Relationship Effort, Satisfaction, and Stability: Differences Across Union Type



Relationship satisfaction and stability are two commonly studied outcomes in marriage and family research. Majority of studies address socio demographic variability and differences across union type in these outcomes. We extend this literature by addressing how the amount of effort one puts into their relationship is associated with stability and satisfaction. Specifically, we focus on how effort impacts these measures of quality in four union types: premarital cohabitation, first marriage, post-divorce cohabitation, and second marriage following divorce. Furthermore, we make union type comparisons in the strength of effort's association with satisfaction and stability. Using data from 8,006 respondents in the Relationship Evaluation Survey, our results show that effort was strongly and positively associated with satisfaction and stability in all four unions. Although effort is more strongly associated with satisfaction in first marriage than cohabiting relationships, no union type differences in the role of effort on stability were observed. Clinical and research implications of these findings are discussed.

Relationship satisfaction and stability are two of the most commonly studied outcomes in family research, and are typically linked to the larger concept of relationship quality (Amato, Booth, Johnson & Rogers, 2007; Hawkins & Booth, 2005). Satisfaction refers to how happy individuals are with several facets of their relationship including intimacy, conflict, and equality, while perceived stability1 can be viewed as attitudes and behaviors associated with dissolution proneness (Amato et al., 2007). Relationship satisfaction and stability are also positively correlated with a number of outcomes, including psychological well-being and physical health (Proulx, Helms & Buehler, 2007; Umberson & Williams, 2005; Umberson, Williams, Powers, Liu & Needham, 2006). Children raised in happy, stable homes are less likely to have internalized and externalized behavioral problems, do better in school, are more likely to get married, and have happy relationships themselves (Amato & Booth, 1997; Amato, Loomis & Booth, 1995; Booth & Amato, 2001; Wolfinger, 2011). Although many sociodemographic and interpersonal explanations have been forwarded to predict relationship satisfaction and stability (Segrin & Flora, 2005), we focus on variation in how the amount of effort put into a relationship affects these outcomes.

We begin with the premise that satisfying and stable romantic relationships do not just happen—they require work. Strong relationships are emotionally supportive, open, involved, and focused on the needs of each partner (Johnson, 2008; Johnson & Greenman, 2006). Couples in high-quality relationships evaluate behavior, set goals, and make changes, as appropriate (Halford, 2011). As such, relationship effort, or how much an individual or couple works at their relationship, has been linked to higher levels of relationship quality (Halford, Lizzio, Wilson & Occhipinti, 2007; Wilson, Charker, Lizzio, Halford & Kimlin, 2005). Yet, with some variability in satisfaction and stability by union type (Brown, 2004; Brown & Booth, 1996; Skinner, Bahr, Crane & Call, 2002), the association between effort and quality may depend on the kind of relationship one is in.

Our article is motivated by the strong association between satisfaction, stability, and dissolution (Amato et al., 2007; Hawkins & Booth, 2005; Johnson, Amoloza & Booth, 1992), and the role effort plays in good relationships (Halford, 2011; Halford, Sanders & Behrens, 1994; Halford et al., 2007; Wilson et al., 2005). We focus on this association in four types of unions: first marriage, never-married cohabitation, divorced cohabitation, and second marriage (remarriage) after divorce. We address not only how effort is associated with quality in each of these unions, but also if there are differences in that association across union type. This question is an important one to marriage and family therapists, social workers, and other clinicians as cohabitation and remarriage have become increasingly prevalent. In fact, recent estimates suggest that up to half of individuals will cohabit and nearly 25% of married men and women will divorce and remarry (Cherlin, 2010). Our data comes from the RELATionship Evaluation survey (often referred to as RELATE), a large commonly-used relationship assessment questionnaire which is uniquely suited to answer our questions. We now turn to a discussion of hypothesized relationships in our article, results from our analyses, and a discussion of the implications from our findings.

Relationship Effort

We focus on relationship effort, which is one aspect of relationship self-regulation (RSR). RSR includes strategies utilized to produce a high-quality romantic relationship and the effort put into a relationship (Halford et al., 1994). Relationship effort consists of individually based behaviors and attitudes that are deemed either effective or ineffective based on a partner's reaction. Thus, self-regulatory actions, such as effort, are both individual and dyadic in nature. Notably, the effort individuals put into their relationships is goal-oriented (i.e., actions will lead to specific results), requires explicit action (i.e., to improve my relationship I will do X or Y), and need to be flexible (i.e., if A does not work, I will do B, C, or D; Meyer, Larson, Busby & Harper, 2012). Although some research has been conducted on the role of effort for relationship satisfaction, no studies of which we are aware address the association between effort and perceived stability in romantic relationships. Notably, we only address the association between effort, satisfaction, and stability, and do not focus on how self-regulatory strategies can help produce high-quality relationships. We do this for three reasons. First, the use of relationship strategies in individuals and couples is quite stable over time (Halford et al., 2007). Second, strategies has a positive effect on relationship satisfaction initially, but has no long-term impact on couples (Halford, 2011). Meanwhile, effort has strong, positive, and significant effects on satisfaction over the course of a relationship. Effort also has a substantially larger association with satisfaction than strategies—even in the first year of a relationship. Finally, many of the strategies utilized by couples are skills learned in marriage education programs, whereas effort is less of a learned skill—although it clearly can be (Halford, 2011).

Satisfaction and Stability

Although we briefly discussed satisfaction and stability in the introduction, it is important to note the relationship between and the potential variability in these two measures of quality.

