Interaction and Relationship Development in Stable Young Couples: Effects of Positive Engagement, Psychological Aggression, and Withdrawal

The formation and continuing negotiation of romantic relationships during adolescence and early adulthood is a crucial developmental task with long-term consequences for social-emotional adjustment (e.g., Bouchey & Furman, 2003; Erikson, 1968), yet there is a paucity of research on the developmental course of romantic relationships and (mal)adaptive relationship patterns during this transitional period (for exceptions, see Florsheim, 2003 and Furman, Brown, & Feiring, 1999). Based on a dynamic developmental systems model (DDS; Capaldi, Shortt, & Kim, 2005), which provides a guide for examining the contribution of both partners to relationship development, the present study addresses an important relationship outcome –relationship satisfaction – in a sample of young adult couples who remained in a stable relationship over approximately 7 years. We examined the effects of each partner’s affective/behavioral styles, including positive and negative forms of engagement as well as disengagement, during conflict interactions structured around both men’s and women’s issues to shed light on the dynamics of successful relationship development during this critical period. By investigating longitudinal associations between partners’ observed interactions and relationship outcomes, this study provides insight into early adaptive processes that foster satisfying relationships during the transition to adulthood.

Young Adult Romantic Relationships in a Developmental Context

As suggested by Brown (1999), both individual orientation toward romantic relationships and the function of such relationships show characteristics of a developmental sequence across adolescence and young adulthood. In the early phases of “initiation” and “status,” younger adolescents focus on the enactment of romantic roles to help build their own sense of identity and competency within their peer group. The quality of the romantic relationship itself is secondary to the purpose it serves in defining the self. By late adolescence, however, the “affection” phase signals a shift to the relationship itself, which becomes a more intense and rewarding experience. In the final “bonding” phase, late adolescents/emerging adults apply more pragmatic and personal concerns to the relationship as they consider long-term commitment to the partner, and possibly in marriage. It is during the latter phases of relationship development that conflict management skills and emotion regulation become important as couples begin building potentially a lifelong pair bond.

In keeping with this model, national statistics have shown that the prevalence of romantic relationships, length of relationship, and levels of intimacy and commitment increase substantially across adolescence (Carver, Joyner, & Udry, 2003). As suggested by Selman and colleagues (Selman, Beardslee, Schultz, Krupa, & Podorefsky, 1986), this evolution at the relationship level is possible because individual partners have developed stronger social-cognitive skills that allow them to consider both self and other needs and to integrate these in a collaborative process. By the end of adolescence, individuals are poised to deepen romantic relationships into committed partnerships such as those studied by marital researchers. Tabares and Gottman (2003) suggest that a marital process perspective may be applied to the study of adolescent romantic relationships, although adolescent themes such as autonomy, the need for emotional support, and negotiation of emotionality would remain salient. In studying young adult couples, then, we consider behavioral growth profiles against the backdrop of normative patterns of individual/relationship development from adolescence to adulthood.

A Dynamic Developmental Systems Model of Romantic Relationships

The DDS (Capaldi et al., 2005) takes into account development both in terms of the individual’s developmental history and stage and the developmental characteristics of the relationship itself, and as such it provides a useful frame for examining the progression of the bonding phase of romantic relationship development. The DDS further incorporates an interactional perspective that considers the active role of both partners in shaping each other’s behavior and the evolution of the relationship. Although the approach shares some characteristics of dynamic systems models, including an emphasis on the importance of time at various levels (Granic, Dishion, & Hollenstein, 2003; van Geert & Lichtwarck-Aschoff, 2005), the dynamic systems model is more molecular, with very specific predictions regarding the course of an interaction, whereas the emphasis of the DDS model is on intersecting systems (e.g., biological, psychological, and social) that are in transaction and all developing over time and also affected by proximal contextual factors (e.g., stress, substance use, men’s versus women’s issues). Mutual influences within couples, rather than one-sided views of men’s and women’s behaviors, are a particular emphasis. In addition, the DDS model underscores the importance of the context in which interactions between partners occur. Thus, adaptive couples interaction patterns might differ depending on contextual variables such as who exhibits the behavior (man or woman partner) and in which conflict structure (his or her issue).

Effects of Couples’ Observed Interaction Styles on Relationship Outcomes

Research on couples’ interactions has tended to show benefits of positive affective and behavioral patterns and a destructive impact of their negative counterparts (see Johnson et al., 2005), yet the meaning and impact of a particular emotional/behavioral display may vary depending on the couple’s developmental stage, as well as the factors described above (i.e., partner and conflict structure). For example, normative patterns of relationship development from adolescence to young adulthood show increases in the use of negotiation and decreases in both coercion (Laursen, Hartup, & Koplas, 1996) and disengagement/minimizing (Tuval-Mashiach & Shulman, 2006), yet individual couples transitioning into adulthood may find more or less “mature” conflict strategies adaptive based on partners’ developmental capacities. Therefore, we discuss possible effects of positive and negative engagement and disengagement in couples’ interactions, with the expectation that all could play a part in successful relationship development depending on when and how they are used.

On the negative end of the conflict engagement spectrum, psychological aggression involves attacking behaviors such as personal criticisms, name calling, derogatory statements, coercion, and threats (O’Leary, 1999; Stets, 1991). Not surprisingly, psychological aggression relates negatively to relationship adjustment and positively to other harmful relationship dynamics, such as physical aggression (e.g., Capaldi & Crosby, 1997). In adolescent samples, dissatisfied couples have also been distinguished by longer periods of observed hostility compared to satisfied couples (Florsheim, Moore, & Edgington, 2003). During adolescence, psychologically aggressive behavior may be part of an effort to assert autonomy and control in relationships, and as such, a more normative part of conflict negotiation. On the other hand, younger couples may be more vulnerable to the negative effects of psychological aggression because they are learning to regulate their emotional responses and to repair damage to a less established relationship compared to married adult couples. Younger partners may also rely more on the relationship to support their developing sense of agency, which could be threatened by psychological aggression within the relationship. There is some evidence from marital studies that women’s psychological aggression, in particular, has deleterious effects on relationship outcomes over time (e.g., Schumacher & Leonard, 2005; Story, Karney, Lawrence, & Bradbury, 2004). Among adolescent couples, female partners have been found to be more sensitive to issues of power and control in the relationship (Welsh, Galliher, Kawaguchi, & Rostosky, 1999), which could give rise to greater sensitivity to and use of psychological aggression by young woman partners.

