Demand-Withdraw Patterns in Marital Conflict in the Home

Abstract

The present study extended laboratory-based findings of demand-withdraw communication into marital conflict in the home and further explored its linkages with spousal depression. U.S. couples (N = 116) provided diary reports of marital conflict and rated depressive symptoms. Hierarchical linear modeling results indicated that husband demand-wife withdraw and wife demand-husband withdraw occurred in the home at equal frequency, and both were more likely to occur when discussing topics that concerned the marital relationship. For both patterns, conflict initiator was positively linked to the demander role. Accounting for marital satisfaction, both demand-withdraw patterns predicted negative emotions and tactics during marital interactions and lower levels of conflict resolution. Spousal depression was linked to increased likelihood of husband demand-wife withdraw.

Communication in close relationships largely determines whether the relationship is healthy for partners (Kiecolt-Glaser, Glaser, Cacioppo, & Malarkey, 1999), is satisfying (Story, Rothman, & Bradbury, 2002), and endures over time (Fincham & Beach, 1999). A communication pattern that receives particular attention is demand-withdraw, in which one partner attempts to discuss a problem, while the other avoids the issue or ends the discussion (Christensen, 1988). This pattern – evidenced typically during disagreements and problem-solving discussions (Sevier, Simpson, & Christensen, 2004) – ranks among the most destructive and least effective interaction patterns in couples’ problem-solving communication repertoires (Heavey, Layne, & Christensen, 1993). Numerous studies have linked couples’ demand-withdraw to relationship dysfunction (Christensen & Shenk, 1991; Eldridge, Sevier, Jones, Atkins, & Christensen, 2007) and individual maladjustment (Malis & Roloff, 2006). Furthermore, relationship therapists have noted the salience of this pattern to impaired couple functioning (Markman, Stanley, & Blumberg, 1994; Shoham & Rohrbaugh, 2002).

Relationship theorists have posited that particular interpersonal patterns may underlie the established connections between relationship dysfunction and individual well-being, in particular, depression (Davila, 2001; Rehman, Gollan, & Mortimer, 2008). Marital conflict and depression have been associated repeatedly (Whisman, 2001), but gaps remain in understanding how specific forms of conflict interactions are linked with depression (Du Rocher Schudlich, Papp, & Cummings, 2004). Greater specification of patterns of behavior is especially important to understanding links between depression and conflict at a process-oriented level of analysis. More specifically, certain communication patterns, including demand-withdraw, may hinder the resolution of problems and expression of support while increasing partners’ anger and other negative emotions, thereby keeping partners engaged in a reciprocal cycle of relationship dysfunction and individual distress. Despite the array of negative implications that demand-withdraw as a problematic interpersonal communication pattern holds for partners’ relationships and individual functioning, we know little about how demand-withdraw is actually expressed during day-to-day conflicts and whether spousal depression is associated with demand-withdraw communication. Accordingly, the goal of the present study is to extend this knowledge by examining couples’ expressions of demand-withdraw in marital conflict that occurs in their homes in relation to other within-conflict processes (such as initiator, topics, emotions, tactics, and resolution) and spousal depression.

Demand-Withdraw Patterns

Demand-withdraw occurs in one of two patterns between marital partners, in which one partner is the demander, seeking change, discussion, or resolution of an issue, while the other partner is the withdrawer, seeking to end or avoid discussion of the issue. Christensen and colleagues have demonstrated repeatedly that the woman demand-man withdraw pattern is more common than man demand-woman withdraw (Christensen, Eldridge, Catta-Preta, Lim, & Santagata, 2006; Christensen & Heavey, 1990). Others have shown that husbands and wives are both more likely to be in the demander role when discussing an issue they desire to change or address (Klinetob & Smith, 1996). This gender hypothesis has received much attention due to its treatment implications for helping couples who display these negative interaction patterns (Vogel, Murphy, Werner-Wilson, Cutrona, & Seeman, 2007). Nevertheless, previous investigations of spouses’ gender differences in demanding and withdrawing have relied on data collected either through questionnaires or through observed discussions in a laboratory setting.

Several lines of research support continued investigation of demand-withdraw as a unique communication process. First, distressed couples are more likely to express the pattern than non-distressed couples; demand-withdraw also has long-term relationship satisfaction implications (Guay, Boisvert, & Freeston, 2003). Thus, improving our understanding of demand-withdraw patterns may aid clinical efforts to identify and treat couples who are entrenched in this destructive pattern (Shoham & Rohrbaugh, 2002). Next, studies show that expression of demand-withdraw is generally not attributable to personality (Heaven, Smith, Prabhakar, Abraham, & Mete, 2006), suggesting it is open to change and likely to vary across contexts. In an empirical test, Caughlin and Huston (2002) examined whether demand-withdraw, indexed by retrospective questionnaire ratings, was a separate factor from general negative behaviors, such as yelling, partners expressed over the course of a week. Their results revealed that general negativity and demand-withdraw were positively related, but separate constructs. Overall, existing research encourages investigation of demand-withdraw patterns as distinct from other negative behaviors, but how the constructs co-vary within the context of everyday conflicts in the home remains unexamined.

Methods of Assessing Demand-Withdraw Patterns

Research on demand-withdraw patterns to date has relied on one of two methodological approaches, self-report questionnaires and laboratory-based observations of couples’ behavior (Eldridge & Christensen, 2002). First, with regard to the questionnaire approaches, the Communication Patterns Questionnaire (CPQ; Christensen & Sullaway, 1984) asks couples to rate the extent to which 35 symmetrical and asymmetrical interaction patterns occur in their relationship conflict. As an example, items tap how often one tries to start a discussion while the other withdraws, trying to avoid a discussion. Demand-withdraw subscales on the CPQ include woman demand-man withdraw, man demand-woman withdraw, and a sum of both. As another example, on the Initiator Style Questionnaire (ISQ; Denton & Burleson, 2007), participants rate their own and their partner's initiating and avoiding responses to relationship problems (e.g., “I usually keep my feelings about our relationship private”). We reverse-scored avoiding responses so the resultant summed scores reflect an individual's rating of their own and their partner's initiator tendencies.

