Couple Classes of Depressive Symptom Trajectories and Associations with Relationship Satisfaction

Abstract

The literature indicates a robust negative relationship between depression and couple relationship satisfaction. However, less is known about the differential effects of romantic relationships on depression over time and whether one or both partners experience depression. Using data from 1215 couples across 4 years, we sought to examine couple classes of depressive symptom trajectories and investigate the degree to which relationship satisfaction predicted class membership. Using joint-probability growth mixture modeling, we found three couple classes of depressive symptom trajectories: women’s moderate, men’s low class, men’s moderate, women’s low class, and men’s and women’s low class. Logistic regression results revealed both men’s and women’s higher relationship satisfaction was associated with women’s moderate, men’s low class membership while both partners’ higher relationship satisfaction was not associated with men’s moderate, women’s low class membership, in comparison to the men’s and women’s low stable class. These findings contribute to the literature by identifying the heterogeneity of patterns of depressive symptom trajectories among couples and the association of relationship satisfaction with couple classes.

Keywords

Depression trajectories joint-probability growth mixture modeling relationship satisfaction

Depression is the leading contributor of disability worldwide (World Health Organization [WHO], 2017). Each year, approximately 3–7% of adults around the globe suffer from depression (Lim et al., 2018; WHO, 2017). The proportion of the population with depression varies by geographic region, with the highest rates in Africa and the Americas and the lowest rates in the Western Pacific. Depression, along with anxiety, carries significant costs to individuals, families, and society including a $1.15 trillion annual global economic burden (Chisholm et al., 2016). The risk of developing depression increases with exposure to adverse social and economic conditions, including relationship conflict and dissolution and poor health (WHO, 2017). Depression is heterogenous, which means that individuals experience different trajectories of depressive symptoms over time and thereby suffer at disproportionate rates (Eaton et al., 2008).

Heterogeneity in the Course of Depressive Symptoms

Long-term trajectories of depressive symptoms are heterogenous, meaning that some people experience depression as a single brief episode, others report recovering from multiple episodes, and some endure recurrent and persistent symptoms for years (Klein & Allmann, 2014). This means that people may experience depressive symptoms as transient, intermittent, or stable (Eaton et al., 2008). One prospective population-based study of participants over 23 years found that about 50% of those who experience a depressive episode recover and have no additional episodes, 35% experience recurrent depression, and 15% experience unremitting symptoms (Eaton et al., 2008). A systematic review of 25 studies on the heterogeneity of depressive symptom trajectories identified variation in terms of severity of symptoms (i.e., low, medium, and high) and stability of symptoms (i.e., stable, increasing, and decreasing; Musliner et al., 2016). Most prior group-based trajectory modeling studies have identified either three or four classes of trajectories of depressive symptoms that vary by severity and stability, with the majority of community-based samples including participants with low or no symptoms. Predictors of poor depressive symptom trajectories include variables such as female gender, low education, and low income. Findings also suggest that women’s symptoms endure longer compared to men (Eaton et al., 2008).

Romantic Couples and the Course of Depressive Symptoms

Romantic relationships are a vital context to examine because they can affect the course of depressive symptoms by improving, maintaining, or worsening symptoms (Whisman et al., 2021). These effects are supported by three main theories: Coyne’s interpersonal model of depression (Coyne, 1976), the marital discord model of depression (Beach et al., 1990), and the stress generation model (Hammen, 1991). Broadly, these theories describe interrelations among depressive symptoms and difficulties in close relationships (Rehman et al., 2008). Coyne’s interpersonal model notes that depressed individuals seek frequent reassurance from their partner, which can lead to cycles of interpersonal problems and depressive symptoms (Coyne, 1976). From a stress generation perspective, dependent stressors – life events influenced by the depressed individual such as getting into an argument – are more strongly associated with depressive symptoms than independent stressors that are outside of an individual’s control (Bos et al., 2007; Davila et al., 1997; Kendler et al., 1999). Finally, the marital discord model (Beach et al., 1990) also explains how relationship conflict can have lingering effects on relationships, which can worsen depressive symptoms in one or both partners over time. Although these theories have distinct differences, they all suggest a link between depressive symptoms and relationship problems. Individual differences among partners give way to differences in how couples experience depression and how depressive symptoms and interpersonal stress unfold for each partner over time.

