A Meta-Analytical Review of the Demand/Withdraw Pattern of Interaction and its Associations with Individual, Relational, and Communicative Outcomes

Abstract

This meta-analysis reviews the findings of 74 studies (N = 14,255) examining the associations between the demand/withdraw pattern of interaction and individual, relational, and communicative outcomes. When both the individual behaviors of demanding and withdrawing and the pattern of demand/withdraw are considered collectively, the cumulative evidence indicates a moderate, meaningful relationship between demand/withdraw and overall outcomes (r=.360). Similar overall effect sizes were observed for wife demand/husband withdraw (r=.380) and husband demand/wife withdraw (r=.392), although the overall effect size for demand/withdraw patterns in studies that included distressed/clinical participants (r=.413) was greater in magnitude than that obtained for studies that included nondistressed participants (r=.345). On average, higher correlations were observed for relational (r=.423) and communicative outcomes (r=.418) than for demographic (r=.239) and well-being outcomes (r=.249).

Keywords:

• Demand
• Withdraw
• Conflict
• Communication
• Distress
• Marital Satisfaction
• Well-being

Notes

[1] One reviewer expressed concerns about our use of the label “outcomes” to describe the five broad categories of variables reviewed within our meta-analysis and the causal implications of interpretations surrounding the variables we reviewed. We readily acknowledge that most of the associations reported within this body of work represent reciprocal associations (rather than causal links) between the DM/W pattern and individual, relational, and communicative correlates. For instance, although researchers have generally examined DM/W as a predictor of relational dissatisfaction, it is certainly plausible that dissatisfied partners are motivated to communicate desires for change that lead to DM/W behaviors. Likewise, some personality variables likely serve as antecedents to DM/W behavior (e.g., the Big Five), whereas others may better represent correlates of DM/W behavior (e.g., attachment styles). The category of demographic variables might include correlates of DM/W behavior (e.g., age, socioeconomic status) and/or moderators of DM/W behavior and other outcomes (e.g., biological sex, education discrepancy between partners, divorce status). Rather than specify the precise theoretical role of each set of variables included within our categorization scheme, we used the term “outcomes” to represent the results of statistical analyses for each of the five categories of variables we analyzed within this meta-analysis. This is consistent with previous meta-analytic reports (e.g., Schrodt, Witt, & Messersmith, 2008) and provides a greater degree of parsimony in the reporting of results.

[2] Some of the variables categorized as relational, wellness, personality, demographic, and communication outcomes may also be included in other categories; that is, these groupings are not entirely mutually exclusive. For example, differences in desired intimacy may be classified as both a relational and communication construct (if such differences are expressed). In such instances, the items used to measure each variable (i.e., their operationalizations) were carefully reviewed to determine where the variable fit within the classification scheme. For instance, if differences in desired intimacy were measured as individual preferences for intimacy rather than as expressions of intimacy, we classified them as a relational outcome.