The second commonly utilized assessment of demand-withdraw involves coding observed couple behavior in the laboratory, typically following the Couple Interaction Rating System (CIRS; Heavey, Gill, & Christensen, 1996; Heavey et al., 1993). The CIRS captures expressions of demand-withdraw in laboratory-based conflict or problem-solving discussion (Sevier et al., 2004). In this coding system, trained coders rate each partner along dimensions of blaming, pressures for change, withdrawing, avoidance, and engagement. Each partner receives a subsequent score on demanding communication (i.e., average of blaming and pressures for change) and withdrawing communication (i.e., average of avoidance, withdrawal, and the reverse score of engagement). These data can further provide individual and couple measures of demand-withdraw patterns. Numerous research studies have supported the reliability, validity, and generalizability of the CIRS to multiple sample types (e.g., partners experiencing violence or alcoholism). Although less commonly utilized, trained coders have also identified avoiders and initiators by rating descriptions of general conflict tendencies partners provide in semi-structured interviews (Denton, Burleson, Hobbs, Von Stein, & Rodriguez, 2001). These classifications were reliably associated with spouses’ physiological reactivity assessed in the laboratory (Denton et al., 2001). Yet although both self-report and observational ratings of couples’ demand-withdraw have supported the validity of one of the most important relational communication patterns and its implications for partners (Caughlin & Vangelisti, 2000), we know little about how couples actually express demand-withdraw in the context of marital conflict in the home.

Spousal Depression and Demand-Withdraw Patterns

Spousal depression has emerged as a particularly robust correlate of impaired partner communication and relationship maladjustment (Davila, Karney, Hall, & Bradbury, 2003; Gotlib & Whiffen, 1989). Identifying specific communication patterns that play a role in the relationship dysfunction-individual well-being linkage may be important to prevention and treatment efforts (Davila, 2001). Research on expressions of individual behaviors (i.e., not patterns) has linked elevated spousal depression to greater use of negative marital conflict behavior, including both withdrawal and demanding, controlling for marital functioning (Du Rocher Schudlich et al., 2004; Jackman-Cram, Dobson, & Martin, 2006). To the extent that a spouse with higher levels of depressive symptoms is more likely to use either withdrawal or demanding in conflict, the partner may be more likely to respond in turn to withdrawal by pursuing, or to demanding by withdrawing, given the cyclical nature of demand-withdraw patterns. In other words, elevated spousal depression places one partner at a higher likelihood for using demand or withdrawal, which, in turn, may elevate the couples’ use of demand-withdraw patterns. Furthermore, preliminary evidence using non-partnered individuals indicates that men and women's symptoms of depression were positively correlated with retrospective ratings of wife demand-husband withdraw communication, but not with husband demand-wife withdraw (Uebelacker, Courtnage, & Whisman, 2003); investigation of this question among couples in the home awaits replication.

The Present Study

For reasons Bolger and colleagues (Bolger, Davis, & Rafaeli, 2003; Laurenceau & Bolger, 2005) outlined, the diary methodology is uniquely suited to testing demand-withdraw in relation to other within-conflict processes. First, diaries permit investigation of whether established findings from laboratory settings replicate in naturalistic contexts or environments. Investigation of demand-withdraw in the home is particularly needed given that constraints of a lab setting may make it difficult to withdraw physically from problem-solving discussions (Bodenmann, Kaiser, Hahlweg, & Fehm-Wolfsdorf, 1998). Next, diary methods facilitate microlevel analysis of the process of interest. Specifically, the current study employed a home-reported diary procedure that captures co-occurring within-conflict correlates such as topics, emotions, behaviors, and resolution, in line with findings that conflict is a multi-dimensional process (Cummings, 1998; Cummings & Davies, 2002; Feldman, 1979). These aspects of conflict predict global relationship satisfaction as well as individuals’ well-being (Kurdek, 1995; Papp, Goeke-Morey, & Cummings, 2007). In addition, diary methodologies encourage participants to record the process of interest soon after it actually occurs. Thus, the diary method addresses problems of the available questionnaires, which heavily rely on partners’ memory of past conflicts or evaluations of hypothetical conflicts. Next, assessment of demand-withdraw in marital conflict in the home using a diary method would provide information that is distinct yet complementary to traditional methods used in this area. For example, while partners’ ratings of demand-withdraw communication in their relationship as a whole provide a foundation for continued study of this pattern, diary methods afford more detailed information, including the frequency of how often partners express this pattern on a daily basis. Finally, relating spouses’ depression to demand-withdraw captured with a diary methodology supports our aim of testing the potential role of demand-withdraw communication patterns in the reciprocal linkage between marital dysfunction and spouses’ psychological maladjustment.

Finally, with regard to controls needed for more precise interpretation of findings, marital satisfaction and demand-withdrawal patterns may be interrelated. As an example, Caughlin (2002) documented concurrent associations between demand-withdraw communication (both observed and self-reported) and spouses’ marital satisfaction. Marital satisfaction may also partially account for how couples handle their differences in the home. Accordingly, all analyses controlled for both partners’ marital satisfaction levels, thereby elucidating how demand-withdraw is uniquely related to within-conflict processes and spousal depression.

