Abstract
Although a large literature links relationship discord with depression, most prior studies have evaluated relationship discord as a continuous variable and have examined this association in younger couples. Based on findings suggesting that the latent structure of marital discord is taxonic, this study evaluated whether taxon status was associated with depressive symptoms in middle-aged and older individuals and whether taxon status accounted for unique variance in depressive symptoms, over and above that which was accounted for by continuous measures of marital quality. This association was examined in a community sample of 502 couples who had been married at least 10 years and in which at least one spouse was 55 to 75 years old. There were statistically significant differences between taxon (discordant) and complement (nondiscordant) women and men in depressive symptoms, with discordant spouses reporting higher levels of depressive symptoms than nondiscordant spouses. For wives but not for husbands, taxon status contributed to symptoms beyond the effect of the mean of continuously distributed marital quality scales, indicating a point at which couple taxon status incrementally predicts a significant increase in depressive symptoms. Implications for clinical assessment and further research are discussed.
Relationship (e.g., marital) discord is associated with the onset, course, and treatment of depression (for a review, see Beach & Whisman, 2011). For example, relationship discord is concurrently associated with depressive symptoms (for a meta-analysis, see Whisman, 2001) and depressive disorders (e.g., Whisman, 2007). Longitudinal research indicates that relationship discord predicts increases in depressive symptoms (e.g., Beach, Katz, Kim, & Brody, 2003) and incidence of depressive disorders (e.g., Overbeek et al., 2006; Whisman & Bruce, 1999) and that relationship discord and depressive symptoms covary over time within individuals (e.g., Davila, Karney, Hall, & Bradbury, 2003).
Although informative, there are two limitations with this body of research. First, existing research on relationship discord and depression has relied exclusively on continuous measures of discord. However, emerging research suggests that discordant couples differ from nondiscordant couples not only quantitatively but qualitatively as well. Taxometric methods, which examine covariation among indicators (e.g., test scores) to identify patterns diagnostic of either latent categories (i.e., taxa) or dimensions, make it possible to evaluate empirically whether dimensional scales are derived from an underlying discrete entity or taxon (Waller & Meehl, 1998). As such, a taxon is a nonarbitrary latent category whose members differ qualitatively from nonmembers. Accordingly, a taxon cut point, when one can be identified, should indicate a point at which the impact of variation in marital quality may relate differently to other outcomes, such as variation in depression, leading to a point at which impact on depressive symptoms is increased. To the extent that such points of noncontinuity are present, they may help to inform theorizing about the mechanisms linking relationship discord to depression and may help guide decisions regarding treatment.
Beach, Fincham, Amir, and Leonard (2005) examined the taxometric structure of marital discord in newlywed couples and found evidence for a marital discord taxon with a base rate of .20. A more recent study involving 1,020 couples who were representative of the U.S. population with respect to race and ethnicity, educational level, and occupation also found that marital discord was taxonic, with a base rate of .32 (Whisman, Beach, & Snyder, 2008).
A second limitation with existing research on relationship discord and depression is that most studies have focused on middle-aged and younger participants. Between 2004 and 2024, the U.S. population in the 55–64 age-group is projected to increase by 43%, and the population in the 65 and older age-group is projected to increase by 70% (Centers for Disease Control and Prevention, 2005). Research suggests that depression in older adults, compared to younger and middle-aged adults, differs in prevalence, symptoms, etiology, and course (for a review, see Fiske, Wetherell, & Gatz, 2009). Existing research suggests that relationship discord is concurrently (e.g., Bookwala & Jacobs, 2004) and longitudinally associated with depression in older adults (e.g., Whisman & Uebelacker, 2009). Relationship quality may be particularly important for understanding depression in older individuals, as the strength of the association between marital discord and major depressive disorder increases in magnitude with increasing age (Whisman, 2007). Results from these studies support continued investigation into the association between relationship discord and depression among older individuals.
This study evaluated whether marital discord taxon status was associated with depressive symptoms in middle-aged and older adults, and, if so, whether this association was comparable to, and potentially incremental to, the association between continuous measures of marital discord and depressive symptoms.
