Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Fam Psychol. Author manuscript; available in PMC 2010 August 1.
Published in final edited form as:
PMCID: PMC2782415

The Differential Association between Change Request Qualities and Resistance, Problem Resolution, and Relationship Satisfaction


Although research supports the negative sequelae of the “demand-withdrawal” pattern, research is scant on the impact of “non-demanding” change requests (e.g., specific, increasing, and “we” requests). We hypothesize that such change requests will be associated with less partner withdrawal/resistance, better problem resolution, and greater relationship satisfaction. Seventy-two conversations between couples who were recruited through random digit dialing were coded for change request qualities. Results indicate that wife specific and “we” requests led to less husband resistance, and husband increasing and “we” requests led to less wife resistance. Greater percentages of wife and husband specific and “we” requests were related to better problem resolution in the conversation, and greater percentages of wife specific and “we” requests were related to greater wife satisfaction. Research and clinical implications are detailed.

According to many theories of close relationships (e.g., attachment theory, Hazan & Shaver, 2004; social exchange theory, Thibaut & Kelley, 1959), people are motivated to engage in relationships to meet needs. Over the course of an intimate relationship, each partner’s needs continue to change (Neff & Karney, 2005); communicating these evolving desires increases the probability that those needs will be met, thus maximizing benefits in the relationship (Epstein & Baucom, 2002). However, communicating needs does not guarantee their fulfillment; the dyadic nature of a close relationship creates a situation in which maximizing one partner’s benefits might come at a cost to the other partner (Impett, Gable, & Peplau, 2005). Therefore, the other partner might respond negatively to a request for change (e.g., with resistance or hostility), which in turn causes conflict (Marcus, 2000). Partner conflict has been the heart of a large portion of couple research (e.g., Heyman, 2001) and treatment (e.g., traditional behavioral couple therapy, Jacobson & Margolin, 1979; integrative behavioral couple therapy, Jacobson & Christensen, 1996). Managing conflict has been recognized as a complicated process that has significant impact on the course and state of the relationship (e.g., Cramer, 2002). Asking for change is often the beginning of this process and each request for change presents the partner with a new opportunity to respond; furthermore, the way change is broached influences the remainder of the conflict discussion (e.g., Gottman, Coan, Carrere, & Swanson, 1998). Asking for change of one’s partner, then, is a central process in relationships, and asking for change in a manner that facilitates getting problems resolved may be critical for relationship success. However, it is this start of the chain that has been largely overlooked. Although start-up of the conversation that is negative in valence has been linked to poor conversation and relationship outcomes, a more specific understanding of how to best ask for change is due; for example, specific ways of “softer start-up” (Gottman et al., 1998) may result in nonnegative responses from the partner. Determining the optimal way to begin this process, as opposed to focusing on dynamics within the body of the conflict, may offer more power for meaningful intervention.

Despite the potential importance of this process of asking for change, the empirical basis for what constitutes effective requests is scant. Instead, research has focused on the conflict process as a whole, and has mostly paid attention to desired changes as markers for unresolved conflict, which can be used to observe the conflict process. In observational studies of couples, for example, researchers often assess for problem behaviors within a relationship that one partner seeks to alter, either with a questionnaire (e.g., Areas of Change Questionnaire; Weiss, Hops, & Patterson, 1973) or through interview (e.g., play-by-play interview; Gottman, 1996). These assessments have been used to elicit topics for video recorded problem-solving conversations to observe couples in conflict in hundreds of studies (see Heyman, 2001), but none have focused specifically on the various ways couples ask one another to change and their differential impacts. As a result, we still know very little about how to request change in a way that promotes closeness, effective problem resolution, and relationship satisfaction.

Instead of attempting to uncover effective approaches of asking for change, the extant research regarding change requests merely describes the possible negative sequelae of this process when it is ineffective. Labeled the “demand/withdrawal” sequence, this change pattern discriminates between distressed and non-distressed couples (Christensen & Heavey, 1993). Direct and implied requests for change (e.g., complaints, criticisms, or requests for change) have been operationalized as “demands” that can trigger the partner’s withdrawal and/or defensiveness (Christensen & Heavey, 1990). Not only has research indicated a correlational link between demand/withdrawal interactions and lower marital satisfaction, but longitudinal studies (e.g., Gottman & Levenson, 2000) have suggested that demand/withdrawal communication is predictive of lower marital satisfaction and divorce.

Although the findings supporting the negative sequelae of the demand/withdrawal process highlight the harmful possibilities of this pattern of interaction, they leave unaddressed the empirically supported clinical implications for partners. On the one hand, research implies that partners must make explicit requests for change on one another to have a closer and more satisfactory relationship; on the other hand, research also implies that requests for change can lead to withdrawal or resistance to change, relationship deterioration, and divorce. What is missing is an empirically-supported distinction between deleterious demands and relationship-enhancing requests. Instead, both the terminology and the agglomerative manner in which “demand” is commonly operationalized cast a negative light on asking for change and ignore significant variations among these requests. For example, in Christensen and Heavey’s (1993) observational measure of demand/withdrawal, demand is operationalized as trying to discuss the problem; requesting change; blaming, accusing, or criticizing; and, demanding, nagging, or otherwise pressuring for change. Other observational systems of demand/withdrawal likewise combine requesting or pressuring for change or discussion with criticizing, complaining, or blaming change requests (Klinetob & Smith, 1996). This conflation of potentially effective and ineffective (and sometimes potentially destructive) demands likely has prevented researchers from discriminating differential effectiveness and impact of requests and hostile demands.