Couple Classes of Depressive Symptom Trajectories

Researchers have historically identified different types of couples by partners’ depressive symptoms or diagnoses and sex. For example, heterosexual couples in which one partner (either male or female) has depression are often compared to heterosexual couples not experiencing depression (e.g., Byrne & Carr, 2000; Gabriel et al., 2016; Gotlib et al., 1989; Leung et al., 2017; Sacco et al., 1993). Generally, this literature supports the idea that depressed couples, regardless of which partner is depressed, experience more relationship problems and stress than non-depressed couples. However, most of these studies recruited couples with depressed women and few studies included couples with depressed men (e.g., Byrne & Carr, 2000; Sacco et al., 1993; Gabriel et al., 2010), which complicates the examination of sex differences among couples. Of the few studies that did examine couples with depressed men and couples with depressed women, depressed women were more vulnerable to greater relationship distress than men (Gabriel et al., 2016). On the other hand, both groups of couples were found to be similar in that both depressed men and depressed women were more critical and demanding of their nondepressed partners (Gabriel et al., 2010).

While the existing findings offer some insight, there are gaps in the literature that warrant further exploration. Prior studies used small samples ranging from 22 to 80 couples (Gotlib et al., 1989; Ruscher & Gotlib, 1988). Although these studies used samples from clinical populations, the small sample sizes limit generalizability. Furthermore, it is unclear whether these differences and similarities between couples with depression exist in large, broad, and representative population-based samples. This is important because presenting with subclinical depressive symptoms is common. Further, studying population-based samples improves the generalizability of findings. Hence, we aim to use a large representative sample of couples in Germany.

Prior studies also did not examine heterogeneity among couples’ depressive symptom trajectories. For example, the literature generally classified partners’ depression as binary (e.g., depressed or not depressed) through the use of cut-off scores from clinician rated or self-reported measures. This limits our understanding of depression because depression can vary by severity and stability over time and severity and stability can vary among partners (American Psychiatric Association, 2013). Heterogeneity is often examined in mixture models that identify various types of depression based on patterns and trends in the variation of the sample. Testing the heterogeneity of depression allows researchers to parse out groups of people who experience different levels of symptomatology over time. We identified one previous study that used mixture modeling on a large sample of German couples that identified four types of couples ranging in level of depressive symptoms and conflict at one point in time (Morgan et al., 2020). The study of a large population-based sample illustrated that partners and couples vary in severity of depressive symptoms.

The prior empirical literature mostly relied upon cross-sectional analyses of couples, so it is not possible to understand changes in the course of depressive symptoms over time among different types of couples. This is a crucial next step in the research since many partners’ depressive symptoms are recurrent and chronic (Klein & Allmann, 2014; Musliner et al., 2016). To capture both the heterogeneity of depressive symptoms and how depressive symptoms change over time, prior researchers have used growth mixture modeling to examine trajectories over time (Musliner et al., 2016). However, there is no known study that has identified various types of depressive symptom trajectories among couples. This is important because the literature remains focused on the average trajectory of depressive symptoms and provides limited understanding of the heterogeneous nature of depressive symptoms among couples. Given that the previous literature supports that couples in which one or both partners have depressive symptoms also experience greater relationship distress, it remains unclear how different types of couples with depressive symptoms differ over time. Understanding the course of depressive symptoms among different types of couples is important because it can provide clinicians and researchers insights about the course of depressive symptoms for different types of couples that often present in clinical settings. Hence, we aim to identify different types of couples with different depressive symptom trajectories.