The Relationship Between Satisfaction and Stability

Relationship satisfaction and perceived stability are often considered “conceptually distinct but empirically correlated” [41] measures (Amato et al., 2007). Scales that combine satisfaction, stability, and other measures of relationship quality are rare because of couple-level variability in the correlation of distinct measures of quality (see Amato et al., 2007 for a full discussion). For example, although satisfaction and stability are strongly correlated with one another (in our data, r = .66), a couple with unstable families-of-origin may be more likely to separate than a couple where both partners were raised by continuously married biological parents (Teachman, 2002). In such cases, perceived instability is common even if relationship satisfaction is high. Similarly, dissolution is not solely the domain of unsatisfied couples. In fact, Amato and Hohmann-Marriott (2007) showed that satisfaction and perceived stability have distinct effects on dissolution, irrespective of how strongly correlated the two are. Viewing satisfaction and stability separately is also important as we use cross-sectional data and linear regression in our article. Although both methodological choices are common in studies of relationship quality (e.g., Addo & Sassler, 2010; Amato & Hohmann-Marriott, 2007; Amato et al., 2007; Skinner et al., 2002), it is difficult to correlate the two measures with one another using such a modeling strategy because it is theoretically unclear if satisfaction leads to stable relationships or stable relationships help generate satisfying unions.2

Differences in Satisfaction and Stability by Union Type

There is some variability in relationship satisfaction and stability by union type—which are commonly compared to first marriage in the literature. Cohabiting couples tend to report lower levels of satisfaction, more conflict, and poorer communication than the first married (Brown, 2004; Brown & Booth, 1996). Yet, there is some variability in satisfaction among cohabiters which impacts the significance of these differences. Cohabiters who plan to marry have higher quality relationships than cohabiters with no such plans and differences between “engaged” cohabiters and married couples in satisfaction are either extremely small or non-existent (Brown & Booth, 1996). However, married couples with a prior cohabitation history do have more unstable relationships than non-cohabiters (Skinner et al., 2002). Similarly, dissolution rates for cohabitation are substantially higher than in first or higher order marriages (Lichter & Qian, 2008; Lichter, Qian & Mellott, 2006). Importantly, comparisons between cohabiting and married couples are almost always between never-married cohabiting and first married couples, while potential comparisons with divorced cohabiters and remarried couples are lacking from the literature. One study notes, however, that cohabitation prior to remarriage is associated with low-marital quality (Xu, Hudspeth & Bartkowski, 2006).

Remarriages appear to be just as satisfying as first marriages. Skinner et al. (2002) found similar levels of happiness, communication, fairness, and disagreement between first and second married couples, while Amato et al. (2007) noted no difference in marital satisfaction in a recent study. Yet, remarried couples have lower perceived stability than the first married. Although slightly less than half of first marriages end in divorce, more than 60% of second marriages will dissolve (Copen, Daniels, Vespa & Mosher, 2012). Furthermore, challenging re-marital stability, divorcès often see divorce as the best solution to marital strife (Amato & Booth, 1991) and maintain low divorce initiation thresholds (Amato et al., 2007). Similarly, most divorcès feel they can handle separation from their romantic partner (Skinner et al., 2002) and have little moral objection to divorce (Booth & Edwards, 1992).

Effort, Satisfaction, and Stability

We expect that effort predicts relationship satisfaction and stability—which is conventional in both effort and RSR models. Working at one's relationship requires one or both partners to reflect on which behaviors produce positive responses, negative responses, and how to change negative behaviors in a way that brings about beneficial change (Halford, 2011; Halford et al., 2007). Couples who work at their relationships do so persistently which, unsurprisingly, has a stronger and more sustainable effect on a relationship than a one-time burst of effort. Hence, our measures of effort are cast in a way that asks respondents to take a long view on their relationship, not a snapshot.

Although we have good theoretical reason to expect that effort leads to increased relationship quality, an alternative specification is possible. In one alternative, relationship quality predicts effort (Halford et al., 1994). This may be due to the attribution of poor and/or unstable relationships to the notion that one or both partners did not work hard enough. In other work, we have run analyses for these two possibilities (our theoretical model and the alternative) using structural equation models (SEM). Although our data were cross-sectional, the best model (according to model fit and highest variance explained) had effort as a predictor of satisfaction and stability. Thus, statistical analyses supported the theoretical model underlying RSR. In choosing such a model we argue, like the majority of RSR scholars, that interpersonal processes are more likely to produce happy marriages than happy marriages are to produce positive interactions. Such a model is congruent with the expectations of marriage and family practitioners, such as social workers or marriage and family therapists.

Potential Union Type Variation in the Association of Effort and Relationship Quality

Building on the notion that RSR, including effort, is shaped by social context (Halford et al., 1994), relationship effort could have a different association with quality in various types of unions. As such, we discuss each union type and how it may mediate the relationship between effort and quality in detail below.

Nearly 90% of individuals will eventually marry—although there is substantial racial, ethnic, and socioeconomic variability in marriage rates (Cherlin, 2010). Although cohabitation, remarriage, and other family forms continue to emerge, marriage is still viewed by most as the best family structure in which to raise children and is often seen as beneficial to both individuals and society (Cherlin, 2004; Smock, 2004). Compared to other relationship types, first marriages are relatively stable, committed relationships (Amato et al., 2007). While slightly less than half of marriages will end in divorce, other union types are more likely to dissolve (Amato, 2010).