Clearly, a refinement in the focus of research on requests for change could offer vital information about such varying efficacy and impact. Instead of lumping any requests for change into an omnibus category of “demands,” this study focuses on identifying the qualities of requests for change that have differential impact on withdrawal/resistance and differential association with problem resolution and relationship satisfaction. It is hypothesized that the qualities of the requests for change will significantly predict relationship satisfaction and problem resolution, and that mere number of requests will not add significant predictive ability.

Most communication skills training programs (e.g., Jacobson & Margolin, 1979; Markman, Floyd, Stanley, & Lewis, 1986) recommend being specific about one’s requests. It is thought that a request must “clearly reflect what the requester intends the target to do” to be effective (Paulson & Roloff, 1997, p.262). Partners’ vagueness is a common problem noted by clinicians; Epstein and Baucom (2002) theorize that vague descriptions of the partner’s bothersome behavior offer the partner inadequate information about what and how to change. Thus, we hypothesized that requests for change that are specific (i.e. that state the action that the person should take to change) will elicit less resistance than requests that are vague (i.e. that simply complain about a situation without specifying what action should be taken). Specific, compared to vague, requests for change are hypothesized to trigger less resistance and to be associated with better problem resolution and greater relationship satisfaction.

Furthermore, the qualities of both the actor and the action of the change request might be significant in affecting couple interactions. The actor is typically specified in the use of a pronoun (e.g., you, we). Generally, partners who use “we” and “us” more have demonstrated greater relationship satisfaction and commitment (e.g., Agnew, Van Lange, Rusbult, & Langston, 1998). The use of “we” pronouns rather than “you” or “I” pronouns is thought to underscore a great regard for togetherness and sharing (Fitzpatrick, Bauman, & Lindaas, 1987) and to show greater integration of identities (Weiner & Mehrabian, 1968). “We” statements highlight the mutuality and connectedness of the relationship, and thus are theorized to decrease emotional distance and create less resistance (Burr, 1990). In general, research using various means of measurement (Inclusion of Other in Self Scale, Aron, Aron, & Smollan, 1992; Oral History Interview, Buehlman, Gottman, & Katz, 1992) has demonstrated that couples who express more closeness or we-ness than separateness are more satisfied with their relationship (Flora & Segrin, 2000). With regard to requests for change, different actor specification (“you” vs. “we”) represents varying amounts of we-ness and likely affects the recipient of the request differently. The positive effects of language that highlights we-ness (e.g., the emphasis on the couple as a unit; the decrease of the implied degree of blaming for the problem) potentially could play an important part in the delivery of a change request such that the partner is less inclined to react defensively. Thus, we hypothesize that requests that use “we” (e.g., “We need to stop giving in to them”) as opposed to “you”(e.g., “You should just tell them no”) will generate less resistance from the recipient, and conversations with a higher ratio of “we” to “you” requests for change will be associated with better problem resolution and higher levels of relationship satisfaction.

With regard to the qualities of the action, the direction of change desired will likely significantly differentiate between requests that lead to resistance and those that do not. Although research on couples has not investigated differential responses to “do” and “don’t” requests, research on children has indicated that requests to stop a behavior (i.e., “don’t” requests) are met with less compliance than requests to increase or initiate a behavior (i.e., “do” requests) (e.g., Kochanska, Askan, & Nichols, 2003). Furthermore, there is evidence to support the belief that “do” and “don’t” requests are functionally different, in that reinforcement of compliance to one type of request does not generalize to increase compliance with the other type. Reinforcement of compliance with “don’t” requests not only increases the rate of compliance to those requests but also increases the use or necessity of the use of “don’t” requests (Houlihan & Jones, 1990).

Since the earliest observational studies of couples (e.g., Hops, Wills, Patterson, and Weiss, 1972), researchers have distinguished between proposals to increase and to decrease behavior (i.e., positive and negative solutions). However, this distinction has never been investigated in terms of differential elicitation of resistance or association with problem resolution or relationship satisfaction. If “do” and “don’t” requests are indeed functionally different, it is likely that this quality of requests will have a significant impact on the recipient. Thus, we hypothesize that requests to increase behavior (i.e. “do” requests) will be met with less resistance by the partner than requests to decrease behavior (i.e. “don’t” requests), and conversations with a higher ratio of “do” to “don’t” requests will be associated with increased problem resolution and higher relationship satisfaction.

In summary, we hypothesize that requests that are specific in action, requests that contain the actor “we,” and requests for increases of behavior will predict less resistance than requests that are vague, that contain the actor “you,” and that ask for decreases of behavior. Finally, conversations with lower percentages of those types of requests that elicit resistance are hypothesized to be associated with better global problem resolution for the conversation and greater relationship satisfaction, and problem resolution and relationship satisfaction will be predicted by the qualities as opposed to quantity of change requests. Because individuals’ requests for change occur within the context of a dyadic interaction, it is possible that how one person asks for change will be influenced not only by his own satisfaction but also by his partner’s relationship satisfaction. Thus, our analytic approach (the Actor-Partner Interactional Model; e.g., Kenny, Kashy, & Cook, 2006) will isolate the actor effects (i.e. how one’s own satisfaction affects one’s own change request qualities) from the partner effects (i.e. how one’s partner’s satisfaction affects one’s own change request qualities).



