Nonverbal Behavior in Couple Relationships: Exploring the Causes and Consequences of Withdrawal

Nonverbal Behavior in Couple Relationships: Exploring the Causes and Consequences of Withdrawal

The importance of communication in close relationships is high-lighted by Wood's assertion that 'communication is not only a cen­tral, generative process of intimacy, but is actually what we experience as relationships' (1995, p.125). Although communica­tion usually consists of both verbal and nonverbal channels, there is evidence that the nonverbal channels may be particularly crucial to relationship processes and outcomes (Gottman, Markman, & Notarius, 1977; Noller, 1984). Burgoon and Dillman (1995) suggest that 'nonverbal relational messages signal how participants regard each other, their relationship and themselves in the relationship.' Similarly, Watzlawick and his colleagues (Watzlawick, Beavin & Jack-son, 1967) suggest that communication involves two levels of mean­ing: the content and relationship levels. The content level involves the literal meaning of the words that are spoken. In contrast, the relation-ship level conveys important information about how the partners are feeling about each other. This relational information is generally con­veyed nonverbally, and can modify the meaning of the words. Our goals in this chapter are to review the literature on the interrelated topics of nonverbal behavior in close personal relationships and

withdrawal in couple interactions, and to report on three of our em­pirical studies in this area.

NONVERBAL COMMUNICATION, POWER, INTIMACY,
AND RELATIONSHIP SATISFACTION

Patterson (1983) has argued that nonverbal communication has five important functions: providing information, regulating interaction, expressing intimacy, exercising social control, and facilitating task or service goals. Of these five functions, the present chapter focuses par ticularly on expressing intimacy and exercising social control. We see these two functions as central to close personal relationships. Burgoon and Dillman (1995) have argued that knowing who wields power, and how that power is expressed through nonverbal behavior, is central to understanding any given relationship. Nonverbal commu­nication is a rich source of power-related messages, as expressed through such behaviors as physical appearance (e.g., 'power dressing', height), touch, gaze, body movements and spacing (Guerrero, Andersen, & Miff, 2001). Similarly, these authors have proposed that feelings of intimacy, including love, passion and interpersonal warmth, lie at the heart of intimate relationships. A study by Gonzaga, Keltner, Londahl and Smith (2001) provides evidence for nonverbal displays of love, including head nods, forward lean, and Duchenne (candid as opposed to contrived) smiles.

Given that expressions of power and intimacy are fundamental to personal relationships, nonverbal behaviors that create psychological closeness or distance between partners are of special interest to re-searchers and clinicians. Such behaviors, termed immediacy or in­volvement cues, have been described by Andersen (1985) as actions that signal warmth, communicate availability, decrease psychological and/or physical distance, and promote involvement in interactions. Immediacy behaviors include gaze, close distance, lean, body orienta tion, touch and smiling. Conversely, a lack of these behaviors is likely to indicate distance or coolness in the relationship.

Despite the key relational roles played by nonverbal expressions of power and intimacy, it is important to recognize that nonverbal com­munication is a relatively unreliable system. That is, there is no dictio­nary that neatly and unambiguously defines the meaning of a given nonverbal behavior. In fact, Manusov (2002, p. 15) has noted that 'one of the most intriguing aspects of nonverbal communication is its abil ity to be interpreted in myriad ways.' There are several reasons for the ambiguity of nonverbal behavior. First, relational messages tend to in volve multiple rather than isolated cues. For this reason, any given cue needs to be understood in the context of other nonverbal cues that may be present (Burgoon & Dillman, 1995).

Second, variables such as relationship satisfaction appear to influ­ence both the specific perceived function of a partner's nonverbal com-

munication and the response to it. For example, couples who are happier with their relationships tend to make more relationship-en­hancing attributions for their partner's nonverbal behavior (Manusov, 1990). Similarly, happy couples are more likely to notice positive part­ner behaviors, and to interpret these behaviors in a positive way (Manusov, Floyd, & Kerssen-Griep, 1997).

Third, another contextual factor that impacts on the understanding of nonverbal behavior is gender. There is considerable evidence that men and women differ on many aspects of their nonverbal displays. For example, women tend to use immediacy cues such as eye contact to express affiliation more than do men. There is also evidence that non-verbal behavior is likely to be interpreted differently, depending on whether it was enacted by a male or female. Burgoon, Coker, and Coker (1986) found that when a male engaged in high levels of eye con tact, this behavior was interpreted as dominance, but when a female behaved similarly, the behavior was seen as submissive. This finding suggests that the interpretation of nonverbal behavior is affected by gender stereotypes.

Finally, attachment security is also likely to affect expressions of in­timacy and power in close relationships. This proposition is sup-ported by the work of Tucker and Anders (1998), who observed dating couples while they discussed positive aspects of their relationships. These researchers found that secure attachment was associated with more emotional expressivity; that is, higher levels of gaze, touch, smil ing, and laughing. Conversely, preoccupied attachment was associated with lower levels of touch and smiling, and avoidant attachment was associated with lower levels of gaze, touch, and smiling. In another study of the interactions of dating couples, Simpson and his col-leagues observed partners interacting while the female member of the dyad believed that she was about to take part in a stressful experiment (Simpson, Rholes, & Nelligan, 1992). In this situation, women's non-verbal responses to their partner's touch varied depending on their at tachment security. Specifically, secure women responded by engaging in touching and kissing, whereas avoidant women tended to resist physical contact. These links between attachment security and non-verbal behavior are likely to reflect the relational goals of the different attachment styles, especially with regard to intimacy and affection.

