<<

Journal of Psychology Copyright 1999 by the American Psychological Association, Inc. 1999, Vol. 13, No. 1,20-32 0893-320O/99/$3.0O

Feeling Controlled in : A Phenomenon Specific to Physically Aggressive Couples? Miriam K. Ehrensaft, Jennifer Langhinrichsen-Rohling, Richard E. Heyman, K. Daniel O'Leary, and Erika Lawrence State University of New York at Stony Brook

Spouses in maritally happy nonaggressive (H; n = 21), distressed nonaggressive (DNA; n = 16), and distressed aggressive (DA; n = 20) were inter- viewed about their perceptions of their spouse as controlling. Four areas of spousal control were assesed: involvement in decision making, relationships with family and friends, freedom to plan activities independently, and sense of competence and self-respect. Overall, as expected, spouses in happy marriages reported feeling less controlled than spouses in the 2 distressed groups. Few gender differences were obtained, with the exception that in aggressive marriages were more likely to report that their controlled their sense of competence and self-respect. Differences between the DA and DNA groups depended on the specific area of control. Wives in the aggressive couples were significantly more likely than their husbands to state that their spouse's aggression was an attempt to control them.

Scholars and practitioners from a variety of between marital aggression and control, defini- theoretical perspectives have asserted that there tions of control produced by each theory, and is a relationship between coercive control and empirical research that has been associated with the occurrence of physical aggression in mar- these definitions. Then a definition of control is riage (Black, 1983; Finkelhor, 1983; Gondolf, offered, drawing from each of these three 1985; Stets, 1988). Efforts to test this assertion, theories. Finally, the relationship between this however, have been hampered by a lack of definition of control and physical aggression is agreement on how to operationally define studied in a sample of distressed aggressive spousal control. Consequently, this article briefly (DA), distressed nonaggressive (DNA), and reviews existing theories explicating the link nondistressed nonaggressive (H) couples. Partner abuse literature points to three impor- Miriam K. Ehrensaft, Jennifer Langhinrichsen- tant theoretical frameworks for understanding Rohling, Richard E. Heyman, K. Daniel O'Leary, and the link between marital aggression and control: Erika Lawrence, Department of Psychology, State attachment, feminist, and social control theories. University of New York at Stony Brook. Bowlby's (1969, 1973, 1977, 1988) attachment Jennifer Langhinrichsen-Rohling is now at the theory postulates that aggression against an Department of Psychology, University of South Alabama. Erika Lawrence is now at the Department attachment figure, whether during childhood or of Psychology, University of California, Los Angeles. adulthood, is a control strategy that functions to This research was supported in part by National regain either physical or emotional proximity to Institute of Mental Health Grant MH19107. Because that figure, when the bond with that figure is the research constituted Miriam K. Ehrensaft's perceived to be endangered. Individuals who are doctoral specialties project, special thanks are ex- insecurely attached to their primary attachment tended to committee members Marvin Goldfried, figure are more likely to perceive subjective David Pomeranz, and Dana Bramel. threats to the bond with their attachment figure, Correspondence concerning this article should be addressed to Miriam K. Ehrensaft, who is now at the as compared with individuals who are securely Department of and Adolescent Psychiatry, New attached (Bowlby, 1977; Hazan & Shaver, York State Psychiatric Institute, 722 West 168th Street, 1987). A definition of control based on attach- Unit 78, New York, New York 10032. Electronic mail ment theory would encompass behaviors in- may be sent to [email protected]. tended to regulate the proximity of the attach- 20 CONTROL IN MARRIAGE 21

ment figure. Consistent with attachment theory, ful behavior (e.g., Gondolf, 1985). A definition clinical reports indicate that battering men are of control based on this theory would describe highly controlling toward their partner and behaviors intended to maintain adherence to display extreme difficulty in tolerating their perceived regulations and norms within the partner's autonomy (Elbow, 1977; Ganley, marriage; violence could be used as a form of 1981; Gondolf, 1985; Shields & Hanneke, 1983; retribution for violations of such norms. Sonkin, Martin, & Walker, 1985; Walker, 1984). In summary, ideas from attachment, feminist, Research suggests that partner-violent men are and social control perspectives each yield more likely to be insecurely attached to and are theoretical support for a relationship between less trusting of their partner (Holtzworth- control and physical aggression in marriage and Munroe, Stuart, & Hutchinson, 1997), display different operational definitions of control. higher needs to control emotional distance in Drawing ideas from each of these three theories, their marriage (Dutton & Strachan, 1987), and we propose that an assessment of control should are more dependent on their partner than are cover the following areas: reducing the spouse's DNA men (Murphy, Meyer, & O'Leary, 1993). power to make decisions, limitations of the Feminist theory provides a second theoretical spouse's relationships with others and indepen- link between control and aggression in romantic dence in daily activities, and diminution of his relationships. This theory views family violence or her self-image and ego strength. Behaviors in each of these domains may be defined as as an outgrowth of a patriarchal hierarchy that controlling when they attempt to or have the sanctions the use of male-female violence as a effect of directing or constraining the spouse's means of maintaining male power within the actions, thoughts, or emotions. We reasoned that marriage (Denzin, 1984; Dobash & Dobash, an individual might control his or her spouse in 1979; Lips, 1991). Consistent with this stance, just one, or several, of these areas and, thus, rates of abuse appear to be lower in included a separate assessment of control over societies in which wives have economic influ- each area. For us to distinguish feeling con- ence within the marriage, compared with those trolled from mutual and normative interspousal in which wives have little or no economic influence, these attempts had to be perceived as influence (Levinson, 1988). Family violence is negative by the recipient. Control was also also less common in societies in which men do differentiated here from psychological maltreat- not expect to be masters of their homes and ment of a spouse (e.g., Marshall, 1992). We tolerance for violence is low (O'Kelly & conceptualized psychological maltreatment as a Carney, 1986). On the basis of feminist theory, broader, umbrella structure, of which control- partner control (especially by