On the other side of the spectrum, positively engaged aspects of conflict interaction tend to predict better relationship outcomes, though some studies have failed to find effects of positive problem-solving behaviors (Kiecolt-Glaser, Bane, Glaser, & Malarkey, 2003) or affect (Smith, Vivian, & O’Leary, 1990) on later satisfaction. Considered in light of adolescent themes, emotional support during conflict should be important, and satisfied adolescent couples have been found to show more nurturant and loving behavior in interactions compared to dissatisfied couples (Florsheim et al., 2003). We might expect positive engagement during conflict to be even more important for young couples’ satisfaction with the relationship than for older married couples, given adolescents’ use of emotional cues to determine whether a relationship is worthwhile (Larson, Clore, & Wood, 1999). Although there is little direct evidence for gender differences in the effects of positive engagement, social structural theories that highlight the adaptive importance of communal qualities and relationship-maintaining behaviors in women (Eagly & Wood, 1999) suggest that women partners’ positive engagement might be most effective in fostering relationship satisfaction.

Another important area of couples’ interactions is the role of disengagement or withdrawal. Withdrawal has often been cast as a destructive strategy (e.g., Heavey, Christensen, & Malamuth, 1995; Smith et al., 1990) involving an implicit refusal to resolve an issue that the partner wishes to resolve. In contrast, Gottman’s (1993) description of three “types” of stable couples included “avoiding” as a working relationship dynamic, and a later validation study (Holman & Jarvis, 2003) found the avoiding type to be the second best adjusted (after “validating,” which resembles the typical standard of high-positive engagement and low-negative engagement) on indices of relationship satisfaction and communication quality. More direct evidence for the reward value of withdrawing from marital conflict, at least for men, comes from a recent study of emotional arousal and negative affect in marital interaction, which found that husbands who withdrew more had lower levels of postconflict arousal (Verhofstadt, Buysse, De Clercq, & Goodwin, 2005). For wives, on the other hand, withdrawal actually increased emotional arousal, and it was demand behavior that exerted a positive effect in the sense of decreasing her own negative affect. During adolescence, male partners have been found to show more autonomy-seeking behaviors in observed interactions (Florsheim et al., 2003), and the use of withdrawal may be an important part of maintaining autonomy and/or regulating levels of negative emotion for younger men, in particular.

Beyond the main effects of particular conflict styles exhibited by young men and women partners, we consider the context of the conflict structure in which those behaviors occur. The individual who raises the relationship problem first often has strong feelings about the issue. Because it usually involves an area where they would like to see change in their partner’s behavior, this tends to place them in the role of critic who expects acknowledgement and some willingness to change from their partner and places their partner in something of a defensive mode (i.e., having to respond to some level of change demand). Some studies have found stronger effects of behavior during the partner’s issue on own satisfaction (Johnson et al., 2005), whereas others have found stronger effects for behavior during discussion of one’s own issue (Kim, Capaldi, & Crosby, 2007). In studies on demand-withdraw patterns, the conflict structure – who is requesting change from whom – has proved an important modifier of men’s and women’s roles in conflict interactions (Vogel & Karney, 2002), with woman-demand/man-withdraw asymmetries only evident during discussion of the woman’s issue. Verhofstadt and colleagues (2005) found that for emotional arousal, men reported higher arousal and negative affect during discussion of their own issue, making the benefits of withdrawal possibly more salient in this context. Some of the inconsistencies in the literature on couple interaction and adjustment could be resolved by distinguishing whose behavior is being examined and during whose problem discussion.

The Present Study

The present study tested longitudinal associations between observed conflict styles --positive engagement, psychological aggression, and withdrawal -- and relationship satisfaction in young adult couples. In keeping with the DDS model, we expected to find distinct effects of young men’s and women’s affect/behavior styles that depended on the conflict structure. The primary study question addressed was the following: Which conflict styles characterize a successful bonding process for young adult couples in the sense of fostering/maintaining relationship satisfaction? For couples at this developmental stage, a satisfying relationship must offer the opportunity to both assert autonomy and to feel supported, to balance negative emotions with positive ones. With these considerations in mind, as well as indications from both the marital literature and the available literature on adolescent couples, we hypothesized that positive engagement would predict higher levels and more positive (or less negative) slopes of relationship satisfaction over time, whereas psychological aggression would predict lower levels and steeper declines in satisfaction. Given the mixed results for withdrawal in previous studies, we approached it in a more exploratory fashion.

Secondary study questions concerned the relative influence of the women’s versus the men’s interaction patterns, and whether the impact of those patterns differed according to whose problem was being discussed. Given the evidence from past couples research with both adolescents and married couples, it was predicted that there would be a greater impact of women’s psychological aggression and positive engagement and men’s withdrawal on both partners’ outcomes. We also expected differential effects of behaviors during discussions of young men’s versus women’s issues. As a general rule, it was expected that positive engagement during one’s own topic would be more closely related to satisfaction as it signifies an ability to set aside strong negative emotions that partners might have about their issue. Psychological aggression during the partner’s issue, on the other hand, was expected to exert more negative effects as it would imply an attack on the partner in a context where negativity is not as expected. Finally, withdrawal during men’s topics was predicted to be most important, given the evidence for both greater arousal and greater benefits of withdrawal for men during discussion of their own issue.

Method

Participants

The present study used data from the Couples Study of a community sample of young couples with at-risk backgrounds. The entire fourth-grade classes (ages 9–10 years) of boys in higher-crime areas of a medium-size metropolitan region in the Pacific Northwest were originally recruited to participate in the Oregon Youth Study (OYS). When the men were ages 17–18 years, the Couples Study was started to examine the OYS men and their intimate partners’ adjustment in young adulthood. The men and their partners were assessed five times over the years; late adolescence Time 0 (T0, ages 17–20 years), emerging adulthood Time 1 (T1, ages 20–23 years), and early adulthood Time 2 (T2, ages 23–25 years); Time 3 (T3, ages 25–27 years); and Time 4 (T4, ages 27–29 years). For the present study, data from T1 through T4 were used, as many men did not participate in T0 assessment because they were not in a relationship. This is in keeping with Brown’s (1999) model in which stable, committed relationships emerge later in development. The present study included 47 couples who participated at all four time points with the same partner (see Descriptive Characteristics of the Sample below for a comparison of these stable couples with the rest of the Couples Study sample). The majority of partners were Caucasian (84%), with smaller proportions of African American (3%), Native American (2%), Latino (4%), Asian American (2%), and other ethnicities (5%). A majority of men (81.1%) had a high school education or less, and 9.5% had completed a 4-year college; the women tended to show somewhat higher educational achievement (48.7% high school or less, 13.8% completed a 4-year college).