Despite the ambiguity of much nonverbal behavior, there is clear ev idence that particular patterns of nonverbal behavior are linked to re lationship satisfaction. That is, relationship satisfaction is likely to be influenced by nonverbal behavior (although, as already noted, rela­tionship satisfaction also affects perceptions of nonverbal behavior). In particular, unhappy couples display more negative nonverbal be­haviors than happy couples. They display less positive emotion, and tend to exchange many more negative nonverbal cues than do those who are more satisfied with their relationships (Burgoon, Buller, & Woodall, 1996). More specifically, Gottman (1996) has argued that

happy couples tend to display five positive behaviors for each negative behavior, whereas unhappy couples have a much lower positive-to­negative ratio. Researchers have also shown that unhappy couples are more likely to reciprocate negative behaviors than are happy couples (Gottman et al., 1977; Pike & Sillars, 1985). These findings highlight the importance of promoting positive nonverbal behaviors between intimate partners.

NONVERBAL BEHAVIORS RELATED TO WITHDRAWAL,
POWER AND INTIMACY

In this chapter, we have chosen to illustrate issues of nonverbal behav ior, power and intimacy, through studies of withdrawal in couple inter-actions. Withdrawal in couple interactions may reflect a lack of intimacy, or attempts to control and manipulate the partner. In either case, withdrawal, by its very nature, tends to involve an absence of nonverbal immediacy or involvement cues (as already noted, the lack of such cues creates psychological distance or lack of closeness be­tween partners).

Withdrawal in marital interaction, particularly in response to con­flict, has long been a topic of interest to researchers and to clinicians who work with distressed couples (Christensen, 1988; Fogarty, 1976; Napier, 1978; Wile, 1981). Withdrawal may be subtle (as in be-coming silent, looking away, changing the topic or diverting atten­tion), or more blatant (as in storming out of the room, or refusing to talk). Because of its consequences, withdrawing during conflict is generally seen as a negative behavior (Fruzzetti, 1996); specifically, issues are not resolved and may cause further conflict, and the per-son who tries to raise the issue may become resentful and angry. This anger and resentment may eventually lead to coldness and distance between the partners, and a reduced level of intimacy. Withdrawal may also lead to ongoing power struggles; as Holtzworth-Munroe, Smutzler and Stuart (1998) have noted, 'a pattern of coercive efforts can gradually develop, creating a rigid pattern of negative, polarized interaction' (p. 732). Similarly, in Blake and Mouton's (1964) model of interpersonal communication, withdrawal is seen as reflecting low concern for self (an unwillingness or inability to clearly express one's own needs and issues) and low concern for the relationship (an un­willingness to work at resolving conflict for the sake of the relation-ship). On the other hand, some researchers have argued that withdrawal from conflict is not necessarily negative, if partners are not introspective, and if they share a strong bond of mutual affection (Raush, Barry, Hertel, & Swain, 1974).

In many couples, a more common pattern than mutual withdrawal is what has become known as demand/withdraw, where one member of the couple wants to talk about the issue and becomes demanding, and the partner withdraws and is unwilling to discuss the issue. Ac-

cording to Schaap and his colleagues (Schaap et al., 1988), the more demanding one partner becomes, the more the other partner is likely to withdraw.

There is evidence for a gender linkage to this interaction pattern, with females being more likely to be the demanders and males more likely to be the withdrawers (Christensen, 1988; Christensen & Heavey, 1990). Gottman and Levenson (1988) have argued that males are more likely to withdraw in the context of intense conflict, because of the high level of physiological arousal they experience in such situa tions. Although both husbands and wives are more likely to demand on issues where they want change and more likely to withdraw on is-sues where the partner wants change, males are more likely to with-draw overall, and particularly likely to withdraw when the female's issue is being discussed (Christensen & Heavey, 1990). In this situa tion, males' withdrawal may be a power play: Refusal to discuss the is sue inhibits its resolution and maintains the status quo. As Klinetob and Smith (1996) have noted 'Because of their different motivations, women demand in order to bring about change, whereas men withdraw in order to avoid change' (p. 946).

A follow-up study (Heavey, Layne & Christensen, 1993) replicated the finding that it is particularly when wives' issues are being dis­cussed that wives demand and husbands withdraw. These researchers also found that wife-demand/husband-withdraw interaction predicted a decline in wives' marital satisfaction over a one-year period, whereas husband-demand/wife-withdraw predicted an increase in wives' mar­ital satisfaction over the same period. In a further study, Heavey. Christensen, and Malamuth (1995) replicated this finding with a dif­ferent sample followed over a period of 2.5 years.

These studies by Christensen and his colleagues were designed to test the social structure perspective, which proposes that gender dif ferences in conflict patterns are a function of men's and women's place in the social structure (rather than individual differences). Although they were able to show the predicted pattern for discussion of wives' is sues, the pattern for discussion of husbands' issues was less clear (see above). Klinetob and Smith (1996) criticized Christensen and Heavey's (1990) study, arguing that limiting the discussion topics to parenting behavior (generally considered the purview of women) could have promoted more of the traditional wife-demand/husband-with­draw pattern than would have occurred if a wider range of topics were allowed. These researchers, using both self-report and observational methodologies, were able to obtain a full reversal of behavior patterns across topics. In other words, there was more wife-demand/husband­withdraw when the wife's issue was being discussed, and more hus­band-demand/wife-withdraw when the husband's issue was being discussed.

Christensen and Shenk (1991) tested the possibility that the de­mand-withdraw pattern is related to partners being discrepant in their

desires for intimacy and independence; that is, the demander seeks greater intimacy and the withdrawer seeks greater independence. As predicted, these researchers found significant correlations between discrepancies in desired independence and reports of wife-demand/ husband-withdraw communication.

