$53.37. ”’— DOMINANCE IN MARITAL INTERACTION Thesis for the Degree ofrPh. D. MICHIGAN ”STATE UNIVERSITY RGBERT H. TINKER 191-? - .5 I..a‘- 'Wv , --.‘ ‘f'"' ' - ' ‘I’ ‘ 0.," ' A - ' A- W ’ fl‘ . V_" " . . a ’ . A . J. 3“"1. .. ...‘v_ ‘ 1- ~ ‘_ l.‘ l -,_. ‘1 J ‘ u I I ' ‘ ‘ - . ‘ r . I I . . s ‘5 I ~ This is to certify that; the thesis entitled DOMI NANCE IN MARITAL INTERACTION I presented by ‘ I I ‘1‘ ROBERT HAMMOND TINKER 4‘ I. . . - ~ 3- ‘3 . ,f‘; has been accepted towards fulfillment . A ; q‘ ‘3 - of the requirements for I . > i ‘ h' 2,. Ph . D . degree in PS 2 Ch 0109! - 'r . d . . ————‘— ' 1 l ///z/m_w Mafor professor/ _ . l $WX’) r> ABSTRACT DOMINANCE IN MARITAL INTERACTION BY Robert H. Tinker The purpose of the present study was to investigate the effects of different amounts of dominant behavior upon the interaction of student married couples. Dominance is a major dimension of inter- personal behavior, but little is known of its interpersonal consequences in a marital relationship. Twenty-four married couples, or 65% of a target sample, randomly selected from all couples living in Michigan State University two- bedroom married housing units, participated in the experimental procedure. This consisted of having each marital partner (1) complete an opinionnaire in a separate room from his spouse; (2) interact with his spouse in Discussion I to reach joint agreement on three lists previously filled out individually on the opinionnaire (a revealed difference technique); (3) complete a marital adjustment measure, again in a separate room from his spouse; (4) interact in Discussion II to complete jointly three more lists from the opinionnaire; (5) interact in Discussion III to reach agreement on a final set of revealed differences from the opinionnaire. The married couples were randomly assigned to one of four groups: a high, medium, or low dominance, or control group. If the couple was assigned to one of the dominance Robert H. Tinker groups, a randomly selected spouse (called the Selected Spouse), was given special instructions for Discussion II. The Selected Spouse was asked to do his "utmost" to get all, or the two most important of his individual choices to be the ones on the couple's joint list, for the high and medium dominance groups respectively. The Selected Spouse from the low dominance group was asked to let his spouse determine the two most important choices on the joint list. The partner of the Selected Spouse was unaware of the special instructions. The discussions were tape recorded with each couple's knowledge. One main rater judged three-minute segments of interaction from the middle of each discussion period. Each speech was rated as to whether it was high, medium, or low on each of four categories: Negative—hostile (N-H), Positive-Affectionate, Dominant, and Submissive. Ratings were also made of the clarity of the interaction segments, and the time needed to reach agreement on each discussion was recorded. Independent ratings by a second rater on every sixth couple furnished evidence that all ratings were highly reliable. Analyses of variance indicated that the couples were only partially able to carry out the instructions to be high, medium, or low dominant, which might have obscured or diminished differences among the four groups. The number of Negative-hostile speeches was the only dependent variable affected by the amount of dominance. Specifically, spouses of either sex allowed a moderate increase in partner dominance (medium dominance group) without an increase in N-H speeches, but when the Selected Spouse was instructed to greatly increase the amount of dominance (the high dominance group), an increase in the number of N-H Robert H. Tinker speeches, but not greater dominance, resulted. Overall, the wives made more Submissive and Positive-affectionate speeches than their husbands, and both spouses gradually became less submissive over the three discussion periods. Correlational analyses, based on the data from Discussion I, supported the analyses of variance findings in that the number of high Dominant speeches significantly correlated with the number of Negative- hostile speeches, while the number of medium Dominant speeches did not. However, the number of medium Dominant speeches was significantly related to the number of Positive-affectionate speeches, which was not supported by the analyses of variance. It was concluded that high dominance had a negative-hostile effect on the married couples' interaction, while moderate dominance did not. These findings, while not at variance with present interactional theories, would not have been predicted by them. Further, the results go beyond much present interactional research in indicating a causal relationship between two interactional variables. L4/Wh%[' z/é/ 7 7//L/7L DOMINANCE IN MARITAL INTERACTION By .34.) Robert HCflTinker A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology l972 To Marti, who still happens to be a lovable wife. And to Mike and Joni, who live in the present, but belong to the future. 11' ACKNOWLEDGEMENTS I am deeply grateful to John R. Hurley, Ph.D., Gary E. Stollak, Ph.D., Dozier Thornton, Ph.D., and John P. McKinney, Ph.D., who as members of my committee, made many suggestions which substantially improved the quality of this research. A Special thanks goes to John R. Hurley, chairman, who was consistently helpful, patient, and encouraging. His expertise was of great help in the writing of the thesis. I would also like to thank Dave Hanley, who served as the second rater, giving many hours of assistance. These are the persons who were directly involved in the conception and execution of the research. There were many others along the way, who have influenced and stimulated me. To them also, I am grateful. TABLE OF CONTENTS Page DEDICATION ........................ ii ACKNOWLEDGEMENTS ..................... iii LIST OF TABLES ...................... vii LIST OF APPENDICES .................... viii INTRODUCTION ....................... 1 REVIEW OF THE LITERATURE ................. 2a Why Study Marital Relationships? ........... 2a Lack of Study .................... 2a. The Relationship between Marital Disturbances and Offspring Disturbances ............... 3 Evaluation of Research which Shows Relationships between Marital Disturbances and Offspring Disturbances .................... 6 Marital Research .................... ll Research in Marital Dominance ............. 14 Studies of Dominant Behavior ............ l5 Theories of Dominant Behavior ............ 16 Definition of Dominance ............... 17 Studies of Parental and Familial Dominance ..... 18 Studies of Dominance with Married Couples ...... 22 Self-report studies ........... . . . . 22 Interactional studies ............... 23 Conclusions about marital dominance ........ 27 Methodological Considerations ............. 28 iv Page METHOD ............................ 32 Subjects .......................... 32 Procedure ......................... 33 The Opinionnaire ..................... 37 The Marital Adjustment Measure .............. 37 Influence Scores .................... 39 The Dominance Index ................... 40 Evaluation of Ability to Carry Out Instructions ..... 40 Evaluation of the Nature of the Interaction ....... 41 Evaluation of the Clarity of the Interaction ....... 45 Evaluation of the Ability to Reach Agreement ....... 47 The Correlational Analyses ............... 47 Individual correlations ............... 48 Couple correlations .................. 49 RESULTS ........................... 50 Interrater Reliability ................... 50 Evaluation of Ability to Carry Out Instructions ...... 50 Influence Scores of the Selected Spouse ......... 50 Dominance Index Scores ................. 52 Evaluation of the Effects of Different Amounts of Dominance upon the Dependent variables ............... 55 Speeches Labeled Negative-Hostile ............ 56 Speeches Labeled Dominant ................ 60 Speeches Labeled Submissive ............... 62 Speeches Labeled Positive-Affectionate .......... 65 Clarity of the Interaction ............... 65 Time Needed to Reach Agreement .............. Summary of Major Findings from the Analyses of Variance The Correlational Analysis ................. Individual Correlations ................. Couple Correlations . . ................. DISCUSSION ....... . . . . .............. Major Findings ....................... The Issue of Causality ................... The Dominance-Hostility Relationship ............ The Difference between High Dominance and Medium Dominance . The Relationship of Dominance with the other Major Variables ........................ Sex Differences ...................... Persistence of Experimental Effects ............ Interactional Changes over Time .............. The Rating Scales ..................... Limitations of Study .................... Comparison of Results with other Studies ........... Directions for Future Research ................ Implications for Therapy ................... Concluding Statement . . . .................. vi Page 65 65 67 73 73 74 76 83 Table 01 h 0) N o o o 0 N03 oo 10. ll. 12. LIST OF TABLES Rankings by the Selected Spouse of Topics Discussed within each Discussion Period .................. Relative Usage of the Speech Categories . . . . Interrater Reliability for Four Couples . . . . Mean Influence Scores of the Selected Spouses Dominance Index Scores (Totals) ........ Total Number of Speeches Labeled Medium N-H Total Number of Speeches Labeled N-H (High plus Medium) ................. Total Number of Speeches Labeled Dominant Total Number of Speeches Labeled Submissive Total Number of Speeches Labeled Positive- Affectionate ............... Product- Moment Correlations Among All Individuals (N = 48)+ ........... Product- Moment Correlations Among All Couples (N = 24)+ ............. vii Page . 38 46 51 53 54 57 59 61 64 LIST OF APPENDICES Appendix Page A. Letter for Soliciting Couples 99 8. Complete Instructions to Couples lOl C. The Opinionnaire l05 D. The Family Concept Inventory 109 E. E_Ratios from Analyses of Variance llO viii "...psychopathology is a product of a power struggle between persons rather than between internal forces (Haley, 1963, p. l56)." INTRODUCTION Since the present research is concerned with dominating behavior in a marital relationship, several pertinent questions about this topic come to mind. First, why study marital relationships? Second, why study the effects of dominance in such a relationship? Third, what is presently known about the marital relationship in general and about marital dominance in particular? The present review of the literature deals with these rather basic questions. Although the author is much indebted to the concepts of communication theory, these concepts will not be reviewed. Rather, the interested reader is referred to the seminal works by Haley (1963), and Watzlavick, Beavin, and Jackson (l967). One of the author's reasons for studying marital dominance, is that he became intrigued by the difficulties which couples in marital therapy showed in trying to reach agreement on almost any issue, no matter how trivial. Each spouse seemed to be putting pressure on the other to come around to his own point of view. The more one spouse tried to convince the other, the more the other resisted, either overtly or covertly. It thus seemed that the more involved that the marital partners became in trying to circumscribe, change or direct the behavior of each other, the more they resisted each other's attempts by refusing to negotiate, compromise, or change. Even the simplest disagreements could become battlegrounds for control. Another way of , v .I‘ . ‘7») ,1. I . \ ‘~ . v f )1 I . .. I i . . ,I"'i-' .. .. 