The participants of the original study were 453 couples (906 individuals) who were living together for at least one year, who were parenting a 3–7 year-old who was the biological child of at least one of the individuals, and who were proficient in English. Participants were recruited through random digit dialing (RDD) of numbers in Suffolk County, New York. Undergraduate research assistants dialed 229,106 unique phone numbers and reached 12,009 consenting individuals who answered at least one question. Those respondents were screened for the eligibility criteria with a telephone survey assessing demographic characteristics and family functioning variables. Of the 1,335 eligible and interested participants who were then recontacted and invited to participate, 453 couples ultimately participated and were compensated with $250 dollars for completing a six hour protocol. Comparisons of the phone survey data with the 2000 U.S. Census data indicate that the participants were representative of the population with regard to ethnicity and income (for further detail on sampling procedures and representativeness, see Slep, Heyman, Williams, Van Dyke, & O’Leary, 2006). Furthermore, comparisons of the participant demographic data with the demographic data of individuals who qualified but did not choose to participate suggest that there were few statistically significant differences between participants and non-participants, and that the significant differences had small effect sizes.

The average age for male participants of the original study was 37.3 (SD = 6.02, range = 21.0 – 57.0) and 35.1 for females (SD = 5.00, range = 21.0 – 48.0). Eighty percent of participants were non-Hispanic white, 6.2% were African-American, 8.6% were Hispanic or Latino, 2.0% were Asian, 0.6% were Native American, and 1.4% were Caribbean American. The median xfamily income for participants was $71,000 ($1,200 – 700,000).

For this project, conversations for 72 randomly-selected couples were coded. The average age of participants was 37.2 for males (SD = 5.97, range = 24 – 50) and 35.1 for females (SD = 5.12, range = 21 – 47). Eighty percent of the subsample was non-Hispanic white, 5.2% was African-American, 9.0% was Hispanic or Latino, 3.7% was Asian, 0% was Native American, and 1.5% was Caribbean American. The median family income for participants was $67,500 (16,000 – 200,000). The participants in this subsample did not differ significantly from those in the larger sample on age t(904) = 0.01, ethnicity χ2(7, N = 905) = 4.40, or income t(853) = 1.75.


Participants were told that they were participating in a study about how families cope with conflict. As part of the larger study, participants came into the lab for a total of 6 hours (usually in two separate sessions), during which they completed a battery of questionnaires about themselves their relationships with their partners and their families, underwent physiological monitoring to standardized stimuli and completed two conflict discussions. In preparation for the video recorded conversations, participants completed a questionnaire that asked them to list topics that (a) they wanted their partners to do, to do differently, or to change, (b) were important to them, (c) they had discussed with their partners in the past year, and (d) they had not resolved. The researchers selected the topic rated as most important for the husband and the wife as the subjects of the two 10-minute conversations. The topic selected was shared with the person whose topic was to be discussed. The couple was brought together from their independent rooms to the video-recording studio, where they were told to discuss it as they usually did at home for ten minutes. This procedure was repeated for the second conversation. Order of husband and wife topics was randomly determined.

After the participants completed the two conversations as well as the questionnaires, they were paid and offered a list of community resources for their families.


Requests for Change Coding System (RCCS)

The Requests for Change Coding System (developed by the authors for this study) is a microanalytic observational coding system designed to study couples’ conversations with regard to the qualities of one partner’s requests for change and the subsequent level of withdrawal/resistance in the listening partner. A request for change is defined as an utterance (a) expressing a problem or some dissatisfaction with a current situation or (b) stating what the speaker would like to change. One group of coders divided the 10-minute interactions into speaker turns (the unit of coding) and identified those speaker turns containing a request for change. Requests for change for either partner were coded in each conversation, regardless of whose topic initiated the conversation. For each speaker turn that contained a request for change, the coders recorded the various qualities of that request (e.g., specific vs. vague, increasing vs. decreasing, actor “you” vs. “we”). The other group coded each speaker turn after a request for change for presence or absence of resistance, which was defined as a partner responding to a request for change in either (a) a manner that demonstrated involvement or openness to change or (b) a manner that suggested withdrawal or resistance to change. The coding of the withdrawal/resistance construct was consistent with other observational coding systems that include demand-withdraw behaviors in couples (e.g., Couple Interaction Rating System; Heavey, Gill, & Christensen, 1998). Because withdrawal behaviors that can be measured through self-report cannot be or are rarely seen in an observational setting (e.g., leaving the room), the construct of withdrawal/resistance for observational coding typically includes lack of verbal production (e.g., being silent, refusing to discuss a topic), avoidance (e.g., denying the existence of a problem, being vague about a problem to confuse the discussion), and lack of engagement in the discussion (Sevier, Simpson, & Christensen, 2004).

Each coding group consisted of three undergraduates who trained for one and a half semesters through 2 one-hour weekly meetings and 4 hours of coding assignments. These coders were blind to the hypotheses of the study. The inter-rater agreement was calculated using Cohen’s kappa for 25% of each coder’s interactions. For Request for Change coders, the inter-rater agreement were κ = 0.61 for presence of request for change, κ = 0.61 for specificity of request, κ = 0.60 for direction of change (increasing vs. decreasing), and κ = 0.59 for pronoun; for Resistance coders, inter-rater agreement was κ = 0.79.