Present Study

We aimed to explore different types of depressive symptom trajectories among couples and their association with relationship satisfaction. More specifically, this study investigated the following research questions: (1) What were the couple classes of men’s and women’s trajectories of depressive symptoms over 4 years, and (2) To what extent were men’s and women’s relationship satisfaction associated with the identified couple classes. Specifically, joint-probability growth mixture modeling (JPGMM; Li et al., 2002) is an extension of growth mixture modeling that can estimate classes of two partners’ trajectories of depressive symptoms simultaneously, instead of only one partner’s trajectory (e.g., Lavner & Bradbury, 2010). Thus, by using JPGMM, we identified dyadic classes of trajectories of depressive symptoms among couples, investigated differences in changes in depressive symptoms by partner and couple, and examined the association between relationship satisfaction and class membership. Finally, we accounted for individual personality factors (i.e., neuroticism), which are potential confounding factors among couples with depression (Cao et al., 2017; Whisman et al., 2006).

Method

In this study, we used five waves of data across 4 years from the longitudinal Pairfam study (release 9.0; Brüderl et al., 2018; Huinink et al., 2011). The Pairfam investigators used a stratified sampling design to collect a representative sample of adults in Germany. Focal participants were recruited by birth year cohorts (i.e., 1971–1973, 1981–1983, and 1991–1993). Focal participants completed in-home interviews annually starting in 2008, and the partners of focal participants were recruited to complete return-by-mail questionnaires. We used Waves 2, 3, 4, 5, and 6 (2009–2013) in this study because depressive symptoms were measured starting in Wave 2. Although there were 2687 couples at the beginning of Wave 2, there was a general attrition in couples over time (Wave 3, n = 2362; Wave 4, n = 2182; Wave 5, n = 2039), with 1992 couples remaining in the study at Wave 6. We included couples who remained together across the waves and excluded data from any new partners of the focal participants at later waves, which resulted in a sample of 1297 couples. Next, we excluded divorced and separated couples (n = 37), one widowed individual, and those with missing values on relationship status (n = 37) at Wave 2 only, which reduced the sample to 1251 couples. At each subsequent wave, there were very few (i.e., 5 or fewer) couples who divorced or separated and any data provided by these couples were retained. We also excluded adolescents (i.e., aged 17 and younger, n = 36) because of the well-documented differences in relationship dynamics among adolescents compared to adults (Lantagne & Furman, 2017), which resulted in a final sample of 1215 couples.

Participants

Demographic data from the first wave of the current study (Pairfam Wave 2), unless otherwise specified, were used to understand the sample composition. Couples reported they had been in a committed relationship an average of 10.08 years (SD = 6.04). Most couples reported being married (70.50%), others were cohabitating (21.80%) or committed, but living in separate residences (7.70%). Most couples had at least one child (67.40%). All couples in the current study reported different-sex relationships. Additionally, the vast majority of couples (81.80%) were native Germans, while the remaining couples (18.20%) were immigrants to Germany, including Aussiedler (former Soviet Union countries; 3.90%), half-German (5.80%), Turkish (2.30%), and other non-German background (6.20%). These immigrant couples had been in Germany for an average 14.6 years (Range: 1–37 years). Couples tended to be in cohort 3 (59.50%; 1971–1973), and, on average, women were 32 years of age (Median = 33, SD = 5.73) and men were 35 years of age (Median = 37, SD = 6.39). Couples reported an average monthly household income of €2974.24 (SD = €1231.85). Generally, couples were educated with men (43.70%) and women (34.00%) reporting a university-level education, while no current students were in the sample. Concerning employment, most men were employed full-time (80.20%) while some women were employed full-time (29.30%), part-time (23.80%), and self-employed (6.40%). These couples on average reported satisfactory levels of health (M = 3.77 for men, M = 3.77 for women) and some were legally disabled (2.2% of men, 5.1% of women, measured at Wave 1). Couples were generally satisfied with their relationship (M = 8.41 for men and M = 8.32 for women) on a range of 0 (very dissatisfied) to 10 (very satisfied). This sample had, on average, lower levels of depressive symptoms for both men (M = 1.57) and women (M = 1.66) on a range of 1 (almost never) to 4 (almost always), which remained stable and slightly elevated by Wave 6 for men (M = 1.66) and women (M = 1.73). See Table 1 for a summary of the descriptive statistics.

Table 1.

Descriptive Statistics (N = 1215 couples).