men) would be ling behavior is only one specific subtype, as not denned in terms of male-dominated decision all psychological maltreatment described in the making, beliefs about men's greater competence literature is necessarily controlling (Marshall, relative to women, and expectations that men's 1992; Tolman, 1989). For instance, verbal needs take priority over women's needs. hostility toward a spouse may be psychologi- Social control theory provides a third theoreti- cally maltreating, whereas its actual or per- cal connection between marital aggression and ceived function might be to express anger, rather spousal control. Black (1983) has suggested that than control, per se. There is some disagreement among domestic violence researchers regarding violent responses to perceived deviance or the differentiation between psychological abuse injustice by others serve as conflict manage- and control. Although some maintain that the ment, as a means of expressing grievances, and function of all psychological abuse is to control as a form of social control. Consistent with this the partner, others disagree. Because it is our theory, most marital assaults do occur in the position that the function and impact of all context of disagreement or a grievance (O'Leary psychological abuse may or may not be to et al., 1989; Vera Institute of Justice, 1977). control the partner, we chose specifically to Social control theory may also be operative in focus our definition of control on behaviors that remarks made by male batterers, implying that do constrain or limit the partner and that are their wives deserved to be beaten for perceived perceived by the partner as unwanted. offenses such as attempts at autonomy, failure to fulfill traditional chores, or disrespect- It is surprising that few empirical studies have 22 EHRENSAFT ET AL. sought to demonstrate the link between control partners. For husbands, our hypothesis was and marital aggression, and most existing based on findings indicating that aggressive men studies suffer from important methodological have stronger needs for control than do nonag- shortcomings. First, most studies that address gressive men (Kimerling & Arias, 1994), control fail to specify just what it is that they are suggesting that aggressive husbands might be measuring. Other studies do not differentiate more likely to interpret their wives' behaviors as between coercive control and decision-making controlling, to perceive a threat to their own power (Blood & Wolfe, 1957). or between control, and thus report feeling more controlled coercive control and noncoercive influence of a than do nonaggressive husbands. Second, accord- spouse. Clearly defining what is meant by ing to feminist theory, wives should feel more control allows us to test whether specific controlled than husbands in general, and wives behaviors and attitudes are more common of aggressive husbands should feel particularly among husbands versus wives, distressed versus controlled. Finally, we expected that wives in nondistressed marriages, and aggressive versus the aggressive group would be more likely than nonaggressive marriages. A second major short- their husbands to believe that their spouse was coming of most research linking control and motivated to use physical aggression as a means spousal aggression is the absence of comparison of maintaining control over them. groups to control for marital discord. Finally, most research on control and aggression is not based on dual spousal reports. To fill the gaps in Method existing research, we tested the idea that spousal control is specifically related to marital aggres- Participants sion, rather than to overall marital discord, by comparing happy nonaggressive, distressed aggres- The sample was composed of 21 maritally happy, nonaggressive couples (H), 16 maritally discordant, sive, and distressed nonaggressive couples. nonaggressive couples (DNA), and 20 maritally The socially undesirable nature of partner discordant, aggressive couples (DA). Previous studies control may pose threats to measurements of using similar samples have found that the occurrence of marital aggression decreases as age and years such behavior. This is particularly true for married increases (Pan, Neidig, & O'Leary, 1994). partner-aggressive men, who tend to deny or For this reason, spouses were included only if they minimize their aversive behaviors toward their were married between 1 and 7 years. This sampling wives (Gondolf, 1985; Sonkin et al., 1985). strategy also increased the likelihood of obtaining a Consequently, we asked participants to report on relatively homogeneous sample and of ruling out their spouse's controlling behavior, rather than third variable confounds. Couples were recruited so on their own behavior. Moreover, an attempt to as to fill three different groups. Happy couples were influence a partner may or may not be perceived required to obtain an average score of at least 115 on as controlling by the targeted partner. For the Marital Adjustment Test (MAT; Locke & Wallace, instance, an individual might perceive a spouse's 1959), neither spouse could obtain a score lower than 100, and neither could have engaged in any act of attempts to limit time spent without him or her physical aggression at any point in the marriage, as as a desirable sign of affection. Thus, we elicited measured by the Conflict Tactics Scales (CTS; Straus, the partner's cognitive-affective response to 1979). Maritally discordant but nonaggressive couples each behavior to distinguish behaviors per- were required to obtain an average MAT score no ceived as controlling from those perceived as higher than 100, neither spouse could obtain a score noncontrolling influence. higher than 115, and neither could have engaged The study tested the following hypotheses in or received any act of physical aggression at any point during the marriage. Maritally discordant, regarding the association between feeling con- aggressive couples were to meet the same MAT trolled and marital aggression on the basis of the criteria as the discordant, nonaggressive couples, but theoretical and empirical literature reviewed were also required to endorse at least two acts of above. First, we predicted that both partners in -to-wife physical aggression on the CTS aggressive marriages should report feel more during the past 12 months (throwing an object at the controlled than partners in nonaggressive mar- partner or more severe acts), as reported by either the riages. We based this hypothesis about wives on husband or the wife. This decision to include only feminist theory's position that wives of aggres- couples reporting at least two acts of husband-to-wife sive husbands are highly controlled by their aggression was to obtain a more homogenous, stably CONTROL IN MARRIAGE 23