In another analysis of the factors linked to destructive conflict pat terns, Newton, Kiecolt-Glaser, Glaser and Malarkey (1995) tested the links between hostility, defensiveness and spouses' communication behavior. They found that high levels of hostility and low levels of de­fensiveness predicted destructive conflict engagement for husbands (criticism, disagreement, interrupting partner), whereas for wives, these same variables predicted withdrawal. The researchers specu­lated that hostility may be expressed in a 'hot' way (expressive and arousing) by husbands, but in a 'cold' way (distant and aloof) by wives. Wives' withdrawal could then be seen not as a way of avoiding issues, but as a different way of expressing hostility.

Roberts (2000) has discussed three different types of withdrawal. Angry withdrawal is seen as an expression of negative affect, whereas conflict avoidance involves 'withdrawal from the conflict without re­jection of the partner' (p. 696). The third type of withdrawal, which Roberts labels intimacy avoidance, involves withdrawal following a partner's intimate disclosure. In a questionnaire study, all three types of withdrawal were related to concurrent marital distress for both husbands and wives.

We now report on three studies from our own work on interaction patterns involving withdrawal. In the first study, we explore the non-verbal concomitants of withdrawal in couple interactions. In the sec­ond study, we explore the associations between attachment security and withdrawal in the context of relationship-centered anxiety. In the third study, we use time-series analysis to compare couples in violent and nonviolent relationships in terms of their withdrawal.

STUDY I : THE NONVERBAL CONCOMITANTS OF WITHDRAWAL

The first study reported here follows on from the work of Christensen and his colleagues on the demand-withdraw pattern of conflict interac­tion, and was carried out by Noller and Christensen (unpublished, 1991). The goal of this study was to explore the nonverbal concomitants of the demand-withdraw pattern, although this chapter focuses specifically on the nonverbal behaviors that accompany with drawal. We expected that withdrawal would be related to low levels of nonverbal behaviors that reflect involvement and immediacy, such as expressiveness and gaze, and to use of avoidance behaviors such as head down and head turn (Patterson, 1983).

The married couples involved in the study engaged in two conflict discussions, one involving an issue of dissatisfaction reported by the

wife, and the other involving an issue of dissatisfaction reported by the husband. Undergraduate research assistants coded the videotaped discussions for the presence or absence of 18 nonverbal behaviors us­ing 15-second time intervals. The behaviors coded included a wide range of nonverbal behaviors (e.g., gaze; open and closed smiles; open, neutral and closed gestures; eyebrow and head movements). A differ­ent group of undergraduate assistants made global ratings of demand­ing and withdrawing for each member of each couple, after watching the entire interaction. The Conflict Rating System (Heavey, Layne, & Christensen, 1993) was used for these ratings. These global ratings were then correlated with the total frequencies of the relevant nonverbal behaviors for each spouse.

The top section of Table 9.1 presents the significant correlations of spouses' nonverbal behaviors with global ratings of withdrawal for wives' issues, and the lower section presents the correlations for hus­bands' issues. As can be seen from the table, the clearest pattern of nonverbal behavior was for husbands withdrawing during discus­sion of their wives' issues. This pattern was characterized by a lack of

Correlations Between Nonverbal Behaviors and Ratings of Withdrawal For Husbands and
Wives' on Wives' and Husbands' Issues

Note: Only correlations significant for at least one partner are included in this table. p<.05*; p<.01**

open gestures and gaze, and by head down and head turn. Wives were rated as withdrawing on their own issue when they used few open ges tures, and withdrawing on their husband's issue when they used few neutral gestures and engaged in head down. Husbands were rated as withdrawing on their own issue when they engaged in head down. Hence the single behavior most characteristic of ratings of with­drawal was head down. It is also interesting to note that when we cor­related husbands' and wives' withdrawal on their own issue with their withdrawal on their partner's issue, there was a strong correla­tion for husbands, r = .70. In other words, husbands' withdrawal was quite consistent, irrespective of whose issue was being dis­cussed. In contrast, there was no such consistency for wives.

There was also evidence, particularly when couples discussed hus bands' issues, that partners' nonverbal behaviors were linked to each other's withdrawal (see Table 9.1). Husbands' open gestures were as sociated with wife withdrawal, and wives' head down and head shake were associated with husband withdrawal. Of course, without time-se ries analysis, we can not be sure about the sequential order involved in these effects.

Analyses of variance with sex, issue and marital satisfaction as the independent variables showed that the frequencies of the nonverbal behaviors were affected by all three of these variables. Wives used more head shake than husbands. There was more gaze on wives' is-sues than husbands' issues, and wives used more head down on their own issues than on husbands' issues. In addition, open gestures were used more by high satisfaction spouses on husbands' issues than on wives' issues, and wives used more open gestures on their own issues than on husbands' issues. High satisfaction spouses used less head down than low satisfaction spouses. These findings are consistent with highly satisfied spouses being more open and less withdrawn in their discussions.

STUDY 2: ATTACHMENT SECURITY AND WITHDRAWAL
DURING CONFLICT

In recent years, several studies have explored the association between dimensions of attachment security (comfort with closeness and anxi ety over abandonment) and withdrawal during conflict. Withdrawal is more common in relationships where the wife is insecure, than in rela­tionships involving two secure partners (Feeney, Noller, & Callan, 1994; Senchak & Leonard, 1992). Links between wives' anxiety over abandonment and the demand/withdraw pattern of interaction have been found both concurrently and longitudinally (Feeney et al., 1994; Fitzpatrick, Fey, Segrin, & Schiff, 1993). In addition, wives high in anx iety over abandonment have been shown to use avoidance strategies in response to marital conflict (Feeney et al., 1994). Links between dis­comfort with closeness and withdrawal/avoidance have been less con-

sistent, with some researchers finding no links (Feeney et al., 1994; Levy & Davis, 1988; Pistole, 1989) and others finding significant links (Fitzpatrick et al., 1993).