Variables	M	SD	Range	α
Women’s depressive symptoms W2	1.66	.48	1–4	.89
Men’s depressive symptoms W2	1.57	.41	1–4	.85
Women’s depressive symptoms W3	1.66	.48	1–4	.89
Men’s depressive symptoms W3	1.58	.42	1–4	.86
Women’s depressive symptoms W4	1.70	.48	1–4	.86
Men’s depressive symptoms W4	1.60	.43	1–4	.86
Women’s depressive symptoms W5	1.72	.47	1–4	.89
Men’s depressive symptoms W5	1.63	.44	1–4	.91
Women’s depressive symptoms W6	1.73	.47	1–4	.87
Men’s depressive symptoms W6	1.66	.44	1–4	.89
Women’s relationship satisfaction	8.32	1.97	0–10	—
Men’s relationship satisfaction	8.41	1.76	0–10	—
German ethnicity^a	81.80%	—	0, 1	—
Household income	2974.24	1231.85	0–7000	—
Women’s age	32.10	5.73	18–53	—
Men’s age	34.84	6.39	18–59	—
Women’s health	3.61	1.00	1–5	—
Men’s health	3.77	.93	1–5	—
Women’ education	5.24	1.60	1–8	—
Men’s education	5.45	1.64	1–8	—
Married^b	71.50%	—	0, 1	—
Cohabitating^b	21.80%	—	0, 1	—
Number of children	1.28	1.14	0–10	—
Women’s neuroticism	2.83	.82	1–5	.72
Men’s neuroticism	2.45	.73	1–5	.67

^aReference group is German migrants.

^bReference group is couples committed to each other, but living in separate residences. Unless specified, all of the variables were measured at Wave 2.

Measures

Depressive Symptoms

Ten items from the State-Trait Depression Scale (STDS; Spaderna et al., 2002) were used to assess depressive symptoms. Only the trait depression items were used, which were intended to assess how participants felt in general as opposed to how they felt at the time of assessment. The STDS specifically assessed cognitive-affective symptoms of depression. The STDS was designed for use with non-clinical samples (Krohne et al., 2002) similar to the current sample. The STDS was highly correlated with common measures of depressive symptoms often used with clinical samples (Spaderna et al., 2002), such as the Beck Depression Inventory (Beck & Steer, 1987) and Zung Self-Rating Depression Scale (Zung et al., 1986). Scale items ranged from euthymia (e.g., “I am happy” and “I feel good”) to dysthymia (e.g., “I am depressed” and “My mood is melancholy”), which were rated from 1 (almost never) to 4 (almost always). The euthymia items were reverse coded so that higher values indicated absence of positive affect, thereby reflecting higher levels of depressive symptoms. The scale had acceptable internal consistency for Wave 2 (men α = .85, women α = .89), Wave 3 (men α = .86, women α = .89), Wave 4 (men α = .86, women α = .86), Wave 5 (men α = .91, women α = .89), and Wave 6 (men α = .89, women α = .87).

Relationship Satisfaction

One item from the German Relationship Assessment Scale (RAS; Sander & Böcker, 1993) measured global satisfaction of the romantic relationship at Wave 2. Specifically, partners rated, “All in all, how satisfied are you with your relationship” with a score ranging from 0 (very dissatisfied) to 10 (very satisfied). This single item has been compared to the full RAS and was found to be a valid and reliable measure of overall relationship satisfaction among couples (Fülöp et al., 2020).

Controls

We used several control variables that were all measured at Wave 2, including age (years), number of children, ethnicity (0 = Non-German native, 1= German native), household income (euros), health (rated 1 [bad] to 5 [very good]) and relationship status (cohabitating [0 = other relationship statuses, 1 = cohabitating], married [0 = other relationship statuses, 1 = married]). Education was measured by the international standard classification of education that ranged from 0 (not currently enrolled in education) to 9 (equivalent to university level education) where higher numbers indicated greater education (UNESCO, 2006). Finally, four items from the Big Five Inventory (Rammstedt & John, 2005) assessed partner’s neuroticism (men α = .67, women α = .72), on a rating scale from 1 (absolutely incorrect) to 5 (absolutely correct).