Dyadic Adjustment Scale (DAS; Spanier, 1976)

The Dyadic Adjustment Scale is a widely-used 32-item measure designed to assess relationship quality by asking partners to rate on 2-, 5-, 6-, and 7-point Likert scales the extent to which each of a broad list of potential couple events describes their relationships in the past months. Summary scores range from 0 to 151, with higher scores indicate greater perceived relationship quality, and scores lower than 98 signifying discordant couples (Heyman, Feldbau-Kohn, Ehrensaft, Langhinrichsen-Rohling, & O’Leary, 2001; Jacobson et al., 1984). Cronbach’s α for the scale was 0.92 for the subsample and 0.94 for the whole sample.

Global Assessment of Problem-Solving (GAPS; Heyman & Slep, 2005)

The Global Assessment of Problem-Solving is a six-item global observational coding system designed to assess the level of problem-solving achieved in a ten-minute couple interaction. This scale includes an item that rates global problem resolution on a Likert scale from 1 (unresolved) to 4 (fairly well/completely resolved). The coders for the GAPS were 5 undergraduates who trained through weekly meetings and coding assignments. These coders were blind to the hypotheses of the study. The inter-rater agreement for the global resolution item on the GAPS was 0.76 as measured by Finn’s r (Whitehurst, 1984).


The means and standard deviations of major study variables are given in Table 1. The remaining analyses and results will be described in order from most proximal to most distal outcome variable; in other words, the results associated with resistance to change will be presented first, followed by those involving problem resolution in the conversation, followed by those associated with relationship satisfaction.

Table 1
Means and Standard Deviations of Major Study Variables

Number of Requests and Request Qualities by Gender and Topic Conversation

Before testing our hypotheses, we first checked for possible effects of the experimental context (i.e., for effects of gender and of topic initiator) on the number of change requests and the qualities of those requests using multilevel modeling (e.g., Kenny et al, 2006). Women, compared with men, made more change requests, b=−2.63, t(64.95)=.4.86, p < .001 and were more likely to resist following men’s change requests, b=−1.95, t(63.34)=.418, p < .001. Further, significantly more “you” requests happened during wives’, compared with husbands’, topic discussions, b=−0.62, t(7.94)=4.43, p = .002.

There were significant interactions between gender and whose topic the conversation was about in the proportion of change requests, b=5.85, t(64.03)=10.93, p < .001, in the proportion of requests that resulted in resistance, b=3.77, t(39.48)=8.86, p < .001, in the proportion of specific requests, b=−.04, t(20.29)= −2.43, p =.024, and in the proportion of “you” requests, b=.04, t(52.61)=2.58, p =.013. Follow-up t-tests revealed that both women and men made significantly more change requests during their own topic conversation than during partner’s topic conversation, t(54.65)=6.97, p < .001 for women and t(54.88)=6.42, p < .001 for men. During female-initiated topics, women made significantly more change requests than did men t(3.78)=9.12, p = .001 and during male-initiated topics, men made significantly more change requests than did women, t(28.70)=3.88, p = .001. Women and men both resisted significantly more in response to a change request during partners’ initiated topic than their own topic, t(30.45)=5.78, p < .001 for women and t(18.54)=3.99, p = .001 for men. Additionally, both women and men made significantly more requests that lead to partner’s resistance during their own, compared to partners’, topic, t(31.99)=7.03, p < .001 for women’s topic and t(38)=4.05, p < .001 for men’s topic. Women made significantly more specific requests during partner’s topic than during their own, t(15.12)=2.69, p = .017. Finally, women made significantly more “you” requests during their own, as compared to partners’, discussions, t(31.32)=4.44, p < .001.

Prediction of Resistance through Request Qualities

Sequential analysis is used to estimate one partner’s influence on the other based on temporally coded events. This type of analysis provides better information about whether one partner’s action indeed elicits the other partner’s response within an ongoing interaction, rather than telling whether those behaviors are simply correlated globally throughout the interaction (Bakeman & Gottman, 1997). Sequential associations for each couple were calculated with the kappa statistic (Wampold, 1989). Like the z-scores often used in sequential analysis (cf. Bakeman & Gottman, 1986), kappa is a measure of sequential patterning (and should not be confused with Cohen’s kappa, the interrater agreement statistic). Like z-scores (Allison & Liker, 1979), kappa reflects the magnitude of dependence between the two behaviors, and range from −1 (the antecedent behavior completely suppressed the consequent behavior) to 1 (the consequent behavior followed the antecedent perfectly), with 0 reflecting chance levels of sequential dependence. Unlike z-scores, kappa is not influenced by the length (total number of transitions in the interaction) nor by the base rates of the two behaviors under examination (Wampold & Kim, 1989). This becomes especially salient when z-scores or kappas are used as copule-level variables (Gottman & Roy, 1990). Variability of the frequency of transitions will result in inconsistent distributions of z-scores across couples, making intercouple comparisons inadvisable; kappa, as a nonparametric statistic, is unaffected by such variability. As shown in Table 2, wives’ decrease requests and “you” requests significantly increased the likelihood of husbands’ resistance, whereas their “we” requests significantly decreased the likelihood of husbands’ resistance. Husbands’ increase requests and their “you” requests increased the likelihood of wives’ resistance. When comparing the likelihood of eliciting resistance within request quality type, wives’ “we” requests were significantly less likely to lead to husbands’ resistance than were “you” requests; the other hypothesized effects were not significant.