aggressive group, rather than one in which the Table 1 aggression was a single occurrence. Demographic Means for Happy (H), Distressed All of the couples were obtained from advertise- Nonaggressive (DNA), and Distressed ments in local newspapers in Suffolk County, Long Aggressive (DA) Spouses Island, NY, inviting "couples married between 1 and 7 years to participate in a study of marriage" and were Variable H DNA DA screened for eligibility by telephone interview, using Years of age the MAT and the CTS. As this advertisement yielded a rapid filling of the H group, the announcement was M 31.00 34.60 31.90 later rerun inviting "couples married 1 to 7 years SD 6.62 10.49 6.82 Years married having problems in their marriage to participate in M study of marriage." Of 168 couples who were 4.45 4.11 4.86 SD screened for the project, 57 were rejected as they did 2.09 2.51 3.96 Family income ($) not fit any group, 35 were accepted but were never successfully scheduled, 75 were accepted and partici- M 48,053 51,767 39,088 SD pated, and 21 of these 75 were being recruited for a 19,409 20,791 21,207 Years of education different study. Thus, the final sample included 54 M couples. 14.40a 13.60b 12.41C SD 2.92 1.89 3.56 No. of children Sample Characteristics M 1.02a 1.23b 2.11C SD 1.30 1.22 1.26 The sample was 84.5% Caucasian, 6.0% Hispanic, 2.4% African American, 2.4% Asian, and 4.8% other. Note. Means with different subscripts are different A 3 (group) X 2 (gender) multivariate analysis of at the p<. 05 level. variance (MANOVA), with gender as the repeated variable, and age, income, years married, years of education, and number of children as dependent Holtzworth-Munroe et al., 1997, reported a mean of variables indicated significant group differences on 12.9, and 8.6 acts of aggression in two community the above variables, Wilks' A (10, 174) = .67, p < samples of male batterers). Hamberger, Lohr, Bonge, .001. Univariate analyses of variance (ANOVAs) and Tolin (1996) reported means of 5.96 to 31.51 acts revealed group differences on education, F(2, 91) = of violence in three groups of court-mandated 5.32, p < .01, and number of children, F(2, 91) = batterers in treatment for spouse abuse. 8.88, p < .001; DA spouses had less education and more children than DNA and H spouses. No group differences were obtained on the other demographic Measures: Phone Screen Instruments variables. Neither gender differences nor Group X CTS (Straus, 1979). The CTS is a scale com- Gender interactions were found. posed of physically and psychologically coercive As expected on the basis of selection factors, a items occurring in the context of a marital conflict. significant group difference was observed for the Participants reported, separately, how frequently they Dyadic Adjustment Scale (DAS; Spanier, 1976), F(2, and their spouse engaged in each tactic during the 92) = 94.40, p < .001. Follow-up univariate past 12 months. The aggressive tactics include ANOVAs revealed that the difference in DAS scores throwing something at the spouse, pushing, shoving, was attributable to a difference between the H group slapping, hitting, kicking, beating up, and using a versus the two distressed groups, and that the two knife or a gun against the spouse. Couples in which distressed groups' DAS scores did not differ from one neither spouse endorsed any of those items were another. The mean DAS scores were 122.20, identified as nonaggressive. (SD = 11.86), 88.30 (SD = 12.81), and 86.50 A factor analysis found the CTS to be structurally (SD = 12.51) for the H, DNA, and DA groups, sound (Pan et al., 1994), and the psychometric respectively. Table 1 shows the demographic means properties of the original CTS are well documented for each group. Consistent with the study acceptance (Straus, Gelles, & Steinmetz, 1980). The CTS is criteria noted above, averaging the reports of both the frequently used as a dichotomous yes-no classifica- DA husbands and wives, husbands had committed an tion measure of aggression (e.g., Anglin & Holtzworth- average of six acts of mild aggression (e.g., pushing, Munroe, 1997; Ehrensaft & Vivian, 1996). shoving, slapping) and one act of severe aggression MAT (Locke & Wallace, 1959). The MAT is a (e.g., beating up) in the 12 months prior to measure of marital adjustment suitable for rapid participation in the study. The frequency of aggres- administration on the phone (O'Leary et al., 1989). It sion perpetrated by husbands in this study is roughly consists of 15 items, 9 of which assess the degree of comparable with, if somewhat lower than, rates disagreement on major marital issues. It is sensitive to reported by community samples of batterers (e.g., changes in marital therapy (cf. O'Leary, 1987), and its 24 EHRENSAFT ET AL. convergent validity has been repeatedly demonstrated four dimensions: (a) decision making (major deci- (e.g., Navran, 1967; Spanier, 1976). Scores can range sions, spending money, etc.), (b) relationships (free- from 2 to 158, with higher scores indicating higher dom to pursue relationships with friends and family), levels of adjustment; a score of 100 has been the (c) activities (freedom to select activities, schedule traditional cutoff point for marital distress. The scale one's day, and engage in activities without the served to classify spouses as maritally discordant or spouse), and (d) self-image (being treated as worthy happy. A longitudinal study of recently married of respect, competent, and able to manage indepen- couples (O'Leary et al., 1989) reported a mean MAT dently). Table 2 lists the topics and sample questions score of 115. On the basis of that study, we classified from each dimension. For each question, if the couples as happily married if they scored above this participant indicated that his or her spouse exercised value. some influence over that topic, the interviewer then asked the participant to describe the effects of that influence (i.e., positive, negative, or neutral) on his or Study Measures her feelings or behaviors, or both. Only responses indicating negative effects were scored as indicating DAS (Spanier, 1976). The DAS is currently the that the respondent felt controlled by the behavior. most widely used measure of overall marital satisfac- tion. Its psychometric properties are well established Items in the Control Interview were originally (Spanier, 1989). Scores range from 0 to 151. selected on the basis of face validity, following a Control Interview (Ehrensaft, 1994). Parts I-m review of the spouse abuse literature. For instance, of the Control Interview were semistructured so that Walker (1984) indicated that residents of shelters for all participants were asked the same questions, but battered women often reported batterers' attempts to could answer open-endedly. isolate mem, forbidding them to pursue close relation- Part I: Participants were queried about the degree ships with others. Thus, questions in the relationships to which their spouse controls them on the following dimension addressed the degree to which the participant