Procedures

Assessment for the Couples Study included a separate interview and questionnaires for the OYS men and their partners and a series of six discussion tasks that were videotaped, including a warm up session. The entire couple assessment lasted approximately 2 hours. For the present study, coding of two problem-solving discussion tasks was used. The man and his partner were each directed to pick an issue in their relationship either from the list provided (the Partner Issue Checklist; Capaldi, Wilson, & Collier, 1994) or of their own choosing. Examples of issues selected included partner’s jealousy, not having enough money for dates or activities, and having a hard time talking to each other. The couples discussed each partner’s issue for 7 minutes for a total of 14 minutes. For further information regarding the discussion tasks, see Capaldi and Crosby (1997) and Capaldi, Shortt, and Crosby (2003).

Coding of the interaction tasks

The Family and Peer Process Code (FPPC; Stubbs, Crosby, Forgatch, & Capaldi, 1998) was used to code the interaction tasks at each time. A computer was used to record, in real time, the interpersonal behavioral and emotive content of each couple’s interaction. The 24 content codes included verbal, vocal, nonverbal, physical, and compliance behaviors that were judged a priori as having a positive, neutral, or negative impact. Six affective ratings described the participant’s emotional tone: happy, caring, neutral, distress, aversive, and sad. An affect rating was assigned to each content code on the basis of tone and inflection of voice, body posture, nonverbal gestures, and facial expressions. Each partner’s behavior was coded continuously with a particular content/affect code combination, which did not change until a change in content and/or affect occurred; in other words, behavior was coded by duration of a codeable state rather than by discrete events. Content and affect kappas on the basis of independent ratings by two trained research assistants ranged from .75 to .85 for T1–T4 (see www.oslc.org/resources/codingsystems.html for a complete description of the coding system).

Measures

Couples interaction patterns

Behavioral clusters defined by pertinent combinations of content codes and affects were created, and duration proportions (time spent exhibiting the behavior cluster/total discussion time) for two of these clusters and a qualifier code were examined in the following analyses.

Positive engagement

This cluster included neutral behaviors (talk, tease, advise, vocal, neutral nonverbal) combined with positive affect, as well as positive behaviors (positive interpersonal, endearment, self-disclose, positive nonverbal, touch/hold) combined with positive or neutral affect. Examples of positive engagement could include a partner telling about something that happened to him/her (talk) with a happy or caring demeanor (positive affect) and a partner complimenting the other (endearment) or giving him/her a hug (touch/hold) with neutral to positive affect.

Psychological aggression

This cluster was defined by negative behaviors (negative interpersonal, verbal attack, coerce) combined with neutral, disturbed, aversive, or sad affect. Behaviors such as name calling, threats, and criticism or humiliation were included in this coding. For example, a partner demanding that the other change his/her behavior while threatening negative consequences if he/she fails to comply, accompanied by a neutral to angry emotional tone, would be coded as psychological aggression. Attacking statements like “drop dead” or “you blew it again, stupid” would also be considered psychological aggression.

Withdrawal

This code was used as a qualifier of the interaction and was defined independent of the content codes as “a lack of engagement with or attention to ongoing interaction.” No single behavior defined withdrawal, rather it was defined from a gestalt of behaviors including such indicators as closed-off body language and failure to respond to partner. For example, a partner in withdrawal might turn away from the partner and ignore his/her questions. Withdrawal also included less obvious forms of disengagement, such as a partner showing intermittent eye contact or a “glazed” expression and making brief responses only after repeated prompting. At least 3 seconds of observed withdrawal were required before a state of withdrawal was coded.

Kappas for the specific content and affect clusters comprising these constructs ranged from .89–.94 for positive engagement, from .59–.72 for psychological aggression, and from .95–.99 for withdrawal across the four time points.

Relationship satisfaction

Both young men’s and their partners’ relationship satisfaction at T1 through T4 were assessed with a 32-item self-report scale designed to measure the quality of marriage and other similar dyads (Dyadic Adjustment Scale [DAS]; Spanier, 1976). The alphas of the total score ranged from .91 to .93 for the men and .88 to .95 for the women. In the current sample, two items (how much partners agree on “handling family matters” and “religious matters”) were not answered by a large proportion of the sample and therefore were dropped. The remaining 30 items were split into two subscales¹ with roughly equal variances for each time point to serve as outcomes in Hierarchal Linear Modeling (HLM) (see Analytic Strategy).

Relationship control variables

Length of relationship and presence of children were assessed as possible control variables relevant to the couple’s satisfaction. The average of the two partners’ reported length of relationship (in weeks) at the T1 assessment indexed initial length of time together. A dichotomous variable indicating whether the couple had children living in the home with them (1) or not (0) was assigned at each time point.

Analytic Strategy

The data under investigation are dependent (i.e., nested, scores over time are clustered within an individual, and individuals’ scores are clustered within a couple); thus, a multilevel modeling strategy, specifically dyadic growth curve modeling as outlined by Raudenbush, Brennan, and Barnett (1995), was used. In this framework, variability is modeled separately for within-couple and between-couple components, taking into account the interdependence of partners’ trajectories. At Level 1, within-couple variation in relationship satisfaction over time is modeled with partner-specific growth parameters that are allowed to vary across couples. At Level 2, these growth parameters from Level 1 are modeled as outcomes to be explained by characteristics of the couple. This modeling approach is preferred for examining couples longitudinal data when both similarities and differences between partners’ trajectories are of interest (see Atkins, 2005).