Table 2
Wampold’s Kappa for Sequences of ‘Change Request Qualities→Withdrawal’

Number of Requests, Request Qualities, and Problem Resolution

To test for hypothesized associations between number and types of requests and problem resolution, zero-order correlations were calculated and are given in Table 3. Significant positive relationships were found between problem resolution and percentage of specific requests (wives: r = 0.24, p = 0.05; husbands: r = 0.31, p < 0.05) and “we” requests (wives: r = 0.27, p < 0.05; husbands: r = 0.27, p < 0.05) requests, and a significant negative relationship between problem resolution and number of wife change requests (r = −0.28, p < 0.05). The number of husband requests and percentage of increasing requests for wives and husbands were not significantly associated with problem resolution.

Table 3
Correlations Between Change Request Qualities and Problem Resolution

A hierarchical regression analysis was performed, in which problem resolution was entered as the dependent variable, the percentages of each quality (i.e. specific, increasing, “we”) were entered together as the first set of predictor variables, and numbers of wife and husband change requests were the second set of predictor variables. The results are presented in Table 4. Model 1 tests the relative contributions of each specific quality. The qualities accounted for a significant amount of variance in problem resolution (19.9%; F [6, 55] = 2.28, p < .05). Model 2 controls for the quantity of husband and wife change requests and shows the unique variance accounted for by each quality and each partner’s quantity of change requests. The addition of the numbers of wife and husband change requests accounted for 27% of the variance in problem resolution F(8, 53) = 2.45, p < .03 and significantly better than qualities alone F(2, 53) = 5.155, p < .01. The improvement in prediction was largely driven by number of wife requests (r = −0.28).

Table 4
Results of Hierarchical Regression Analysis of Qualities and Quantities of Change Requests as Predictors of Problem Resolution, Wife Relationship Satisfaction, and Husband Relationship Satisfaction

Number of Requests, Request Qualities, and Relationship Satisfaction

To test for hypothesized associations between types of requests and relationship satisfaction, a percentage was calculated for the frequency of each type of request (i.e., specific/vague, increasing/decreasing, you/we) compared with the total number of requests per conversation for both the wife and the husband. The relationships between these request quality percentages, the number of wife and husband requests, the percentage of resistance responses, and the scores on the DAS were analyzed using the actor-partner interdependence model (APIM), which controls for nonindependence of partners’ data (Kashy & Kenny, 2000). APIM isolates the effects of an individual’s independent variable on his own dependent variable as well as his partner’s dependent variable. To evaluate the hypothesized partial relations among variables, maximum likelihood structural equation modeling (SEM) procedures were applied using AMOS 5.0 (Arbuckle, 2003). Although in most circumstances, SEM is not an appropriate method of analysis with less than 100 participants, APIM is a just-identified model, in that as many parameters as are available are set and fit is not evaluated. Therefore, this method simply represents a convenient way to isolate effects while controlling for other relationships between variables. The results of these analyses are summarized in Table 5 (Graphical representation of APIM path model results can be found in supplemental figures 1–5 online). Wives’ specific requests were significantly positively associated with wives’ relationship satisfaction, even after controlling for a similar path from husbands’ specific requests (r = 0.31, p < 0.01). Likewise, wives’ “we” requests were significantly positively associated with wives’ relationship satisfaction, even after controlling for a similar path from husbands (r = 0.34, p < 0.05). Wives’ number of requests was significantly negatively associated with wives’ relationship satisfaction, even after controlling for a similar path from husbands (r = −0.28, p < 0.05). Finally, wives’ withdrawal/resistance was significantly negatively associated with wives’ relationship satisfaction, even after controlling for a similar path from husbands (r = −0.37, p < 0.01). The other hypothesized effects were not significant. The “we” constrained model was a significantly worse fit than the just-identified model χ2 (1) = 4.39, p < .05, indicating that the path between satisfaction and “we” requests is stronger for women than for men; the increasing χ2 (1) = .41 and specific χ2 (1) = 2.04 constrained models did not differ from the just identified models, implying no gender differences for satisfaction’s associations with these request qualities.

Table 5
APIM Models for Relationship Satisfaction and Request Qualities

To follow up on the association between satisfaction and request qualities, hierarchical regression analyses for men and women were also performed. In the first, wife relationship satisfaction was entered as the dependent variable, the percentages of each of her own qualities (i.e. specific, increasing, “we”) were entered together as the first set of predictor variables, and number of wife change requests was the second predictor variable. The results are given in Table 4. The wife qualities accounted for a significant amount of variance (13.5%) in wife relationship satisfaction F(3, 64) = 3.32, p < .03. The addition of number of wife change requests did not significantly add to the prediction of wife relationship satisfaction (Fincrement = 1.17, ns). This hierarchical regression was repeated for husband relationship satisfaction with husband qualities and number of requests as predictors, but these variables did not significantly predict husband relationship satisfaction (see Table 4).


The results of this study indicate that both husbands’ and wives’ change requests framed as “you” result in an increased likelihood of immediate partners’ resistance. Increase/decrease requests had differential effects, with wives’ decrease requests and husbands’ increase requests eliciting a significantly increased probability of immediate resistance. Only wives’ “we” requests resulted in significant suppression of resistance.

Specific, increasing, and “we” requests had all been hypothesized to lead to less resistance; however, only wives’ statements that highlight togetherness (Flora & Segrin, 2000) suppressed resistance immediately (i.e., “we” statements were followed by significantly less resistance than expected by chance). Contrary to clinical recommendations (e.g., Epstein & Baucom, 2002) and our hypotheses, being specific versus vague did not impact resistance immediately; however, as discussed below, overall specificity was associated with lower overall resistance. Somewhat puzzlingly, our extension of research findings about parents’ “do” (versus “don’t”) commands (e.g., Kochanska, Askan, & Nichols, 2003) being superior not only weren’t supported, but operated differently for men and women: wives’ decrease (“don’t do”) requests increased husbands’ resistance, whereas husbands’ increase (“do”) requests increased wives’ resistance. These results imply that seemingly sensible clinical recommendations may not only be ineffective, but may be counterproductive in many cases; they need to be replicated before speculation about mechanisms is warranted.