Roberts and Noller (1998) found evidence that communication pat-terns mediate the association between attachment and couple vio­lence. In other words, attachment insecurity seems to affect violence through couples' destructive communication patterns. These re-searchers found that demand/withdraw communication mediated the association between anxiety over abandonment and violence for both males and •females. These authors note:

These results are consistent with a situation in which one partner's fear of being abandoned leads to the development of destructive patterns of communication within a relationship, such as one partner making de­mands while the other withdraws, which, in turn, fosters an environ­ment in which couple violence is more likely to occur. (p. 337)

Our second study follows on from the work of Raush and his col-leagues (1974), and was designed to assess the link between attach­ment security and withdrawal in the context of relationship-centered anxiety Relationship-centered anxiety (that is, anxiety about the na­ture and viability of the couple relationship) is likely to be particularly problematic for insecure individuals; paradoxically, these individuals may react in hostile or avoidant ways, which may threaten their rela­tionships and hence exacerbate their insecurities.

The study employed an observational methodology. Couples who had been dating for at least 12 months engaged in three interactions: one in which the male partner was asked to act cold and distant (show low levels of immediacy behaviors) and the female partner was asked to try to reconcile; one involving a reversal of those roles; and one in­volving an issue-based interaction about leisure activities (Feeney, 1998). Our report here focuses on the 'partner-distant' interaction, which was expected to create relationship-centered anxiety for the individuals involved.

Emotional reactions were rated using 7-point scales assessing five emotions: anger, anxiety, sadness, disgust and happiness. Nonverbal behaviors were coded using a scheme based on the one used by Simpson and his colleagues (Simpson et al., 1992). Behaviors coded in­cluded hugging, holding hands, turning head or body toward or away from partner, and resisting contact. Factor analysis was used for each construct, to reduce the number of variables. Two attachment factors were obtained from questionnaire items: Comfort with closeness and Aiixiety over abandonment; these are the two factors most commonly found in measures of romantic attachment. Two negative affect factors were found: Worry and hostility, and there were also two factors of non-verbal behavior: Touch (e.g., hugging, holding hands) and avoidance (turning head or body away from partner, resisting contact).

Analyses revealed several significant correlations among attach­ment scales, affect and nonverbal behavior, although these were gener ally modest in size. For females, comfort with closeness was inversely related to the measures of negative affect (r = -.23, p < .05 for hostile affect and r = -.30, p < .01 for worried affect), and anxiety over aban donment was related to nonverbal behaviors, with anxious females en-gaging in high levels of avoidance (r = .38, p < .01) and low levels of touch (r = .23, p < .05).

For males, comfort with closeness was related to high levels of touch (r = .23,p < .05) and low levels of avoidance (r = -.30,p < .05), but at­tachment dimensions were unrelated to ratings of affect. Interestingly, other measures and analyses (beyond the scope of this chapter) indi­cated that males' anxiety over abandonment was related to their de­structive verbal behavior (such as coercion) in response to partner's distancing.

For both genders, satisfaction with the relationship was negatively correlated with avoidance in the female-distant scene in response to partners' primed distancing behaviors, r = -.23,p < .05 for males and r = -.29, p < .05 for females. Thus, those in more satisfying relation-ships were less likely to respond with avoidance behaviors when the partner acted cold and distant.

These findings illustrate the roles of both attachment security and relationship satisfaction in predicting low levels of withdrawal/avoid­ance. Consistent with a number of other studies (e.g., Kirkpatrick & Davis, 1994), the relevance of partner attachment dimensions de­pended on gender. Males who were uncomfortable with closeness, and females who feared abandonment, responded to partners' distancing behaviors by withdrawing physically from the partner. Further, fe­males who were uncomfortable with closeness were rated as display­ing both worry and hostility in the face of partners' distancing. Interestingly, the results suggest that insecure and distressed partners are likely to act in ways that increase the distance between them, rather than create a climate in which reconciliation can occur.

STUDY 3: WITHDRAWAL IN THE INTERACTIONS
OF VIOLENT COUPLES

Little is known about the use of withdrawal by violent couples, al-though Murphy and O'Farrell (1997) found that behaviors related to withdrawal and avoidance discriminated between couples in which the husband was alcoholic and violent and those in which the husband was alcoholic but not violent. Specifically, more withdrawal occurred where the husband was violent.

Holtzworth-Munroe, Smutzler, and Stuart (1998) reported two studies assessing the links among marital distress, violence, and de­manding and withdrawing during conflict. Study 1 relied on hus­bands' self-reports, and found that distressed groups reported

significantly higher levels of wife-demand!husb and-withdraw than did the nonviolent nondistressed group. In addition, violent groups re-ported more husband-demand/wife-withdraw communication than did nonviolent groups. The second study compared the demand and withdraw interactions of violent and nonviolent couples, based on the coding of videotaped interactions. The violent distressed group tended to engage in the most withdrawal, and the nonviolent nondistressed group in the least withdrawal.

Roberts and Krokoff (1990) used time-series analysis to examine withdrawal in couple interactions. They explored sequence and pat­terning in the interactions of distressed and nondistressed couples, fo­cusing on three variables: withdrawal/involvement, hostility/friend­liness and displeasure/pleasure. The strongest finding was that in dis­satisfied marriages, the husband's withdrawal predicted the wife's hostility, suggesting that the destructive process seems to be initiated by the husband's withdrawal (rather than by the wife's demanding). In this study, there were no differences between satisfied and dissatisfied couples in terms of their overall use of withdrawal.