Table 2 Control Interview Topics and Questions Topic Question Decisions Major decisions How much influence do you have over major decisions, e.g., moving? Social decisions How much influence do you have over social decisions, like what to do together on the weekend? Spending money How much influence do you have regarding spending money? Personal decisions How much influence do you have over personal decisions that don't have anything to do with your spouse, like handling a problem at work or with a friend? Relationships Family/friends How does your spouse feel about you spending time with your family/friends? Is there anyone whom he discourages you from spending time with or being close to? Constraints Are there any constraints on whom you spend time with, or on what you do with them? regarding Has your spouse ever accused you of having an , or acted suspicious about you affair having an affair? Activities Freedom How much freedom does your spouse give you to do the things you want, go where you want, etc.? Schedule Does one of you have to schedule his or her day around the other, or consult the other about how you schedule your time? Alone How does spouse feel about you going places alone? Forbidden Is there anything you are "not allowed" to do? Awareness How aware is your spouse of what you do and where you go on a daily basis? Self-image Respect How much does spouse seem to respect you as a person, by what he/she says or does? Competence How much does spouse treat you like an independent adult, like you can function as a and worker? Self-sufficiency How much does spouse act as though you would be able to manage without him/her? Does he/she tell you that you couldn't manage without him/her, or that no one else would want you? CONTROL IN MARRIAGE 25

felt controlled by his or her spouse in developing and Spearman's rhos (rs) for negative control responses maintaining relationships with family and friends. Follow- were as follows: decisions, T-S = .35, p < .01; ing the original selection of interview items, a list of all of relationships, rs = .27, p < .05; activities, rs = .32, the items was given to 10 psychologists and psychiatrists, p < .05; and self-image, rs = .16,/? > .10. For wives, who were asked to sort them into categories. Items were the correlations were decisions, rs = .52, p < .01; deleted (n = 4) if raters obtained less than 80% agree- relationships, rs = .24, p = .07; activities, rs = .53, ment on their dimension classification. Agreement was p < .01; and self-image, rs = .25, p = .08. high, and items that were conserved obtained interrater Part II: Participants were then asked to indicate agreement ranging from 80% to 100%. whether they felt that their spouse tried to control Finally, the list of items was submitted to a test of them (yes or no), and to quantify, on a 10-point scale internal scale reliability using the current sample. ranging from 0 (not at all) to 10 (completely), how Items that significantly reduced item-scale alphas controlled by their partner they felt, regardless of were deleted (n = 2). Cronbach's alphas for the whether the partner was trying to control them. retained items in each dimension were as follows: Part III (only for participants who report receiving decision making, a = .83; relationships, a = .76; aggression from their spouse): Participants were activities, a = .80; and self-image, a = .75. Moderate asked to describe one incident in which they received but not high correlations with marital satisfaction marital aggression. They were also asked what they were expected. Consistent with this prediction, the thought their partner's motivation was when he or she DAS and the four dimensions of the Control was aggressive. Last, they were asked (yes or no) Interview correlated as follows: decisions, r = —.43; whether they thought the aggression was conducted in relationships, r = —.46; activities, r = —.47; and order to control them or in response to losing control self-image, r = —.56. over them. All of the responses to Part I were coded according All of the responses were coded directly from to a set of mutually exclusive, exhaustive categories. videotapes of the interviews. Five undergraduates A code of +2 was assigned to responses in which a were trained to code the interviews through an participant suggested that he or she either (a) had intensive training curriculum. During this training, more control than the spouse, or (b) had complete disagreements regarding categorization of responses freedom over the topic (e.g., "I make 90% of the were resolved by consensus. Reliability was calcu- social decisions"). A code of +1 was assigned when lated on a random sample of 25% of all interviews. the participant stated that he or she was somewhat Kappas-by-category were .94 for +2, .81 for +1, .53 for influenced by the spouse, but that this had a positive 0, .61 for -1, and .87 for -2. When the -2 and -1 codes effect on either the participant or on the marriage were combined, kappa-by-category was .85. (e.g., "I must schedule my day around him, but I feel that this is good for our marriage"). If a participant's response indicated that the spouse had some influ- Design and Procedure ence, but indicated that the effect was neutral, the response was coded 0 (e.g., "Having to come home The present study was part of a larger study on both before midnight does not really affect the way I the positive and difficult aspects of marriage (Langhin- feel"). richsen-Rohling, Schlee, Monson, Ehrensaft, & Hey- man, 1998). At the laboratory, participants completed Responses in which the participant suggested a demographic questionnaire, all self-report instru- feeling somewhat controlled were coded — 1 (e.g., "I ments, and several other short interviews before feel somewhat restrained from seeing my friends, and commencing the Control Interview. Spouses were this bothers me a little"). Those in which the separated during self-report measure administration participant suggested feeling quite controlled were and during the Control Interview. Interviewers and coded —2 (e.g., "I can't see my friends anymore and coders were blind to the couples' group assignment. it really bothers me"). Uncategorizable responses The order of husband and wife Control Interviews were coded other. was counterbalanced across couples. Specific guide- As a concurrent validity check for the Control lines about the extent to which probing was permitted Interview, the four dimensions of the interview were was provided to interviewers to ensure consistency correlated with the Dominance/Isolation subscale of across participants' responses. Participants were Tolman's (1989), Psychological Maltreatment of asked only about controlling behaviors that they had Women Inventory (PMWI). This self-report scale experienced from their partner. They were not asked differentiates battered women in domestic violence about the control behaviors in which they had shelters from nonbattered women and includes items engaged. The Control Interview was conducted by such as "My partner ordered me around," "My one of two trained graduate-level interviewers and partner monitored my time and made me account for lasted approximately 30 min per spouse. Participants where I was," and "My partner tried to keep me from were debriefed in full and paid $60 per couple for 3 hr seeing or talking to my family." For husbands, of participation. 26 EHRENSAFT ET AL.