'I J _ . . I . , , . .I I 9 Y stating this, is that these couples seemed to have problems in handling controlling, or dominating behavior. From such observations, the author tentatively formulated the idea that marital dominance would be a worthwhile topic to investigate, in that it seemed to be a highly problematic area for married couples having difficulties. REVIEW OF THE LITERATURE Why_Study Marital Relationships? Lack 9f_$tudy There is a great deal of ignorance about marital relationships. This is in spite of surveys on various aspects of marriage, which have been shown to be of doubtful validity because married persons have been repeatedly shown to be poor observers and reporters of their own behavior (Kenkel and Hoffman, 1956; Levinger, 1963; Olson, 1969; and Weller and Luchterhand, 1969). It is also in spite of numerous marriage manuals which are short on data, but long on prescriptive standards. Further, this ignorance persists even though the vast majority of Americans get married, and a large percentage (about one-third) subsequently get divorced. It persists even though the family is the basic social unit in almost all societies. In short, it persists in the face of an exceedingly pervasive institution. As Lederer and Jackson (1968) state: "On their wedding day, a young man and a young woman, standing before the priest, minister, or justice of the peace, usually have a high Opinion of one another. They overflow with joyous thoughts. Each has a firm intention of pleasing and nourishing the cherished person who is 2a about to become a partner for life. Some years later (the highest incidence of divorce, excluding teen- agers, is after ten or so years), these same two people may be living in a chronic situation of hate, fear, and confusion. Each spouse in such a marriage may blame the other and defensively emphasize how hard he tried to be loving, tried to make the marriage a success, and tried to keep the other from sabotaging the effort. What causes such frightful changes? What brings about such start- ling emotional and behavioral metamorphoses (p. 15)?" The above description of what frequently happens in marriage is doubly unfortunate, in that the couple involved likely has only a vague and inaccurate idea of what went wrong. Several pieces of research (Kenkel and Hoffman, 1956; Levinger, 1963; Olson, 1969; and Weller and Luchterhand, 1969) indicate that most couples lack even rudimentary awareness of their interaction with each other. In fact, Kenkel and Hoffman state: "The couples were unpracticed at even the relatively superficial degree of analysis necessary to recognize the part they play in a simple and structured interaction (p. 316)." Thus, the first basic reason to study marital relationships is that they are pervasive, but relatively unstudied. ‘ The Relationship between Marital DiSturbances and Offspring Disturbances Another major reason to study marital relationships is that there is an association between marital disturbances and psychological disturbances in their offspring. While there is a large body of literature supporting this statement, only a few illustrative studies will be reviewed here. For more complete reviews, the reader is refer- red to Oaklander (1971), Hoffman (1969), and MacKenzie (1968). Such studies have indicated that schizophrenic, neurotic, and normal parents differ from each other on measures of clarity and agreement. With respect to clarity, Morris and Wynne (1965), Singer and Wynne (1965a & b, 1966), Beavers, Blumberg, Timken and Weiner (1965), Lerner (1964), and Levin (1966) have found that different measures of clarity differentiate between parents of schiZOphrenics, neurotics and normals, with parents of schizophrenics being least clear in their communication and parents of normals most clear. With respect to agreement, Fisher, Boyd, Walker, and Sheer (1959), Caputo (1963), and Lerner (1964), have found that parents of schizo- phrenics have most difficulty in reaching definite agreement on various tasks, and that parents of normals are most able to reach definite agree- ment (with parents of neurotics intermediate). More recent family interaction research has been getting more SOphisticated and complicated. Earlier studies have been found to have possible methodological shortcomings. Thus despite a careful attempt at replicating Singer and Wynne's research (1965a & b, 1966), Hirsch and Leff (1971), were unable to come up with similar results. Singer and Wynne correctly identified 85-90% of the parents of typical schizophrenics and neurotics from their Rorschach protocols, and there was no overlap of the scores of the parents of schizophrenics and neurotics. In contrast, Hirsch and Leff correctly identified only 61% of the parents on the basis of their Rorschachs, with a 97.5% overlap in the distribution of the two groups. They suggest the discrepant findings were a result of methodological shortcomings in Singer and Wynne's research. Another methodoloqical error has likely been resolved by Waxler and Mishler (1970), who found that when participation rates are controlled, there is no difference between normal and schizophrenic families in predict- ibility of act sequences. Previously there had been disagreement in the literature on whether schizophrenic families were more rigid in their act sequences. Sojit's studies (1969 & 1971) are good examples of the more recent pieces of research. In the earlier study, Sojit compared the interaction of 20 sets of parents of ulcerative colitis children, eight sets of parents of delinquents, nine sets of parents of normal children, nine sets of cystic fibrosis children, and found basic interactional differences. The total number of interchanges of ulcerative colitis couples was small and disagreement between them was rare. They seldom expressed understanding of the other's verbalizations. They seemed highly restrictive, Sojit concluded. Delinquent couples seemed to encourage indiscriminate action, to be perceptive about others to gain control, and to use disqualifications and disaffirmations to disagree covertly. They disagreed less than normal parents, more than ulcer- ative colitis parents. Cystic Fibrosis parents weren't different from normal parents who were characterized by frequent disagreement, and less use of disqualifications and disaffirmations. In the more recent study (1971), Sojit also compared parents of schiZOphrenics with these previous groups. O'Connor and Stachowiak (1971) compared patterns of interaction in families with low adjusted, high adjusted, and mentally retarded children, and found basic differences. Novak and van der Veen (1970) studied the families of disturbed adolescents where there was a normal sibling, and suggested on the basis of their results that rather than bad parents "causing" a disturbed child, a more appropriate model would be that a particular type of child in interaction with a particular set of parents becomes disturbed. This model fits in very well with research by Thomas, Chess, and Birch (1968), who found individual temperamental differences among neonates which persisted over several years. Some of these temperamental differences, in interaction with parental behaviors, gave rise to behavioral disorders in the children. Thomas, Chess, and Birch conclude that a given child needs to be handled in a manner appropriate to his temperamental characteristics. Evaluation of Research which Shows Relationships between Marital Disturbances and Offspring Disturbances Two kinds of articles predominate in the family interaction literature: the kind which demonstrates a linkage between abnormal parental behavior and abnormal child behavior, and the kind which advocates the use of family therapy and various techniques of family therapy. The two go hand in hand. If the parents and children are influencing each other in some sort of system, as the research articles seem to indicate, it seems logical to make changes with several persons in the svstem, rather than with just one (i.e. the disturbed child). However, the research articles which demonstrate the association between abnormal parental and abnormal child behavior are almost all of one type, where different groups of parents are compared with each other. This research has great value in that it specifies what the connections are. Lerner's research (1964) is a good example of how specific parental interactions seem tied to certain offspring behaviors, and not others. Lerner studied 36 pairs of parents: 24 sets of parents of schizophrenic sons. and 12 sets of parents of sons hospitalized for nonpsychiatric reasons. Initially the parents of the 24 schizophrenic sons were dichotomized on the basis of their son's social competence scores, and later on the baSis of their son's "thought disorder". When the schizophrenic sample was dichotomized on the basis of a social competence score, disproportionate patterns of parental dominance and submission appeared, and the sex of the dominant parent appeared to be related to the patient's premorbid level of social maturity, such that excessive maternal dominance was associated with low social competence on the part of the schizophrenic sons, and excessive paternal dominance with higher social competency. However, when the schizophrenic sample was dichotomized on the basis of the severity of "thought disorder" similar patterns of disproportionate dominance and submission failed to emerge, but severity of "thought disorder" was related to the degree of parental distortions. Lerner interpreted this as indicating that the entire intrafamilial environment did not exert an indiscriminant influence, but that role structure was specifically related to the offspring's social competence, while parental distortion was related to offspring "thought disorder". By relating specific parental behaviors to specific child behaviors, psychotherapists can be helped with setting therapeutic interventions. For example, as Sojit (1969) found that children with ulcerative colitis had overly restrictive parents, a psychotherapist might direct his efforts at getting the parents to be less restrictive, and then the child's ulcerative colitis should improve. All this seems simple and straightforward. Find out what parental behaviors are causing the child's problems, and then intervene appropriately. The problem is that none of the family interaction research shows causality, but it is very tempting to infer such causality. That this can lead to unwarranted conclusions can be suggested by the following example: Let's say that two parents are mentally retarded for genetic reasons, and that they have a child who is retarded for the same reasons. If their family interaction were studied, one would find that the parents were interacting with each other in deviant ways, and that their child's interaction was deviant in the same way. On the face of it, if one didn't know the origin of the mental retardation, one would be tempted to infer that the child "learned" his retardation from his parents. In a similar way, one is tempted to infer that schizophrenic children "learned" their deviant communication patterns from their parents. This may not be entirely so, in that the causes of schi20phrenia may be genetic, constitutional and environmental, with the environment festering learned maladaptive interpersonal behaviors, as well as furnishing the specific environmental stresses. Sarnoff Mednick has begun longitudinal research on schi20phrenics with fascinating preliminary results (Mednick, 1970; Mednick, 1971; Mednick and McNeil, 1968; Mednick and Schulsinger, 1968). Noting that schizophrenic women tended to have a higher incidence of schizophrenic children than nonschizophrenic women, he followed schizophrenic women over eight years, starting before they had children. He found that the combination of pregnancy and birth complications plus a schizophrenic mother was highly likely to produce a schizophrenic child. More specifically, in 1962, Mednick examined 207 children who were likely to become schizophrenic (they had schi20phrenic mothers). He also tested 104 controls (no mental illness for three generations), matched for sex, age, social class, education, residence, and child- ren's home experience. Ages were between nine and twenty. The subjects were given personality and IQ tests, a psychiatric interview, physiological measures (heart rate, muscle tension, respiration, and GSR). Mednick also interviewed parents, obtained school reports and midwife's report. After eight years, 27 in the high risk group became psychotic. These subjects were matched with a high risk subject who did not become psychotic and with a control from the low risk group. Five critical variables were found: 1. Future psychotic §s lost schizophrenic mothers to hOSpitalization at an earlier age, and these mothers were more severely schi20phrenic. 2. Future psychotic §fs tended to be disruptive in class: they were domineering, aggressive, and posed disciplinary problems, (more than 50% versus 18% in the well group, and 11% in the control group). 3. 0n the word association test, the future psychotics gave series of words that drifted away from the stimulus word, or restricted themselves to one or two words. 4. In nine stress trials, the well and control groups' GSR habituated, but the group that later became psychotic became more irritable; the latency of their GSR decreased 75%. In extinction trials, the future psychotics showed great resistance to extinction. 10 They also showed remarkably fast recovery from autonomic imbalance (rate GSR returned to normal). These measures almost perfectly discriminated between the future psychotics and well group. 