There are two possible relationships between effort and relationship quality among first married couples. In one respect, first married couples are in highly committed relationships and have low levels of relational uncertainty which could minimize the link between effort and quality when compared to cohabiting and remarital relationships. In other words, the amount of effort put into their relationships may be less critical for first married couples because they are already in unions which are typically built on good communication skills, high commitment, and compassionate ideals such as love, friendship, and shared interests (Amato et al., 2007; Burgess & Locke, 1960). On the other hand, low relational uncertainty may yield high returns on effort. For example, Young, Curran and Totenhagen (2012) found that highly committed couples see significantly larger returns on positive interpersonal processes (such as effort) than couples in less stable relationships (e.g., cohabitation or remarriage). However, large positive returns of effort may also mean that individuals and couples who put little effort into their relationships may have poor quality marriages.

People's motivation to cohabit may lead to less stable relationships than exists for first married couples. Many people who choose to cohabit do so because they fear a marriage will end in divorce, and they want uncomplicated options to end a relationship (Miller, Sassler & Kusi-Appouh, 2011). Other times the decision to cohabit is less conscious. Some slide into cohabitation because they need housing, want to benefit from economies of scale, pregnancy, or simply out of convenience. Cohabiters are less likely than married couples to enter their unions on the basis of compassionate ideals or the desire for a stable partnership (Brown, 2004; Brown & Booth, 1996; Huang, Smock, Manning & Bergstrom-Lynch, 2011). They also tend to hold much more liberal attitudes about divorce, sex, the importance of marriage, and the desire for long-term committed relationships than married men and women (Willoughby & Carroll, 2012). On the one hand, this may lead cohabiters to benefit more from effort than married couples because it can reduce relational uncertainty, signal high commitment, and illustrate a desire to take the relationship to the next step (e.g., marriage and/or childbearing). Although we expect that effort is positively associated with relationship quality among cohabiters, it is also possible that the association is less positive than in first marriage because of high relational uncertainty present in cohabitation (Young et al., 2012).

Remarried couples face a number of unique challenges which may be met with increased effort to build high quality relationships. Unlike first married and never-married cohabitating couples, remarried people often have close ties to an ex-spouse, non-normative stepparent-stepchild relationships, must negotiate family roles and boundaries, and regularly try to integrate family cultures (Brimhall, Wampler & Kimball, 2008; Coleman, Fine, Ganong, Downs & Pauk, 2001; Falke & Larson, 2007). Other remarital characteristics, such as heterogeneous partnerships (Shafer, 2012), mean partners may have difficulty working together to build positive relationships. Unclear behavioral norms in remarriage (Sweeney, 2010) may lead to misdirected and unhelpful efforts which fail to improve satisfaction or stability (Meyer et al., 2012). The selective attributes of divorcès, such as their inclination toward divorce and high gender/partner distrust due to first marriage experiences, may also undermine the work each partner puts in (Amato & Booth, 1991). While we expect effort to have smaller effects on satisfaction and stability for remarried couples because of these contextual factors, as they may lead to unstable or uncertain relationships, they could also be associated with a strong, positive relationship between effort and quality.

Overall, our expectation is that effort will be positively associated with satisfaction and stability in all union types. Although our prior discussion focused on the possibility that this positive effect will differ across unions, it is also possible that no or very small differences may be observed. This pattern would be consistent with the notion that strong, stable, and satisfying relationships are possible if couples continuously work at their partnership, are emotionally available to one another, involved, and responsive to each other's needs (Johnson, 2008; Johnson & Greenman, 2006). If such a finding is observed, it suggests that context matters less than the act of putting effort into the relationship. As such, working at one's relationship may lead to trust, love, relational security, interdependence, and beneficial behaviors such as good communication and agreed-upon boundaries (Huston, Caughlin, Houts, Smith & George, 2001; Johnson & Greenman, 2006).

Research Questions

To summarize, we focus on two main research questions:

  1. What impact does relationship effort have on stability and satisfaction in first marriages, pre-marital cohabitation, post-divorce cohabitation, and remarriage?
  2. How does the association between relationship effort and relationship quality, as measured by satisfaction and stability, vary across union type?


Sample and Procedures

We use the RELATE, a secondary data set gathered from an online relationship assessment questionnaire taken by individuals in committed romantic relationships (Busby, Holman & Taniguchi, 2001). RELATE offers feedback to couples on potential problem areas in their partnerships and is available to researchers as rich, in-depth data on romantic relationships. Respondents were asked to complete the survey independently of their partner. Importantly, the data are not from a random sample, but instead primarily consists of college students in family-focused courses, participants in couple workshops, couples seeking therapy, and individuals who happened upon the questionnaire via web search. These procedures prevent us from claiming that our results are generalizable in any way. However, data are large, diverse, includes batteries of questions not found in other data sources and are commonly used (e.g., Busby, Holman & Niehuis, 2009; Busby, Holman & Walker, 2008; Busby et al., 2001; Meyer et al., 2012). In fact, the sample we use is large and includes cohabiters, although prior studies had only previously focused only on married couples (Halford et al., 2007, 2010; Wilson et al., 2005). Therefore, it is likely data comes from a less select group than many other studies of effort, satisfaction, and perceived stability.