Klinetob and Smith (1996) also used time-series analysis of obser­vational data, and found that demand and withdraw behaviors were temporally associated across spouses during the course of a discus­sion. These researchers also explored the direction of influence, but found that the results were more complex than suggested by Roberts and Krokoff's (1990) findings: Patterns of dependency between one partner's demanding and the other partner's withdrawal varied de-pending on whose issue was being discussed.

In our third study, we compared four groups of couples in terms of their use of withdrawing behaviors: a satisfied nonviolent group, a dis tressed nonviolent group, a distressed violent group, and a satisfied vi­olent group (Noller & Roberts, 2002). Couples engaged in four interactions: Discussion of a serious issue proposed by the wife, a seri ous issue proposed by the husband, a trivial issue, and the sharing of a recent disappointment or sadness. Only discussions of the serious conflicts will be reported in this chapter. These videotaped interac­tions were coded by trained coders for expressions of affect and communication behaviors.

We found men used more withdrawal overall than women, espe­cially if they were in unhappy relationships. In addition, when the fe­male's issue was being discussed, men in unhappy relationships withdrew more than women in unhappy relationships. We also found that withdrawal was used more during the female's than the male's is-sue by couples in violent distressed relationships.

In this study, time-series analysis was also used. Two physiological measures were included: Inter-beat interval and galvanic skin re­sponse. Participants also reported on their level of experienced anxi­ety. Using these data, Noller and Roberts were able to explore the links between anxiety/arousal and withdrawal. Bivariate time-series analy-

sis was used to quantify, in the form of a Z-score, the extent to which one time-series (e.g., heart-rate) accounted for a second time-series (e.g., withdrawal), controlling for the second time-series own past. The number of lags was limited to three (i.e., 30 seconds). Anxiety/arousal in this study involved the average of three measures of the extent to which one variable (e.g., female-anxiety/arousal) could be predicted by another variable (e.g., withdrawal). The Z-scores were then used as the dependent variables in ANOVAs exploring the variables affecting these associations. Measures of behavior such as withdrawal and hostility were averaged across the two discussions (his and her issues).

The anxiety/arousal of men in violent relationships was linked to their later withdrawal (within 30 seconds), supporting Gottman and Levenson's (1988) contention that high levels of arousal are causally related to male withdrawal, although their claim was not specific to vi­olent relationships. The withdrawal of males in violent relationships was also linked to the later anxiety of their female partners, as well as to their female partner's later withdrawal. Thus male withdrawal dur­ing serious conflict is affected by his own anxiety, and in turn, affects the anxiety of his partner, as well as her own withdrawal. In addition, hostility of the male partner was associated with the later withdrawal of his partner.

One problem with this method of data analysis is that, although it can show us the associations between different behaviors, and even tell us the order in which particular behaviors occur, it cannot specify whether the later behavior increases or decreases as a result of the prior behavior. In other words, although this Gottman-Ringland pro­cedure (Williams & Gottman, 1981) has the advantage of controlling for auto-correlation effects, the statistics provide a measure of the strength of the association between two variables, but not the direction of the effect. In the discussion section, we will discuss the most plausi­ble interpretations of these findings, based on theory and previous research.

GENERAL DISCUSSION

Our initial study showed that there are distinctive patterns of nonver­bal withdrawal, and that these behaviors are used particularly by men in distressed relationships while discussing issues raised by their wives. Behaviors such as head down, head turn, lack of gaze and lack of open gestures were particularly characteristic of husbands with-drawing during discussion of their wives' issues. In short, as expected, husbands responded to their wives' pressures for change by engaging in low levels of involvement behaviors. The pattern was not nearly so clear for wives' withdrawal, or for either spouse's withdrawal when husbands' issues were being discussed.

Husbands' withdrawal on their own issues was related to their use of head down, but also to nonverbal behaviors of their wives, particu-

larly head shake and head down. These results suggest that hus­bands may withdraw in response to these signals that their wives are not accepting their arguments or suggestions. Overall, these findings are suggestive of a power struggle, in which husbands respond to their wives' disagreement by refusing to discuss the issue further. There was also some evidence that husbands and wives were involved in mutual withdrawal, both with their heads down, while discussing husbands' issues.

The findings for sex, issue, and marital satisfaction indicate that all three of these variables affect nonverbal behavior, with sex and issue often having interactive effects. Wives seem to be both more open and more withdrawn on their own issues; perhaps they are more open in expressing their feelings but more withdrawn in response to hus­bands' counter-arguments. High satisfaction spouses were generally more open and less withdrawn than dissatisfied spouses, suggesting an emotional climate of openness and intimacy, in which issues are more likely to be aired and resolved.

The second study showed links between withdrawal behavior and both relationship satisfaction and attachment security. Consistent with the first study, partners in more satisfying relationships were less likely to respond to conflict with nonverbal behaviors reflecting avoid­ance. In line with the findings of 'Ricker and Anders (1998), females who were anxious about abandonment tended to engage in high levels of avoidance and low levels of touch. Further, for males, comfort with closeness was associated with high levels of touch and low levels of avoidance. Hence, in the context of a partner acting cold and distant, insecure and distressed individuals seem to act to create more dis­tance, rather than working towards achieving intimacy, understand­ing, or reconciliation.

The finding that both attachment dimensions were related to with­drawal (avoidance) differs from the findings of Simpson et al. (1992), who reported no relations between attachment anxiety and responses to partner distancing. However, it is important to note that Simpson et al.'s study focused on support processes, rather than conflict. The cur-rent findings demonstrate the salience of conflict, especially for those who are anxious about their relationships, and particularly when the conflict centers on the core issues of closeness and distance.