5. Seventy per cent of the future psychotics' mothers had suffered one or more serious pregnancy and birth complications, in contrast to 15% of the Well groups, and 33% of the controls (these were such things as anoxia, prematurity, prolonged labor). The 5'5 with the pregnancy and birth complications were almost all the ones with the abnormal GSR's. Mednick interprets these findings as evidence that genetic predisposition from the schizophrenic mother and pregnancy and birth complications interact to produce a schi20phrenic. He also suggests that the pregnancy and birth complications damage the body's ability to regulate its stress response mechanisms, and notes that rats with hippocampal lesions parallel the behavior of the subjects who had pregnancy and birth complications. Mednick generally advocates the use of the "high risk" methodology in social science research; i.e. define a group of high risk subjects, and then design a prevention program, into which half of the high risk subjects could be put. The two groups of high risk subjects could then be compared to evaluate the effectiveness of the preventative programs. This kind of methodology could be used with some of the presently available data on parental and family interactions. For example, if parental restrictiveness is associated with ulcerative colitis in children, couples who interact in highly restricted ways could be followed longitudinally, and their children evaluated. When some children developed the disorder, 11 they could be compared, and their family interactions could be compared with the high risk children and controls who did not develop the disease. Later, family therapy (and other programs) could be evaluated with respect to their efficiency in reducing the incidence of ulcerative colitis in the high risk children. The major point, however, is that the descriptive and comparative family interaction research which has been gathering data for the last ten years is not enough. There is an additional need for research which will indicate causal relationships. If one compares different groups of parents, there is practically an infinity of different groups that can be compared, and a large number of variables. Just because significant differences are found, it does not mean that these differences are mean- ingful or causal. Even when differences are found between groups, these differences need to be cross-validated. Marital Research Since the present research is an interactional study, concerned with the effects of different amounts of dominance on the quality of the interaction, the previous research on marital dominance, and the previously completed studies actually observing marital interaction will be most pertinent to the present review. In order to place these topics in some kind of context, however, the present review will take a look at some of the general conclusions reached in marital research over the past 20 or 30 years. Ryder (1970), divided marital research into two major areas: research which contrasts "happy" or "successful" marriages with their opposites, and research which focuses on marital power. Ryder noted 12 that there are problems with the orientation which contrasts happy with unhappy marriages, in that it is too gross a distinction by which to evaluate marriages, and in that such research tries to treat such evaluations as descriptive facts. Cuber and Harroff's research (1965) supports Ryder's contention that contrasting successful or happy marriages with their opposites is too gross a comparison. In their study of the marriages of 437 upper-middle—class Americans, they found five major kinds of marital relationships, with successful and un- successful marriages within most types. Ryder also noted problems with the research on marital power. It seems almost impossible to determine an answer to such questions as "Who has more power, the husband or wife?" Safilios-Rothschild (1970), in her review of family power structure, indicated that even though there are a large number of studies on family power structure, one would have an impossible task describing the power structure in the American family, unless one were willing to accept the results of just one major study. She further noted that the present data on power structure are based predominantly on wives' answers, and rarely on data from both spouses. Barry's (1970) Review divides marital research into studies of mate selection, and studies of marital adjustment. Tharp (1963) has written a definitive review of the research in mate selection, which indicates that individuals getting married choose each other on the basis of homogamy (like choosing like), not only for cultural and social variables (race, age, religion, ethnic origin, and social class), but also in regard to personality variables as measured by tests. Even if it is known that like chooses like, the question re- l3 mains as to how this is accomplished. Barry (1970) points out that there are several theories as to how this is accomplished, but one of the most interesting is one proposed by Coombs (1966). His theory (based on a review of research on interpersonal attraction), is that value consensus fosters mutually rewarding interaction and leads to interpersonal attraction. To test the theory, Coombs sponsored a dance, pairing computer-selected college students of varying degrees of value consensus, and obtained results in the hypothesized direction concerning interpersonal attraction. In considering the studies of marital adjustment, Barry (1970), like Ryder (1970), also notes problems with defining marital "success". It has been variously defined on the basis of number of years married, on the absence of marital counseling, or on reported or judged happiness. The trouble with the first two standards is that a marriage may be in deep difficulty, and yet not end in divorce, separation, or even at the marriage counselor. Reported happiness is also a problem in that such ratings are subject to halo effects from areas outside the marital relationship itself. In spite of these limitations, Barry feels that the studies of marital adjustment have some value in terms of tentative data, and in terms of developing more sophisticated methodology. He concludes from these studies that factors pertaining to the husband appear to be crucial to marital success. He writes: "Background factors generally considered to lead to a stable male identity, such as happiness of the husband's parents' marriage and the husband's close attachment to his father, are related to happiness in l4 marriage. The higher the husband's socioeconomic status and education- al level, the greater the marital happiness. The more stable and non- neurotic the husband portrays himself on personality inventories at the time of marriage and the more consistent he is in such self-portrayal over the course of the marriage, the happier the marriage. The higher the wife rates him on emotional maturity as well as on fulfilling his role as husband in conformity to cultural expectations, the happier the marriage. The more the wife comes to resemble her husband on attitude and personality inventories over time, the happier the marriage. It would appear--to generalize a bit--that a solid male identification, established through affectional ties with the father and buttressed by academic and/or occupational success and the esteem of his wife, is strongly related to happiness in marriage for the couple. (p. 47)." General conclusions then, from the most researched questions in marital relationships are that homogamy is the rule in mate selection, family power structure is still obscure, and that marital success is more related to qualities of the husband than to qualities of the wife. Research in Marital Dominance Much of the research which has been completed on marital or family power structure has had serious shortcomings, as has been congently pointed out by Safilios-Rothschild (1970). A good part of the difficulty has been the nature of the question which has been asked. Instead of asking the question, "Who really has the power here?", which is held to be relatively unanswerable (it seems likely that husbands are in charge in some areas, wives in others, and that these areas vary from couple to couple, and probably within each couple over time), the present research 15 instead asks the question, "What happens to the interaction when one spouse acts in a highly dominant fashion toward the other spouse?". It would seem that answers to the latter question would be less situation specific, than answers to the former, and hence more researchable. Some studies have sought to determine if aberrant dominance patterns on the part of the parents are associated with offspring deviancy of various sorts, by comparing different groups of parents with each other on measures of dominance. A second type of research has dealt primarily with couples, rather than families, and typically has focused upon how dominance affects marital happiness or stability. Before reviewing these two groups of research, however, it would be pertinent to take a look at dominance itself as an interpersonal variable. Studies 9f_Dominant Behavior A substantial number of studies and reviews indicate that dominant behavior is a major category of interpersonal behavior. Hurley (1972), in an extensive review of relevant literature, concluded that there was a strong evidence for two prepotent interpersonal dimensions: Acceptance/ Rejection of Others (often labeled affection-hostility), and Self- Acceptance/Rejection (often referred to as a dominance-submission dimension). A number of other reviews of the literature (Adams, 1964; Foa, 1961, and Shaefer, 1959, 1961) have suggested that a large number of studies converge in their findings toward an arrangement of interpersonal behaviors in a circumplex around two orthogonal axes: a dominance-submission axis and a hostility-affection axis. Leary (1957) who used such a circumplex model in his theory of interpersonal behavior (16 interpersonal reflexes arranged around a circumplex), suggested that certain interpersonal behaviors 16 on the part of one person, tended to "pull" interpersonal behaviors of a specific sort from another person. For example, dominant behavior would tend to "pull" submissive behavior from another person. He wrote, "The facade of power and control provokes others to obedience, defer- ence and respect (p. 315)." However, Leary did go on to say that the nature of the response was altered by the personality of the other person. For example, he noted that two managerial types might generate a power struggle. In general, though, the theory stated that dominance pulled submission, except in special circumstances. The important thing is that Leary's hypotheses are testable. Does dominance pull submission in all relationships or only in some? Does submission pull dominance? Does dominance pull submission in a marital relationship? Leary's hypotheses remain largely unexplored, although there is some research which bears on these questions. Theories 9f_Dominant Behavior In addition to the evidence that dominant behavior is a major dimension of behavior, a number of communication theorists have accorded dominance a major place in their theories, either implicitly or explicitly. Haley (1963) argued that each and every communication that one person makes to another contains an element in it which re- lates to the control of that relationship. The other person, in each response that he makes, has the choice of accepting or rejecting the first person's control maneuver. Also, in any relationship, two people are faced with deciding what kinds of behavior are to take place in the relationship, and with deciding who is to control what is to take place in the relationship and thereby control the definition of the relationship. 17 Thus, the more one person controls the relationship, the more he is dominant in it. The present study poses the question: What happens when one spouse tries to be highly in control, or highly dominant in the relationship? Haley also argues that when the struggle for control between two people becomes extreme, it may lead to symptom formation. Watzlawick, Beavin, and Jackson (1967) state that each communication has a report and a command function. The report function conveys information, and is synonymous with the content of the message. The command aspect refers to the relationship between communicants. They further state that relationships are only rarely defined deliberately or with full awareness, and that the healthier a relationship, the more the relation- ship aspect of communication recedes in importance. 0n the other hand, less healthy or satisfactory relationships are characterized by a constant struggle about the nature of the relationship, with the content aspect of communication becoming less and less important. Bach and Wyden (1968), while devoting very little space to the concept of dominance in marriages, in their book on how intimates can fight effectively, do mention that all COUples must fight, because each person has his own ideas, some of which will conflict with the other person's idea. Each person wants his own ideas to be dominant, and all marital struggles can be reduced to the issue of who or whose ideas are to be dominant in a particular situation. Definition gf_Dominance While the concept of dominance is being discussed, it might be well to provide an indication of how the term is used in the present research. 