Missing Data

We examined the skewness and kurtosis of the data and the amount of missing data. Household income as well as men’s and women’s relationship satisfaction had high kurtosis values (i.e., greater than 5; Kline, 2016). Consequently, we recoded the values outside 3 standard deviations to be 3 standard deviations from the mean, which resulted in acceptable kurtosis levels. Missing data were low and ranged from 0% (e.g., men’s and women’s ages) to 9% (e.g., men’s relationship satisfaction). Little’s test revealed that missing was not missing completely at random (χ² (2489) = 3141.56, p < .01). We further evaluated missing data associations by coding all the outcomes, predictors, and controls as missing variables (e.g., missing = 1, observed = 0). We then tested bivariate correlations of the missing data variables (5 controls had no missing and were excluded from bivariate correlations). Forty-five percent of the 200 bivariate correlations were correlated at p < .05, ranging from r = −.06 (women’s relationship missing was negatively associated with men’s depressive symptoms missing at Wave 4) to r = .97 (women’s health missing was positively associated with women’s neuroticism missing at Wave 2). Full-information maximum likelihood estimation was used to handle missing data by estimating parameters from all of the available data in the variance and covariance matrices (Acock, 2005).

Data Analysis

Using Mplus 8.0 (Muthén & Muthén, 1998–2017), we began with a series of unconditional latent growth models. Specifically, we examined unconditional latent growth models of men’s and women’s trajectories of depressive symptoms separately. Latent constructs of women’s initial (i.e., intercept) and rates of change (i.e., slope) of depressive symptoms were composed of women’s depressive symptom scores at Waves 2 through 6. Wave 2 was coded 0 with Wave 6 coded as 1 while Waves 3–5 were freely estimated to account for nonlinear trends in the trajectory. Next, we assessed a conjoint model composed of both partner’s trajectories of depressive symptoms simultaneously with men’s and women’s initial and rates of change covarying. Each model was evaluated by a common fit index (CFI) greater than .95, and root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) less than .05 (Kline, 2016).

We then added a categorical latent construct for the couple classes of men’s and women’s initial and rates of change for trajectories of depressive symptoms. Relationship satisfaction for men and women were then added as predictors of the couple class latent construct; control variables were also added (see Figure 1). With Waves 3–5 being freely estimated, we examined change in the slopes over time by calculating percent changes in trajectories at specific waves in addition to the overall rates of change estimate (i.e., slope). Similar to growth mixture modeling procedures (Soloski & Durtschi, 2020), we then used an iterative approach to test the model from one to five classes, to assess the best fitting number of classes. Specifically, we evaluated if models improved when latent variances and covariances between and within dyadic classes were constrained or freely estimated, which were similar to homogeneity of variance tests in ANOVA (Soloski & Durtschi, 2020). The number of classes were evaluated by the following indices: model convergence and warnings, Loglikelihood, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Sample-size Adjusted BIC (ABIC), Lo-Mendell-Rubin Adjusted Likelihood Ratio Test (LMR-RT), and entropy. Fit indices, theory, and parsimony all informed the class selection. Finally, logistic regression was used to examine the degree to which relationship satisfaction and the controls predicted class membership.

Figure 1.

Joint-probability growth mixture model.

Results

Unconditional Models

We ran unconditional models for both men’s and women’s trajectories of depressive symptoms separately, which each had good model fit (men: CFI = .99, RMSEA = .04, SRMR = .04; women: CFI = .99, RMSEA = .05, SRMR = .04). Next, we examined men’s and women’s trajectories of depressive symptoms conjointly, which also had good model fit (CFI = .99, RMSEA = .03, and SRMR = .03). Both partners reported low levels of depressive symptoms (men: M = 1.58, SE = .01, p < .01; women: M = 1.67, SE = .01, p < .01) that had small increases over time (men: M = .08, SE = .01, p < .01; women: M = .05, SE = .01, p < .01) with variances at p < .01 for both partners’ initial levels (men: σ² = .12, SE = .01; women: σ² = .17, SE = .01) and rates of change (men: σ² = .06, SE = .01; women: σ² = .04, SE = .01). This conjoint unconditional model established a single class of couples with average trajectories of depressive symptoms for both men and women.