Our sample consisted of 8,006 heterosexual males (38%) and females (62%) in one of four relationship types: married for the first time (N = 3,745, 47%), cohabiting with no previous marriages (N = 2,693, 33%), cohabiting following their first and only divorce (N = 1,005, 13%), and married for the second time (N = 563, 7%). Participants were between 17 and 77 years of age with a mean age of 31.4 (SE = 0.106). Table 1 provides the sociodemographic characteristics of the entire sample and for each group.

Table 1. Descriptive Statistics for Independent Variables for Full Sample and by Union Type
 Full sample (n = 8,006)First married (n = 3,745)Never-married cohabit (n = 2,693)Divorced cohabit (n = 1,005)Remarried (n = 563)
RangeImputed (%)MeanSEMeanSEMeanSEMeanSEMeanSE
  1. Source: RELATE.

    Note. Superscript letters next to mean values indicate significant mean differences between that group and the other specified group(s). a = first married, b = never-married cohabit, c = divorced cohabit, and d = remarried. These mean differences are then provided for effort, satisfaction, and stability.

Relationship effort1–503.2130.0083.148bc0.0113.245ac0.0133.366abd0.0223.216c0.027
Relationship satisfaction1–50.133.6660.0103.563bcd0.0163.818ad0.0153.818ad0.0243.349abc0.042
Relationship stability1–50.234.0530.0094.130bcd0.0134.012ad0.0154.022ad0.0253.798abc0.039
Sociodemographic characteristics
Relationship status
First marrieda0–100.468    
Never-married cohabitationb0–100.333    
Divorced cohabitationc0–100.130    
Educational attainment
High school or less0–100.0530.0460.0520.0660.074
Some college0–100.3840.4580.2950.3450.374
Bachelor's degree0–100.2370.1990.2980.2610.158
Graduate education0–100.3260.2960.3530.3270.394
Non-Hispanic Black0–100.0500.0470.0570.0420.051
Non-Hispanic White0–100.8050.8260.7750.7930.831
Other race/ethnicity0–100.0540.0450.0630.0640.054
Latter-day Saint0–100.2430.4770.0160.0140.178
Other religion0–100.1510.1100.1740.2190.192
No religion0–100.2290.1220.3630.2950.185
Frequently attend church0–100.3190.5690.0630.0850.302
Family characteristics
Marital orientation1–503.4340.0103.8330.0153.0580.0153.0190.0233.3270.036
Parents divorced0–10.300.2990.2360.3480.3880.314
Family-of-origin quality1–502.9000.0102.9550.0152.9330.0162.7860.0292.5870.041
Biological child0–119.490.4320.5590.1170.5770.812
Non-biological child0–119.490.2040.0940.1220.5790.638
Relationship duration1–1104.7870.0215.1720.0404.5690.0204.4790.0333.8110.058

All four groups had similar educational and race/ethnicity compositions. About 81% of the sample was Caucasian, 5% African American, 5% Latino, 4% Asian, and 5% some other racial or ethnic identification. The sample was better educated than the U.S. as a whole, with nearly 24% of the sample having earned a Bachelor's degree and 33% attending or completing a graduate or professional school, 39% attended college but lacked a bachelor's degree, with the remaining 5% having earned a high school diploma or less.

Our sample was also not representative on religious affiliation. Twenty-three percent of the participants were Protestant, 24% were Latter-day Saints (LDS; Mormon), 15% were Catholic, 15% identified with other groups, and 23% were not affiliated with any religion. Table 1 shows that the proportions of religious groups were notably different between the first married, never-married cohabiting, post-divorce cohabiting, and remarried groups. As a result, we controlled for religious differences in our models. Finally, nearly 56% of first married respondents had biological children, as did 58% of divorced cohabiters and 81% of remarried respondents. Respondents with non-biological children were rare in first marriage and never-married cohabitation and much more common among previously-married individuals.


Relationship effort

Relationship effort measured how willing and able individuals put forth effort into their relationships. The scale comes from a modified version of questions used by Halford et al. (1994) and Wilson et al. (2005) on relationship effort and has been utilized in prior research (Meyer et al., 2012). This variable is a four-item scale, using the following questions: “If things go wrong in the relationship I tend to feel powerless”, “I tend to fall back on what is comfortable for me in relationships, rather than trying new ways of relating”, “Even when I know what I could do differently to improve things in the relationship, I cannot seem to change my behavior”, and “If my partner doesn't appreciate the change efforts I am making, I tend to give up.” We reverse coded the variables from 1 to 5 so that higher values indicated more effort. Cronbach's alpha is .72 for this measure, indicating good consistency and reliability. We provide the reliability coefficients for this measure, satisfaction, and stability by group in Table 2.

Table 2. Descriptive Statistics, Reliability Coefficients, and Correlations Between Latent Exogenous and Endogenous Variables For All Groups
  1. Note. *p < .05, **p < .01. The variables in this table are composite scales for the purpose of assessing overall correlations, means, and standard deviations.

Relationship satisfaction
First married.513**  
Never-married cohabiting.468**  
Post-divorce cohabiting.461**  
Relationship stability
First married.409 **.697 ** 
Never-married cohabiting.396 **.653 ** 
Post-divorce cohabiting.425 **.672 ** 
Remarried.368 **.684 ** 
Reliability (Cronbach's Alpha)
First married.720.920.779
Never-married cohabiting.716.869.801
Post-divorce cohabiting.703.881.805

Relationship satisfaction

Relationship satisfaction measured how happy respondents were with various dimensions of their relationship. Although standardized scales of satisfaction are not common in secondary data, the questions available in RELATE have been used frequently in the satisfaction literature (e.g., Amato & Hohmann-Marriott, 2007; Amato et al., 2007). Individuals were asked about their satisfaction with seven different aspects of their relationship: physical intimacy, love, conflicts, equality, time spent together, communication, and a global measure of satisfaction. Items were scored on a 1–5 scale with higher scores indicating greater satisfaction. Cronbach's alpha for this scale was .91 for the full sample.