18 It is used exactly in the way that Safilios-Rothschild uses the term "influence": "the degree to which formal or informal, overt or covert pressure exerted by the one spouse upon the other is successful in imposing that spouse's point of view about a pending decision, despite initial opposition (p. 540)." Studies of Parental and Familial Dominance Interactional studies on parental and familial dominance have yielded inconsistent results. Some studies have shown no differences in dominance patterns between normal and abnormal families, while others have reported that normal families are more equalitarian in their dominance patterns, abnormal families being more authoritarian and yet others have reported the opposite result that normal families are more authoritarian, abnormal families more equalitarian. Comparisons between studies, however, are quite difficult, in that the studies differ from each other in several major respects. First of all, different normal and abnormal groups are compared. For example, one study might compare dominance relationships between normal and schi20phrenic families, while another might compare normal and neurotic families, while a third study might compare high-adjusted-offspring families with low-adjusted-offspring families. Secondly, few studies except those reported by the same author have used the same measures of dominance. Thirdly, different interactional tasks have been used to elicit family interaction. Barry (1968), Swain (1969), Smith (1970), and Hofman (1969) have all shown that the nature of the interaction task has a significant effect upon the interaction obtained. Finally, some studies have not controlled for relative participation rates, 19 which Waxler and Mishler (1970) have shown to produce results of spurious statistical significance. No differences in dominance have been reported by Caputo (1963), who found no differences in the dominance patterns of parents of schizophrenics and parents of disabled veterans; Farina and Holzberg (1968), who found no difference in dominance patterns between the families of good premorbid schizophrenics, poor premorbid schizophrenics, and nonschizophrenic psychiatrically impaired families; Becker and Iwakami (1969), who found no differences in dominance patterns between nonclinic families and nonpsychotic clinic families, after adjusting for verbal activity; and Hofman (1969), who found no dominance differences between clinic and nonclinic families. Of the studies reporting equalitarian dominance patterns in normal families and authoritarian dominance patterns in abnormal families, both Farina (1960) and Lerner (1964) found that parents of poor premorbid male schizophrenics were characterized by excessive maternal dominance, parents of good premorbid male schizophrenics by excessive paternal dominance, with controls being intermediate. Haley (1962) found that normal families shared equally in wins, while in schizophrenic child families, the father won most, with the schizophrenic child winning hardly at all. Haley (1964) also reported that normal families participated more equally than abnormal families. The preponderance of evidence, however, seems to lie with the studies which report at least a moderate trend toward dominance on the part of one parent in normal families, with less dominance noted in abnormal families. 20 Bachove and Zubaly (1965) found that in nonclinic families, the father was generally the dominant figure. Mishler and Waxler (1968) reported that with respect to "attention control" families with a normal son demonstrated a pattern with the father most powerful, and the son least powerful. In families with a schi20phrenic son, mothers and sons took high power with fathers last. With respect to "person control" normal families directly confronted each other to maintain the recognized power structure, while the power structure was less clear in families with schizophrenic sons. MacKenzie (1968), found that clinic mothers were aggressively dominant compared to nonclinic mothers who were friend- ly dominant. Pienaar (1969) found that parents of high-adjusted children provided more overall dominance, and tended toward paternal dominance, compared to parents of low-adjusted children. Murrell and Stachowiak (1967) found nonclinic families distributed their inter- action less evenly than clinic families. Schuhan (1970) reported that normal families had a clearer power structure than clinic families, with the father winning most, the child least. In the clinic families, the child won more than the father. A family interaction study by Cancian (1971) deserves particular attention, in that its methodology and results are especially relevant to the present research. As with the present study, the major variables of interest were a dominance-submission axis and an affection-hostility axis. Fourteen upper-middle class, two-children families were observed in their homes for 14 hours over several days and their interaction was coded by raters unfamiliar with the study's hypotheses. An interaction was coded into five modes (seek, give, deprive, accept, reject), three 21 resources (information, support or affection, direction). The initiater and recipient of the act was also recorded. It was hypothesized that the greater the dominance from A to B, the greater the submissiveness would be from B to A; the more the affection was from A to B, the more the affection would be from B to A. These hypotheses are of special interest both because they are tests of Leary's formulations, and because they explore the relationships among the four previously mentioned variables. In contrast, the present study is an attempt to determine whether dominance level is causally related to the other three interpersonal variables. Cancian's study does not show causality, but attempts to determine whether there is an association between affection and affection, and between dominance and submissiveness. Cancian found that for the mother-father dyad, there was a .53 correlation between A to B and B to A dominance, a result opposite to their hypothesis (however, the hypothesis was supported by one mother-child, and for the child-child dyad). Cancian's affection hypothesis was generally supported. Such findings suggest that Leary's idea that dominance pulls submission may be true for unequal status dyads, and may not be true for relatively equal persons, such as marital partners. Cancian's research also deserves note in that it is one of the few studies on family interaction which has broken out of the mold of comparing one group of parents with another group, in order to find interactional differences. Rather, he seeks to find associations among major interaction variables, surely an area which deserves more attention in interactional research. It would also be of interest to explore causal relations among major interactional variables as well as association, and the present research is an attempt in that direction. 22 Studies 9f_Dominance with Married Couples In some respects, it is surprising that marital interaction has attracted less attention than family interaction. A dyadic marital system is much simpler to study, it is the initial family relationship, it is usually more stable and enduring than the other family relationships, and it is often the most intimate family relationship. However, one reason family interaction has likely attracted more attention and research, is that the interest seems to have grown out of clinical practice, where it became more and more clear that disturbed children had come from a disturbed family. Thus, the study of the family seems to have grown out of the pressing need to help disturbed children. Self-report studies. Because of the poverty of research on marital interaction, there are relatively few studies of marital dominance. There are a few studies which show that husbands and wives who say they share decision making in an equalitarian fashion are happier (Jansen, 1952; Blood and Wolfe, 1960; Wolfe, 1962). Also it appears that wife- dominated couples are the most unhappy with marriage (Jansen, 1952; Blood and Wolfe, 1960; Wolfe, 1962). Of course, a problem with the self-report method of determining dominance, is that it does not answer the question of whether couples who say they are equalitarian, really act that way. Jacobson (1952) found that a discrepancy between spouses; ideas on dominance was associated with divorce. Studying 100 randomly selected divorced couples, and 100 randomly selected married couples by interviews and questionnaires, he found that the divorced males had scores indicating the greatest husband dominance, and divorced females 23 had scores indicating the strongest equalitarian views. The difference between the mean scores for divorced couples was four times larger than for married COUples. Studying 641 problem marriages presented by one or two spouses at the Marriage Guidance Council in Australia, Krupinski, Marshall, and Yule (1970), found three main dimensions of marital maladjustment: a dominance-submission power struggle, the alienation pole of an affection-alienation dimension, and the avoidance pole of an intimacy- avoidance dimension. The studies by Jacobson (1952), and Krupinski, Marshall and Yule (1970), suggest that the problems concerning the dominance-submission dimension are of major importance in marriage, and in the dissolution of marriage. Further study could well be directed toward this dimension. Interaction studies. With respect to interactional studies of marital dominance, Bauman and Roman (1966), using a revealed difference technique (RDT), obtained a split-half reliability for dominance of .54 (statistically significant), and concluded that couples could be reliably characterized as to degree of dominance. Bahr and Rollins (1971) found that the more equalitarian the conjugal power structure in a noncrisis situation, the more likely the relative marital power was to change during a crisis. They concluded that a democratic couple adjusted to a crisis more flexibly than a dominant couple. Comparing 20 alcoholics and their wives with 20 nonalcoholics and their wives, Gorad (1971) found that alcoholics and their wives did not have a dominant-submissive relationship, but were locked in a "one- ups—man-ship" battle, with both spouses highly competitive, and with 24 the alcoholic's wife accepting responsibility, while her husband avoided it. Control couples did not show such extreme patterns. An N.I.M.H. project has been in existence over the past several years which has been studying the early stages of family development. A number of published studies document the work of this group which has been following newlywed couples longitudinally (Goodrich, 1968; Goodrich and Boomer, 1963; Goodrich, Ryder, and Raush, 1968; Olson and Ryder, 1970; Raush, Goodrich, and Campbell, 1963; Ryder, 1966; Ryder 1970a; Ryder, 1970b; Ryder and Goodrich, 1966). Much of this research has been descriptions of their data collecting, their techniques for eliciting interaction, and descriptions of the types of newlywed marriage that they have found. As part of this larger project, Barry (1968) studied among other things the expression of dominant behavior in a marital relationship. Barry had 48 white, middle-class newlywed couples role-play in four different conflict situations. The inter- action was tape recorded, transcribed, and coded into 36 categories, which was narrowed down to six categories (cognitive acts, resolving the conflict acts, reconciling the relationship acts, appeals, and coercion or personal attack) for the actual data analysis. Barry presents evidence that the coding system was highly reliable, even after a year's lapse between codings. Six-by-six contingency matrices were computed for each individual, and each couple, for each scene and all scenes combined. These contingency matrices tallied the frequencies with which each category was used in response to each category. Barry found that cognitive acts followed cognitive acts; resolving acts followed resolving acts; reconciling acts followed reconciling acts; appeals, and coercion or personal attack followed 25 rejection; rejection, and coercion or personal attack followed coercion or personal attack. Barry also found that husbands used more resolving and reconciling acts, while wives used more coercion or personal attacks, and more appeals, which he related to previous research which shows that middle-class wives are more dependent and anxious, less self-confident, self-sufficient, and self-accepting. Barry hypothesized that those factors in wives might lead them to be more concerned with winning in conflicts, especially at the newlywed stage, which also is probably a more difficult transitional stage for them than their husbands. Independent work at N.I.M.H. on all data except Barry's yielded four bi-polar factors and scores on each factor for each couple (Ryder, 1970a). Barry put extreme scores (one standard deviation or more from the mean) on each factor into factor groups and combined the contingency matrices of the individuals of each group. The factor III+ group were six couples who reported trouble with the marriage, and frequent disagreements and whose marriage was negatively evaluated by the interviewers. Wives in this group reported difficulties with their families of origin. Three of the six couples were divorced as of 1968 (the only three marital breakups in the whole sample as of that time.) The most noteworthy thing about the contingency matrices of these couples was the way the husbands exceeded all husbands and their own wives (thus reversing the general sex difference found) in the proportionate use of coercion or personal attack. These husbands were very punitive, especially in response to a coercion or personal attack. Their wives, on the other hand, did not differ much from all wives in 26 their matrices. In addition, these couples averaged significantly more acts than did all couples, thus indicating a prepensity to extended conflicts. Barry related this finding to research which indicates that husband characteristics are more crucial to marital success (than wives' characteristics.) Seven couples comprised the group at the opposite end of factor III. These couples were characterized by the factor scores as show- ing great mutual empathy and support, an easy decision process, and a couple rather than individual identification. These COUp1eS were found to be polar opposites to the factor III+ group in Barry's interaction situations. Both spouses avoided the relationship-threatening categories (rejection and coercion or personal attack) and were high on resolving and reconciling acts, with both of these things being much more true of husbands than wives. Moreover, they averaged less acts than the average of all couples, thus corroborating the ratings from other data of