Method
Participants
As described elsewhere (Sandberg & Harper, 2000), married couples in which at least one partner was 55 to 75 years old were recruited to participate in a study on the impact of retirement on physical and mental health. Questionnaires were sent to 9,328 addresses selected at random from a sample of couples from each state in the United States. Of the 9,328 mailed questionnaires, 1,611 were returned. The current analyses are based on 502 couples who had been married at least 10 years (M = 38.8 years, SD = 8.9) and in which at least one spouse was 55 to 75 years old. The mean age was 65.3 years (SD = 4.2) for husbands and 63.6 years (SD = 4.2) for wives, and couples had an average of 3.0 children (SD = 1.6). Most of the sample (98%) was White.
Measures
Marital Satisfaction Inventory, Revised
The Marital Satisfaction Inventory–Revised (MSI-R) scale (Snyder, 1997) consists of 150 True–False items that assess the nature and extent of conflict within a marriage or close romantic relationship. To classify couples as members of the taxon (i.e., discordant) or complement (i.e., nondiscordant), we used the definition provided by Whisman, Beach, and Snyder (2008). We first computed the mean of husbands’ and wives’ scores on Global Distress (GDS), Time Together (TTO), Sexual Dissatisfaction (SEX), Affective Communication (AFC), and Problem-Solving Communication (PSC). Whisman et al. selected these scales as indicators of the taxon because they met the dual criteria of being applicable to most types of couples and focused primarily on current relationship adjustment, broadly conceived. We calculated the mean of these 10 scales and classified couples as discordant (i.e., members of the taxon) if their mean score was ≥54T and as nondiscordant (i.e., members of the complement) if their mean score was <54T; this cut score was derived by Whisman et al. based on receiver operating characteristic analyses.
Center for Epidemiological Studies Depression Scale
The Center for Epidemiological Studies–Depression (CES-D) scale (Radloff, 1977) is a commonly used measure of depressive symptoms. A 14-item version of the CES-D was used in the current study. The scale demonstrated good internal consistency (α = .89 for wives and .86 for husbands) in this sample.
Results
Using Whisman et al.’s (2008) cutoff of ≥54T on the mean of the 10 MSI-R scales to define the taxon, the base rate of marital discord was .19. 1 Compared with their nondiscordant counterparts, discordant husbands were younger (M = 64.2 years, SD = 4.0 vs. M = 65.5 years, SD = 4.2, t(500) = 2.79, p = .005) and discordant wives were less likely to be White (94.6% vs. 98.8%, χ2(1) = 6.70, p = .01); there were no differences between groups on wives’ age, husbands’ race, number of children, or length of relationship. 2 Mean scores across the entire sample for the 14-item CES-D were 5.97 (SD = 6.32) for wives and 5.55 (SD = 5.75) for husbands.
To examine whether discordant spouses differed from nondiscordant spouses on depressive symptoms, we regressed CES-D scores on taxon status (0 = nondiscordant and 1 = discordant) separately for wives and husbands. Results indicated that taxon status accounted for 16% of the variance in wives’ CES-D scores, B = 6.51, SE = 0.67, β = .40, p < .001, and 7% of the variance in husbands’ CES-D scores, B = 3.94, SE = 0.64, β = .27, p < .001. Discordant wives (M = 11.28; SD = 8.92) reported higher levels of depressive symptoms than nondiscordant wives (M = 4.77; SD = 4.81; d = 1.03), and discordant husbands (M = 8.75; SD = 7.44) reported higher levels of depressive symptoms than nondiscordant husbands (M = 4.82; SD = 5.02; d = .68).
To examine whether taxon status was incrementally associated with depressive symptoms, we first regressed the CES-D on the mean of wives' and husbands’ GDS, TTO, SEX, PSC, and AFC scales. The mean of these 10 scales accounted for 16% of the variance in wives’ CES-D scores, B = 0.34, SE = 0.04, β = .40, p < .001, and 9% of the variance in husbands’ CES-D scores, B = 0.23, SE = 0.03, β = .30, p < .001. We then simultaneously regressed the CES-D on (a) the mean of the 10 MSI-R scales and (b) taxon status. For wives, depression severity was significantly associated with both (a) the mean of the 10 MSI-R scales, B = 0.20, SE = 0.06, β = .23, p < .001 and (b) taxon status, B = 3.59, SE = 1.06, β = .22, p < .001. For husbands, depression severity was significantly associated only with the mean of the 10 MSI-R scales, B = 0.18, SE = 0.05, β = .23, p = .001; for husbands, the association between depression severity and taxon status was not statistically significant after adjusting for the continuous scales, B = 1.33, SE = 1.01, β = .09, p = .19.