HLM (Raudenbush & Bryk, 2002) was used to test models of relationship satisfaction that included no explanatory predictors, as well as a series of models that included couples’ interaction scores during men’s and women’s problem trials to predict satisfaction. An advantage of HLM is that it separates true score variability in the outcome from measurement error; the presence of multiple scores over time and/or at each time point (the two DAS scores) allows for this calculation. Predictors were added at both levels of the model. At Level 1, predictors were time-varying covariates (centered around each partner’s mean) indicating the extent to which changes in partners’ interactions patterns were associated with changes in relationship satisfaction. At Level 2, time-invariant effects indicated the extent to which each partner’s mean interaction patterns (centered around the men’s or women’s mean for the sample) predicted individual satisfaction growth parameters.

As an example, in the model testing the effect of positive engagement on relationship satisfaction, a significant Level 2 effect would indicate that the men’s or women’s average level of positive engagement across time predicted a higher level (intercept at T4) and/or less decline (slope) in satisfaction for them. A significant Level 1 effect in this example would indicate that high positive engagement (relative to the individual’s average level) at any of the four time points was associated with higher than average satisfaction at the same time point.

The two-level model for positive engagement (with the underlying baseline growth model highlighted in bold) is as follows:

Level 1 (within-couple)

$$\begin{array}{l}\text{DAS}={(\mathbf{man})}_{\mathit{it}}[{\mathbf{\pi }}_{\mathbf{m}\mathbf{0}\mathit{i}}+{\mathbf{\pi }}_{\mathbf{m}\mathbf{1}\mathit{i}}{(\mathbf{linear})}_{\mathit{it}}+{\pi }_{\text{m}2i}{(\text{man}\text{'}\text{s}\text{positive}\text{engagement})}_{it}+{\pi }_{\text{m}3i}{(\text{woman}\text{'}\text{s}\text{positive}\text{engagement})}_{it}]\\ +{(\mathbf{woman})}_{\mathit{it}}[{\mathbf{\pi }}_{\mathbf{w}\mathbf{0}\mathit{i}}+{\mathbf{\pi }}_{\mathbf{w}\mathbf{1}\mathit{i}}{(\mathbf{linear})}_{\mathit{it}}+{\pi }_{\text{w}2i}{(\text{man}\text{'}\text{s}\text{positive}\text{engagement})}_{it}+{\pi }_{\text{w}3i}{(\text{woman}\text{'}\text{s}\text{positive}\text{engagement})}_{it}]+{\mathit{e}}_{\mathit{it}}\end{array}$$

Level 2 (between-couple)

$${\mathbf{\pi }}_{\mathbf{m}\mathbf{0}\mathit{i}}={\mathbf{\beta }}_{\mathbf{m}\mathbf{0}\mathbf{0}}+{\beta }_{\text{m}01}(\text{avg}.\text{man}\text{'}\text{s}\text{positive}\text{engagement})+{\beta }_{\text{m}02}(\text{avg}.\text{woman}\text{'}\text{s}\text{positive}\text{engagement})+{\mathit{u}}_{\mathbf{m}\mathbf{0}\mathit{i}}$$

(Similar equations describe π_m1i and π_w0−1i ; π_m/w2−3i are simply modeled with intercepts and no variability.)

Results

Descriptive Characteristics of the Sample

Table 1 presents descriptive information for the sample, including relationship satisfaction scores, across the four time points. Comparisons of the present sample of stable couples with the couples from the larger study who did not remain together across T1–T4 revealed several differences at T1; the couples who would remain stable were in more committed relationships (mean relationship status = 1.98 for stable vs. 1.62 for nonstable couples, t[156] = −2.86, p = .005), and they were less likely to have children (29% of stable couples vs. 54% of nonstable couples had children in the home, χ²[1] = 5.43, p = .03). Not surprisingly, stable couples also reported higher relationship satisfaction at T1 (mean DAS = 112.04 for stable vs. 102.99 for nonstable men, t(156) = − 2.84, p = .005; mean DAS = 110.21 for stable vs. 103.71 for nonstable women, t(156) = − 2.39, p = .02). Table 2 presents correlations among men’s and women’s mean relationship satisfaction and observed interaction scores (averaged across times) separately for men’s and women’s topic interactions.

Table 1.

Descriptive Information for Intact Couples Across all Time Points

Assessment	T1	T2	T3	T4
Men’s Age (mean)	21.5	24.0	26.0	28.0
Women’s Age (mean)	20.8	23.3	25.3	27.4
Relationship Status
Dating	13	5	1	1
Living together	22	15	10	9
Married	12	27	36	37
Of those couples living together or married, % with children in home	29%	57%	67%	72%
Length of relationship (mean weeks)	93	224	324	423
Men’s relationship satisfaction (M, SD)	112.04 (18.50)	107.72 (17.75)	108.98 (16.28)	107.26 (17.76)
Women’s relationship satisfaction (M, SD)	110.21 (15.59)	108.34 (17.64)	110.72 (16.71)	109.14 (20.45)

Note. Data are drawn from 47 couples who remained intact across the four time points of the study.

Table 2.

Correlations Among Partners’ Relationship Satisfaction (DAS) Scores and Observed Interaction Patterns During Men’s (above the diagonal) and Women’s (below the diagonal) Topics, Averaged Across Times

Variable	1	2	3	4	5	6	7	8
1. Man’s DAS	--	.68**	.23	.55**	−.46**	−.55**	.05	−.17
2. Woman’s DAS	.68**	--	.36*	.42**	-.43**	−.57**	.20	−.05
3. M positive engagement	.24	.33*	--	.46**	−.24	−.13	−.26†	.02
4. W positive engagement	.51**	.50**	.45**	--	−.19	−.25†	−.11	−.16
5. M psychological aggression	−.41**	−.37*	−.13	−.30*	--	.59**	−.16	.25†
6. W psychological aggression	−.52**	−.27†	.07	−.31*	.52**	--	−.12	.28†
7. M withdrawal	.06	.12	.01	−.06	−.08	.10	--	−.14
8. W withdrawal	−.08	−.17	−.20	−.24	.44**	.29*	.05	--

Note. Data are drawn from 47 couples who remained intact across the four time points of the study. M = Man. W = Woman.

^†p < .10.

^*p < .05.

^**p < .01.