Hypotheses

Hypothesis 1

Taking advantage of the availability of home-reported data on actual demand-withdrawal interactions, we had predictions about when a partner would be more likely to initiate demand-withdraw patterns in the home and the relationship themes most likely to involve this pattern. Specifically, we expected demand-withdraw to be linked with which spouse desires change or initiated the issue (Heavey et al., 1993; Klinetob & Smith, 1996), such that we predicted wife demand-husband withdraw to be more likely when discussing an issue wives initiated and husband demand-wife withdraw to be more likely when discussing an issue husbands initiated. In a recent study, Eldridge and colleagues (2007) found demand-withdraw to occur more often during discussions of intimate relationship problems than personal problems. Thus, we further hypothesized that conflicts concerning marital-themed topics (i.e., intimacy, communication, commitment, habits, personality) would include greater likelihood of demand-withdraw patterns, whereas conflicts concerning such issues as other relationships (i.e., children, friends) might be less likely to include demand-withdraw patterns.

Hypothesis 2

The use of demand-withdraw patterns in general correlates with problematic relationship outcomes (Caughlin & Huston, 2002; Rehman & Holtzworth-Munroe, 2006), including partner hostility and aggression (Holtzworth-Munroe, Smutzler, & Stuart, 1998; Sagrestano, Heavey, & Christensen, 1999). As such, we expected that the use of these patterns in the home would be associated positively with problematic conflict expressions (i.e., emotions of anger, sadness, fear, and tactics of threat, physical distress, verbal hostility, and aggression), and in a negative direction to more constructive conflict expressions (i.e., positive emotion and tactics of affection, support, problem solving, compromise, and apology). Although less explored in previous research on general demand-withdraw patterns, we tentatively predicted that expression of demand-withdraw in the home would be associated with lower levels of conflict resolution because this interaction pattern engages partners in behaviors that we did not expect to lead to the successful resolution or handling of differences.

Hypothesis 3

Finally, on the basis of initial reports linking men and women's depressive symptoms to expressions of demand-withdraw (Uebelacker et al., 2003), we predicted that spouses’ depressive symptoms would be positively associated with the use of demand-withdraw patterns in marital conflict in the home. Although past research is limited, we drew on theoretical foundations that link partner depression to relationship dysfunction (Davila, 2001; Rehman et al., 2008) to tentatively expect both husband and wife depression to be related to greater use of both husband demand-wife withdraw and wife demand-husband withdraw patterns.

Method

Sample and Procedures

The current study includes a sample of 116 couples who completed diary ratings of instances of marital conflict occurring at home as part of their first year of participation in a longitudinal investigation of family relationships and child development. Couples resided in the area surrounding a private university located in a relatively rural Midwestern area of the United States. According to the 2000 U.S. Census, this county was 80.8% White, 11.1% Black, 4.6% Latino, 1.3% Asian, and 2.2% multiracial and other races. Of the study's husbands, 105 (90.5%) were Caucasian, seven (6.0%) were African American, and one (0.9%) was biracial. Of the study's wives, 105 (90.5%) were Caucasian, seven (6.0%) were African American, and one (0.9%) was biracial. Three husbands and three wives did not provide this demographic data. Husbands’ ages ranged from 25 to 57 years (M = 40 years, SD = 6 years) and wives’ ages ranged from 25 to 50 years (M = 38 years, SD = 5 years). On average, couples had been married for 13 years (SD = 5.5 yrs) and had 2 to 3 (range = 1-6) children. Spouses completed an item that asked respondents to indicate which of 6 categories captured their approximate yearly combined family income. Husbands’ and wives’ reports of family income were not significantly different, t(115) = 0.58, p = .57. Husbands reported that one couple earned less than US$10,000 per year, four between US$10,001-25,000, 20 between US$25,001-40,000, 52 between US$40,001-65,000, 24 between US$65,001-80,000, and 15 earned more than US$80,000.

The use of a community-based convenience sample was consistent with the broader investigation's aims of assessing typical and problematic family relationships and child development. We recruited participating families through letters sent home with children from local schools; postcards mailed to community residents; referrals from other participating families; flyers distributed at churches and community events; and newspaper, television, and radio advertisements. The university's committee for the protection of human subjects approved the project, and we obtained informed consent. Families attended two private laboratory sessions scheduled approximately 15 days apart lasting 2 to 2.5 hours each. We describe procedures relevant to the present study below. During the first laboratory session, we taught husbands and wives to complete a checklist-style diary of marital conflict (described below). Couples then completed these records following each instance of marital conflict over a period of 15 days, and returned the completed records during the second laboratory session. During the first laboratory visit, couples completed the demographics questionnaire and marital quality and depression questionnaires used in the present analyses. Couples received monetary compensation (US$100) for their participation.

Measures

Marital Conflicts in the Home

As part of their first of two laboratory visits, a graduate-level research assistant led husbands and wives through a 1-hour training protocol designed to instruct them when and how to complete diary reports of marital conflicts that occurred in the home. Consistent with a goal of the larger study to examine daily interparental differences that occur in families’ homes, we defined marital conflict “as any major or minor interparental interaction that involved a difference of opinion, whether it was handled in a mostly negative or even mostly positive way.” Thus, every diary reflected a conflict of some sort between the spouses. During the training, the research assistant provided oral, written, and video examples of the conflict expressions of interest. We tested husbands and wives on their ability to identify the conflict behaviors in the lab, and both were able to do so (Papp, Cummings, & Goeke-Morey, 2002). We clarified all questions during the training and provided spouses with written instructions to bring home. During the designated 15-day reporting period, husbands and wives separately completed diary records at home following each naturally-occurring instance of marital conflict; thus, completion of multiple diaries per day was possible. See Cummings, Goeke-Morey, and Papp (2003) for a full description of the contents of the diary and training protocol.

Couples brought their completed diaries when they returned for a second laboratory visit. We tested husbands and wives again on their ability to identify the diary behaviors and were both still able to identify the intended behaviors (Papp et al., 2002). Both husbands and wives provided diary ratings of marital conflict in the home during the 15-day reporting period (H: range = 0-41 interactions, M = 11.06, SD = 8.41; W: range = 1-53 interactions, M = 14.13, SD = 10.39). Six husbands indicated that no conflicts occurred in the home and we therefore excluded them from subsequent analyses.