Relationship stability

Relationship stability measured the respondent's perception of relationship insecurity. Like with satisfaction, standardized measures of stability are not commonly available in secondary data. However, RELATE includes a battery of questions used in prior studies which are empirically associated with dissolution (e.g., Amato & Hohmann-Marriott, 2007; Amato et al., 2007). This measure consisted of three questions asking respondents how often they thought about ending their relationship, discussed dissolution, and frequency of break-up, separation, and reunion. Scores ranged from 1 to 5 with higher scores indicating less frequency or more stability. Cronbach's alpha for this scale was .79 (See Table 2 for reliability coefficients by group).

Control variables

We control for various individual characteristics because they may be associated with relationship quality and effort (Amato et al., 2007; Meyer et al., 2012). This includes several sociodemographic characteristics such as race, socioeconomic status, age, and gender. We control for self-identified race and ethnicity with a set of dichotomous variables for non-Hispanic Black, non-Hispanic White, Latino, Asian, and other racial/ethnic group. We include control variables for two socioeconomic characteristics: educational attainment and income. Education was measured through dichotomous measures for high school degree or less, some college, including an Associate's degree, Bachelor's degree, and graduate/professional education (degree or attending). Income is an ordinal variable with values ranging from 0 = none to 11 = $300,000 or greater. Age was measured with a continuous variable whereas gender was identified with a dichotomous variable with 0 = male and 1 = female.

Religion and other attitudes have also been linked to relationship quality. Self-identified religious affiliation was measured with dichotomous variables for Catholic, Protestant, LDS (or Mormon), other religious affiliation, and no religion. Notably, our sample is disproportionately LDS which is likely due to the initial focus of RELATE on the Intermountain West, where over half of all American Mormons reside. Furthermore, the majority of LDS respondents in our sample are first married. The lack of self-identified Mormons in the cohabiting and remarried after divorce groups likely coincides with LDS beliefs about pre-marital sex, non-marital cohabitation, and substantially lower divorce rates among Mormons, compared to the national average (The Church of Jesus Christ of Latter-day Saints, 2012).

Commitment to marriage and family, as an institution, has been associated with relationship quality (Amato et al., 2007) and may influence effort, as well. We use a four-item standardized scale (α = .74), consisting of the following items: “Being married is among the one or two most important things in life”, “If I had an unhappy marriage and neither counseling nor other actions helped, my spouse and I would be better off if we divorced”, “Marriage is a covenant, not just a legal contract recognized by the law”, and “Living together is an acceptable alternative to marriage.” All variables were coded on a 1–5 scale with higher scores indicating greater commitment to marriage. Intergenerational transmission of marital behaviors and attitudes is also extremely common (Teachman, 2002). As a result, we include a measure for family relationship quality in the family-of-origin, which is a four item scale asking if the respondent's family growing up was safe and warm, confusing and unfair, happy, and if their childhood was happy. Items were scored 1–5, with higher scores indicating greater satisfaction. Cronbach's alpha for this scale was.86.

Finally, we control for three relationship-specific characteristics. Relationship duration was specified by participants selecting one of the various duration options ranging from 1 = 0 to 3 months and 11 = More than 40 years (outcomes not interpretable in year units). The presence of children has been linked to low levels of relationship quality—especially among remarried couples (Amato, Kane & James, 2011; Cole & Cole, 1999; Goldscheider & Sassler, 2006; Golish, 2003). As a result, we include dichotomous variables for the presence of biological and non-biological children, relative to the respondent, in the household with dichotomous variables.

Analysis Strategy

We use ordinary least-squares (OLS) multiple regression in our analysis of effort, satisfaction, and stability. Alternative analyses using SEM group comparison models, not presented here, were also run and are available from the authors upon request. However, we present OLS models because they had substantively similar results, allowed us to integrate more control variables, and simplified the interpretation of union type comparisons. We used multiple imputation in Stata 12 to account for a small number of missing cases on satisfaction, stability, and some of our control variables. The percent of missing cases on which we imputed are available in Table 1. The data appeared to be missing at random or missing completely at random, allowing for the use of multiple imputation procedures (Enders, 2010). Imputing on the dependent variable can be controversial (Von Hippel, 2007), but is now common practice—especially in cases like ours where <2% of cases were missing. The most commonly imputed variable was the presence of biological children, where <25% of cases were imputed. Comparisons to models using list-wise deletion were substantively similar to the results we present using imputation methods.

Our analytic strategy proceeds in the following manner: first, we analyze data by running union type specific regression models for satisfaction and stability; second, we include all union types in a model which provides differences in our measures of quality across relationships; finally, we include interactions between union type and effort to assess differences in the effect of effort on satisfaction and stability by relationship.