Relationship Satisfaction: Baseline Model

An unconditional model of relationship satisfaction allows for examination of couples’ mean trajectories without the influence of predictor variables. A linear growth model was first tested to describe couples’ relationship satisfaction across the four times of data collection spanning approximately 7 years; men’s and women’s intercepts and slopes were modeled at Level 1, which were in turn entered as outcomes at Level 2. A quadratic model was also tested, but ultimately rejected because the addition of quadratic terms resulted in extremely unreliable linear and quadratic terms with no significant variability to be explained. The baseline and subsequent models were centered at the final time point (T4); thus, intercept terms represent men’s and women’s relationship satisfaction at the end of the period under study. This centering allows us to survey partners’ satisfaction after a 7-year relationship, as well as to pinpoint predictors of satisfaction at this relatively mature stage of the relationship.

Substantial associations between partners’ intercept (tau = .79) and slope (tau = .95) terms underlined the dependency of the men’s and women’s relationship satisfaction within couples. A planned contrast of coefficients showed that the women’s intercept of 51.40 (representing close to one half of the total DAS score) was significantly higher than the men’s intercept of 50.38, χ²(2) = 1927.95, p < .001. The average men’s slope depicted a significant decline in satisfaction over time, but the average women’s slope was nonsignificant, and the former (−.83) was significantly more negative than the latter (−.26), χ²(2) = 6.58, p < .05. Thus, over 7 years, men declined more in satisfaction and ended with lower levels than women. At the same time, significant variability around the average for both men’s and women’s intercepts and slopes suggests that this pattern varied across couples, and the addition of explanatory predictors could reveal differing patterns across time.

Control Variables

Before entering the behavioral predictors of interest, two couple-level characteristics that might need to be controlled were tested (see Method). Length of relationship at T1 was significantly related to relationship satisfaction; longer relationship length was associated with lower satisfaction at T4 for both partners. Apparently, couples who began their relationships earlier in adolescence were less satisfied by the end of the study period. This variable was included as a control for predicting satisfaction in all subsequent analyses.

Positive Engagement During the Men’s Topic

The top panel of Table 3 shows prediction to men’s and women’s relationship satisfaction from both partners’ levels of positive engagement during the discussion of the men’s chosen problem topic. Both partners’ positive engagement scores during the men’s topic were entered as predictors of relationship satisfaction at both the within-couple (Level 1) and between-couple (Level 2) levels of analysis. The women’s positive engagement emerged as a significant predictor of both their own and their partners’ satisfaction at Level 2 (line 4), indicating that in couples with women who showed generally high levels of positive engagement, both partners were more satisfied at T4. In addition, the women’s positive engagement related to the men’s (and, marginally, to their own) relationship satisfaction at Level 1 (line 1). Thus, controlling for women’s overall level of positive engagement across time, relatively higher levels of the woman’s positive engagement at any individual time point related to men’s higher satisfaction at that same time point. The men’s positive engagement, on the other hand, failed to predict either his or his partner’s relationship satisfaction (line 2). A deviance test showed the overall model for the men’s topic to be a significant improvement in fit over baseline, χ²(14) = 50.68, p < .001, and it explained 35.44% of the variance in men’s and 27.32% of the variance in women’s intercepts, 12.88% of the variance in men’s and 14.57% of the variance in women’s slopes.

Table 3.

The Effect of Positive Engagement on Couples’ Relationship Satisfaction

	Women’s Satisfaction				Men’s Satisfaction
Predictor	Estimate	SE	t(df)	p	Estimate	SE	t(df)	p
Men’s Topic
Level 1 Effects
1. Women’s positive engagement	5.60	3.25	1.72 (734)	.08	11.42	3.24	3.52 (734)	.00
2. Men’s positive engagement	4.23	4.32	0.98 (734)	.33	1.02	4.30	0.24 (734)	.81
Level 2 Effects(Positive engagement as predictor averaged across four time points)
3. Intercept (sample average T4)	51.94	1.22	42.37 (43)	.00	50.40	1.00	50.33 (43)	.00
4. Women’s positive engagement	39.05	17.98	2.17 (43)	.04	43.74	14.58	3.00 (43)	.00
5. Men’s positive engagement	45.47	26.26	1.73 (43)	.09	18.86	21.26	0.89 (43)	.38
6. Slope (average)	−0.10	0.36	−0.29 (44)	.78	−0.61	0.35	−1.76 (44)	.09
7. Women’s positive engagement	5.78	5.32	1.09 (44)	.28	−1.66	5.07	−0.33 (44)	.74
8. Men’s positive engagement	10.31	7.76	1.33 (44)	.19	11.65	7.40	1.58 (44)	.12
Women’s Topic
Level 1 Effects
9. Women’s positive engagement	9.57	3.63	2.63 (734)	.01	10.22	3.62	2.82 (734)	.00
10. Men’s positive engagement	−1.53	4.24	−0.36 (734)	.72	1.07	4.22	0.25 (734)	.80
Level 2 Effects
11. Intercept (sample average T4)	51.74	1.22	42.29 (43)	.00	50.43	1.00	50.58 (43)	.00
12. Women’s positive engagement	58.86	19.62	3.00 (43)	.01	64.49	15.81	4.08 (43)	.00
13. Men’s positive engagement	18.24	28.93	0.63 (43)	.53	−9.14	23.30	−0.39 (43)	.70
14. Slope (average)	−0.15	0.36	−0.41 (44)	.68	−0.67	0.33	−2.01 (44)	.05
15. Women’s positive engagement	9.04	5.80	1.56 (44)	.13	7.86	5.43	1.45 (44)	.15
16. Men’s positive engagement	0.54	8.52	0.06 (44)	.95	−0.31	7.97	−0.04 (44)	.97

Positive Engagement During the Women’s Topic

During the discussion of the women’s problem topic (Table 3 lower panel), the women’s average positive engagement again predicted higher self and partner intercepts (line 12). Furthermore, the women’s positive engagement showed a positive association with both their partner’s and their own satisfaction over time (line 9). Not only did partners in couples with more positively engaged women enjoy higher relationship satisfaction at T4 compared to other couples, but within the course of the relationship, both partners reported higher satisfaction at times when the woman was more positively engaged relative to her own within-couple average level. Again, the men’s positive engagement showed no predictive relationships. This model was a significant improvement in fit compared to baseline, χ²(14) = 40.81, p < .001, explaining 37.44% of the variance in men’s and 28.53% of the variance in women’s intercepts, and 21.72% of the variance in men’s and 18.30% of the variance in women’s slopes.