Overall, these results indicate that there is at least one way to ask for change that lessens withdrawal/resistance (wives’ requests framed as “we”), which has quite different implications than findings involving the less nuanced operationalizations of “demands.” That the demand-withdrawal pattern is harmful to relationships and leads to worse marital outcomes (Christensen & Heavey, 1993; Heavey, Christensen, & Malamuth, 1995) implies that to avoid those negative effects, one might lessen or terminate either “demands” or withdrawal/resistance. However, this is an unpalatable solution to an individual who desires one or many changes from his or her partner, or to the partner who feels accosted by complaints or attacks and withdraws as a consequence. Further, simply because the pattern of demand-withdrawal as depicted in the research literature has been linked to poor outcomes does not necessarily imply that requests or withdrawal/resistance alone are inherently negative. As stated earlier, the heading under which “demands” are grouped is an omnibus category that includes quite varied types of requests, from trying to discuss the problem and requesting change of one’s partner to blaming, accusing, and pressuring the partner for change, and empirically teasing this category apart will allow us to make better recommendations for partners. Thus, the finding that certain “demands” or ways to request change — in particular, wives’ “we” requests — are less likely to lead to withdrawal/resistance offers researchers and clinicians a viable strategy that allows individuals to ask for what they want while lessening the negative impact of those requests. Further, both men and women should be discouraged from using “you” requests, as these increased resistance.

As hypothesized, greater percentages of wife and husband specific and “we” requests were related to better problem resolution in the conversation. However, fewer wife requests were also related to better overall problem resolution, and this contributed additional significant predictive ability for problem resolution when considered with the qualities. Considering the association between specific and “we” requests and fewer total requests for wives, it could be that specific and “we” requests are more efficient ways to ask for change that immediately result in better problem resolution, which obviates the need for the repetition or rephrasing of those requests and results in a lower overall number of requests. Future research would be well-advised to explore this possibility regarding the relationship between number and qualities of requests, as they relate to more distal outcomes, such as problem resolution.

Results also indicated that greater percentages of wife specific and wife “we” requests were related significantly to greater wife satisfaction; this relationship was in the hypothesized direction. Although the number of wife change requests was significantly negatively associated with wife satisfaction, it did not add predictive ability of wife relationship satisfaction over the qualities with which the wife delivered the requests. These findings suggest that the qualities with which change requests are made, especially at whom the request is directed, are predictors of relationship satisfaction, and ones that can more easily be manipulated than mere number of change requests. This reiterates the notion that it is less advantageous to think of demands as a nuisance to be extinguished and more appropriate to regard “demands” as necessary requests for change that can vary in their deliverance, reception, and impact. In other words, at least as far as relationship satisfaction is concerned, it is less important whether a change request is made but more critical how it is made.

Finally, our examination of gender and topic initiation warrants attention. Consistent with self-report studies (e.g., Doss, Simpson, & Christensen, 2004; Margolin, Talovic, & Weinstein, 1983), women were observed making more change requests. Contrary to demand-withdrawal studies and clinical beliefs (e.g., Gray, 1992), women, not men, were more likely to respond to change requests with withdrawal/resistance. Further, women made more of the pernicious “you” requests and these requests were more likely to occur during their own topic conversations.

One of the limitations of this study is that it did not use an experimental design, and thus we cannot conclude that the relationships discussed are causal. It could be that couples who are better problem-solvers and are more satisfied with their relationships use specific and “we” requests, but that it is not those particular requests that cause those levels of problem resolution and satisfaction. With these tentative results in mind, it would be helpful to design an experimental study in which individuals were instructed to make requests in a specified way, so that the impact on problem resolution could be ascertained. Furthermore, a similar experiment in a clinical setting would help test for a more long-term impact on satisfaction. Finally, although the sample is representative, the DAS scores reflect a generally satisfied sample, with relatively few individuals in the dissatisfied range; therefore, it is not clear to what extent this sample is similar to treatment-seekers, so replication in that population is necessary.

One quality that was not explored but that might have an impact on the recipient of the request and the nature of the subsequent conversation is the affect with which a change request is delivered. It has been found that negative affect at the start of a conversation is predictive of divorce (Gottman et al., 1998). It seems likely, then, that request quality and affect might interact in their association with resistance, problem resolution, and satisfaction. Certain types of requests might be naturally associated with more or less negative affect, and it is possible that purposefully delivering a certain type of request could help the speaker to regulate emotion and affect and achieve the “softer start-up” that has been found to be associated with better conversation and relationship outcomes (Gottman et al., 1998). Future studies should examine how affect and request types interact and whether controlling request qualities could impact affect as well as resistance and problem resolution.

Several study strengths instill confidence in the findings. Although studies of communication patterns such as demand-withdrawal often rely on individual self-reports (e.g., Communication Patterns Questionnaire; Christensen & Sullaway, 1984), these results may be tainted by memory biases and confounded by trait-like negativity that would likewise impact reports of demand-withdrawal as well as relationship satisfaction. This possibility was avoided in this study by observing communication behaviors. Further, the use of dyadic data analytic strategy assured that actor and partner effects were isolated. Finally, the couples rated the conversations they had as being quite typical to ones that they had at home about similar kinds of topics, indicating that these video recorded conversations are valid analogues to usual conflicts.