Results relationships versus activities dimensions, F(2, 51) = 3.70, p < .05, and when comparing For each participant and for each dimension responses with the activities versus self-image of the Control Interview, a proportion of dimensions, F(2,51) = 3.78, p < .05. responses indicating perceived negative spousal Univariate analyses for each dimension re- control was calculated. Analyses were con- vealed that on the decisions dimension there was ducted by summing the number of —1 and —2 a significant group effect, F(2, 51) = 7.60, p < (negative control) responses given for that .001, with the H group obtaining significantly dimension and dividing this number by the total lower scores than the two distressed groups number of questions answered for that dimen- (95% contrast confidence interval lower sion. For instance, of the total five questions bound = -10.02, upper bound = -39.43). On posed on the decision-making dimension, if a the relationships dimension, a significant participant gave two answers that were coded as Group X Gender interaction was obtained, F(2, — 1, that participant would receive a proportion 51) = 3.87, p < .03, indicating that DNA and of 2/5 X 100 = 40%. DA husbands felt more negatively controlled in Table 3 shows the mean percentages of felt the relationships domain than did DNA and DA negative control across each of the four wives, whereas this difference was not obtained dimensions, for husbands and wives in H, DA, among H couples. On the activities dimension, and DNA marriages. no gender effect was obtained, F(l, 51) = .92, Correlations between husband and wife data p > .10, but a significant group effect was were significant, as were correlations between obtained, F(2,51) = 13.64, p < .0001. Spouses responses on each of the four dimensions. Thus, in H marriages scored lower than spouses in the a three-way general linear model (GLM), two distressed groups (95% contrast confidence repeated measures analysis was run, using interval lower bound = -34.96, upper gender (2) and dimension (4) as the repeated bound = -13.06). Spouses in the DA group variables and group (3) as the between-subjects reported more felt control on the activities variable.1 A priori contrasts were requested, dimension than spouses in the DNA group (95% comparing the percentage of responses indicat- contrast confidence interval lower bound = 3.73, ing felt negative control by spouses in H upper bound = 31.10). No significant Group X marriages with those in the two distressed Gender interaction was obtained on the activi- groups, and then comparing responses by spouses in the DA and DNA marriages to each other.2 1 So as to ensure that the assumptions of ANOVA Results of the GLM suggest no overall gender were not violated when analyzing percentages, effect, but an overall group effect, Wilks' A = examinations were made of the heterogeneity, skew, .53, F(2, 51) = 22.24, p < .0001, collapsing and kurtosis of the percentage of perceived control across the four control dimensions. The H group responses for each of the four Control Interview scored significantly lower (felt less negative dimensions (Tabachnik & Fidell, 1989). These three control) than the two distressed groups, Wilks' variables were examined for distributions of hus- bands' and wives' responses in each of the three A = .57, F(l, 51) = 38.56, p < .0001. As groups (H, DA, and DNA), thus looking at a total of hypothesized, the DA group scored significantly six distributions. In all of these distributions and on higher (felt more negative control) than the all four of the interview's dimensions, the highest DNA group, Wilks' A = .92, F(l, 51) = 4.31, variance was less than 4 times the lowest variance, p < .05. There was no significant Group X which indicates that the heterogeneity did not differ Gender interaction, Wilks' A = .97, F(2, 51) = problematically across the different distributions. .70,p>.10. Similarly, for all four Control Interview dimensions, each of these six distributions met appropriate standards few- Participants seem to have responded differ- skew and kurtosis (Tabachnik & Fidell, 1989). Thus, ently to the various dimensions; an overall assumptions of ANOVA were not violated. dimensions main effect was obtained, Wilks' 2 Although years of education and number of A = .48, F(3, 49) = 17.80, p < .0001. A children differed across the three groups, they were Group X Dimensions interaction was obtained not significantly correlated with the dependent when comparing participants' answers with the variables and thus were not included as covariates. CONTROL IN MARRIAGE 27

Table 3 Mean Percentages for Perceived Negative Control and Means for 0-10 Perceived Control Scale for Happy (H), Distressed Aggressive (DA), and Distressed Nonaggressive (DNA) Couples Husbands Wives H DNA DA H DNA DA Dimension (n = 21) (« = 15) (n = 18) (n = 21) (n = 15) (n = 18) Decisions M 8.10 16.33 29.72 5.00 20.33 29.72 SD 18.06 18.94 33.28 11.18 29.31 38.60 Relationships M 7.62 45.67 54.17 13.68 28.00 25.19 SD 14.80 43.46 31.82 11.67 27.04 31.15 Activities M 8.73 26.00 43.61 6.67 20.00 37.22 SD 17.90 27.20 36.77 13.45 25.07 34.90 Self-image M 10.71 38.33 58.33 16.67 55.00 63.88 SD 20.27 33.89 32.08 24.15 27.06 37.60 0-10 Perceived Control scale M 1.83 4.69 5.17 2.18 3.65 4.03 SD 2.17 2.32 2.66 2.00 2.49 3.39 ties dimension, F(2, 51) = .08,/? > .10. Finally, Perceived Motivation for Spouse's on the self-image dimension, there were no Aggression (DA Spouses) gender, F(l, 51) = .10, p > .10, or Group X Gender, F(2, 51) = 1.79, p > .10, effects, but a Mixed-design ANOVAs with gender as the significant group effect was obtained, F(2, 51) = repeated variable on perceptions of the spouse's 24.16, p < .0001. The H group scored motivations for physical aggression revealed significantly lower than the two distressed that wives were more likely than husbands to groups (95% contrast confidence interval lower report that their spouse was aggressive to get bound = -52.49, upper bound = -27.91). Also, them to stop doing something, F(l, 19) = 4.75, DA wives reported significantly more felt p < .05. Wives were also significantly more control of their self-image than did the DNA likely than husbands to answer affirmatively wives (95% contrast confidence interval lower when asked directly whether they thought that bound = 3.19, upper bound = 47.92). This last their spouse was aggressive in order to control effect was not significant for husbands. them,F(l, 19) = 4.13,/? < .05. Finally, on the 0-10 scale of overall perceived control, a group main effect was obtained, F(2, Perceived Influence 48) = 10.39, p < .0001; DNA and DA participants felt more controlled than H partici- After examining whether group and gender pants (95% confidence interval lower differences existed with regard to experiencing bound = —4.91, upper bound = -1.84). Unex- control from one's spouse as negative, a second pectedly, DA couples did not report feeling more set of post hoc questions arose regarding controlled than DNA couples (95% confidence perceived spousal control. Instead of looking interval lower bound = —1.33, upper only at spousal control that is perceived as bound = 2.53). No gender main effect was negative, are there group or gender differences, obtained, F(l, 48) = 1.54, p > .05, nor was or both, in the amount of overall spousal there any Group X Gender effect, F(2, 48) = influence the participants experience? A new 1.07, p > .10. Table 3 shows the means on the summary variable was created for each dimen- 0-10 Perceived Control scale. sion, calculating the proportion of responses in 28 EHRENSAFT ET AL. which the individual indicated feeling influ- percentage of influence reported by each group, enced by their partner, whether they stated that for each dimension, and Table 5 shows the this influence