To summarize, when it comes to the positive end of the relationship interaction spectrum, women’s behavior appeared to constitute a greater influence on couples’ satisfaction. Women’s positive engagement during both their own and their partners’ problem topics predicted the couple’s relationship satisfaction at both within-couple and between-couple levels.

Psychological Aggression During the Men’s Topic

Women’s psychological aggression during the men’s problem trial (Table 4 top panel) predicted lower satisfaction for themselves at both Level 1 (line 1) and Level 2 (line 4). Women who exhibited higher psychological aggression reported lower satisfaction at T4. In addition, higher relative levels of psychological aggression on the women’s part coincided with lower satisfaction over time. Marginally significant effects of women’s psychological aggression were found for men partners’ satisfaction. This model improved significantly over baseline in fit, χ²(14) = 42.32, p < .001. It explained 27.60% of the variance in men’s and 25.86% of the variance in women’s intercepts, and 12.32% of the variance in men’s and 13.10% of the variance in women’s slopes.

Table 4.

The Effect of Psychological Aggression on Couples’ Relationship Satisfaction

	Women’s Satisfaction				Men’s Satisfaction
Predictor	Estimate	SE	t(df)	p	Estimate	SE	t(df)	p
Men’s Topic
Level 1 Effects
1. Women’s psych. aggression	−21.65	8.46	−2.56 (734)	.01	−16.31	8.49	−1.92 (734)	.06
2. Men’s psych. aggression	−1.05	6.52	−0.16 (734)	.87	−0.03	6.56	−0.00 (734)	1.0
Level 2 Effects (Psychological aggression as predictor averaged across four time points)
3. Intercept (sample average T4)	51.31	1.21	42.25 (43)	.00	50.18	1.02	49.09 (43)	.00
4. Women’s psych. aggression	−128.00	61.07	−2.10 (43)	.04	−101.51	51.49	−1.97 (43)	.06
5. Men’s psych. aggression	−48.23	52.05	−0.93 (43)	.36	−47.51	43.83	−1.08 (43)	.28
6. Slope (average)	−0.37	0.36	−1.02 (44)	.31	−0.90	0.35	−2.62 (44)	.01
7. Women’s psych. aggression	−1.00	18.11	−0.06 (44)	.96	4.52	17.21	0.26 (44)	.79
8. Men’s psych. aggression	−4.78	15.79	−0.30 (44)	.76	2.81	15.04	0.19 (44)	.85
Women’s Topic
Level 1 Effects
9. Women’s psych. aggression	5.11	5.40	0.95 (734)	.34	4.97	5.44	0.91 (734)	.36
10. Men’s psych. aggression	−16.22	6.72	−2.42 (734)	.02	−12.12	6.75	−1.80 (734)	.07
Level 2 Effects
11. Intercept (sample average T4)	51.36	1.30	39.50 (43)	.00	50.29	0.98	51.11 (43)	.00
12. Women’s psych. aggression	−26.50	45.18	−0.59 (43)	.56	−90.86	34.17	−2.66 (43)	.01
13. Men’s psych. aggression	−96.58	59.12	−1.63 (43)	.11	−69.39	44.72	−1.55 (43)	.13
14. Slope (average)	−0.31	0.38	−0.83 (44)	.41	−0.86	0.36	−2.39 (44)	.02
15. Women’s psych. aggression	−1.86	13.06	−0.14 (44)	.89	−0.50	12.45	−0.04 (44)	.97
16. Men’s psych. aggression	−5.72	17.50	−0.33 (44)	.74	−7.59	−9.07	16.74 (44)	.59

Psychological Aggression During the Women’s Topic

Women’s average psychological aggression predicted lower satisfaction at T4 for their partners (Table 4, line 12), but their aggression during their own problem trial did not relate to lower satisfaction for themselves at either Level 1 or Level 2. Instead, relatively higher levels of the men’s psychological aggression related to lower satisfaction for the women over time (line 10). This model proved a significant improvement in fit over baseline, χ²(14) = 40.38, p < .001. With the exception of the men’s intercept term, this model tended to explain less variance in growth parameters than previous models -- 33.80% of the variance in men’s and 13.67% of the variance in women’s intercepts, 1.22% of the variance in men’s and 3.62% of the variance in women’s slopes.

As for positive engagement, the women’s psychological aggression related most reliably to both partners’ evaluations of the relationship at the between-couple level, though this effect was specific to the topic under discussion. Not surprisingly, higher levels of this negative interaction pattern were associated with lower satisfaction. At the within-couple level, greater psychological aggression on either the man’s or the woman’s part during discussion of the partner’s topic signified a disturbance in the relationship that affected women partners in particular.

Withdrawal² During the Men’s Topic

Withdrawal during the men’s topic showed a positive relationship with satisfaction over time; both partners’ average withdrawal predicted a more positive slope for men’s satisfaction across times (Table 5 top panel, lines 7 and 8). Thus, men in couples that showed greater withdrawal during discussion of his issue also tended to exhibit less decline in relationship satisfaction. At the within-couple level, similarly, relative increases in men’s withdrawal paralleled increases in their own satisfaction (line 2). This model did improve significantly in fit compared to baseline, χ²(14) = 38.77, p = .001. Not surprisingly, most of the model’s explanatory power lay in the slope terms; it explained 3.56% of the variance in men’s and 7.34% of the variance in women’s intercepts, 21.86% of the variance in men’s and 19.31% of the variance in women’s slopes.

Table 5.