The major implication of these findings is that research that lumps all requests for change together is indeed mixing differentially resistance-provoking requests. Results indicate that in general, making specific and “we” requests leads to less resistance and is associated with better problem resolution and greater relationship satisfaction for women. One significant point about these findings is that they lend empirical support to current clinical theory and interventions — such as that specific requests to one’s partner will minimize conflict (e.g., traditional behavioral couple therapy, Jacobson & Margolin, 1979; PREP, Markman, Stanley, Blumberg, Jenkins, & Whitely, 2004). In addition, these findings offer more detailed strategies to clinical recommendations that exist. For example, although most couples therapy approaches instruct partners to make “I-statements” to state what they feel, think, or want (e.g., Baucom & Epstein, 1990), which could be applied to change requests, that recommendation still leaves much room for variation in terms of the specificity and actor of the change. Thus, recommendations to make “we” and specific requests could potentially lead to better proximal and distal couple outcomes, such as less resistance, better problem resolution, and eventually higher relationship satisfaction. An experimental study that taught couples to make these requests and measured the outcomes of interest would help determine the actual clinical impact of such an intervention.

Supplementary Material

Suppl Data


Preparation of this article was supported by National Institute of Mental Health Grant R01MH57985. We also wish to acknowledge the monumental efforts of all those involved in this project, including Cheryl Van Dyke, Susan O’Leary, Bonnie Rainey, Camilo Ortiz, Michael Lorber, Debbie Leung, Evelyn Flaherty, Patti Fritz, Jeff Snarr, Mat Williams, and Heather Foran. We would also like to thank the hundreds of undergraduate research assistants who coded the videos, conducted the random digit dialing survey, and managed the data. Finally, we would like to thank the hundreds of families who volunteered to expose some of their most vulnerable issues for the long-term benefit of helping others.


Publisher's Disclaimer: The following manuscript is the final accepted manuscript. It has not been subjected to the final copyediting, fact-checking, and proofreading required for formal publication. It is not the definitive, publisher-authenticated version. The American Psychological Association and its Council of Editors disclaim any responsibility or liabilities for errors or omissions of this manuscript version, any version derived from this manuscript by NIH, or other third parties. The published version is available at