Conflict initiator and topics

For each conflict instance, spouses rated whether the husband (0) or wife (1) initiated it. Spouses also indicated (0 = not endorsed, 1 = endorsed) the topic(s) of the conflict instance, including habits, relatives, leisure, money, friends, work, chores, children, personality, intimacy, commitment, and communication. Thus, spouses could endorse multiple topics during a single interaction. For analysis purposes, we created composites of conflict topics identified in past research (Cummings, Goeke-Morey, & Papp, 2004): Child (children), Marital (intimacy, communication, commitment, habits, personality), Social (relatives, leisure, friends, chores), and Work (work, money). See Table 1 for descriptive statistics of these composite variables used in subsequent analyses. Wives reported that 36.1% of conflicts in the home included child themes, 44% included marital relationship themes, 49.4% included social themes, and 30.6% included work themes. Husbands reported that 33.3% of conflicts in the home included child themes, 38.8% included marital relationship themes, 49.9% included social themes, and 29.9% included work themes.

Table 1.

Linkages between Demand-Withdraw Patterns in Marital Conflict in the Home and Topics

			Husband demand-Wife withdraw		Wife demand-Husband withdraw
Topic	M	SD	γ 10	t-value	γ 10	t-value
Husband report (N = 110 husbands / 1,284 conflict instances)
Child	0.33	0.47	-0.154	-1.02	0.082	0.48
Marital	0.53	0.79	0.398	4.18*	0.457	3.29*
Social	0.58	0.65	-0.156	-1.70	0.079	0.34
Work	0.35	0.57	0.167	1.35	0.339	2.42
Wife report (N = 116 wives / 1,638 conflict instances)
Child	0.36	0.48	-0.125	-0.93	-0.342	-3.06*
Marital	0.62	0.85	0.410	5.36*	0.645	8.56*
Social	0.60	0.69	-0.391	-4.26*	0.229	2.11
Work	0.36	0.58	0.355	3.31*	-0.069	-0.60

Note. Analyses control for husbands’ and wives’ marital satisfaction scores. M and SD presented for topic composites: Child (child); Marital (intimacy, communication, commitment, habits, personality); Social (relatives, leisure, friends, chores); and Work (work, money).

^*p ≤ .013 (.05/4).

Conflict emotions and tactics

Spouses reported the emotions and tactics they and their partners displayed during and at the end of conflict between them. Spouses rated their own and their partners’ emotions of positivity, anger, sadness, and fear, during and at the end of interactions on scales ranging from 0 (none) to 9 (high). Spouses also indicated (0 = not endorsed, 1 = endorsed) tactics they and their partners expressed throughout marital conflict in the home (i.e., physical affection, verbal affection, support, problem solving, defensiveness, pursuit, withdrawal, threat, physical distress, verbal hostility, personal insult, physical aggression toward an object, physical aggression toward a person, change topic, compromise, and apology).

For the present analyses, we summed emotions and tactics that could have occurred during and at the end of conflict. Also, we combined tactics that overlapped empirically and theoretically: Specifically, we summed verbal affection and physical affection to create an affection score, and we summed physical aggression toward a person and physical aggression toward an object to create an aggression score. As a final step, we summed husbands’ and wives’ expressions to parallel the dyadic communication pattern of demand-withdraw. We examined husbands’ and wives’ reports of the following dyadic conflict expressions (presented with their possible ranges) in relation to demand-withdraw patterns in the home: positivity (0-36), anger (0-36), sadness (0-36), fear (0-36); affection (0-4), support (0-2), problem solving (0-2), threat (0-2), physical distress (0-2), verbal hostility (0-4), aggression (0-4), compromise (0-2), and apology (0-2).

Conflict resolution

For each conflict instance, spouses rated by answering on scales that ranged from 0 (not at all) to 9 (completely), “How much was the problem solved at the end for you?” and “How much was the problem solved at the end for your spouse?”

Coding demand-withdraw in marital conflict in the home

To capture the demand-withdraw process, we coded conflict instances as including demand-withdraw when different spouses in the same interaction expressed demand and withdraw tactics. We indexed demand on the diary by tactics of pursuit (defined as “just not letting it go; not wanting to drop it, even if the other person wants it to stop; nagging; or following the other person when they walk away”) and personal insult (defined as “insulting your spouse; saying something hurtful about your spouse; or hurting your spouse's feelings with something you say, including making accusations, name calling, put downs, blaming, rejecting, using sarcasm”). We indexed withdraw on the diary by tactics of defensiveness (defined as “trying to avoid blame or responsibility by justifying yourself; making excuses; saying things like ‘Yes, but...’; defending your position; cutting your spouse off instead of listening; responding to a criticism/complaint with a criticism/complaint; or protecting your point of view”), change topic (defined as “changing the topic to avoid the interaction”), and withdraw (defined as “creating a physical or emotional distance between you and your spouse; for example, leaving the room/house, refusing to talk to your spouse, turning away, avoiding eye contact, silent treatment, backing away, asking to stop the interaction, not responding, or pretending not to care”). Thus, each interaction received a husband demand-wife withdraw code of 1 if, in that interaction, spouses rated the husband as demanding (i.e., pursuing, using personal insult) and the wife as withdrawing (i.e., withdrawing from the discussion, changing topic of the discussion, using defensiveness). All other interactions received a husband demand-wife withdraw code of 0. In addition, each interaction received a wife demand-husband withdraw code of 1 if, in that interaction, spouses rated the wife as demanding and the husband as withdrawing. All other interactions received a wife demand-husband withdraw code of 0.