easy decision process. Barry found that the eight couples who comprised the factor II- (role-sharing couples) had just as easy a decision process as the III- couples, but, unlike the III- couples. they engaged in heated exchanges which were almost required by Barry's instructions. Goodrich (1968) speculated that the 11- couples in contrast to the III- couples might have the greater degree of marital harmony and stability, because of their ability to express hostility constructively. Barry's research is significant for a number of reasons: First of all, he studied (along with others in the N.I.M.H. project), a 27 sample of normal couples, instead of comparing couples of differing pathologies. As such, for the first time, normative data on married couple interaction is becoming available. Secondly, through this normative interactional data, beginning attempts at classifying couples through their interaction are being made. With respect to specific findings, it is interesting that Barry's study indicates that dominance is associated with dominance, at least with respect to married couples (in Barry's language coercion or personal attack is associated with coercion or personal attack). Also that when the husband acts in highly dominant (coercive) ways, it is associated with relationship difficulties and even divorce, and in fact appears to be the main factor associated with such difficulties in this small sample of relative newlyweds. Conclusions about marital dominance. Although there has been little research on marital dominance, that available, suggests that dominance is a major variable which is strongly related to the stability and happiness of marriages. Although it seems clear that dominance is associated with dominance and not submissiveness (in contrast to Leary's formulation) on the following act (Cancian, 1971; Barry, 1968), and with relationship difficulties in the long run, there is no information available as to how different amounts or degrees of dominance affect the quality of the subsequent interaction beyond the act immediately following the highly dominant act, or for how long the quality of the subsequent interaction is altered. The present study 28 is an attempt to provide some answers to these questions, by experiment- ally inducing different degrees of dominanting behavior in married couples, and determining how this affects the quality of the sub- sequent interaction. Although the familial research on dominance has widely differing results, there is some suggestion that moderate amounts of dominance are associated with normal or well-adjusted offspring behavior. For this reason, moderate, as well as high and low dominance levels were included in the present study. Methodological Considerations The present study takes into account some of the methodological considerations which have been mentioned in the preceding review, as well as some which have not yet been mentioned. As it was suggested that the group comparison method has been over-used, the present study randomly sampled couples in married housing at Michigan State University. A "normal" sample was thus obtained, although the sample was drawn from a highly restricted population. The present study uses a Revealed Differences Technique (RDT), which has been employed in a large number of interaction studies to elicit interaction. The RDT appears to have been successful in generating involved participation from families and couples, but it must be borne in mind that the setting in which the interaction is obtained, the effect of the observer, and the nature of the inter- actional task itself, have all been demonstrated to have a significant effect on the interaction sample. O'Rourke (1963) found that in the 29 lab context, there was more disagreement between spouses, more activity, less efficiency at decision making, and less emotionality registered than in the respondents' home. Harris (1969) found that the known presence of an observer in the home affected family interaction rates. Kenkel (1961) found that the sex of the observer greatly influenced the decision making process and affected the observed power structure. When the 9_was a woman, wives tended to take a more powerful and active role in the decision making. Barry (1968), Swain (1969), Hofman (19691, and Smith (1970) have found that the nature of the interaction task has a significant effect upon the interaction obtained. Even after a sample of marital interaction is obtained, a decision has to be made on what to attend to in that sample of interaction. At the present time, different investigators have developed a wide variety of interactional measures, and there seems to be little agreement (and little knowledge) on the utility of these different rating systems. What is perhaps needed sometime in the near future, is an omnibus study, where a large number of these rating systems would be applied to a given sample of couples. Basically, there seems to be three major types of interaction analysis reported in the literature. One is a logical analysis, which generally focuses on distortions and inconsistencies in logical process, such as if a person says one thing, and then disqualifies it shortly after. Sojit (1969 & 1971) and Singer and Wynne (1965a & 1965b) are examples of researchers who have used this type of interactional analysis. A second general type could be called a 3O structural analysis, which deals with variables such as spontaneous agreement, choice fulfillment, who wins the conflict, the number of interruptions, the amount of time a person spends talking, who speaks to whom. Winter and Ferreira (1969), Farina (1960), and Farina and Holzberg (1968) are examples of researchers who have used this type of analysis. The third type of analysis pays attention to the type of relationship implied in a communication. The relationship analysis has been infrequently used (Crowder, 1972; Cancian, 1971; Barry, 1968; MacKenzie, 1968; Tinker, 1967), although it seems to give a wealth of interpersonal data, and has produced highly interesting results. For example, MacKenzie (1968) found that clinic fathers primarily responded to their wives' hostile dominant behavior with passive hostility and next most often with dominant hostility. Clinic fathers' passive hostility and dominant hostility most frequently elicited active hostility from their wives. When clinic fathers did send dominant friendly behavior, clinic mothers were most likely to respond with dominant hostility. In contrast, nonclinic fathers sending dominant friendly communications elicited friendly dominant or friendly passive behavior from their wives. She also found that both nonclinic and clinic mothers manifested dominant behavior with fathers and sons, but clinic mothers were more dominant than the nonclinic mothers. Also clinic mothers were aggressively dominant while nonclinic mothers were friendly dominant. Tinker (1967) found that by attending to the relationship aspects of communication, families could be ranked accurately from most to least disturbed in their communication patterns, 31 on the basis of only two to three minutes of interaction. Crowder (1972) used Leary's system to study therapist-client interactions, with reliable and interesting results. The present study uses a relationship analysis of communication, because of the apparently productive and relevant type of information it provides. The rating system in the present study has been simplified as much as possible, as previous research with relationship analyses has generally started with overly complex rating systems, which had to be simplified to raise interrater reliability, and to provide interpret- able results. The need for causal research, instead of just correlational research, prompted the analysis of variance design of the present study. METHOD Subjects The subjects were 24 married couples who were living in Michigan State University married housing two-bedroom apartments. The fact that they were living in two-bedroom apartments indicated that they had at least one child, and thus were not newlywed couples, but persons who had likely spent at least the better part of a year establishing a marital relationship. Five married housing units were randomly selected from all two-bedroom units, and letters sent to all couples living in each of the five buildings. A copy of the letter requesting participation from each couple is included as Appendix A. Within a few days after receiving the letter, each couple was phoned and encouraged to participate in the study. If the couple agreed, an appointment was arranged. Each couple was paid $15.00 for participating. In all, 47 letters were sent, with seven couples being dropped from the sample for the following reasons: could not be contacted (2), foreign students (2), divorced or separated (2), married less than one year (1). Twenty-six couples out of the remaining 40 couples participated, a 65% participation rate. Of the 14 couples who did not participate, the following were the reasons given: moving away within two weeks (eight couples), unwilling (six couples). Since the letters requesting participation were sent out just prior to the end of summer term, it is 32 33 possible that eight couples were actually moving. 0f the 26 couples who participated, two were used to train raters. The data from these two couples were not included in the present study, leaving 24 couples who were the actual subjects. With respect to these 24 couples, the average age of the husband was 27; the average age of the wife was 26; the average length of marriage was five years; the average number of children was 1.5, the average number of years of education for the husband was 17.6; for the wife, 14.5. Ninety-six per cent of the husbands were part-time or full-time students, while 75% of the wives were working and were not students. When the couples were randomly assigned to four different groups for the purposes of the study, analyses of variance indicated that there were no significiant differences between the groups on any of the variables contained in the couple correlation matrix (see Table 12). Procedure Only one couple at a time participated in the experimental procedure. This consisted of having each spouse individually complete an opinionnaire, a marital adjustment scale, and having them engage in three discussions which were tape recorded. After the COUp1e entered the discussion room, §_explained the presence of the tape recorder, and the order of the experimental procedure. After the initial orientation, §_asked the couple to complete an opinionnaire, with the husband completing it in one room, the wife in another. The exact instructions, read to each COUple, are given in Appendix B. The opinionnaire consisted of nine lists (see Appendix C). 34 Each list required four Opinions to be given, in order of declining preference. Also, after the spouse had completed the nine lists, he ("he" is used throughout the present section without respect to sex) was required to rank them in order of declining importance to him. Part of these rankings were later used to determine which tapics the couple would discuss in each discussion period. When the opinionnaire had been completed, §_randomly assigned the couple to one of four groups, using a table of random numbers: High Dominance (HD), Medium Dominance (MD), Low Dominance (L0), or Control (C). Six couples were assigned to each group. §_also selected one spouse from each couple. This "Selected Spouse" would later receive special instructions, if in one of the three experimental groups. .E used the topic rankings of the Selected Spouse to determine which topics would be discussed in each of the three discussion periods. The Selected Spouse was randomly selected, again using a table of random numbers, but within the restrictions that three were the husbands and three were the wives, within each of the four groups. After each spouse from the couple had completed the nine lists, but was still in a separate room from his spouse, §_told each of them that the next task would be to compose a joint list. This would be accomplished through discussion with his spouse, on three of the tepics they had just completed. §_then brought the couple into the room with the tape recorder, started it, and advised the couple he would be in an adjoining room. This first discussion is called Discussion I. When the couple indicated they had completed discussing the three joint 35 topics by calling E, he asked them to fill out the Family Concept Inventory (FCI), a measure of marital adjustment, again in separate rooms. The FCI is presented in Appendix D. When each had completed the FCI, E asked the partners to compose joint lists through discussion with each other on three additional topics from the nine lists they had completed previously. Again these instructions were given to each spouse while he was still in the separate room in which he had filled out the FCI. Unknown to his partner, the Selected Spouse was given additional instructions while still in the separate room, for each of the couples that had been assigned to one of the three dominance groups. §_asked each Selected Spouse from the HD Group to do his "utmost" to get all four of his individual choices to be the ones on the joint list, for each of the three topics he was to discuss. He asked each Selected Spouse from the MD Group to do his "utmost" to get two of his four individual choices to be the first two on the joint list, for each of the three topics the couple was to discuss. With respect to the LD Group, §_asked each Selected Spouse to let his spouse determine the first two most important choices on the joint list for each of the three topics to be discussed. Couples in the C Group received no additional instructions. After both spouses had completed the FCI and received their particular instructions, E_brought them together in the room with the tape recorder, where they interacted to reach agreement on three lists (Discussion II). E_then returned, and explained what the procedure had been on the previous discussion period. He then asked the couple to 36 reach agreement without any special instructions on the final three tepics (Discussion III). §_asked couples in the C Group merely to interact as before to produce three more joint lists. When the couple had completed Discussion III, §_returned to the room, discussed the reactions of the couple to the experimental procedure, answered their questions, and explained to them the full extent of the research. He asked them not to talk to any of their neighbors about any part or aspect of the research, for if another COUp1e had prior knowledge of the research, he would not be able to use them as subjects. Each couple was also asked if they had heard anything about the research from their neighbors who had already part- icipated, and if so, what (they were assured they would be paid no matter what their answers were). Because no couple reported hearing any significant information from their neighbors, none was eliminated. In summary, each couple went through the following sequence of events: 1. The marital partners filled out the opinionnaire, each spouse in a separate room. Then the couple was randomly assigned to one of four groups (HD, M0, L0, or C Group). 