had a positive, neutral, or negative correlations between husbands' and wives' effect on the marriage (i.e., collapsing —2, — 1, reports on mean percentage of control and + 1, and 0 responses). A Group X Gender influence responses, respectively. MANOVA was conducted, with gender as the To test the theoretical assumption that feeling repeated variable and the four dimensions controlled should be defined as feeling influ- (decisions, relationships, activities, and self- enced and reacting with negative affect to that image) as dependent variables. An overall group influence, partial correlations between the PMWI main effect was obtained, Wilks' A = .45, F(8, and influence (any influence, whether seen as 96) = 5.89, p < .0001, but neither a gender, negative, neutral, or positive) responses were Wilks' A = .27, F(4,48) = 1.35, p > .05, main calculated, partialing out the negative control effect nor Group X Gender interaction, Wilks' responses. For men, the only significant influen- A = .82, F(8, 96) = 1.24, p > .05, was ce-PMWI partial correlation was for the self- obtained. ANOVAs indicated that the H group image, rs = .32, p < .05. For women, the partial felt significantly less influenced than the two correlations were significant for activities, rs = distressed groups on the relationships (95% .70, p < .01, and self-image, rs = .68, p < .01. confidence interval lower bound = —32.83, Thus, it appears that for husbands, one can upper bound = —1.39) and self-image (95% generally infer that their wives' influencing confidence interval lower bound = -74.97, behaviors is perceived as controlling only if the upper bound = -38.77) dimensions; the H husband responds with negative affect. An group did not differ from the two distressed exception would be for self-image, where any group on the decisions (95% confidence interval influence over this seems to be perceived as lower bound = -3.31, upper bound = 26.71) controlling. For wives, the picture is slightly or activities dimensions (95% confidence inter- more complex. Whereas influence over deci- val lower bound = -23.08, upper bound = sions and relationships is seen as controlling 3.79). Also, the DA group felt significantly more only if paired with negative affective response, influenced than the DNA group on the activities any influence over activities or over self-image dimension (95% confidence interval lower is seen as controlling. bound = 3.59, upper bound = 37.18), but not on the other dimensions (confidence intervals Discussion for decisions = -1.09 to 36.44; for relation- ship =-15.68 to 23.64; and for self-im- Happy couples felt markedly less controlled age = -3.84 to 41.42). Table 4 shows the mean across all dimensions investigated than did the

Table 4 Mean Percentages for Perceived Influence for Happy (H), Distressed Aggressive (DA), and Distressed Nonaggressive (DNA) Couples Husbands Wives H DNA DA H DNA DA Dimension (n = 21) (« = 15) (n = 18) (« = 21) (n = 15) (n = 18) Decisions M 70.24 57.67 73.06 63.81 47.33 56.94 SD 26.25 32.34 29.66 30.70 21.87 37.46 Relationships M 20.00 39.67 48.33 34.29 36.00 32.96 SD 18.97 33.30 28.34 27.67 25.30 33.57 Activities M 50.16 50.33 70.00 45.95 45.00 54.17 SD 20.38 25.53 30.24 16.10 19.91 30.88 Self-image M 16.67 60.00 64.35 17.86 41.67 63.88 SD 21.41 26.39 30.63 23.90 32.27 37.60 CONTROL IN MARRIAGE 29

Table 5 Two surprising findings emerged in the Husband-Wife Pearson Correlations (r) on present study. First, no differences were found Perceived Control and Influence between DNA and DA couples on the decisions Control interview Perceived Perceived and relationships control dimensions. This is not dimension influence control to say that couples in aggressive marriages do not have conflict over decision making or Decisions -.17 .21 Relationships -.002 .66* relationships with others outside the marriage. Activities .04 .53* On the contrary, compared with H couples, Self-image .49* -.03 scores for DA couples were high. However, it seems that those issues are not unique to aggressive couples and may be more closely related to marital discord. A second unanticipated finding was that DA and DNA couples. On the other hand, in all husbands in both distressed groups felt more groups, most spouses reported feeling influ- controlled than wives regarding the freedom to enced by their spouse to some degree. Clearly, pursue relationships with family and friends most marriages entail a certain level of compro- (i.e., the relationships dimension). This finding mise and reduced independence. However, these is particularly surprising in view of numerous results suggest that spouses in distressed mar- clinical reports by battered women that docu- riages tend to perceive this influence as negative ment jealous and isolating behavior by battering and controlling, whereas spouses in happy men (Dobash & Dobash, 1979; Walker, 1984). marriages are more likely to appraise spousal However, two important caveats must be stated influence as neutral or even positive. Thus, we here. First, the context of husbands' reports may conclude that spouses in happy marriages regarding their wives' limitations of their are either obliged to make fewer compromises relationships with others is unclear. For in- or they have fewer negative perceptions of those stance, wives who request that their husbands compromises. spend less time with friends and more time As predicted, results indicate that spouses in fulfilling parenting responsibilities may be maritally DA marriages feel more controlled viewed by aggressive husbands as controlling. than spouses in maritally DNA marriages, but in Also, aggressive husbands may have a particu- specific areas. They perceive less freedom to larly poor choice of friends if they are antisocial, independently select activities, to spend time may have highly dysfunctional-conflictual fami- alone, or to schedule their time independently. lies of origin, or may have very real tendencies Moreover, wives in DA marriages, relative to to be sexually promiscuous. Viewed in this wives in DNA marriages, tend to feel that their manner, those men's wives' attempts to limit husbands have more control over their sense of their interactions with friends and family may be self-respect and over their feeling of being efforts to protect themselves or their children. competent workers, , or both. These The second caveat to our findings on the wives report that their husbands make them feel as though they would be unable to manage relationships dimension concerns the nature of adequately without them. That wives of aggres- our sample. The sample interviewed in this sive husbands felt more controlled in such areas study is considerably less violent than samples may suggest that these husbands feel threatened of men arrested for abusive behavior and by their partner's autonomy and are motivated women seeking aid from battered women's to control their activities outside the marriage. shelters. On the other hand, several studies now This finding is consistent with Dutton and indicate that jealousy, rejection sensitivity, and Strachan's (1987) clinical observation of partner- insecure attachment are elevated among women assaultive men's need to regulate the socioemo- and men in mildly aggressive relationships (e.g., tional distance within the marriage. In fact, to an Downey, 1997; Downey, Feldman, & Ayduk, in insecurely attached individual, the spouse's press). In sum, part of the dynamic in both attempts to engage in activities outside the DNA, and DA couples may involve both marriage may be perceived as signaling distanc- spouses perceiving that their partner wishes to ing and potential abandonment (Murphy et al., limit their involvement with parties outside the 1993). marriage, albeit for potentially different reasons. 30 EHRENSAFT ET AL.