Discussion
This study evaluated the association between marital discord taxon status and depressive symptoms in middle-aged and older adults. Compared to nondiscordant spouses, discordant wives and husbands reported higher levels of depressive symptoms. Furthermore, the variance in depressive symptoms accounted for by taxon status was comparable in magnitude to the variance accounted for by husbands’ and wives’ mean scores on the MSI-R scales used to define the taxon. This finding is particularly noteworthy because the mean of the 10 MSI-R scales was dimensional, whereas taxon classification was dichotomous. A general rule of thumb is that dichotomizing normally distributed predictor variables reduces the squared multiple correlation to about 64% of what it otherwise would have been (Cohen, 1983). Consequently, we would have expected that taxon status would have accounted for considerably less variance in depressive symptoms than that which was explained by the continuously distributed scales.
The present finding that the marital discord taxon is associated with depressive symptoms is consistent with a study which found that a categorical measure of marital discord, based on a cutoff score reflecting Beach et al.’s (2005) estimated .20 base rate of the taxon, was associated with general distress and functional impairment (Whisman & Uebelacker, 2006). This finding is also consistent with the results from a study evaluating the association between a categorical measure of marital discord and asymptomatic coronary artery disease, which found that discordant couples, defined by cluster analysis, had greater coronary artery calcification than nondiscordant couples (Smith, Uchino, Berg, & Florsheim, 2012). Taken together, these results suggest that a nonarbitrary categorical classification of couples as discordant versus nondiscordant is associated with important indicators of individual health and well-being.
The significant association between taxon status and depressive symptoms may have implications for the marital discord model of depression. Gottman (1993) proposed that when negativity exceeds some threshold, perceptions of the relationship shift and a new cognitive–emotional process is engaged, leading people to recast their relationship history negatively. One consequence of a discontinuous relationship shift could be effects on spouses’ individual functioning, that is, when negativity exceeds some threshold, people may become helpless and hopeless about improving their relationship, increasing their risk for depression. Longitudinal research is needed to evaluate whether taxon status predicts the development of depression or, conversely, if depression contributes to a generalization of problems from one area of the relationship to other areas, thereby contributing to the emergence of taxonicity in marital discord.
Extending beyond the significant association between marital discord taxon status and depressive symptoms, the pattern of results differed for men and women. For men, the association between taxon status and depressive symptoms was no longer significant when controlling for the mean of the 10 MSI-R continuous scales. By contrast, for women, the association between taxon status and depressive symptoms remained significant even when controlling for the mean of the continuous scales. For wives, marital discord taxon status was incrementally associated with depressive symptoms, explaining additional variance in wives’ depressive symptoms than could be explained solely by the mean of these scales. At a descriptive level, therefore, it appears that depression is more closely connected to marital discord taxon status among women than among men. These results parallel research with continuous measures of relationship discord, which has found that relationship discord is more strongly associated with depressive symptoms for women (for a meta-analysis, see Whisman, 2001). The current results also parallel Whisman et al.’s (2008) findings that taxon status was uniquely associated with therapists’ ratings of couples’ distress across five areas of their relationship, controlling for partners’ scores on the GDS. Thus, taxon status may explain unique variance in both individual and relationship outcomes, over and above what is accounted for by continuously distributed measures.
One limitation of the current study was that marital discord and depressive symptoms were both measured by self-report. Therefore, it is possible that shared method variance may have contributed to the magnitude of their association. Future research would benefit from indicators and outcomes based on methods in addition to self-report.
Results from the study add to the literature on marital discord in middle-aged and older adults in suggesting that approximately 20% of such couples are likely to be members of a marital discord taxon, a base rate that is comparable to another sample of couples in which one partner was 55 to 75 years old. 1 Furthermore, results suggest that taxon classification is associated with medium-to-large effect size differences in level of depressive symptoms and, for wives, accounts for additional variance over that accounted for by continuous measures of marital discord.
The assessment of a marital discord taxon has implications for clinical evaluation. For example, screening for the marital discord taxon may be helpful in identifying individuals having an elevated risk for depression. In a similar vein, Whisman and Beach (in press) suggested that couple therapy might be appropriate for depressed individuals in discordant relationships, defined in terms of the marital discord taxon. Finally, the results from this study support continued research in relationship discord and depression in middle-aged and older individuals, including research on the association between taxon status and the onset, course, and treatment of depression.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