2. The couple interacted in Discussion 1. 3. The marital partners filled out the FCI, each in a separate room. The Selected Spouse was given special instructions for Discussion II, if the couple was in one of the three experimental conditions. 4. The couple interracted in Discussion II. 5. The couple interacted in Discussion III. 37 The Opinionnaire There were nine lists of four opinions or preferences that each spouse completed on the opinionnaire (see Appendix C). Each spouse listed his opinions in order of declining preference, on each of the nine topics. When he had completed the nine lists, each spouse ranked the t0pics from one to nine, in order of declining importance to him. The rankings of the Selected Spouse was used to equalize the intensity of the topics discussed from one discussion period to another. For example, in Discussion I, a particular couple in the HD group would discuss the topics that the Selected Spouse had ranked one, five, and nine; in Discussion II, topics ranked two, six, and seven; and in Discussion III, topics ranked three, four, and eight. Each of these rank sums totals 15. Table I indicates how the t0pic rankings were distributed within groups and across Discussion periods. Within the dictates of this experimental design, E randomly assigned the couples to each of the different orders within each grOUp. Pilot research with seven couples had indicated that the couples became involved in discussions on these nine topics, and found them interesting. The Marital Adjustment Measure The Family Concept Inventory (FCI) was chosen as a quickly administered marital adjustment test, which has reasonable evidence of its reliability and validity. It is a 48 item, five choice (strongly agree-strongly disagree) marital adjustment measure based on Q-sort developed by van der Veen, Huebner, Jorgens, and Neja (1964), who reported significant mean differences between high and low marital 38 Table 1 Rankings by the Selected Spouse of Topics Discussed within each Discussion Period Discussion Period Group I II III High Dominance N = 2 159 267 348 N'= 2 267 348 159 E = 2 348 159 267 Medium Dominance N = 2 267 348 159 N'= 2 348 159 267 E = 2 159 267 348 Low Dominance N = 2 348 159 267 N'= 2 159 267 348 E = 2 267 348 159 Control N = 2 159 267 348 N'= 2 267 348 159 1_1 = 2 348 159 267 39 adjustment groups, which was further substantiated with another low adjustment group (van der Veen, 1965). Van der Veen et a1. (1964, 1965) reported median test-retest reliabilities of .71 and .80. Hofman (1966) administered both the O-sort and a true-false version of the test to a sample of 50 and reported a correlation of .72 between the two forms and an internal consistency measure of .84 for the true-false version. Palonen (1966) used the five-choice, 48 item form and reported a split- half reliability of .85. The FCI correlated .73 with the Locke-Wallace Marital Adjustment Test (Palonen, 1966) using 40 couples, while the Family Concept True-False version correlated .76 with the Locke-Wallace Test on a sample of 25 COUples (Hofman, 1966), and the Family Concept Q-sort correlated .68 (Hofman, 1966) and .67 (van der Veen et al., 1964) with the Locke-Wallace. Hofman (1969) also found that on a sample of 15 clinic and 15 nonclinic couples, that the FCI discriminated between the two groups better than the Locke-Wallace. Influence Scores The Influence Scores were computed by comparing the couple's joint list with one spouse's individual list, and scoring one to four points for each choice on the individual list which appeared on the joint list. For each of the nine topics discussed, four, three, two, or one points were scored if the spouse's individual choices were ranked first, second, third, or fourth, respectively on the joint list. TWo Influence Scores were computed for each couple, one for the Selected Spouse and one for the Non-selected Spouse. 40 The Dominance Index A Dominance Index was also computed for each couple by subtracting the Non-selected Spouse's Influence Score from the Selected Spouse's Influence Score. This gave an Index of Dominance which was based solely on the difference between the two scores. A negative score indicated that the Selected Spouse was the less dominant. Evaluation of Ability to Carry Out Instructions Two measures were used to evaluate the Selected Spouse's ability to follow instructions: an analysis of variance for the Influence Scores obtained by the Selected Spouse, and an analysis of variance for the Dominance Index. The Scheffe method (Hays, 1963) was used for post-hoc comparisons. Both Influence Scores and an Index of Dominance were used because of the different information each provided. The Influence Scores were used to indicate how closely the Selected Spouse was able to come to an ideal score, or range of scores on Discussion II. For the HD group, 30 (4+3+2+l = 10 points X 3 topics = 30 points) was the ideal score, as the Selected Spouse was asked to try to get all of his individual choices to be the ones on the joint list. For the MD group, 21-30 points was the ideal as the Selected Spouse was instructed to try to get two of his individual choices to be the first two on the joint list (4+3 = 7 points X 3 topics = 21 points). Zero to nine points was the ideal range for the LD group, as the Selected Spouse was asked to allow the other spouse to determine the first two choices on the joint list (1+2 = 3 points X 3 t0pics = 9 points). The Index of Dominance, on 41 the other hand, indicated how much the couple's dominance pattern was changed by the Selected Spouse's responding to the instructions in Discussion II. For example, in the HD group, if the less dominant spouse (as determined by the Influence Score in Discussion I) were the Selected Spouse, and managed to become much more dominant in Discussion II, to cut down on the other's dominance, but not come close to 30 points, this increase in relative dominance would be reflected in the Dominance Index scores. Evaluation 9f_the Nature 9f_the Interaction In order to evaluate the nature of each couple's interaction, a three-minute segment was selected from the middle of each of the three discussion periods. Thus, a total of nine minutes of interaction was rated for each couple. This small amount of interaction was judged sufficient, both from the pilot research, and from previous evidence (Tinker, 1967), that small amounts (two to four minutes) of interaction which have been minutely analyzed furnish sufficient information for accurate global judgments. Each "speech" was rated as to whether it was high, medium or low in four categories: Negative-hostile, Positive-affectionate, Dominant, and Submissive. Terrill and Terrill's (1965) definition of a "Speech" was followed: "A scorable speech consists of a relatively continuous utterance by an individual which is either uninterrupted, or if interrupted, apparently uninfluenced by the interruption. In this definition, the term 'relatively continuous' refers to situations where a person begins a 42 sentence of a train of thought, is briefly interrupted, but continues with the same sentence of thought as if he had not heard the interruption (p. 264). Also, if the speaker changed topics without pausing, a separate speech was scored each time the topic was changed, because changing topics often implied a change in the relationship. The criteria for rating the speeches are listed below. Some categories have several statements, only one of which is necessary for a speech to be rated in that category. Negative-hostile High: Angry shouting. Strong condemnation of other. Disparage- ment. Disagreement with negative vocal inflection. Medium: Disagreement without negative vocal inflection. Negative tone in voice without disagreement. Low: No negative tone. Can't tell. Positive-affectionate High: Direct positive comment. Friendly laughing. Agreement with positive vocal inflection. Medium: Agreement without positive vocal inflection. Affectionate tone, not related to agreement. Low: No positive tone. Can't tell. Dominant High: Firm, spirited vocal emphasis on own point of view. No concession. Demanding other person change. Medium: Expression of own point of view, respecting other. Asks question. 43 Low: Clarifies the other's point of view. No dominance expressed. Can't tell. Submissive High: Acquiesces to other (a terminal act). Medium: Yes, but... Answers question. Asks clarification from other. Disparages own view in favor of other's. Low: No submissiveness expressed. Can't tell. The scale used in the present research are modifications of the scales develOped by Terrill and Terrill (1965). Their scales were modified on the basis of previous research (Tinker, 1967), and on the basis of pilot work for the present study. The present scales differ from Terrill and Terrill's scales, in three important respects: Firstly, they are greatly simplified. Secondly, they contain a measure of intensity. Thirdly, they allow speeches to be rated independently with respect to the four main categories. The present scales permit a given speech to be rated as both medium Hostile and medium Affectionate, or as both medium Dominant and Submissive, for example. This feature helped to deal with some of the conflicting elements within a speech. For example, if a spouse answered his partner's question, at the same time expressing his own opinion, should that be rated dominant, because he was expressing his own opinion, or submissive, because he was answering his partner's question? The present scales made it possible to rate it as both medium Dominant and medium Submissive. In pilot research, two raters independently rated seven couples' interaction on the above scales (these couples were not used in the results of the present study). A total of 23 minutes of interaction was 44 rated, approximately three to four minutes of interaction per couple. Using a speech-by-speech comparison of ratings for establishing interrater agreement, it was found that the two raters had an overall agreement rate of 83%. If one rater rated speech number one as high in Dominance, high in Hostility, medium in Submissiveness, and low in Affection, and the other rater rated the speech the same, except medium in Affection, it would be scored as three agreements and one disagreement. It can be seen that the speech-by-speech comparison is a considerably more stringent measure of establishing interrater reliability, than the method of comparing how many speeches each rater scored high Dominant, medium Dominant, and so on. With the present method of determining interrater reliability, the overall agreement rate of 83% seemed to be an acceptable level of agreement. It was much superior to the 40% average agreement previously reported with Terrill and Terrill's scales (Tinker, 1967). Only one main rater E was used in the present research, with a second rater rating every sixth couple as a reliability check on the main rater. The main rater scored the tapes blindly. He had not previously heard the taped discussions, and did not know to which experimental group the couple had been assigned. Each couple was identified by a randomly assigned letter of the alphabet. §_rated the couples in alphabetical order. The rated speeches were compared statistically by the use of analyses of variance. Analysis of covariance would have been possible if the particular dependent variable had been linearly correlated with marital adjustment scores on the FCI. In the present study no significant correlations appeared (see Tables 11 and 12), with analyses of variance 45 consequently being the statistic employed. The form of the analyses of variance was that of a three-factor analysis with repeated measures, with sex nested within each group (Winer, 1962). The Scheffe method for post-hoc comparisons was used for individual comparisons. Twelve separate analyses concerned the number of speeches labeled: (1) high Negative-hostile, (2) high Positive-affectionate, (3) high Dominant, (4) high Submissive, (5) medium Negative-hostile, (6) medium Positive- affectionate, (7) medium Dominant, (8) medium Submissive, (9) total (high plus medium) number of speeches labeled Negative-hostile, (10) total number of speeches labeled Positive-affectionate, (ll) Dominant, (12) Submissive. The I'1ow" categories were of the wastebasket variety, used when the speech could not definitely be rated high or medium, so no analyses of variance were made of that data. Table 2 indicates the relative usage of the different categories. Evaluation 9f_the C1arity_gf_the Interaction Each of the three-minute periods of interaction, which had been selected from each of the three discussion periods, was given a global rating on a four part clarity scale: 1. All speeches are clear in meaning. 