Based on 110 husbands’ ratings of 1,284 conflicts, husband demand-wife withdraw occurred in 5.6% (n = 72) of conflicts, and wife demand-husband withdraw occurred in 3.7% (n = 49) of conflicts. Based on 116 wives’ ratings of 1,638 conflict instances, husband demand-wife withdraw occurred in 5.6% (n = 91) of conflicts, and wife demand-husband withdraw occurred in 6.8% (n = 111) of conflicts.

Depression

We assessed symptoms of depression using the Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). The CES-D is a 20-item scale designed to measure depression in the general population, with an emphasis on affective and somatic symptoms (Shafer, 2006). Respondents indicated how frequently the listed depressive symptoms have bothered them during the past week on a scale ranging from 0 (less than one day) to 4 (5 days). Higher resultant scores indicate higher levels of depressive symptoms, with scores of 16 and above reflecting potentially diagnosable depression (e.g., Pandya, Metz, & Patten, 2005). The CES-D has well-established psychometric properties, including high internal consistency, test-retest reliability, and convergent validity with clinical and self-report measures of depression (Radloff, 1977). In the current study, Cronbach's indices of internal consistency (α) were .88 for husbands and .92 for wives. Average CES-D scores were 8.53 (SD = 7.57) and 9.49 (SD = 8.96) for husbands and wives, respectively. Approximately 16% of husbands and 18% of wives indicated potentially diagnosable depression.

Covariate: Marital Satisfaction

Husbands and wives reported their global marital satisfaction by completing the Marital Adjustment Test (MAT; Locke & Wallace, 1959). The MAT is a widely-used 15-item instrument aimed at capturing levels of relationship satisfaction or adjustment. Respondents answered questions concerning areas of disagreement, relationship compatibility, and overall satisfaction with the relationship. Possible scores range from 2 to 158, with higher scores indicating greater satisfaction and scores below 100 indicating marital distress. The MAT has demonstrated good content and concurrent validity (Locke & Wallace, 1959). In the current study, MAT scores ranged from 50.83 to 156 (M = 110.94, SD = 21.95) for husbands and 46 to 150 (M = 111.76, SD = 23.20) for wives. Thirty percent of husbands and 26.7% of wives reported marital distress, with nearly 13% of couples consisting of two maritally-distressed partners. Husbands’ and wives’ marital satisfaction levels were negatively associated with the amount of demand-withdraw the couple expressed, such that spouses with higher levels of marital satisfaction reported lower summed use of demand-withdraw patterns as occurring in conflict in the home (r_H = -.217, r_W = -.404, ps < .03).

Analytic Plan

When relating demand-withdraw patterns to within-conflict correlates and spouses’ depression, we utilized hierarchical linear modeling (HLM; Raudenbush & Bryk, 2002) to accommodate the multilevel structure of the data. Specifically, within-person diary ratings comprised Level 1 and between-person variables (i.e., spouses’ marital quality scores) comprised Level 2. An advantage of HLM is that it accommodates multiple types of outcomes. In subsequent analyses, standard HLM modeled continuous dependent variables (e.g., ratings of conflict resolution), whereas hierarchical generalized linear modeling (HGLM) modeled binary variables (e.g., occurrence of husband demand-wife withdraw, with a value of 0 or 1) and count variables (e.g., occurrence of apology, with a value of 0, 1, or 2).

We used variants of the equations below, which test whether marital-theme topics predict husband demand-wife withdraw:

Level 1 Model:

$$\begin{array}{cc}\hfill & \text{Prob}(\text{husband demand-wife withdraw}=1\mid \beta )=\Phi \hfill \\ \hfill & \text{Log}\left[\Phi ∕1\text{-}\Phi )\right]={\beta }_0+{\beta }_1\left(\text{Marital Topic}\right)\hfill \end{array}$$

where the two equations indicate a Bernoulli model for a binary outcome, β₀ is the intercept and is interpreted as the average likelihood of husband demand-wife withdraw reported by each respondent, and β₁ is the average likelihood of husband demand-wife withdraw in conflicts in which couples discuss marital topics.

Level 2 Model:

$$\begin{array}{cc}\hfill & {\beta }_0={\gamma }_{00}+{\gamma }_{01}\left(\text{Husband MAT}\right)+{\gamma }_{02}\left(\text{Wife MAT}\right)+{\mathrm{U}}_0\hfill \\ \hfill & {\beta }_1={\gamma }_{10}+{\mathrm{U}}_1\hfill \end{array}$$

where the γs are Level 2 coefficients and U₀ and U₁ are random person effects. The Level 2 model provides an average of all respondents’ intercepts and slopes or the grand mean of the likelihood of husband demand-wife withdraw (γ₀₀), the grand mean of husbands’ and wives’ marital satisfaction related to the likelihood of husband demand-wife withdraw (γ₀₁ and γ₀₂, respectively), and the grand mean of the likelihood of husband demand-wife withdraw in conflicts in which couples discuss marital-themed topics (γ₁₀). Accordingly, the parameter of interest in this sample analysis is the γ₁₀ coefficient, or the association between discussing marital issues and the likelihood of husband demand-wife withdraw during marital conflict in the home, net of the associations between both spouses’ marital quality ratings and demand-withdraw.

Results

Demand-Withdraw Patterns in Marital Conflict in the Home: Links with Conflict Initiator and Topics

Consistent with expectations (Hypothesis 1), the initiator (0 = husband, 1 = wife), or which partner raised the conflict issue, was linked to demand-withdraw patterns, such that when husbands initiated, husband demand-wife withdraw was more likely to occur than not (H report: γ₁₀ = -1.532, t = -6.08, p < .001; W report: γ₁₀ = -1.028, t = -6.46, p < .001) and when wives initiated, wife demand-husband withdraw was more likely to occur than not (H report: γ₁₀ = 0.955, t = 5.13, p < .001; W report: γ₁₀ = 1.043, t = 6.52, p < .001).

