 0.09 6 0.01 ns Social Class * by Topics 0.05 2 0.02 ns Sex by Topics 0.12 2 0.06 1.72 ns Dominance Difference by Social Class 0.14 6 0.02 * ns by Topics Error 3.64 108 0.03 Not accounted for** 2.48 98 Total 13.61 335 *1ess than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 235 Table 31. Proportion of message type 33U (unsuccessful talkover as extension) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.78 3 0.26 2.73 <0.05 Social Class 0.17 1 0.17 1.79 ns Sex 0.29 l 0.29 3.04 ns Dominance Difference by Social Class 0.31 3 0.10 1.08 ns Dominance Difference by Sex 0.32 3 0.11 1.13 ns Social Class by Sex 0.04 1 0.04 * ns Dominance Difference by Social Class 0.05 3 0.02 * ns by Sex Error 9.18 96 0.10 Topics 0.54 2 0.27 4.86 <0.01 Dominance Difference * by Topics 0.25 6 0.04 ns Social Class by * Topics 0.01 2 0.01 ns Sex by Topics 0.01 2 0.01 * ns Dominance Difference by Social Class 0.05 6 0.01 * ns by Topics Error 6.01 108 0.06 Not accounted for** 5.43 98 Total 23.44 335 *1ess than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, and for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 236 Table 32. Proportion of message type 338 (successful talkover as extension) by dominance difference, social class, sex and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 1.08 0.36 1.87 ns SOcial Class 0.37 0.37 1.93 ns Sex 0.24 0.24 1.22 ns Dominance Difference * by Social Class 0'39 3 0'13 ns Dominance Difference * by Sex 0.35 3 0.12 ns Social Class by Sex 0.10 l 0.10 * ns Dominance Difference by Social Class 0.36 3 0.12 * ns by Sex Error 18.53 96 0.19 Topics 0.32 2 0.16 2.39 ns Dominance Difference by Topics 1.17 6 0.19 2.93 <0.02 Social Class by Topics 0.13 2 0.07 1.01 ns Sex by Topics 0.27 2 0.13 1.98 ns Dominance Difference by Social Class 0.29 6 0.05 * us by Topics Error 7.19 108 0.07 Not accounted for** 6.80 98 Total 37.59 335 *less than 1. **the sum of squares for Dominance Difference by Sex by Topics interaction, for Social Class by Sex by Topics interaction, for for Dominance Difference by Social Class by Sex by Topics interaction were not analyzed. 237 Table 33. Silences (code 000) by dominance difference, social class and correlated topics. L —.—-: _—‘— T f r Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 44.05 3 14.68 * ns Social Class 372.02 1 372.02 9.93 <0.003 Dominance Difference * by Social Class 95'45 3 31'82 Error 1797.71 48 37.45 Topics 82.33 2 41.17 8.96 <0.0005 Dominance Difference * by Topics 10.81 6 1.80 ns 5°C131.C1355 31.48 2 15.74 3.43 <0.04 y Top1cs Dominance Difference by Social Class 19.19 6 3.20 * ns by Topics Error 440.86 96 4.59 Total 2893.90 167 *less than 1. APPENDIX VI I Table 34. Proportion of competitive symmetrical trans- actions by dominance difference, social class and correlated topics. -1—‘ ‘ Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.36 3 0.12 1.58 ns Social Class 0.11 1 0.11 1.50 ns Dominance Difference by Social Class 0'11 3 0'04 * ns Error 3.68 48 0.08 Topics 0.65 2 0.33 7.50 <0.001 Dominance Difference by Topics 0.32 6 0.05 1.24 ns Social Class by Topics 0.05 2 0.02 * ns Dominance Difference by Social Class 0.42 6 0.07 1.62 us by Topics Error 4.18 96 0.04 Total 9.88 167 *less than 1. Table 35. Proportion of submissive symmetrical transac- tions by dominance difference, social class and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.15 3 .05 * ns Social Class 0.11 1 .11 1.23 ns Dominance Difference * by Social Class 0'06 3 '02 n5 Error 4.41 48 .09 Topics 0.01 2 .05 1.38 ns Dominance Difference by Topics 0.75 6 .12 3.17 <0.001 Social Class by Topics 0.22 2 .11 2.74 ns Dominance Difference by Social Class 0.24 6 .04 1.02 ns by Topics Error 3.78 96 .04 Total 9.73 167 *less than 1. 240 Table 36. Proportion of neutralized symmetrical trans- actions by dominance difference, social class and correlated topics. A— T _L i :— _ — m - -_ Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.19 3 0.06 * ns Social Class 1.28 l 1.28 13.21 <0.001 Dominance Difference by Social Class 0.32 3 0.11 1.09 ns Error 4.65 48 0.09 Topics 0.14 2 0.07 1.02 ns Dominance Difference by Topics 0.40 6 0.07 * ns Social Class by Topics 0.22 2 0.11 1.62 ns Dominance Difference by Social Class 0.10 6 0.02 * ns by Topics Error 6.53 96 0.07 Total 13.83 167 *less than 1. 241 Table 37. Proportion of transitory transactions by domi- nance difference, social class and correlated topics. Analysis of Variance Table Source of Variance SS df MS F p Dominance Difference 0.27 3 0.09 * ns Social Class 0.13 1 0.13 1.29 ns Dominance Difference * by Social Class 0'03 3 0'01 ns Error 5.01 48 0.10 Topics 0.19 2 0.10 1.51 ns Dominance Difference by Topics 0.53 6 0.09 1.38 ns Social Class * by Topics 0.02 2 0.01 ns Dominance Difference by Social Class 0.10 6 0.02 * us by Topics Error 6.16 96 0.06 Total 12.44 167 *less than 1. i! 1IIIIIIIIIIIII