The Effect of Withdrawal (log-transformed) on Couples’ Relationship Satisfaction

	Women’s Satisfaction				Men’s Satisfaction
Predictor	Estimate	SE	t(df)	p	Estimate	SE	t(df)	p
Men’s topic
Level 1 Effects
1. Women’s withdrawal	−0.13	0.24	−0.55 (734)	.58	0.05	0.24	0.22 (734)	.83
2. Men’s withdrawal	−0.43	0.30	−1.46 (734)	.14	0.70	0.30	2.37 (734)	.02
Level 2 Effects (Withdrawal as predictor averaged across four time points)
3. Intercept (sample average T4)	46.90	2.12	22.08 (43)	.00	54.92	2.01	27.29 (43)	.00
4. Women’s withdrawal	−0.42	2.02	−0.21 (43)	.84	−1.82	1.74	−1.04 (43)	.30
5. Men’s withdrawal	3.02	2.22	1.36 (43)	.18	0.18	1.91	0.10 (43)	.92
6. Slope (average)	−0.34	0.36	−0.96 (44)	.34	−0.72	0.33	−2.15 (44)	.04
7. Women’s withdrawal	1.03	0.53	1.92 (44)	.06	1.08	0.50	2.15 (44)	.04
8. Men’s withdrawal	0.90	0.59	1.54 (44)	.13	1.53	0.55	2.78 (44)	.01
Women’s Topic
Level 1 Effects
9. Women’s withdrawal	0.02	0.30	0.07 (734)	.95	−0.09	0.30	−0.31 (734)	.75
10. Men’s withdrawal	−0.13	0.26	−0.51(734)	.61	0.19	0.26	0.73 (734)	.47
Level 2 Effects
11. Intercept (sample average T4)	50.70	2.28	22.27 (43)	.00	51.06	2.17	23.57 (43)	.00
12. Women’s withdrawal	−2.54	2.18	−1.16 (43)	.25	−2.05	1.87	−1.10 (43)	.28
13. Men’s withdrawal	2.13	2.26	0.94 (43)	.35	1.25	1.93	0.65 (43)	.52
14. Slope (average)	−0.27	0.38	−0.72 (44)	.47	−0.82	0.35	−2.37 (44)	.02
15. Women’s withdrawal	1.02	0.63	1.62 (44)	.11	0.78	0.59	1.33 (44)	.19
16. Men’s withdrawal	0.21	0.62	0.33 (44)	.74	0.55	0.57	0.96 (44)	.34

Withdrawal During the Women’s Topic

During the women’s topic, none of the links between withdrawal and relationship satisfaction proved significant. Despite the lack of significant effects, this model as a whole improved significantly in fit over baseline, χ²(14) = 25.89, p = .03. As for the men’s topic model, it explained little variance in intercepts -- 5.05% of men’s and 6.88% of women’s -- and somewhat more in slopes -- 15.22% of men’s and 8.80% of women’s.

Although withdrawal has been cast as a negative factor in couples’ interactions, these results suggest that partners’ withdrawal during discussion of the men’s issues can be protective against steep declines in satisfaction for men.

Discussion

The present study adds to both the couples and developmental literatures by tracing the unfolding bonding process in a unique sample of at-risk couples who maintained a stable relationship across early adulthood. The uniqueness of this sample precludes broad generalizations to the large proportion of couples who dissolve early adult relationships, yet it provides an ideal context for identifying behavioral growth patterns that allow vulnerable young couples to cement a lasting bond. Bridging the gap between research on adolescent romance and adult marital relationships, our findings offer a window into the development of romantic relationships and of individual partners in relationship making the transition from adolescent concerns and capacities to adult roles. At the relationship level, the couple is building a more intimate and committed partnership that elicits but also contains and integrates strong emotions (Brown, 1999). At the individual level, partners are striving for both autonomy and support in a way that balances self and other needs, even in the face of conflict (Selman et al., 1986). To interpret the effects of young men’s and women’s conflict behaviors on relationship satisfaction, we consider how these behaviors meet or thwart developmental needs of the partners and the couple.

Current findings suggest that young adult couples may still show adolescent characteristics as they move into adulthood, with distinct roles played by male and female partners. Specifically, we found that couples’ behavioral convergence with normative shifts in conflict engagement from adolescence to adulthood -- greater use of positive engagement and less use of psychological aggression – was generally adaptive, but that the disengagement strategies thought to characterize adolescent couples served a function for young men. These broad effects may be interpreted in light of young adults’ needs for support (satisfied by positive engagement and threatened by psychological aggression) and autonomy (through withdrawal), as well as for an emotional climate that balances negative with positive emotion and allows for regulation of arousal. As suggested by DDS theory, mutual influences on relationship course were apparent in that both partners’ conflict styles impacted changes in relationship satisfaction, with differing effects depending on who exhibited the behaviors and when (during whose problem topic).

The importance of positive engagement as a marker of successful couples development may lie in the fact that it reflects the emergent capacity for negotiation and integration of self and other needs seen in mature relationships (Selman et al., 1986). In particular, young women’s positive engagement benefited couples in this study, and this benefit extended to both partners. At the between-couple level, the women’s average level of positive engagement predicted higher satisfaction, regardless of whose problem was under discussion. Similarly, at the within-couple level, increases in the woman’s positive engagement related to increasing satisfaction for the couple. For the young men, the influence of women partners’ positive engagement may be attributed to greater reliance on partners to provide support and to regulate their conflict-related arousal, as implied by emotional reactivity models of gendered communication (Gottman & Levenson, 1986). The fact that women’s positive engagement was also important for their own satisfaction may speak to a greater sense of accountability for resolving relationship issues; alternatively, this link could reflect greater sensitivity to the relationship climate, with women finding it easier to be positively engaged when they feel good about the relationship. These associations likely point to a remaining gap in socialization of boys and girls, with the latter taught to be the caretaker and to be more attuned to attachment aspects of relationships (e.g., Eagly & Wood, 1999; Shulman & Scharf, 2000). Thus, young women in the study appeared to bear the burden of upholding more “mature” forms of relationship negotiation that maintained couples’ satisfaction.