2. Most speeches are clear in meaning. 3. Some speeches are clear in meaning, most are not. 4. All speeches are unclear in meaning. The statistical analysis here was a two-way analysis of variance. Again, the Scheffe method was used for individual comparisons where appropriate. 46 TABLE 2 Relative Usage of the Speech Categories Category Low Medium High Total Negative-Hostile 1429 426 49 1904 Positive-Affectionate 1466 411 27 1904L— Submissive 1202 537 165 1904 Dominant 45 1668 191 1904 Total 4142 3042 432 7615‘ + 7616 ratings were made on 1904 speeches (4 ratings for each speech) 47 In the pilot research, the two raters made these ratings on seven different couples for a total of 23 minutes. TWenty-three different ratings of clarity were made by each rater. The raters were in exact agreement 83% of the time, and on 17% of the ratings they had one-step errors. Since this degree of interrater agreement was comparable to that for the interaction categories, it was decided again to use one main rater in the actual research, with a second rater checking reliability by rating every sixth couple on the clarity categories. Evaluation 9j_the Ability tg_Reach Agreement The length of time in minutes that was required for a couple to reach agreement on a joint list was used as the measure of ability to reach agreement. Again, a two-way analysis of variance was the statistic used, with the Scheffe method used for post-hoe comparisons. The Correlational Analyses The data from Discussion I, prior to any instruction as to how the couples should interact, are presented in two intercorrelational matrices. One matrix was based on individual data, and one from couple data. It could be determined whether there were similarities between the significant relationships from the experimental manipulations, and the significant relationships from the correlations. It could be argued that the experimental manipulations created an artificial situation, in that forcing one spouse to be highly dominant would have negative effects on the couple's communication, which would not be found in a couple where one spouse was naturally highly dominant, or that it was the disruption of normal communication patterns which produced the significant findings. An intercorrelation matrix showing for example, what variables were 48 correlated with the number of high dominant speeches would help resolve this problem. If there were correspondences between the analyses of variance and the correlations, it would buttress the experimental findings, indicating that they were not due to an unnatural situation. If there were no correspondence between the analysis of variance and the correlations, it would call into question the validity of the experimental findings. TWo intercorrelation matrices were completed in order to be able to compare the effects of individual dominance with dominance expressed bilaterally by both members of the COUple. No correlational analysis was completed on Discussions II or III, because of the likelihood that the special instructions (in Discussion II), or residual effects from the instructions (in Discussion III), distorted the interaction patterns of the COUp1es. The information sought from the correlations pertained to whether the same relation- ships would appear in the correlations when the couples were inter- acting "naturally" in Discussion I, as appeared through the experimental manipulations. Individual correlations. The following variables were included in the individual correlation matrix: 1. Number of speeches labeled high Negative-hostile (HNH) Number of speeches labeled medium Negative-hostile (MNH) Total number of speeches labeled Negative-hostile (TNH) Number of speeches labeled high Positive-affectionate (HPA) Number of speeches labeled medium Positive-affectionate (MPA) (”01wa Total number of speeches labeled Positive-affectionate (TPA) 7. 8. 9. 10. 11. 12. 13. 14. 15. 49 Number of speeches labeled high Dominant (HD) Number of speeches labeled medium Dominant (MD) Total number of speeches labeled Dominant (TD) Number of speeches labeled high Submissive (HS) Number of speeches labeled medium Submissive (MS) Total number of speeches labeled Submissive (TS) Sex Age Education (Ed) Couple Correlations. In addition to variables 1-12 of the individual correlation matrix, the couple correlation matrix included the following: 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Average Age (Age) Average education (Ed) Education discrepancy between husband and wife (EdD) Number of children (# Chdrn) Child Density (CD) .Degree of Dominance (oDom) Husband (+) or wife (-) dominance (H or W Dom) Spontaneous agreement (SA) Clarity Time Choice of fulfillment (CF) FCI total It should be kept in mind that variables 1-12 on the couple correlation matrix are total number of speeches for the couple, and not the individual RESULTS Interrater Reliability E rated the tapes, with a second rater rating every sixth couple as a reliability check. Both raters thus rated the interaction of four couples. They rated one, three-minute segment of interaction taken from the middle of each of the three discussion periods, for each couple. In this way, 905 ratings of speeches were independently made by both raters, with agreement on 806 ratings, and disagreement on 99 ratings. This is an 89.1% agreement rate, comparable to the 83% agreement rate obtained in pilot research. It suggests that the couples continued to be rated in a reliable fashion throughout the rating process. Table 3 which indicates the results of the rating process for each of the four couples, provides further evidence that the rating process remained stable and reliable, despite the stringency of the speech-by-speech method of comparison. No decline in reliability is evident over the ratings of the four couples. With respect to the interrater reliability on ratings of the clarity of interaction, each rater made one rating for each segment of interaction. As there were three segments of interaction for each of the four COUples, a total of twelve clarity ratings were made by both raters. The raters agreed on eleven of the twelve ratings, for 91.7% agreement. Only one disagreement, a one-step discrepancy, occurred on this four category scale. Evaluation pf_Abi1ity 59 Carry Out Instructions Influence Scores pf_the Selected Spouse The mean Influence Scores for the Selected Spouses are presented in Table 4. Significant differences between means are designated by letter 50 51 Table 3 Interrater Reliability for Four Couples Number of Ratings % of Ratings Couple Agree Disagree Total % Agree % Disagree F 253 27 280 90.4 9.6 L 212 33 245 86.5 13.5 R 198 18 216 91.7 8.3 X 143 21 164 87.2 12.8 Tbtal 806 99 905 89.1 10.9 52 superscripts. For example, the two means with an "a" superscript are significantly different from each other, the two numbers with a "b" superscript are significantly different, and so on. The analysis of variance and post-hoc comparisons indicate that the HD and MD groups are significantly higher than the LD group on the Influence Scores for Discussion II. The Selected Spouses in the HD and MD groups increased their Influence Scores (but not significantly over the C group) and the Selected Spouses in the LD group decreased their Influence Scores (but not significantly under the Control group), although sufficiently that the HD and MO groups differed significantly from the LD group. Examination of Table 4 reveals that the Selected Spouses in the HD group were unable to score at 30 on Discussion II as desired, and the Selected Spouses in the LD group were unable to score close to the nine point mark or less as desired. However, the Selected Spouses in the MD group did obtain a score in the desired range between 21 and 30 points. Dominance Index Scores The Dominance Index scores, presented in Table 5, indicate that the Selected Spouses were partially able to carry out the instruction. In fact, the MD group seems to have been completely able to carry out its instructions, in that its Dominance Index scores were significantly greater than those of the C and LD groups for Discussion II. Further, the MD Selected Spouses significantly increased their Dominance Index scores from the first to the second discussion period. Also, in Discussion II, the HD and LD groups were significantly different from 53 TABLE 4 Mean Influence Scores of the Selected Spouses Discussion Period GrOUp I II III HD 16.17 21.67"’c 14.72‘c MD ' 16.67 22.00b 17.17 LD 17.50 13.17""b 14.50 c 17.33 15.83 18.83 a-a p;<:.05 b-b p;<:.05 C-C 54 TABLE 5 Dominance Index Scores (Totals) Discussion Period Group I II III HD -7 44a -12 MD -9d 52b’c’d 5 ,e.f LD 116 -6Ba’b -9f 0 -10 -17C -8 a-a p_<:.05 b-b p_4(.05 C-C p_ (.05 d-d p_<:.05 e-e p_<(.05 f-f 55 each other, but not from the C group, indicating that the Selected Spouse had been influential in the expected direction, but not strongly enough to obtain significance over the C group. However, in Discussion II, the Selected Spouses in the LD group significantly reduced their Dominance Index scores under the first and third discussion periods. This finding suggests that the LD group had moderate success in carrying out their instructions on Discussion II. In general, the Influence Scores and the Dominance Index scores indicated that the eXperimental grOUps were at least partially successful in carrying out the instructions for Discussion II. Presumably, if they had been completely successful, greater differences would have been found on the dependent variables. Therefore, differences manifested between the groups on the dependent variables, were observed under less than optimal conditions. The limited impact of these instructions may also have obscured other differences. Evaluation of the Effects of Different Amounts of Dominance upon the Dependent Variables The effect of the independent variable (amount of dominance: high, medium, and low) was measured upon six main dependent variables: 1. Number of speeches labeled Negative-Hostile (N-H) Number of speeches labeled Dominant (D) . Number of speeches labeled Submissive (S) 2 3 4. Number of speeches labeled Positive-Affectionate (P-A) 5 The clarity of the interaction 6 . The amount of time in minutes needed to reach agreement 56 Within the category N-H, three separate analyses of variance were completed: (1) the number of speeches labeled high N-H, (2) the number of speeches labeled medium N-H, and (3) the total number of speeches labeled N-H, obtained by summing the high and medium categories. As mentioned previously, no analyses were completed for speeches classified low (all speeches which could not be rated high or medium). The same procedure was followed for the number of speeches labeled Dominant, Submissive, and Positive-Affectionate, with three separate analyses of variance completed for each main category. The summary tables for all analyses of variance are presented in Appendix E. The design of the analyses for the speech ratings, was a three-factor analysis with repeated measures (four groups X three discussions X two sexes), with sex nested within each group. Two-factor analyses of variance were used (group X discussion) for the clarity ratings, and for the Time Needed to Reach Agreement. The Scheffe method for post-hoe comparisons was used in all cases for individual comparisons. Speeches Labeled Negative—Hostile There were significant differences among the four groups, with respect to speeches labeled medium N-H, as indicated in Table 6. The HD group had significantly more of these speeches than the LD group in Discussion II. This finding indicates that the HD condition increased the number of medium N-H speeches (but not significantly over the C group), and the LD condition decreased the number of these speeches (but not significantly under the C group), but enough that the HD and LD groups differed significantly from each other. It was also found that the HD group decreased in the number of medium N-H TABLE 6 Total Number of Speeches Labeled Medium N-H Discussion Period A. ran Group I II III Total HD 47 589'b 27b 132c MD 37 27 21 85 L0 33 15a 30 78C 0 s1 28 52 131 a-a p_‘<.05 b-b p_‘:.05 C-C 58 speeches, from Discussion II to Discussion III. Further, the HD group had more of these speeches than the LD group. While this finding is statistically