Relationship Satisfaction

Table 3 reports the regression results for relationship satisfaction on relationship effort and control variables by union type. We find that relationship effort was positively associated with satisfaction in all four relationship types we examined. In first marriage, a one point increase in effort is associated with a 0.599 point increase in satisfaction. The corresponding effects for non-marital cohabitation, post-divorce cohabitation, and remarriage are 0.465, 0.452, and 0.588, respectively. To understand the magnitude of these relationships, we provide standardized regression coefficients of effort.3 In all relationship types, a one standard deviation increase in effort was associated with an approximately 0.4 standard deviation increase in satisfaction, indicating a relatively strong association between effort and satisfaction. All effects are significant at p < .001. Some of the control variables are also significantly associated with satisfaction. In first marriage, women and Blacks have lower levels of satisfaction than men and whites. Religious affiliation, the presence of biological children, and relationship duration are also negatively associated with satisfaction. High levels of marital orientation and family-of-origin quality are associated with higher levels of satisfaction among the first married. Similar effects for race/ethnicity, family quality, biological children, and relationship duration are seen in most other relationship types. The presence of non-biological (step) children in a home is associated with lower levels of quality among remarried couples—potentially because parental expectations for stepparents are significantly lower than for biological parents (Coleman et al., 2001).

Table 3. OLS Regression of Relationship Satisfaction on Relationship Effort and Control Variables
 First marriageNon marital cohabitingDivorced cohabitingRemarriage
  1. Source: RELATE.

    Notes. aParenthetical values are standardized regression coefficients for relationship effort; bReference is graduate education; creference is non-Hispanic White, dreference is Latter-day Saint. ***p < .001, **p < .01, *p < .05 (two-tailed tests).

Relationship efforta0.599 (0.419)0.019***0.465 (0.404)0.019***0.452 (0.403)0.033***0.588 (0.381)0.060***
Sociodemographic characteristics
Educational attainmentb
High school or less−0.0260.0650.0340.063−0.1450.095−0.0560.154
Some college0.0170.034−0.0120.035−0.0530.056−0.0720.093
Bachelor's degree−0.0280.0360.0250.0310.0020.060−0.1150.110
Non-Hispanic Black−0.2130.063**−0.1320.058*−0.3650.106**−0.7040.173***
Other race/ethnicity−0.1210.060*0.0180.052−0.0930.087−0.3270.162*
Other religion−0.1010.051*−0.0030.092−0.0100.1760.0280.145
No religion−0.0510.059−0.0280.090−0.0820.1780.0710.164
Frequently attend church−0.0090.0390.0380.0530.1430.079−0.1330.103
Family characteristics
Marital orientation0.1480.023***0.0350.019−0.0160.0320.1910.057**
Family-of-origin quality0.1270.015***0.1370.015***0.0770.025**0.1420.040***
Biological child−0.3080.033***−0.3610.067***−0.2730.077**−0.4770.133**
Non-biological child−0.0210.045−0.0070.067−0.0600.0790.2280.087**
Duration of relationship−0.0280.009***−0.1190.012***−0.1190.021***0.0360.030
Constant1.351 2.822 3.099 0.988 
N 3,745 2,693 1,005 563 
R 2 .413 .311 .299 .262 

Relationship Stability

The analogous results for relationship stability are reported in Table 4. Like with satisfaction, effort has strong positive effects on stability in all union types. In first marriage, a one point increase in effort is associated with a 0.372 point increase in stability. The corresponding results for other relationship types is 0.411 in non-marital cohabitation, 0.422 in divorced cohabitation, and 0.447 in remarriage. Like with satisfaction, the standardized coefficients of effort indicate that effort is strongly related to stability in all union types. All effects are significant at p < .001. The results for the control variables are also reported in Table 4 and are very similar to those observed for satisfaction.

Table 4. OLS Regression of Relationship Stability on Relationship Effort and Control Variables
 First marriageNon marital cohabitingDivorced cohabitingRemarriage
  1. Source: RELATE.

    Notes. aParenthetical values are standardized regression coefficients for relationship effort; breference is graduate education; creference is non-Hispanic White; dreference is Latter-day Saint. ***p < .001, **p < .01, *p < .05 (two-tailed tests).

Relationship efforta0.372 (0.304)0.017***0.411 (0.339)0.021***0.422 (0.370)0.034***0.447 (0.318)0.056***
Sociodemographic characteristics
Educational attainmentb
High school or less−0.2200.059***0.0290.069−0.2680.097**−0.0950.144
Some college−0.0590.0310.0200.038−0.0950.058−0.1550.087
Bachelor's degree−0.0320.0330.0500.0340.0220.061−0.0430.103
Non-Hispanic Black−0.4190.057***−0.3900.064***−0.6260.109***−0.8280.162***
Other race/ethnicity−0.1070.055*−0.0670.056−0.2260.090*−0.2600.151
Other religion−0.0870.0460.2120.100*0.0840.181−0.0070.135
No religion0.0850.0530.2800.098**0.1030.1840.2190.152
Frequently attend church0.1040.036**−0.0180.0570.0360.081−0.0270.096
Family characteristics
Marital orientation0.1890.020***0.0400.0200.0290.0330.1720.053**
Family-of-origin quality0.1080.013***0.1170.017***0.0810.026**0.1640.037***
Biological child−0.2090.041***−0.2530.075**−0.1410.115−0.3710.124**
Non-biological child−0.2240.041***−0.1660.060*−0.1330.0990.0010.083
Duration of relationship−0.0300.009**−0.1470.014***−0.1050.021***−0.0050.028
Constant2.179 2.799 2.993 1.584 
N 3,745 2,693 1,005 563 
R 2 .341 .261 .284 .225 

Differences in Satisfaction and Stability by Union Type

Table 5 shows the main effects of relationship type on satisfaction and interactive effects between effort and union type.4 Although we present only the key variables in this table, the full set of controls presented in Tables 3 and 4 are included in each interactive model. Our main effects model shows that both non-marital and divorced cohabiters have slightly higher relationship satisfaction than first married couples in our data. Supplementary analyses found that relationship duration and age differences significantly mediated these differences5.