The absence of gender differences on the should routinely question spouses, in individual majority of the dimensions of the Control interviews, about these types of controlling Interview suggests that, overall, husbands and behavior. Therapists should also carefully ex- wives feel equally controlled in those areas. It plore the context of behavior that is described as also suggests that aggressive marriages may be controlling. For instance, a wife who tries to characterized by a general perception of power- prevent her husband from going out on frequent lessness, each spouse feeling that the other is drinking episodes may be perceived as control- attempting to control them. Results from this ling by the husband, m such cases, the therapist study cannot be used to conclude that husbands should work to reframe the wife's behavior as an and wives actually are equally controlled by indication of concern and frustration regarding their spouse, as this study exclusively measured her partner's problematic habit. perceptions of spousal control. Although one When a clinician obtains a description of a may never be able to objectively assess the true behavior, has explored the context of that degree of spousal control in a laboratory setting, behavior, and deems it to represent a harmful observational assessments of control will be an instance of control, he or she should then work essential methodological addition to this area. to help the clients to recognize these behaviors One gender difference that did emerge, as we and to understand the negative consequences of predicted, was that wives in aggressive mar- the controlling behaviors on their partner's riages tend to perceive their husband's aggres- affect. In turn, the therapist should help the sion as motivated by a wish to control them, clients learn to stop engaging in the controlling whereas this was reported significantly less behavior. Clinicians should also attempt to often by husbands. This finding is consistent uncover the perceived cause of the controlling with other research in which men describe their behavior. For instance, controlling behaviors own marital aggression as instrumentally moti- may reflect a feeling of being threatened by the vated (Hamberger, 1993). The finding is also partner's autonomy and a fear of losing the consistent with social control theory, which partner. Clinicians may point out that control- posits that aggression is used as a means of ling behavior has a paradoxical effect of redressing perceived deviations from expecta- emotionally driving the partner away and may tions. For instance, Gondolf (1985) observed assist spouses to seek alternate, positive means that aggressive husbands often become aggres- of rapprochement with their partner. In addition, sive in response to a perception that their wife clinicians may work with aggressive spouses to failed to fulfill a role obligation. increase their tolerance for the partner's au- tonomy. Note that further research is needed to Implications for Application determine whether the above treatment applica- and Public Policy tions may also extend to individuals who are court mandated for treatment or other popula- The primary clinical implication for the tions of severely partner-aggressive men, where treatment of marital violence, at least in predominantly male-to-female control patterns maritally aggressive couples presenting with a have been reported (Gondolf, 1985). complaint of marital discord, is the need to reduce both the need or sense of entitlement to Study Limitations control one's spouse and the frequency of controlling interchanges between spouses. More- This study has several limitations. Most over, it would be important to teach alternate important, the results cannot assert whether ways of obtaining changes in their partner's feeling controlled in marriage is a cause or result behavior or redressing perceived injustices, of spouse abuse. Clearly, factors other than instead of using controlling behaviors. Aggres- feeling controlled, such as witnessing or endur- sive individuals are more likely to feel that their ing physical aggression in the family of origin, partner is attempting to reduce their personal demographic variables, alcohol use, attitudes autonomy. Furthermore, wives in physically about the use of aggression in marriage aggressive relationships are particularly likely (Rosenbaum & O'Leary, 1981), and psycho- to complain that their husbands control then- physiological variables (Jacobson et al., 1994), sense of personal competence and self-esteem. can contribute to the likelihood of engaging in Clinicians working with marital aggression or receiving marital aggression. Future research CONTROL IN MARRIAGE 31 is needed to clarify the role that control plays in References the prediction of aggression, relative to these other factors, as well as the process by which Anglin, K., & Holtzworth-Munroe, A. (1997). Com- control and aggression become linked. Further paring the responses of maritally violent and study, which is currently underway, is also nonviolent spouses to problematic marital and needed to determine whether spouses' percep- nonmarital situations: Are the skill deficits of tions of being controlled would be supported by physically aggressive husbands and wives global? outside observers. This would allow a determina- Journal of Family Psychology, 11, 301-313. tion of whether spouses in aggressive marriages are Black, D. (1983). Crime as social control. American Sociological Review, 48, 34-45. actually more controlled or simply misinterpreting Blood, R. O., & Wolfe, D. M. (1960). Husbands and their partner's behaviors as intended to control wives. New York: Free Press. diem. Bowlby, J. (1969). Attachment and loss: Vol. 1. Another limitation of the study pertains to the Attachment. New York: Basic Books. generalizability of the findings. As noted earlier, Bowlby, J. (1973). Attachment and loss: Vol. 2. the sample may be representative only of Separation: Anxiety and anger. New York: Basic couples in the early years of intact marriages Books. who volunteer for a study on marriage. Future Bowlby, J. (1977). The making and breaking of research is needed to determine the extent to affectional bonds. British Journal of Psychiatry, which these results generalize to couples who 130, 201-210. Bowlby, J. (1988). A secure base. London: Basic are separated or divorced, refuse to participate in Books. a study on marriage, have never been married, or Denzin, N. K. (1984). Towards a phenomenology of have been married for much longer periods of domestic violence. American Journal of Sociology, time, or where the aggression is so severe as to 90, 483-513. motivate the wife to seek protection at a battered Dobash, R. E., & Dobash, R. P. (1979). Violence women's shelter. Also, given the small sample against wives. New York: Free Press. size, particularly of the DNA group, replication Downey, G. (1997, May). Rejection sensitivity and of the study would be important. intimate relationships. Paper presented at the annual meeting of the American Psychological Society, Finally, perhaps the most important validity Washington, DC. issue pertains to the particular definition of Downey, G., Feldman, S., & Ayduk, O. (in press). control used in this study. We drew from three Rejection sensitivity and male violence in romantic important theories that we found clinically and relationships. Personal Relationships. theoretically useful in defining control and in Dutton, D. G., & Strachan, C. E. (1987). Motivational clarifying the potential relationship between needs for power and spouse-specific assertiveness control and aggression. Other studies taking in assaultive and nonassaultive men. Violence and alternative theoretical views of control may or Victims, 3, 145-156. may not find similar results. We believe this Ehrensaft, M. K. (1994). Control Interview. Unpub- point underscores the importance of clearly defining lished manuscript, State University of New York at the construct of control in empirical studies. Stony Brook, University Marital Therapy Clinic. Ehrensaft, M. K., & Vivian, D. (1996). Spouses' Important questions regarding the relation- reasons for not reporting existing marital aggression ship between control and marital aggression are as a marital problem. Journal of Family Psychology, raised by this study. Is this relationship consis- W, 443-453. tent across different levels of severity and Elbow, M. (1977). Theoretical considerations of chronicity of marital aggression or across age violent marriages. Social Casework, 58, 515-526. and duration of the marriage? Does the sense of Finkelhor, D. (1983). Common features of family feeling controlled by one's spouse relate to the abuse. In D. Finkelhor, R. Gelles, G. T. Hotaling, & spouse's behaviors per se or to the individual's M. A. Straus (Eds.), The dark side of : own hypersensitivity to control? What is the Current research on family violence (pp. 77-88). process by which perceived imbalances in Beverly Hills, CA: Sage. control develop over time? Do such imbalances Ganley, A. L. (1981). Court-mandated counseling for develop in parallel with aggression and marital men who batter: A three day workshop for mental discord or precede them? Ultimately, answers to health professionals: Participant's manual. Washing- these questions are likely to impact treatment ton, DC: Center for Women's Policy Studies. Gondolf, E. W. (1985). Men who batter: An strategies for marital aggression and certainly integrated approach for stopping wife abuse. warrant continued research attention. Holmes Beach, CA: Learning Publications. 32 EHRENSAFT ET AL.