It is important to note that the two attachment dimensions were dif­ferentially related to responses to relationship stress for males and fe­males (Feeney, 1998). For females, negative affect was linked to discomfort with closeness, and nonverbal behavior indicative of dis­tancing was linked to anxiety over abandonment. In contrast, males' nonverbal behavior was affected by their comfort with closeness, with those low in comfort reporting higher levels of avoidance and lower lev­els of touch. Anxiety over abandonment was unrelated to males' affect, but predicted more negative verbal behaviors, such as coercion and aggression. Together, these results suggest that in the face of a part-

ner's distancing, anxiety over abandonment predisposes females to re ciprocate by avoiding intimacy themselves, and predisposes males to respond by exerting power through coercive remarks. Here, as in other studies of romantic attachment (e.g., Feeney et al., 1994), anxious part ners tend to engage in behaviors that exacerbate conflict and distance, and that are likely to bring about the outcomes that they fear most.

In the third study, men were particularly likely to engage in with­drawal during conflict discussions, especially if they were in dis­tressed relationships. Also, withdrawal was used more during discussion of the females' issues, particularly by those in violent rela tionships. When we take into account the findings from the time-series analysis, the sequential pattern becomes clearer. For males in violent relationships, withdrawal was temporally linked to feeling anxious, as argued by Gottman and Levenson (1988). Males' withdrawal was also linked to the later anxiety of the female partner. The most plausible in­terpretation of this finding is that females' anxiety increases when their partners withdraw. In this situation, females may feel powerless because the probability of their issues being resolved decreases; they are also likely to feel distressed at the disruption of intimacy. Roberts and Krokoff (1990) found that wives tended to become hostile after husbands withdrew, suggesting that the termination of the discussion is aversive for wives (however, these researchers had no way of show­ing whether wives' hostility was linked to their anxiety). The with­drawal of females in violent relationships was also linked to the partner's hostility. It seems that females are more likely to withdraw when the husband becomes hostile, perhaps as a way of avoiding conflict, and even violence.

Thus, both males and females in violent relationships tend to be highly reactive to one another's withdrawal, in the sense that they change their own behavior in response to partner withdrawal. They seem to walk a fine line between avoiding violence on the one hand, and avoiding a build-up of resentment and explosive anger on the other. Again, issues related to power and intimacy are likely to underlie the reciprocation of withdrawal behavior in these couples.

In a diary study comparing violent and nonviolent men's reactions to daily stresses, Umberson, Anderson, Williams, and Chen (2003) found that unlike nonviolent men, violent men did not seem to experi­ence changes in their emotional experience in response to daily stresses and relationship conflict. These researchers suggest that rela tionship stresses that elicit negative emotional reactions in nonviolent men may result in violent men repressing their emotions in order to re duce their arousal, and subsequently expressing the ensuing build-up of emotion through violent acts that reassert their masculine identity and sense of power and control. This finding was also supported by an interview study in which men reported on their responses to relation ship stress (Umberson, Williams, & Anderson, 2002). It is important to note that the violent men included in this study were recruited

through a violence program, and this sample is likely to include men involved in serious, rather than 'common couple' violence.

Further research on withdrawal in violent couples should seek to differentiate between different types of withdrawal, as suggested by Roberts (2000), and to examine more closely the links between mascu­linity/femininity, withdrawal, and violence in couples as suggested by Umberson and her colleagues (Umberson et al., 2002, 2003). Another important research question centers on Newton et al.'s (1995) sugges­tion that withdrawal may be different for males and females: With­drawal for husbands may be 'hot' (that is, expressive and arousing, like stomping out of the room), whereas for wives, it may be 'cold' (dis­tant and aloof, such as epitomized in 'the silent treatment'). In both of these contexts, withdrawal was seen as expressing anger, but in very different ways. More research is needed to clarify these issues.

IMPLICATIONS OF THE FINDINGS

The findings of these studies suggest many ways in which therapists may be able to help couples to understand the use and misuse of with­drawal in their conflict interactions. First, discussing models of con­flict behavior (e.g., Blake & Mouton, 1964; Rusbult, 1987) may help couples think more seriously about alternative ways of handling con­flict, and key issues such as their concern for self and their concern for the relationship. Problem solving can be presented as the most con­structive way of dealing with conflict, because it involves concern for both the self and the relationship, and hence promotes both individual and relational adjustment.

Second, couples may need help to understand the possible negative consequences for the relationship brought on by withdrawing during conflict. For example, when withdrawal occurs, conflict issues do not get recognized and aired, and opportunities for achieving greater inti­macy are lost. In addition, partners' needs and desires cannot be in­cluded in a possible solution unless they are expressed in the context of a problem-solving discussion. Clearly, if only one partner makes their desires known, that partner tends to have more power in relation to problem-solving and decision-making. Couples may need help in expressing their needs and concerns in a nonblaming and non-defensive way (both verbally and nonverbally). Couples also need to be made aware that unresolved problems are likely to recur, creating more conflict and distress for the couple. In addition, unresolved con­flict may lead to a build-up of resentment, with one of two possible neg­ative consequences: Either violence as an expression of suppressed rage, or increasing coldness and distance in the relationship.

Third, couples may need assistance to recognize the more subtle signs of withdrawal (such as head down and averted gaze), and begin to talk about their meaning. They may also need to decide what to do when one partner wants to withdraw from discussing an issue. There

are legitimate reasons for withdrawing from or postponing a conflict discussion, such as a lack of information to resolve the issue, or spouses being too tired and/or upset to continue a fruitful discussion. In these situations, partners may need to make a definite 'appoint­ment' for a time when they can consider the issue further.

Fourth, spouses may need help to become more aware of their own withdrawal, the reasons behind it and its implications for the relation-ship. In some cases it may be important to focus on attachment or other issues that may be driving the withdrawal. For example, those who are anxious about abandonment may require help to explore and face their fears and perhaps talk to the partner about his or her com­mitment to the relationship. Clearly, unmet needs for intimacy and se­curity underlie many relationship issues and conflicts.