significant, it is not really meaningful as the Group X Discussion interaction revealed that all the significant differences among the groups occurred on the second discussion period. This overall difference among the groups indicates only that the differences on Discussion II were large enough to retain significance across all three discussion periods. With respect to the total number of speeches labeled N—H, again, the interaction of Group X Discussion was significant, with the HD group having significantly more total N-H speeches than the MD and LD groups in Dicusssion II. It is interesting that even though the N-H speeches were significantly increased by the HD group (over the MD and LD groups, but not over the C group), there was not increased dominance over the MD group as measured by the Influence Scores or Dominance Index, or by the number of dominant speeches (as will be shown later). This suggests that although the HD Selected Spouses may have tried to be highly dominant, their efforts did not result in greater dominance, but rather in increased hostility in their inter- action. It also suggests that a spouse cannot be increasingly dominant unless the other spouse allows it; otherwise it will result in increased friction, but not increased dominance. This interpretation is somewhat further supported by the increase in dominance of the MD group over the C group (as measured by the Dominance Index scores), while the MD group did not increase the number of N-H speeches. It 59 TABLE 7 Total Number of Speeches Labeled N-H (High plus Medium) Discussion Period Ti— Group I II III IGrand Total HD 50 683’b’c 29C 147 M0 42 29a 22d 93 L0 34 16b 31 81e di 6 57 33 64 1548 a-a p3¢:.05 b-b p<.05 c-c ‘: p. .05 d-d p_<.05 e-e pg<:.05 60 appeared that a spouse of either sex would allow moderate dominance by the other spouse, without an increment in hostile interaction, but that higher amounts of spouse dominance generated increased hostility rather than more dominance. Further attention will be given this issue in the Discussion section. In the Group X Discussion Interaction, it was also found that the HD group decreased significantly in total N-H speeches from the second to the third discussion period suggesting that the increased hostility was rather transitory. Further, the C group had significantly more total N-H speeches in Discussion III than the MD grOUp. This finding does not seem interpretable. Finally, the C group had significantly more total N-H speeches than the LD group, probably because of the low initial incidence of such speeches in the LD group, plus the Discussion II decline in these speeches within the LD group. On speeches labeled high N-H, there were no significant F ratios, indicating that the amount of dominance had no effect on the number of high N-H speeches. Speeches Labeled Dominant There were some significant f_ratios on the total number of Dominant speeches, as indicated in Table 3 . First of all, the Group X Discussion interaction was significant, with the C group having significantly more total dominant speeches than the MD group and the LD group in Discussion II. This finding is despite the fact that the MD group actually became more dominant in Discussion II (i.e. more dominant than the C or L0 groups, and more dominant than itself in Discussion I, as measured by the Dominance Index scores). It poses an 61 TABLE 8 Total Number of Speeches Labeled Dominant Discussion Period Group I II III Grand Total HD 172 148 155 475 _f MD 162 128a 129 419 L0 163 134b 166 463 c 165 1798’ 158 502 Total 1859 a-a b-b p < .05 62 interesting and perplexing problem: If the number of speeches labeled dominant does not reflect the actual amount of dominance which was occurring in the interaction, what interactional measure would? And why doesn't the number of speeches labeled dominant reflect the actual amount of dominance going on? These questions will be dealt with in the Discussion. Also with respect to total dominant speeches, there was a significant Group X Sex Interaction in which C group males had more of these speeches than MD group females. This does not seem to be meaningful. For speeches labeled high Dominant and speeches labeled medium Dominant, there were no significant £_ratios, indicating that the amount of dominance did not affect the number of speeches that were labeled high Dominant or medium Dominant. Speeches Labeled Submissive While there were a number of findings that were statistically significant, generally the number of speeches labeled submissive was not affected by the independent variable. With respect to speeches labeled high Submissive, the wives had more speeches labeled high Submissive than the husbands, which corresponds to a cultural stereotype. While the Group X Discussion Interaction was significant, it was not meaningful: The MD group in the first discussion period had more high Submissive speeches than the C group in the third discussion period. On speeches labeled medium Submissive, the number significantly declined from Discussion I to Discussion II, and while there was a further decline from the second to the third period, it was not 63 significant. At any rate, it was apparent that the groups were less submissive on Discussion II and III than they were on Discussion I. Also the groups differed across the three discussion periods on these speeches with the C group having significantly more of them than the HD group. Examination of Table 9 indicates that with respect to the total number of speeches labeled Submissive, the C group significantly exceeded the HD group. This finding does not appear to be a result of the experimental manipulations, as the C group initially started off at a higher level. For example, the C wives made more total Submissive speeches than the HD wives in the first discussion period (not shown on Table 9). The wives i__tptp_made more submissive speeches than the husbands. Again this fits in with commonly accepted cultural norms that wives are more submissive, and supports the validity of the submissive classification. Further, the groups had less Submissive speeches on Discussion III than on Discussion I. While there was a general decline from Discussions I to II to III, only the I-III difference was significant. The groups generally became less submissive the longer they interacted, which could suggest a facade which wore off, or increasing fatigue as the experiment wore on. There was also a three- way interaction, which indicated that on Discussion I, the C wives gave more submissive speeches than the HD wives. The C and HD husbands differed in the same direction on Discussion I, but not significantly so. Further the C wives had significantly less submissive speeches on Discussion III than on Discussions I and II. 64 TABLE 9 Total Number of Speeches Labeled Submissive Sex Group Husband Wife Total HD 70 71 ITDTTT MD 84 94 178 L0 68 95 163 c 91 129 220a Total 313b 389b 702 <.05 b-b 2' p_‘=.05 65 Speeches Labeled Positive-Affectionate Generally, the number of speeches labeled Positive-affectionate was not affected by the independent variable. With respect to the total number of speeches labeled P-A, the wives had a higher number than the husbands, as indicated by Table 10. Generally speaking then, the wives were more submissive and Positive-affectionate in their discussions than their husbands. This finding seems to fit with cultural stereotypes, and suggests further face validity for the scales. For speeches labeled high P—A, the Sex X Discussion Interaction was significant: wives in Discussion III had more high P-A speeches than husbands in Discussion 1. The significant effects for speeches labeled medium P—A were that the LD group had more speeches labeled P-A than the C group, and the wives had more medium P-A speeches than the husbands. Clarity of the Interaction Clarity was not influenced by the independent variable. Time Needed to Reach Agreement The amount of time needed to reach agreement was not influenced by the independent variable. Summary_of Major Findings from the Analyses of Variance l. The experimental groups were partially, but not completely, able to follow the dominance instructions. This partial success may have obscured or diminished differences between groups. 2. Only N-H speeches were affected by the independent variable. Spouses of either sex allowed a moderate increase (MD group) in partner dominance 66 TABLE 10 Total Number of Speeches Labeled Positive-affectionate Sex Group Husband Wife Total H0 49 58 107 M0 46 54 100 L0 56 87 143a c 39 49 88a Total 19ob 248b 438 a-a :1<:.05 b-b ’ 67 without an increase in N-H speeches, but when the Selected Spouse was instructed to greatly increase (HD group) the amount of dominance, an increased number of N-H speeches, but not greater dominance, was observed. This indicates that in couples where one spouse is not normally highly dominant, attempted high dominance had a negative- hostile effect on the interaction. An examination of the intercorrelation matrices later on in the Results section will show whether this same conclusion was true with couples where one member was naturally highly dominant. 3. The number of speeches classified Dominant did not reflect the actual amount of dominance going on (as measured by the Dominance Index scores). Assuming the validity of the Dominance Index, the actual dominance must have been expressed through some means other than by those speeches classified as high or medium dominant. 4. The wives were more submissive and Positive-affectionate than the husbands. 5. The quality of the interaction changed over the three disussion periods, with both spouses gradually becoming less submissive. 6. Clarity and Time needed to reach agreement were not influenced by the independent variable. The Correlational Analysis The information sought from the correlations pertained to whether the same relationships would appear in correlations when the couples were interacting "naturally" in Discussion I, as appeared through the experimental manipulations. The reader is referred to p. 48 for a complete listing of the variables contained in the intercorrelation matrices. 68 Individual Correlations In Table 11, the expected correlations between speeches labeled high, medium, and total Dominant (for example) were bracketed by a dashed line. This was done for each of the four major speech categories. Further, as most of the significant correlations were related to speeches labeled high and medium Dominant, these two categories were enclosed by solid lines. The same procedure was followed in Table 12, for the couple correlations. As Table 11 reveals, the number of high Dominant speeches correlated significantly with the number of Negative-hostile speeches. Specifically, the number of high Dominant speeches correlated significant- ly with the number of high N-H speeches, the number of medium N-H speeches, and the total number of N-H speeches. These findings correspond to the experimental finding that the HD group had more medium and total N-H speeches than the MD and LD groups, in Discussion II. It was also found that the number of medium Dominant speeches was significantly correlated with the number of P-A speeches. More specifically, the number of medium Dominant speeches was significantly correlated with the number of medium P-A speeches, and the total number of P-A speeches. The fact the high Dominant speeches correlated significantly with N-H speeches, but medium Dominant did not, exactly parallels the findings from the analyses of variance, where the HD group had more high N-H speeches than the MD group in Discussion II. However, unlike the correlational analysis, the experimental analysis did not find a relationship between medium Dominant and P-A speeches. 69 No mm mm NF- «no mo mm mp- mo- FN-. em mo- mo- op-_ 3N8 8 8- 8.. mm- [a I bl I I E Po mo mo- mo «N _F mm “.mm RT 8 E- 2 E m. 2.- _ gnarl me my- RP em- 8_ no- .mm mm spa op 3’. 82 mm- mp No- «Fm Pm .83 mo- oo o, mo- so- o_- m_ eN mm mu ac- my- me e, ,N FF- an no MN mo- mp- No Q, em mo- Pm mo 8N op- mo- no No- _o ON- 8N mo 8, mp 3_ MP NF __ CF 8 m + «mm sac .1 888 8:365 6.: 9.5.: 8.vm k. umuumso mpcwoq Fmswumu FP<+ Hug zomhpncH FP< mcoe< mcovmemesou pcmEoZquauoca —— m4m>>>>>>>>>>>>> IStI'OtIsly Agree SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA SA mmmmmmmmmmmmmmt Head to Agree hflflbbflflflfi mmmmmmmmmmmm mmmmmmmmmmu 12 222222222 222222222222222 'Neither Agree 22222222222 222222222222 2 Nor Disagree n-n-n-n-n-n-n-n-n. nn-n-n-e-n-n-n-n-n-mo-n-n-n- [Tend to Dosagree In. D-O-O-D-O-DoO-O-D-D-D- fififi-G-D-O-D-O-O-G-D-C- D. {3 IStrongly Disagreelg SD APPENDIX E F RATIOS FROM ANALYSES 0F VARIANCE HO §,Ratios from Analyses of Variance 111 Analysis Source of Variation A 8 AB 6 AC BC ABC (Group) (Sex) (Discussion Period) Influence - * Scores 1.52 4.01 1.34 1.63 3.96 .85 2.11 Dom1 11a nce .1 Index 1.78 2.72 .04 .37 5.37 1.77 1.94 HNH 2.41 .07 .18 .08 1.69 .10 .44 __. MNH 3.16* .02 .15 2.00 2.42' .10 .26 * TNH 3.44* .00 .07 1.27 2.72 .09 .15 MFA .50 -2.49 1.20 1.46 .78 3.51' .23 * MPA 3.38* 4.46 .83 .19 1.70 1.84 .49 __4 TPA 3.24' 4.86* .70 .02 1.54 1.06 .70 HD .33 .59‘ .36 .41 1.28 .16 .80 MD 1.22 .00 .11 2.03 1.72 .00 .56 ** *fi 10 -207.03 .11 208.17 3.10 376.25 .03 -374.13 i HS .10 24.07 .53 1.21 2.54' 1.12 .75 i MS 5.55 .06 1.09 8.89” 1.02 .42 .40 * ** * if 15 3.64 301.86 -98.14 7.38 . 2.03 -241.15 81.17 Clarity .65 .04 I .21 Time 1.10 2.08 .04 * 2_<::.05 * * p_<(_.Ol