Also consistent with Hypothesis 1, HGLM analyses indicated that conflicts concerning marital relationship issues were more likely to include both husband demand-wife withdraw and wife demand-husband withdraw patterns than not, according to both spouses’ reports (see Table 1). Wives also reported that discussing social topics in marital conflict in the home was associated with lower likelihood of husband demand-wife withdraw and that discussing work topics was associated with higher likelihood of husband demand-wife withdraw. Discussing children in marital conflict in the home was not reliably associated with either spouse's ratings of demand-withdraw patterns (Table 1).¹

Demand-Withdraw Patterns in Marital Conflict in the Home: Links with Conflict Emotions, Tactics, and Resolution

As shown in Table 2 (husbands’ diaries) and Table 3 (wives’ diaries), demand-withdraw patterns in marital conflict in the home consistently related in the predicted directions to other dyadic conflict emotions and tactics (Hypothesis 2). Both husband demand-wife withdraw and wife demand-husband withdraw patterns were related to lower levels of positivity and to higher levels of anger and sadness, according to husbands’ (Table 2) and wives’ (Table 3) diary reports. Husbands’ and wives’ indicated that wife demand-husband withdraw only was linked to higher levels of fear in marital conflict.

Table 2.

Linkages between Demand-withdraw Patterns in Marital Conflict in the Home and Conflict Emotions and Tactics: Husband Report (N = 110 husbands / 1,284 conflict instances)

			Husband demand-Wife withdraw		Wife demand-Husband withdraw
Dyadic Conflict	Descriptive Statistics		γ 10	t-value	γ 10	t-value
Positivity	20.35	9.75	-6.867	-6.82*	-8.895	-7.83*
Anger	6.87	8.18	7.995	6.21*	8.772	6.02*
Sadness	3.89	6.89	0.373	3.03*	0.711	8.89*
Fear	1.39	3.80	0.212	0.95	0.667	3.24*
Affection	0.44	1.00	-0.408	-1.36	-1.339	-3.48*
Support	0.45	0.74	-0.933	-3.73*	-1.754	-3.50*
Problem Solving	0.80	0.92	-0.901	-3.93*	-0.861	-3.40*
Threat	0.03	0.17	1.089	3.29*	1.064	3.47*
Physical Distress	0.06	0.27	0.508	1.71	0.267	0.92
Verbal Hostility	0.39	0.78	1.037	6.10*	1.099	6.54*
Aggression	8.0	0.6%	1.756	3.83*	1.298	1.91
Compromise	0.85	0.94	-0.678	-3.91*	-1.164	-4.44*
Apology	0.16	0.42	0.393	2.53	0.904	4.14*

Note. Analyses control for husbands’ and wives’ marital satisfaction scores. Descriptive statistics include M and SD for continuous variables and frequency and % for binary variables.

^*p ≤ .004 (.05/14).

Table 3.

Linkages between Demand-withdraw Patterns in Marital Conflict in the Home and Conflict Emotions and Tactics: Wife Report (N = 116 wives / 1,638 conflict instances)

			Husband demand-Wife withdraw		Wife demand-Husband withdraw
Dyadic Conflict	Descriptive Statistics		γ 10	t-value	γ 10	t-value
Positivity	20.50	10.24	-10.124	-10.30*	-10.281	-10.93*
Anger	6.85	8.46	8.962	9.07*	8.422	8.21*
Sadness	3.73	6.78	0.615	4.55*	0.687	5.97*
Fear	1.51	4.45	0.326	2.59	0.488	3.00*
Affection	0.47	1.04	-0.370	-1.75	-1.025	-3.83*
Support	0.46	0.72	-1.593	-6.09*	-1.208	-5.71*
Problem Solving	0.81	0.92	-0.915	-4.94*	-0.842	-6.22*
Threat	49.0	3.0%	1.466	6.70*	1.273	5.90*
Physical Distress	0.11	0.34	0.636	2.78	0.747	4.02*
Verbal Hostility	0.42	0.79	0.914	6.60*	1.107	8.59*
Aggression	0.01	0.12	1.761	4.17*	2.037	5.72*
Compromise	0.78	0.93	-0.590	-3.43*	-1.059	-6.14*
Apology	0.14	0.39	0.688	3.63*	0.431	2.51

Note. Analyses control for husbands’ and wives’ marital satisfaction scores. Descriptive statistics include M and SD for continuous variables and frequency and % for binary variables.

^*p ≤.004 (.05/14).

In terms of associations between demand-withdraw patterns and other tactics used in marital conflict in the home, husband demand-wife withdraw was linked to less use of support, problem solving, and compromise, on the one hand, and to greater use of threat, verbal hostility, and aggression, on the other hand, according to husbands’ (Table 2) and wives’ (Table 3) diary reports. Wives reported that husband demand-wife withdraw was also associated with more use of apology in marital conflict in the home (Table 3). The pattern of wife demand-husband withdraw was associated with less use of affection, support, problem solving, and compromise, and greater use of threat and verbal hostility, across both husbands’ and wives’ diary reports (see Tables 2 and 3, respectively). Wives further reported associations between wife demand-husband withdraw and increased use of tactics of physical distress, and aggression (Table 3), while husbands reported an association between wife demand-husband withdrawal and greater use of dyadic apology in marital conflict in the home (Table 2).

Consistent with predictions set forth in Hypothesis 2, husbands and wives, respectively, reported that husband demand-wife withdraw was associated with lower conflict resolution for both husbands (γ₁₀ = -1.661, t = -4.42, p < .001; γ₁₀ = -2.324, t = -7.25, p < .001) and wives (γ₁₀ = -1.253, t = -3.21, p = .002; γ₁₀ = -2.210, t = -8.70, p < .001). Husbands and wives, respectively, also both indicated that wife demand-husband withdraw was linked with lower conflict resolution for husbands (γ₁₀ = -1.371, t = -3.32, p = .002; γ₁₀ = -2.11, t = -7.01, p < .001) and wives (γ₁₀ = -2.069, t = -5.10, p < .001; γ₁₀ = -2.467, t = -8.90, p < .001).