Let us hasten to add that in focusing on directions for individual and couple therapy, we are not suggesting that teaching people to deal with withdrawal will necessarily solve all problems in their relationships. In fact, some caution should be exercised in relation to the reduction of withdrawal behaviors. For couples who tend to become violent when discussing serious conflict issues, withdrawal may function to defuse tension, at least in the short term. These couples need a focus on con­structive skills for resolving conflict, as well as on reducing with­drawal. It is also important to remember that all relationships are different, and any intervention wi11 need to take into account the specific characteristics of each partner.

We also need to bear in mind that most of the violent couples in our third study were involved in 'common couple' violence (Johnson & Ferraro, 2000), involving pushing, shoving and throwing things, and that the violence was often mutual. These couples are very different from the 'batterers' included in some studies, and may respond posi­tively to couple therapy that helps them deal with their conflicts more constructively. Batterers, in contrast, are generally considered unsuit­able for couple therapy, given the risk of provoking violence through raising serious issues in the therapy session.

Generally, however, helping couples to deal with their issues and re-solve their problems is likely to be the most constructive path. Of course, not all relationship problems are easily resolved, and not all partner behaviors, no matter how annoying they may be, are easily changed. In some situations, partners may need to accept differences between them if they want a satisfying, supportive relationship (Christensen & Jacobson, 2000). As Christensen notes in the preface to his book:

We [Christensen & Jacobson] had developed ways in which partners could experience and accept the normal vulnerabilities that we all have and the very natural incompatibilities that crop up between two unique individuals. With this acceptance, something paradoxical often oc­curred: many of the needs and demands for change evaporated, and

each partner became more amenable to making the changes that were truly important to the other. (p. xiv)

REFERENCES

Andersen, P A. (1985). Nonverbal immediacy in interpersonal communication. In A. W. Siegman & S. Feldstein (Eds.), Multichannel integrations of nonver­bal behavior (pp. 1-36). Hillsdale, NJ: Lawrence Erlbaum Associates.

Blake, R.R., & Mouton, J. S. (1964). The managerial grid. Houston: Gulf Pub­lishing Company.

Burgoon, J. K., Buller, D. B., & Woodall, W. G. (1996). Nonverbal communica­tion: The unspoken dialogue (2nd ed.) New York: McGraw-Hill.

Burgoon, J. K., Coker, D. A., & Coker, R. A. (1986). Communicative effects of gaze behavior: A test of two contrasting explanations. Human Communica­tion Research, 12, 495-524.

Burgoon, J. K., & Dillman, L. (1995). Gender, immediacy and nonverbal com­munication. In P. J. Kalbfleisch & M. J. Cody (Eds.), Gender, power and communication in human relationships (pp. 63-81). Hillsdale, NJ: Law­rence Erlbaum Associates.

Christensen, A. (1988). Dysfunctional interaction patterns in couples. In P. Noller & M. A. Fitzpatrick (Eds.), Perspectives on marital interaction (pp. 31-52). Clevedon and Philadelphia: Multilingual Matters.

Christensen, A., & Heavey, C. L. (1990). Gender and social structure in the de­mand/withdraw pattern of marital conflict. Journal of Personality and So­cial Psychology, 59, 73-81.

Christensen, A., & Jacobson, N. J. (2000). Reconcilable dierences. New York: The Guilford Press.

Christensen, A., & Shenk, J. (1991). Communication, conflict and psychologi­cal distance in nondistressed, clinic and divorcing couples. Journal of Con­sulting and Clinical Psychology, 59, 458-463.

Feeney, J. A. (1998). Adult attachment and relationship-centered anxiety: Re­sponses to physical and emotional distancing. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 189-218). New York: The Guilford Press.

Fleeney, J. A., Noller, P, & Callan, V. J. (1994). Attachment style, communica­tion and satisfaction in the early years of marriage. Advances in Personal Relationships, 5, 269-308.

Fitzpatrick, M. A., Fey, J., Segrin, C., & Schiff, J. L. (1993). Internal working models of relationships and marital communication. Journal of Language and Social Psychology, 12, 103-131.

Fogarty, T. F (1976). Marital crisis. In P J. Guerin (Ed.), Family therapy: The­ory and practice (pp. 325-334). New York: Gardner Press.

Fruzzetti, A. E. (1996). Causes and consequences: Individual distress in the couple interaction. Journal of Consulting and Clinical Psychology, 64, 1192-1202.

Gonzaga, G. C., Keltner, D., Londahl, E. A., & Smith, M. D. (2001). Love and the commitment problem in romantic relations and friendship. Journal of Personality and Social Psychology, 81, 247-262.

Gottman, J. M., & Levenson, R. L. (1988). The social psychophysiology of mar­riage. In P Noller & M. A. Fitzpatrick (Eds.), Perspectives on marital inter-action (pp. 182-200). Clevedon and Philadelphia: Multilingual Matters.

Gottman, J. M., Markman, H. J., & Notarius, C. I. (1977). The topography of marital conflict: A sequential analysis of verbal and nonverbal behavior. Journal of Marriage and the Family, 39, 461-477.

Guerrero, L., Andersen, P, & Afifl, W. (2001). Close encounters: communicat­ing in relationships. New York: McGraw-Hill.

Heavey, C. L., Christensen, A., & Malamuth, N. (1995). The longitudinal im­pact of demand and withdrawal during marital conflict. Journal of Consult­ing and Clinical Psychology, 63, 797-801.

Heavey, C. L., Layne, C., & Christensen, A. (1993). Gender and conflict struc ture in marital interaction: A replication and extension. Journal of Consult­ing and Clinical Psychology, 61, 16-27.