Similarly, psychological aggression may serve as an impediment to couples development because it reflects less mature coercive tactics and an inability to balance self/other needs effectively. This strategy may be particularly threatening for young adult partners because they are still building a sense of autonomy and assertiveness in the face of conflicting partner needs. Again, women’s behavior figured more prominently as a predictor of relationship satisfaction, but these effects depended on whose problem was under discussion and whether effects were examined within or between couples. The more psychologically aggressive women were during the men’s topic, both in the sense of higher average levels and increases over time, the less satisfied they were, and the more aggressive they were during their own topics, the less satisfied their partners were. If we follow the thread of explanation introduced for positive engagement above, it could be that women’s negativity was particularly distressing for men because they had been counting on their partners to set the tone for the resolution of conflict issues. Similarly, the impact of women’s psychological aggression on their own satisfaction may reflect greater responsibility for, or responsiveness to, the quality of the relationship and the balance of power as they perceive them. Given the cost to the attachment relationship incurred by aggression as opposed to negotiation, the woman’s reversion to relatively immature tactics could signal a more severe disturbance in her sense of the relationship and her ability to meet her autonomy needs in other ways.

Interestingly, this link between young women’s negative behavior and their own satisfaction was limited to behavior during the men’s problem discussion. In the context of women’s own problem discussions, it was men’s psychological aggression that related to lower satisfaction for women. These effects highlight the importance of examining both within-couple and between-couple levels of analysis; without this distinction, the more subtle effect of men’s psychological aggression would have been missed. Although women’s behavior may be more important for the couple’s adjustment overall, these findings suggest that either partner’s aggression while discussing the other person’s problem signals a harmful relationship dynamic. Given that the partner’s issue likely calls for change on one’s own part, individuals might be expected to be more defensive (but not necessarily attacking) during the partner’s problem discussion, and psychological aggression in this instance may signify a desire to control the partner rather than to help in resolving a problem that is important to him/her, threatening the function of the relationship as a mutually satisfying and collaborative partnership.

While the findings for positive engagement and psychological aggression underline the adaptiveness of more mature conflict resolution styles and the importance of the women’s role, the less mature tactic of withdrawal appeared to serve these young adult couples, particularly the men. Similar to the disengagement/minimization pattern found in adolescent couples (Tuval-Mashiach & Shulman, 2006), this strategy may actually help to preserve relationships when negotiation skills and relationship security are relatively undeveloped. Rather than exerting a destructive effect, partners’ withdrawal during the men’s topics predicted more positive outcomes for men in the sense of a more positive satisfaction slope over time. In addition, increases in men’s withdrawal during their topics related to increasing satisfaction for the men. These findings indicate that young men may be closer to adolescents in their capacity to effectively engage in conflict, again due to a gender gap in relationship socialization or in general social-cognitive maturity. Differences in the regulation of arousal may also come into play, as suggested by the Verhofstadt et al. (2005) study; men experiencing greater arousal and negative affect during a discussion of their own problem may find it more beneficial to withdraw from the conflict, decreasing their own arousal. It could be that men’s and women’s expectations for resolution of their issues dictate different interaction styles, with women more comfortable when they are able to voice demands and men finding it more adaptive to use periods of disengagement to regulate the discussion of their issues. As suggested by previous research with young couples (i.e., Florsheim et al., 2003; Welsh et al., 1999), these styles may address young women’s sensitivity to issues of power and young men’s need to exert their autonomy. This would help to explain the lack of negative effects of women’s psychological aggression, and the presence of positive effects of men’s withdrawal, during their own respective trials.

In comparing these results to other demand-withdraw studies, it is important to keep in mind several methodological distinctions. First, most prior studies have not involved moment-by-moment coding of withdrawal, but rather overall ratings of the interaction or codes that include withdrawal along with other observed behaviors. The present study examined duration proportions of both partners’ withdrawal during each discussion topic, which helps to define and quantify the construct of withdrawal, but it does not take into account when the withdrawal occurred. Building on the current findings, sequential analysis of withdrawal patterns might further clarify the role of withdrawal. It could be the case that a pattern of the men’s withdrawal in response to their partners’ requests and demands harms relationship adjustment, but that other patterns (i.e., the man backs down from certain demands he has made, which is more likely during discussion of his issue) are beneficial for the workings of the relationship. Another important point is that the effects on relationship satisfaction in this study were not evident as cross-sectional associations, but rather in the form of slope and time-varying covariate effects. Thus, we cannot state that withdrawal constitutes an absolute good, but rather that it seems to act as a buffer against more serious maladjustment over time. Perhaps most importantly, differences in the functionality of withdrawal could have to do with developmental issues; the couples in the present study were at a less mature stage of relationship and individual development than adult married couples, making withdrawal a potentially more adaptive method to contain negative affect and protect the relationship from dissolution. Replication with different samples would be needed to determine whether this effect holds for most couples at this stage of development or is specifically adaptive for at-risk couples, in which the men’s withdrawal may simply replace more aggressive behavior.

On the basis of this study, we cannot conclude that women are the major shapers of young adult couples’ relationship development in all areas, but rather that men and women tend to exert effects in different aspects of conflict interactions. These differing roles overlap with stereotypical masculine and feminine roles, with men viewed as more disengaged and stoic and women as both more nurturing and more demanding in relationships. This does not mean that men and women are comparatively higher in these characteristics, but rather that they seem to impact the relationship primarily by fulfilling (or departing from) these roles. As suggested above, young women may be more socialized in relationship caretaking and/or further in social-cognitive development, making the more active conflict strategies a part of their role, whereas a combination of less relationship socialization and emotional immaturity/arousability could make withdrawal part of the men’s role. In this way, the results of this study offer insight into the gendered distinctions that still operate in many romantic relationships and how they aid or impair young couples’ navigation of the bonding phase of development.

Findings from the present study underline several points that should guide the future development of couples research. First, observed interaction measures should encompass a range of affect/behavior patterns and should be examined separately for men’s and women’s chosen problem topics; in order to determine the impact of both partners, we must examine not only positively or negatively engaged patterns but also patterns of disengagement in the context of different conflict structures. We can also gain by refining our methodology to examine both within-couple and between-couple levels of analysis in a longitudinal framework. Although some effects are obvious at an overall, between-couple level, others need to be teased out over time within the couple. These levels of analysis also have meaning for prevention/intervention efforts. Although between-couple effects can help to identify young couples at risk for maladjustment, within-couple effects shed light on adaptive and maladaptive processes within the relationship that could be targeted in treatment. Finally, we need to consider the developmental stage of both the individual partners and of the relationship itself to inform study questions and interpretation of results; the structure and function of a “good” relationship can and should change across the lifespan, and the relevance of a particular couples study can be enhanced by clarifying where it fits in the course of development.