  • Agnew CR, Van Lange PAM, Rusbult CE, Langston CA. Cognitive interdependence: Commitment and mental representation of close relationship. Journal of Personality and Social Psychology. 1998;74:939–954.
  • Arbuckle JL. Amos 5.0 [Computer software] Chicago, IL: SPSS; 2003.
  • Aron A, Aron EN, Smollan D. Inclusion of other in the self scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology. 1992;63(4):596–612.
  • Bakeman R, Gottman JM. Observing interaction: An introduction to sequential analysis. 2. New York: Cambridge University Press; 1997.
  • Baucom DH, Epstein N. Cognitive-behavioral marital therapy. New York: Brunner/Mazel, Publishers; 1990.
  • Buehlman KT, Gottman JM, Katz LF. How a couple views their past predicts their future: Predicting divorce from an oral history interview. Journal of Family Psychology. 1992;5(3–4):295–318.
  • Burr WR. Beyond I-statements in family communication. Family Relations. 1990;39:266–273.
  • Christensen A, Heavey CL. Gender and social structure in the demand/withdraw pattern of marital conflict. Journal of Personality and Social Psychology. 1990;59(1):73–81. [PubMed]
  • Christensen A, Heavey CL. Gender differences in marital conflict: The demand/withdraw interaction pattern. In: Oskamp S, Costanzo M, editors. Gender issues in contemporary society. Thousand Oaks, CA: Sage Publications, Inc; 1993. pp. 113–141.
  • Christensen A, Sullaway M. Communication Patterns Questionnaire. University of California; Los Angeles: 1984. Unpublished questionnaire.
  • Cramer D. Linking conflict management behaviours and relational satisfaction: The intervening role of conflict outcome satisfaction. Journal of Social and Personal Relationships. 2002;19:425–432.
  • Doss BD, Simpson LE, Christensen A. Why do couples seek marital therapy? Professional Psychology: Research and Practice. 2004;35:608–614.
  • Epstein N, Baucom DB. Enhanced cognitive-behavioral therapy for couples: A contextual approach. Washington, DC: American Psychological Association; 2002.
  • Fitzpatrick MA, Bauman I, Lindaas M. A schematic approach to marital interaction. Paper presented at the Intenrational Communication Association; Montreal. 1987.
  • Flora J, Segrin C. Relationship development in dating couples: Implications for relational satisfaction and loneliness. Journal of Social and Personal Relationships. 2000;17:811–825.
  • Gottman JM. What predicts divorce? The measures. Hillsdale, NJ: Erlbaum; 1996.
  • Gottman JM, Coan J, Carrere S, Swanson C. Predicting marital happiness and stability from newlywed interactions. Journal of Marriage and the Family. 1998;60:5–22.
  • Gottman JM, Levenson RW. The timing of divorce: Predicting when a couple will divorce over a 14-year period. Journal of Marriage and the Family. 2000;62:737–745.
  • Gray J. Men Are from Mars, Women Are from Venus. New York: HarperCollins; 1992.
  • Hazan C, Shaver PR. Attachment as an organizational framework for research on close relationships. In: Reis HT, Rusbult CE, editors. Close relationships: Key readings. Philadelphia, PA: Taylor & Francis; 2004. pp. 153–174.
  • Heavey CL, Christensen A, Malamuth NM. The longitudinal impact of demand and withdrawal during marital conflict. Journal of Consulting and Clinical Psychology. 1995;63:797–801. [PubMed]
  • Heavey CL, Gill DS, Christensen A. The Couples Interaction Rating System. University of California; Los Angeles: 1998. Unpublished manuscript.
  • Heyman RE. Observation of couple conflicts: Clinical assessment applications, stubborn truths, and shaky foundations. Psychological Assessment. 2001;13:5–35. [PMC free article] [PubMed]
  • Heyman RE, Feldbau-Kohn SR, Ehrensaft MK, Langhinrichsen-Rohling J, O’Leary KD. Can questionnaire reports correctly classify relationship distress and partner physical abuse? Journal of Family Psychology. 2001;15(2):334–346. [PMC free article] [PubMed]
  • Hops H, Wills TA, Patterson GR, Weiss RL. Marital Interaction Coding System. Eugene, OR: University of Oregon and Oregon Research Institute; 1972.
  • Houlihan D, Jones RN. Exploring the reinforcement of compliance with “do” and “don’t” requests and the side effects: A partial replication and extension. Psychological Reports. 1990;67:439–448. [PubMed]
  • Impett EA, Gable SL, Peplau LA. Giving up and giving in: The costs and benefits of daily sacrifice in intimate relationships. Journal of Personality and Social Psychology. 2005;89:327–344. [PubMed]
  • Jacobson NS, Christensen A. Acceptance and change in couple therapy: A therapist’s guide for transforming relationships. New York: Norton; 1996.
  • Jacobson NS, Follette W, Reventsdorf D, Baucom DH, Hahlweg K, Margolin G. Variability in outcome and clinical significance of behavioral marital therapy: A reanalysis of outcome data. Journal of Consulting and Clinical Psychology. 1984;52:497–504. [PubMed]
  • Jacobson NS, Margolin G. Marital therapy: Strategies based on social learning and behavior exchange principles. New York: Brunner/Mazel, Publishers; 1979.
  • Kashy DA, Kenny DA. The analysis of data from dyads and groups. In: Reis HT, Judd CM, editors. Handbook of research methods in social and personality psychology. New York, NY: Cambridge University Press; 2000. pp. 451–477.
  • Kenny DA, Kashy DA, Cook WL. Dyadic data analysis. New York, NY: Guilford Press; 2006.
  • Klinetob NA, Smith DA. Demand-withdraw communication in marital interaction: Tests of interspousal contingency and gender role hypotheses. Journal of Marriage and the Family. 1996;58:945–957.
  • Kochanska G, Aksan N, Nichols KE. Maternal power assertion in discipline and moral discourse contexts: Commonalities, differences, and implications for children’s moral conduct and cognition. Developmental Psychology. 2003;39:949–963. [PubMed]
  • Marcus EC. Change processes and conflict. In: Deutsch M, Coleman PT, editors. The handbook of conflict resolution: Theory and practice. San Francisco, CA: Jossey-Bass; 2000. pp. 366–381.
  • Margolin G, Talovic S, Weinstein CD. Areas of change questionnaire: A practical approach to marital assessment. Journal of Consulting and Clinical Psychology. 1983;51:944–955.
  • Markman HJ, Floyd F, Stanley SM, Lewis H. Prevention. In: Jacobson N, Gurman A, editors. Clinical Handbook of Marital Therapy. New York: Guilford Press; 1986. pp. 174–194.
  • Markman HJ, Stanley SM, Blumberg SL, Jenkins NH, Whiteley C. Twelve hours to a great marriage. New York: Wiley; 2004.
  • Neff LA, Karney BR. Gender differences in social support: A question of skill or responsiveness? Journal of Personality and Social Psychology. 2005;88:79–90. [PubMed]
  • Paulson GD, Roloff ME. The effect of request form and content on constructing obstacles to compliance. Communication Research. 1997;24:261–291.
  • Sevier M, Simpson LE, Christensen A. Observational coding of demand-withdraw interactions in couples. In: Kerig PK, Baucom DH, editors. Couple observational coding systems. Mahwah, NJ: Erlbaum; 2004. pp. 159–172.
  • Slep AMS, Heyman RE, Williams MC, Van Dyke CE, O’Leary SG. Using random telephone sampling to recruit generalizable samples for family violence studies. Journal of Family Psychology. 2006;20(4):680–689. [PubMed]
  • Spanier GB. Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family. 1976;38:15–28.
  • Thibaut JW, Kelley HH. The social psychology of groups. New York: Wiley; 1959.
  • Wampold BE. Kappa as a measure of pattern in sequential data. Quality & Quantity. 1989;23:171–187.
  • Weiner M, Mehrabian A. Language within language: Immediacy, a channel in verbal communication. New York: Appleton-Century-Crofts; 1968.
  • Weiss RL, Hops H, Patterson GR. A framework for conceptualizing marital conflict: A technology for altering it, some data for evaluating it. In: Handy LD, Mash EL, editors. Behavior change: Methodology concepts and practice. Champaign, IL: Research Press; 1973. pp. 309–342.
  • Whitehurst GJ. Inter-rater agreement for journal manuscript reviews. American Psychologist. 1984;39:22–28.