Holtzworth-Munroe, A., Smutzler, N., & Stuart, G. L. (1998). Demand and

withdraw communication among couples experiencing husband violence.

Journal of Consulting and Clinical Psychology, 66, 731-743.

Johnson, M. P, & Ferraro, K. J., (2000). Research on domestic violence in the

1990s; making distinctions. Journal of Marriage and the Family, 62, 948-963. Kirkpatrick, L. A., & Davis, K. E. (1994). Attachment style, gender and rela‑

tionship stability: A longitudinal analysis. Journal of Personality and So‑

cial Psychology. 66. 502-512.

Klinetob, N. A., & Smith, D. A. (1996). Demand-withdraw communication in marital interaction: Tests of interspousal contingency and gender role hy­potheses. Journal of Marriage and the Family, 58, 945-957.

Levy, M. B., & Davis, K. E. (1988). Lovestyles and attachment styles compared: Their relations to each others and to various relationship characteristics. Journal of Social and Personal Relationships, 5, 439-471.

Manusov, V. (2002). Thought and action: connecting attributions to behaviors

in married couples' interactions. In P. Noller & J. A. Feeney (Eds.). Under‑

standing marriage: Developments in the study of couple interaction (pp.

14-31). New York: Cambridge University Press.

Manusov, V., Floyd, K., & Kerssen-Griep, J. (1997). Yours, mine and ours: Mu tual attributions for nonverbal behaviors in couple interactions. Communi­cation Research, 24, 234-260.

Murphy, C. M., & O'Farrell, T. J. (1997). Couple communication patterns of maritally aggressive and nonaggressive male alcoholics. Journal of Studies on Alcohol, 58, 83-90.

Napier, A. Y. (1978). The rejection-intrusion pattern: A central family dynamic. Journal of Marriage and Family Counseling, 4, 5-12.

Newton, T. L., Kiecolt-Glaser, J. K., Glaser, R., & Malarkey, W. (1995). Conflict and withdrawal during marital interaction: The roles of hostility and defen­siveness. Personality and Social Psychology Bulletin, 21, 512-524.

Noller, B, & Christensen, A. (unpublished). Nonverbal behavior and the de­mand/withdraw pattern of marital interaction. University of Queensland

Noller, P., & Roberts, N. D. (2002). The communication of couples in violent and nonviolent relationships: Temporal associations with own and part­ners' anxiety/arousal and behavior. In P Noller & J. A. Feeney (Eds.), Under-standing marriage: Developments in the study of couple interaction (pp. 348-378). New York: Cambridge University Press.

Patterson, M. L. (1983). Nonverbal behavior: A functional perspective. New York: Springer-Verlag.

Pike, G. R., & Sillars, A. L. (1985). Reciprocity of marital communication. Journal of Social and Personal Relationships, 2, 303-324.

Pistole, M. C. (1989). Attachment and adult romantic relationships: Style of conflict resolution and relationship satisfaction. Journal of Social and Per­sonal Relationships, 6, 505-510.

Raush, H. L., Barry, W. A., Hertel, R. K., & Swain, M. E. (1974). Communica­tion, conflict and marriage. San Francisco: Jossey-Bass.

Roberts, L. J. (2000). Fire and ice in marital communication: Hostile and dis­tancing behaviors as predictors of marital distress. Journal of Marriage and the Family, 62, 693-707.

Roberts, L. J., & Krokoff, L. J. (1990). A time-series analysis of withdrawal, hostility and displeasure in satisfied and dissatisfied marriages. Journal of Marriage and the Family, 52, 95-105.

Roberts, N. D., & Noller, P (1998). The associations between adult attachment and couple violence. In J. A. Simpson & W. S. Rholes (Eds.),Attachment the­ory and close relationships (pp. 317-350). New York: The Guilford Press.

Rusbult, C. E. (1987). Responses to dissatisfaction in close relationships: The exit-voice-loyalty-neglect model. In D. Perlman & S. Duck (Eds.), Intimate relationships: Development, dynamics and deterioration (pp. 209-237). Beverly Hills: Sage.

Schaap, C., Buunk, A. P, & Kerkstra, A. (1988). Marital conflict resolution. In P. Noller & M. A. Fitzpatrick (Eds.), Perspectives on marital interaction (pp. 203-244). Clevedon and Philadelphia: Multilingual Matters.

Senchak, M., & Leonard, K. E. (1992). Attachment styles and marital adjust­ment among newlyweds. Journal of Social and Personal Relationships, 9, 51-64.

Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support-seeking and support-giving within couples in an anxiety-provoking situation: The role of attachment. Journal of Personality and Social Psychology, 62, 434-446.

Tucker, J. S., & Anders, S. L. (1998). Adult attachment style and nonverbal closeness in dating couples. Journal of Nonverbal Behavior, 22, 109-124.

Umberson, D., Anderson, K. L., Williams, K., & Chen, M. (2003). Relationship dynamics, emotional state and domestic violence: A stress and masculini­ties perspective. Journal of Marriage and the Family, 65, 233-247.

Umberson, D., Williams, K. L., & Anderson, K. (2002). Violent behavior: A measure of emotional upset. Journal of Health and Social Behavior, 43, 189-206.

Watzlawick, P., Beavin, J., & Jackson, D, (1967). Pragmatics of human com­munication. New York: Norton & Co.

Wile, D. B. (1981). Couples therapy: A nontraditional approach. New York: Wiley. Williams, E., & Gottman, J. M. (1981). A user's guide to the Gottman-Wil‑

liams time-series analysis computer programs for social scientists. New

York: Cambridge University Press.

Wood, J. T. (1995). Relational communication: Continuity and change in personal relationships. Belmont, CA: Wadsworth Publishing.