Determining Classes

Each predictor and controls were added to the unconditional conjoint model. To examine the variances and covariances between classes, we investigated each of the following models from one to five classes: (1) constraining all latent variances and covariances to be zero (i.e., latent class growth analysis), (2) constraining latent variances to be equal across the classes and constraining covariances to be zero (i.e., class-invariant, diagonal GMM), (3) freely estimating latent variances and constraining covariances to be zero (i.e., class-varying, diagonal GMM), (4) constraining latent variances to be equal across the classes and freely estimating covariances (i.e., class-invariance, non-diagonal GMM), and (5) freely estimating latent variances and covariances across the classes (i.e., class-varying, non-diagonal GMM; Soloski & Durtschi, 2020). The top two models with the lowest LL, AIC, BIC, and ABIC and similar entropy were selected (see Table 2). For parsimony and interpretability, we selected the three-class model instead of the four-class model because the four-class model did not replicate with twice the random starts. The selection of the three-class invariant-diagonal model means that the variance in women’s depressive symptom trajectories was the same across the three classes and the variance in men’s depressive symptom trajectories was the same across the three classes, but the covariances between men’s and women’s depressive symptom trajectories varied between the three classes. The LL in the three-class model was replicated when we tested the model with twice the random starts and a specific OPTSEED start, which increased confidence in the three-class model.

Table 2.

Descriptive Statistics of Indicators for Number of Class in the Conditional Models (N = 1215 couples).

Latent Class Growth Model
Classes	Con	LL	AIC	BIC	ABIC	LMR-RT	Ent	C1%	C2%	C3%	C4%	C5%
1	Y	−37,816.81	75,733.63	75,988.75	75,829.93	—	—	1.0	—	—	—	—
2	Y	−4101.64	8295.28	8515.05	8368.96	2288.52	.88	.71	.29	—	—	—
3	Y	−3509.06	7162.13	7506.12	7277.46	1178.46	.90	.20	.59	.21	—	—
4	N	—	—	—	—	—	—	—	—	—	—	—
5	N	—	—	—	—	—	—	—	—	—	—	—
Latent Growth Model Class Invariant – Diagonal
Classes	Con	LL	AIC	BIC	ABIC	LMR-RT	Ent	C1%	C2%	C3%	C4%	C5%
1	Y	−34,390.19	68,888.38	69,163.92	68,992.39	—	—	1.0	—	—	—	—
2	Y	−2349.01	4798.01	5036.89	4878.10	709.54*	.82	.72	.28	—	—	—
3	Y	−2185.71	4523.42	4886.52	4645.16	324.75	.81	.19	.65	.16	—	—
4	Y/W	−2060.62	4325.24	4812.55	4488.63	248.77	.80	.07	.20	.23	.50	—
5	N	—	—	—	—	—	—	—	—	—	—	—
Latent Growth Model Class Variant – Diagonal
Classes	Con	LL	AIC	BIC	ABIC	LMR-RT	Ent	C1%	C2%	C3%	C4%	C5%
1	Y/W	−34,390.19	68,888.38	69,163.92	68,992.39	—	—	1.0	—	—	—	—
2	Y/W	−2205.53	4519.06	4777.05	4605.56	995.62*	.73	.60	.40	—	—	—
3	N	—	—	—	—	—	—	—	—	—	—	—
4	N	—	—	—	—	—	—	—	—	—	—	—
5	N	—	—	—	—	—	—	—	—	—	—	—
Latent Growth Model Class Invariant – Nondiagonal
Classes	Con	LL	AIC	BIC	ABIC	LMR-RT	Ent	C1%	C2%	C3%	C4%	C5%
1	Y	−34,299.99	68,719.98	69,026.13	68,835.55	—	—	1.0	—	—	—	—
2	Y	−2326.98	4765.78	5033.32	4855.48	615.06*	.83	.72	.28	—	—	—
3	Y	−2154.18	4472.36	4864.13	4603.72	343.47*	.80	.21	.23	.56	—	—
4	Y/W	−2036.51	4289	4805.02	4462.04	233.99	.80	.06	.22	.21	.50	—
5	N	—	—	—	—	—	—	—	—	—	—	—
Latent Growth Model Class Variant – Nondiagonal
Classes	Con	LL	AIC	BIC	ABIC	LMR-RT	Ent	C1%	C2%	C3%	C4%	C5%
1	Y/W	−34,299.99	68,719.98	69,026.13	68,835.55	—	—	1.0
2	Y/W	−2168.71	4469.43	4784.75	4575.15	931.08*	.73	.43	.57	—	—	—
3	N	—	—	—	—	—	—	—	—	—	—	—
4	N	—	—	—	—	—	—	—	—	—	—	—
5	N	—	—	—	—	—	—	—	—	—	—	—