Hamberger, L. K. (1993, November). The function of O'Leary, K. D. (Ed.). (1987). Assessment of marital aggression is different for men and women. Poster discord. Hillsdale, NJ: Erlbaum. presented at the Annual Meeting of the Association O'Leary, K. D. (1988). Physical aggression between for the Advancement of Behavior Therapy, Atlanta, spouses: A social learning theory perspective. In V. GA. B. Van Hasselt, R. L. Morrison, A. S. Bellack, & M. Hamberger, L. K., Lohr, J. M., Bonge, D., & Tolin, Hersen (Eds.), Handbook of family violence (pp. D. F. (1996). A large sample empirical typology of 31-55). New York: Plenum Press. O'Leary, K. D., Barling, J., Arias, I., Rosenbaum, A., male spouse abusers and its relationship to dimen- Malone, J., & Tyree, A. (1989). Prevalence and sions of abuse. Violence and Victims, 11, 277-292. stability of physical aggression between spouses: A Hazan, C, & Shaver, P. (1987). Romantic longitudinal analysis. Journal of Consulting and conceptualized as an attachment process. Journal of Clinical Psychology, 57, 263-268. Personality and Social Psychology, 52, 511-524. Pan, H. S., Neidig, P. H., & O'Leary, K. D. (1994). Holtzworth-Munroe, A., Stuart, G. L., & Hutchinson, Male-female and aggressor-victim differences in G. (1997). Violent versus nonviolent husbands: the factor structure of the Modified Conflict Tactics Differences in attachment patterns, dependency and Scale. Journal of Interpersonal Violence, 9, 366- jealousy. Journal of Family Psychology, 11, 314- 382. 331. Rosenbaum, A., & O'Leary, K. D. (1981). Marital Jacobson, N. S., Gottman, J. M, Waltz, J., Rushe, R., violence: Characteristics of abusive couples. Jour- Babcock, J., & Holtzworth-Munroe, A. (1994). nal of Consulting and Clinical Psychology, 49, Affect, verbal content, and psychophysiology in the 63-71. arguments of couples with a violent husband. Shields, N. M., & Hanneke, C. R. (1983). Attribution Journal of Consulting and Clinical Psychology, 62, processes in violent relationships: Perceptions of 982-988. violent husbands and their wives. Journal of Kimerling, R., & Arias, I. (1994, October). The role of Applied Social Psychology, 13, 515-527. Sonkin, D., Martin, D., & Walker, I. (1985). The male power and control in violence. Poster batterer. New York: Springer. presented at the annual meeting of the Association Spanier, G. B. (1976). Measuring dyadic adjustment: for Advancement of Behavior Therapy, Atlanta, New scales fort assessing the quality of marriage GA. and similar dyads. Journal of Marriage and the Langhinrichsen-Rohling, J., Schlee, K., Monson, Family, 38, 15-28. C. M., Ehrensaft, M. K., & Heyman, R. E. (1998). Spanier, G. B. (1989). Dyadic Adjustment Scale Perceptions of marital positivity as a function of manual. North Tonawanda, NY: Multi-Health Sys- husband-to-wife physical aggression and marital tems. distress. Journal of Family Violence, 13, 197-212. Stets, J. E. (1988). Domestic violence and control. Levinson, D. (1988). Family violence in cross- New York: Springer-Verlag. cultural perspective. In V. B. Van Hasselt, R. L. Straus, M. A. (1979). Measuring intrafamilial conflict Morrison, A. S. Bellack, & M. Hersen (Eds.), and violence: The Conflict Tactics (CT) scale. Handbook of family violence (pp. 435-455). New Journal of Marriage and the Family, 45, 75-88. York: Plenum. Straus, M. A., Gelles, R. J., & Steinmetz, S. K. Lips, H. M. (1991). Women, men and power. (1980). Behind closed doors: Violence in the Mountain View, CA: Mayfield. American family. New York: Anchor Press/ Locke, H. J., & Wallace, M. (1959). Short marital Doubleday. adjustment and prediction test: Reliability and Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics (2nd ed.). New York: Harper- validity. Marriage and Family Living, 21, 251-255. Collins. Marshall, L. L. (1992). Psychological abuse: Viola- Tolman, R. M. (1989). The development of a measure tion, harassment, battering. Unpublished manu- of psychological maltreatment of women by their script, University of North Texas. male partners. Violence and victims, 4, 159-177. Murphy, C. M., Meyer, S. L., & O'Leary, K. D. Vera Institute of Justice. (1977). Felony arrests: Their (1993). Dependency characteristics of partner as- prosecution and disposition in New York City's saultive men. Journal of Abnormal Psychology, courts. New York: Author. 103, 1-7. Walker, L. (1984). The battered woman syndrome. Navran, L. (1967). Communication adjustment in New York: Springer. marriage. Family Process, 6, 173-184. O'Kelly, C. G., & Carney, L. S. (1986). Women and Received October 7,1997 men in society: Cross-cultural perspectives on Revision received August 11, 1998 gender stratification. Belmont, CA: Wadsworth. Accepted August 12,1998 •