Table 5. Main Effects and Interactive Models of Union Type Differences in Satisfaction and Stability
β (SE)β (SE)β (SE)β (SE)
  1. Source: RELATE.

    Notes. Full controls are included in both the main effects and interactive models. Controls had substantively similar results to those presented in Tables 3 and 4. ***p < .001, **p < .01, *p < .05 (two-tailed tests).

Relationship effort0.540*** (0.013)0.604*** (0.018)0.405*** (0.013)0.385*** (0.018)
Union type (vs. first marriage)
Non-marital cohabitation0 191*** (0.024)0 591*** (0.090)−0.084** (0.027)−0.193* (0.089)
Divorced cohabitation0.294***(0.032)0.836*** (0.126)0.043(0.031)−0.014 (0.123)
Remarriage−0.043(0.039)−0.083 (0.167)−0.168*** (0.039)−0.454** (0.162)
Effort*non-marital cohabitation −0.126***(0.027) 0.034(0.026)
Effort*divorced cohabitation −0.166*** (0.037) −0.014 (0.123)
Effort*remarriage 0.011(0.050) 0.089 (0.049)
N 8,0068,0068,0068,006
R 2 .368.371.292.292
Figure 1.

Union type differences in relationship effort on satisfaction. Only statistically significant effects are shown in the figure.

Our interactive model, which indicates union type differences in the effect of effort shows that effort has smaller positive effects on relationship effort in non-marital and divorced cohabiting relationships. However, at the lowest levels of effort cohabiters appear to have slightly higher satisfaction than first married couples. This may be due to low expectations in cohabiting couples. In other words, due to the nature of cohabitation and why people cohabit, effort may be less expected among cohabiters than married couples. Yet, at higher levels of effort, married couples see slightly higher satisfaction than cohabiters. This is illustrated in Figure 1 which shows the expected value of relationship satisfaction by effort in first marriage, never-married cohabitation, and divorced cohabitation. Remarriage is excluded because of non-significant differences. Perhaps most importantly, however, is the fact that effort has substantial positive effects on relationship satisfaction among all groups and the differences, although significant, are less substantial than we expected.

Effort-relationship type interactions for relationship stability are also provided in Table 5. Notably, the main effects model indicates that relationship stability is significantly lower in never-married cohabiting and remarried couples than first married couples. Yet, the interactive model shows no differences in the slope of effort on stability. Thus, effort has a similar positive effect on all relationship types. But, unlike satisfaction, the initial differences in stability are not overcome by additional relationship effort.

Discussion and Conclusion

Stability and satisfaction are two common ways to think about how couples are doing. Prior research has shown these outcomes are affected by sociodemographic characteristics and interpersonal processes (e.g., Amato et al., 2007). Although it is commonly understood among scholars, practitioners, and the general public that good relationships require work, research addressing this notion is uncommon. The important research and clinical implications of effort for happy and stable relationships underscores our article, which focused on the association between effort and relationship satisfaction/perceived stability in different romantic relationships and possible differences across union type.

We found that working at your relationship is positively associated with satisfaction and stability, no matter the type of committed relationship. Although our findings are similar to the results of prior studies, our major contribution is in the use of a large sample and our focus on differences in the association between effort and quality in various types of relationships. Prior studies were limited almost exclusively to married couples, the vast majority of which were in first marriages (e.g., Halford et al., 2007; Wilson et al., 2005). Thus, relationship effort seems important to develop high-quality relationships, particularly in relationships that are traditionally thought of as unstable and unsatisfying—such as cohabitation and remarriage.

We expected the association between effort, satisfaction, and stability to vary across union type. Notably, we argued that much of this might be associated with the relational uncertainty inherent to cohabiting and remarital unions, when compared to first marriages. With respect to satisfaction, we did find differences by union type. Despite slightly higher satisfaction at the lowest levels of effort for cohabiters, the returns on relationship effort are slightly lower for cohabiters than first married couples. The higher levels of satisfaction at very low levels effort for cohabiters may be linked to comparatively low relationship expectations, instability, and other characteristics which distinguish cohabiters from married couples (Brown, 2004; Cherlin, 2010). Yet, the most important takeaway from our article should be the nearly universal positive effects that effort has on satisfaction, regardless of union type.

We also found a positive association between effort and relationship stability in all four relationship types. But, we found no statistically significant differences in the effect of effort on stability by union type. Although we acknowledged this possibility, we are somewhat surprised by these findings given the comparative instability of cohabiting and remarital unions and the potential positive returns which could result from effort in such relationships. As a result, we can say that relationship effort is positively associated with perceived stability in all relationship forms, and the size and strength of that association is similar regardless of one's union status.