Spousal Depression and Demand-Withdraw Patterns in Marital Conflict in the Home

The final analysis explored associations between spouses’ depression symptoms and the likelihood of expressing demand-withdraw patterns in marital conflict in the home (Hypothesis 3). HGLM analyses indicated that, controlling for both partners’ marital satisfaction levels, husband and wife depression symptoms were both linked to greater reported likelihood of expressing husband demand-wife withdraw in marital conflict in the home (H report: γ₀₁ = 0.043, t = 2.10, p < .05; W report: γ₀₁ = 0.022, t = 2.14, p < .05). Husband and wife depression symptoms were not related to the reported likelihood of using wife demand-husband withdraw in marital conflict (H report: γ₀₁ = 0.0004, t = 0.02, p > .05; W report: γ₀₁ = 0.003, t = 0.25, p > .05). Thus, the results partially supported Hypothesis 3.

Discussion

The present study extended our understanding of demand-withdraw communication into the context of marital conflict that occurs in families’ homes. This investigation of demand-withdraw in day-to-day marital conflict contributed to the existing literature – which, to date, has been based on couples’ self-reports of global tendencies and laboratory-based observations – in several ways. First, the present results indicated that, indeed, couples do express demand-withdraw in marital conflict that occurs in the home, although at relatively low frequencies. The comparatively narrow definition of demand-withdraw employed in the present study may contribute to the low frequencies. Whereas laboratory-based coding includes initiating as a behavior in the demand dimension, we considered it as a separate within-conflict process. Thus, our results likely represent conservative estimates. Another explanation for the low occurrence of demand-withdraw in the home aligns with the relatively high levels of positivity expressed during everyday marital conflict and with observations that couples utilize a wide range of emotions and tactics when resolving marital differences in the home (Cummings et al., 2003).

Interestingly, both husband demand-wife withdraw and wife demand-husband withdraw patterns were displayed at nearly equal frequencies, a finding that counters others’ demonstrations that wife demand-husband withdraw is more commonly expressed (Christensen & Heavey, 1990). Three possible reasons for this inconsistency merit discussion. First, fuller consideration of the diary methodology utilized may offer one insight. In the present study, spouses rated their own and their partners’ use of demand and withdraw behaviors along with a list of other possible conflict tactics. The reporters received no indications that we expected the two behaviors to be inter-connected. Methods such as the CPQ and CIRS, which both contain multiple items about who withdraws and who pursues, may invoke for both respondents and coders, respectively, the commonly held notions that women hold “pursuer” roles and men hold “avoider” roles. Thus, if couples do express relatively equal expressions of husband demand-wife withdraw and wife demand-husband withdraw patterns in the home, the diary method likely captured this due to its unbiased assessment strategy. A second reason for our observing relatively equal rates of husband demand-wife withdraw and wife demand-husband withdraw patterns in the home may lie in the types of conflicts examined. In the present study, couples reported all inter-spousal differences that occurred in the home, whether handled in a mostly positive or mostly negative way. Problem-solving discussions in a laboratory setting, however, may elicit couples’ more difficult or negative issues, especially given that researchers often instruct couples to discuss such issues in laboratory-based communication studies. It is also possible that wives raise more weighty issues (i.e., dealing with the marital relationship), which therefore may include higher rates of wife demand-husband withdraw. Third, although our rates of marital dissatisfaction are similar to those documented in other community samples (e.g., Whisman & Bruce, 1999), it is possible that the level of distress in the present sample contributes to this finding. These tentative explanations await further investigation.

Despite their relatively infrequent occurrences, both husband demand-wife withdraw and wife demand-husband withdraw patterns were reliably related to within-conflict processes. Including multiple dimensions of marital conflict permitted examination of demand-withdraw in relation to interparental differences in the home broadly defined (Cummings, 1998). As expected, the present results revealed that the conflict initiator, or which partner raised the conflict issue, reliably predicted demand-withdraw patterns. Specifically, husband-initiated conflict in the home was more likely to include the husband demand-wife withdraw pattern, while wife-initiated conflict was more likely to include the wife demand-husband withdraw pattern. These findings align with lab-based results that indicate that one way partners attempt to address their own issues is to pursue or push for change during relational disagreements, even against their partner's wish for non-involvement (Klinetob & Smith, 1996).

Also, certain topics of marital conflict were linked to higher and lower likelihood of expression of demand-withdraw in marital conflict in the home. Couples consistently demonstrated greater use of rigid and deadlock patterns when their disagreements or desired changes concerned their marital relationship, while disagreements about other children, friends, and events were not likely to invoke such communication patterns. These findings suggest that spouses may be more motivated or better equipped to address changes that concern people outside of the marital relationship. Wives also reported that discussing family finances invoked the husband demand-wife withdraw pattern, perhaps pointing to couples’ previous experiences of money disagreements as especially heated or unsolvable. Our findings that one partner pursues certain areas of disagreement while the other partner withdraws warrant further study as potentially polarizing issues with long-term implications for relationship functioning.

Both husband demand-wife withdraw and wife demand-husband withdraw patterns related to dyadic conflict tactics and emotions in the predicted directions. Demand-withdraw patterns were consistently related to greater likelihood of negative tactics (i.e., threat, physical distress, verbal hostility, aggression) and higher levels of negative emotions (i.e., anger, sadness, fear), and to lower likelihood of constructive tactics (i.e., affection, support, problem solving, compromise) and lower levels of positivity. It is worth noting that these associations revealed unique within-conflict linkages between demand-withdraw and other conflict expressions, accounting for associations between spouses’ marital satisfaction levels and the dyadic conflict expressions. Although we originally expected demand-withdraw patterns to be associated with reduced likelihood of couples’ use of apology in marital conflict in the home, a post-hoc interpretation of this finding merits consideration: Couples were more likely to express demand-withdraw communication along with other negative tactics that prompted eventual apology, providing demand-withdraw more statistical opportunity to co-vary with apology. Notably, apology is weakly and inconsistently related to family members’ perceptions of marital conflict resolution in the home (Goeke-Morey, Cummings, & Papp, 2007). Demand-withdraw patterns were also reliably associated with lower levels of conflict resolution.