Importantly, we do not advocate for a particular therapeutic model or form of relationship education. Instead, a common factors approach (Sprenkle & Blow, 2004) might be most appropriate for understanding and applying our findings since our data suggests a positive association between effort and relationship quality—which could be achieved in various ways. A common factors framework suggests that no one therapy or model is necessarily better for clients than another. Instead, the act of getting relevant, applicable, and competent therapy has positive effects. A recent meta-analysis by Hawkins, Blanchard, Baldwin and Fawcett (2008) showed that various relationship education programs have very similar effects on increases in satisfaction between pre- and post-test. This suggests to us that various programs and approaches are valid for producing positive relationships and committing to improving one's relationship is a potential key to worthwhile relationship education and marital therapy (Markman & Rhoades, 2012). Similarly, the notion that one is doing something to develop positive relationships might be a key component to any connection between effort and relationship quality (Lebow, Chambers, Christensen & Johnson, 2012).

However, it is worth noting that although we do not advocate for one particular therapeutic or educational model, there are still important aspects of the client-therapist relationship which should not be ignored. For example, according to advocates of the common factors approach (Blow, Davis & Sprenkle, 2012; Simon, 2012a,b), marital and family therapists, social workers, relationship educators, and other practitioners should align their therapeutic goals and efforts with those of clients. In other words, a therapist's notion of change should match that of their client, which in turn improves the chances of having a positive impact on a relationship. Thus, to the extent that the amount of effort a couple puts into their relationship is something they are willing to change, it may prove a helpful strategy to create satisfying and stable relationships. Of course, this recommendation should be tempered by the fact that our data does not come from therapy or educational programs and, as a result, cannot be construed as causal. As such, our recommendations are speculative and require testing in educational and therapeutic settings.


In addition to the representativeness of our data, which we discussed in the method section, our study is limited in other ways as well. We can only view relationship effort, satisfaction, and stability through the lens of one partner. Although we include a question about the respondent's behavior and reaction to their partner's behavior, we simply lack a full assessment of respondent behavior, partner reactions to this behavior, and vice-versa. We suspect that relationship effort is really a combined effort of both partners and their reactions to these efforts. An additional limitation of our study is that we lack longitudinal data on the relationship between effort, stability, and satisfaction. This is likely a dynamic association, changing over time, with significant life events, and with the general ups and downs of life in a committed relationship. Unfortunately, we can only provide a snapshot measure in our study. Finally, cohabiting relationships are heterogeneous, which we cannot capture in our study. Some people are engaged cohabiters, others are in trial marriages, whereas others have slid into living together. This sort of variability may have important implications for how effort affects stability and satisfaction.


To summarize our results for practitioner applicability, we find that couples who work at their relationships have good relationships. This work includes recognition of potential problem areas in a relationship, how one's behavior and attitudes contribute to the problem, effort put forth to change behaviors, and flexibility in strategies after feedback. These are things that practitioners can help couples learn. How big can the positive effect of effort be? Certainly, our study shows that effort is associated with substantial positive increases in satisfaction and perceived stability. In prior research, Halford et al. (2007) found that effort had even stronger effects on relationship satisfaction than communication quality. In this way, relationship effort may be a foundational behavior for other positive relationship attributes. We want to caution, however, that relationship effort is not a catch all. Substantial variability in satisfaction and stability remain. And, although effort was positively correlated with quality in all union types, we did find some union type differences in its association with satisfaction. Therefore, our study may also signal that a relationship cannot be improved simply by working harder or having more flexible strategies to create satisfying and stable relationships—particularly in unique and challenging union types.

Interestingly, we find that effort is strongly associated with perceived stability in the least stable union types. As a result, practitioners working with cohabiters who desire long-lasting, positive relationships could help these couples learn how to put forth greater effort in their relationships. Relationship effort could also improve the satisfaction and stability of remarried couples. However, remarried couples have the lowest levels of effort, which may be influenced by ties to their former spouse or issues with stepchildren (Meyer et al., 2012). Given this, it appears that remarried couples need to put more effort into their relationships than first married couples and take a holistic approach to their new families by working on their own relationships and relationships with children.

For researchers, our study focused on an overlooked interpersonal process in romantic relationships. To date, effort and RSR have been largely ignored in the marital quality literature. This is unfortunate, because a more substantial understanding of interpersonal dynamics and variability in these dynamics could help us develop richer and more robust theories of satisfaction, stability, and other aspects of relationship quality. Therefore, we see our article as a jumping-off point for future inquiry into the effect of effort and other aspects of RSR in marital and non-marital unions.


  1. 1

    We use perceived stability and stability interchangeably for sake of parsimony.

  2. 2

    Nevertheless, we did use stability as a control variable in our analysis of satisfaction and vice-versa. The association of effort with measures of quality was substantively similar in both magnitude and statistical significance. In supplemental analyses we used structural equation modeling and correlated the two measures together which produced substantively similar results, as well. However, we do not report the results here because the inclusion of different measures of quality in models predicting other dimensions of quality is extremely uncommon in the literature (Amato et al., 2007).

  3. 3

    Standardized values are not reported for control variables because they are not appropriate for dichotomous or categorical variables (Fox, 1997). Standardized values for continuous control variables are available upon request.

  4. 4

    Notably, standardized regression coefficients cannot be fit for interaction effects (Fox, 1997). In order to show the size of differences across union type, we provide Figure 1 which we discuss in detail below.

  5. 5

    Interactions between union status and these variables showed there were no significant differences in intercept and slope between first married and never-married cohabiters. Although we control for age and relationship duration in all models, these results suggest that this unexpected result is likely due to differences in the subsamples of cohabiters and married couples. Although not presented here, these results are available from the authors upon request.


  • Kevin Shafer, PhD, Todd M. Jensen, BS, and Jeffry H. Larson, PhD, School of Social Work, Brigham Young University.