The present results also contribute to theories that support reciprocal associations between relationship dysfunction and partner well-being by revealing associations between a specific type of problematic interpersonal communication (i.e., demand-withdraw) and spouses’ symptoms of depression. Interestingly, both husbands’ and wives’ depressive symptoms were associated with greater likelihood of the husband demand-wife withdraw pattern in conflict in the home, whereas depression was not linked to wife demand-husband withdraw. One possibility is that husbands are more motivated to engage these patterns when they or their partners express elevated depressive symptoms, which, to them, may reflect a greater potential for marital relationship problems. At the same time, these results differ from those in Uebelacker and colleagues (2003), who reported positive associations between men and women's depressive symptoms and wife demand-husband withdraw only. It is difficult to compare the findings directly as Uebelacker and colleagues analyses used men and women's (not participating as couples) retrospective ratings of demand-withdraw. Nevertheless, both sets of findings encourage study of demand-withdraw communication and depression over time among partners to discern causal pathways that link the processes.

Limitations

Several limitations warrant consideration. First, the present analyses were cross-sectional in nature and, therefore, temper our ability to discuss causal processes. For example, demand-withdraw occurring in one conflict may reflect a previous conflict that remained unresolved. We were not able to test the directions of effects in the present study due to the low frequency of repeated demand-withdraw expressions within couples. Such temporal hypotheses may be more feasible to test using a clinic-based sample that is more likely to have frequent demand-withdraw expressions throughout a reporting period (Eldridge et al., 2007). In addition, the diary methodology still relies on partners’ self-reports, which may introduce bias as well as inflate statistical associations with depression due to method overlap. Nonetheless, the similarity in husbands’ and wives’ findings strengthens our confidence in the diary results.

Second, our probability sample of couples was ethnically homogeneous and, therefore, limits the general applicability of the findings. Although questions about demand-withdraw in the home among culturally-diverse families remain unexplored, recent cross-cultural studies have supported associations between demand-withdraw communication and marital distress among Pakistani couples living in Pakistan and the U.S. (Rehman & Holtzworth-Munroe, 2006) and among couples from Brazil, Taiwan, and Italy (Christensen et al., 2006). Encouraged by the present study's results, investigation of demand-withdraw in the home and its within-conflict correlates awaits replication with demographically similar populations as well as samples of couples who are more diverse ethnically as well as along dimensions of relationship distress, aggression, and clinical levels of psychological disorders.

Implications

The results of the present study offer implications for clinical treatment of both relationship distress and depression. First, given the consistent likelihood that conflicts including demand-withdraw patterns included fewer positive emotions and tactics, more negative emotions and tactics, and lower levels of conflict resolution, couples who express demand-withdraw are at heightened risk for experiencing a cycle of increasingly hostile and unresolved conflicts. This cycle, in turn, may be linked to the development and maintenance of relationship distress. It follows that couples with already distressed relationships may be more likely to engage in demand-withdraw patterns during marital conflict in the home. Further, for couples who are also parents in a family context with children, as were the couples in the present study, other family and child processes are also placed at increased risk for maladjustment (Cummings, Davies, & Campbell, 2000).

In applying the present results to the treatment of spousal depression, we caution clinicians that depressed individuals are more likely to engage in this negative communication pattern, particularly husband demand-wife withdraw, thereby adding stress to couples in which one partner is already experiencing symptomatic distress. Further, demand-withdraw patterns of depressed spouses may be linked to broader relationship processes. As an example, demand-withdraw patterns might indicate struggles concerning relational power – or the process of how couples make important decisions. In a study comparing communication of couples with a wife with depression, couples with a wife with panic disorder with agoraphobia, and non-affected control couples, both partners in the depressed-wife couples reported more demand-withdraw expressions (on the CPQ) compared to the other groups (Byrne, Carr, & Clark, 2004). Although Byrne and colleagues’ analyses did not examine separate patterns of demand-withdraw, the authors suggested that couples with a depressed wife may express more demand-withdraw for two possible reasons: the wives demand due to a depressive belief that change would only happen through nagging or shouting, or the wives withdraw due to a depressive belief that change was not possible. It follows that couples may utilize demand-withdraw when depression compromises effortful problem-solving resources (e.g., realizing multiple solutions to a problem), thereby attempting to restore their relational power balance when addressing problematic areas in marriage and conflicts.

In sum, couples may find themselves stuck in unsuccessful or disrupted communication patterns when they are either dissatisfied in their relationship or when one or both partners exhibit elevated depressive symptoms. In fact, poor interpersonal communication is one factor that maintains a bi-directional association between relationship distress and partners’ maladjustment. This study elucidated ways in which a specific problematic communication pattern, demand-withdraw, is implicated in such a cycle. Specifically, demand-withdraw communication that occurred in marital conflict in the home was more likely to be expressed along with negative emotions and destructive tactics and less likely to be expressed along with positivity and constructive tactics. Demand-withdraw in conflicts was also related to lower levels of resolution. Taken together, demand-withdraw communication may underlie a cycle of increasingly negative and hostile conflicts that are likely to recur as challenges for couples. Identifying ways to reduce demand-withdraw communication may eventually prevent or reduce couples’ risk of experiencing a reciprocal linkage between relationship distress and partner maladjustment.