Note. Con = Converge, LL = Loglikelihood, AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion, ABIC = Sample-size adjusted BIC, LMR-RT = Lo-Mendell-Rubin Adjusted Likelihood Ratio Test, Ent = Entropy, C%= Percentage of sample in class indicated. The bolded row is the model that was selected as optimal. *p < .05.

Three-Class Model

The three-class model reveled three dyadic classes of both partners’ trajectories of depressive symptoms (see Figure 2). The majority of couples in our sample were in the men’s and women’s low stable class (56%). Participants in this class reported low initial levels of depressive symptoms (b = 1.38, p = .01 for men; b = 1.46, p = .01 for women) that increased an average of 6.84% for women (b = .10, p < .01) and 6.54% for men (b = .09, p < .01) across the five waves.

Figure 2.

Three couple classes of dyadic trajectories of Men’s and Women’s depressive symptoms across four years. Note. Wm, ML = Women’s Moderate, Men’s Low class; MM, WL = Men’s Moderate, Women’s Low class; MWL = Men’s and Women’s Low class.

About 21% of couples were in the women’s moderate, men’s low class. This class included women who reported moderate levels of depressive symptoms (b = 2.16, p < .01) that remained stable over time (b = .01, p = .59). Women reported their depressive symptoms worsened late in the study period, with a 1% increase in symptoms at Wave 5. The men in this class reported low levels of depressive symptoms (b = 1.62, p < .01) that increased an average of 5.80% over time (b = .09, p < .01).

The remaining couples (23%) were in the men’s moderate, women’s low class. This class was comprised of men with moderate initial levels of depressive symptoms (b = 1.99, p < .01) that remained stable over time (b = .00, p = .96). Men reported their depressive symptoms worsened late in the study period, with a .01% increase in symptoms at Wave 5. Women reported low initial levels of depressive symptoms (b = 1.60, p < .01) that remained stable and did not vary over time (b = .08, p = .11). The variances, which were constrained to be the same across the classes, were small for the intercepts (women: σ2 = .08, p < .01; men: σ2 = .06, p < .01) and rates of change (women: σ2 = .03, p < .01; men: σ2 = .04, p < .01).

Class Membership

Multinomial logistic regression was used to test the degree to which relationship satisfaction and controls predicted class membership (see Table 3). The results reported here used the reference group men’s and women’s low stable class as well as cutoff for significance of p < .05. Higher relationship satisfaction among women was associated with a 41% decrease in the odds of being in the women’s moderate, men’s low class in comparison to the men’s and women’s low stable class. Men’s higher relationship satisfaction was associated with a 38% decrease in the odds of being in the women’s moderate, men’s low class, in comparison to the men’s and women’s low stable class. Concerning the controls, being of German ethnicity was associated with a 74% decrease in the odds of being in the women’s moderate, men’s low class than the men’s and women’s low stable class. Men’s higher education was associated with a 40% decrease in the odds of being in the women’s moderate, men’s low class than the men’s and women’s low stable class. Women’s health was associated with a 60% decrease in the odds of being in the women’s moderate, men’s low class while men’s health was associated with a 50% decrease in the odds of being in the women’s moderate, men’s low class. Finally, women’s higher neuroticism was associated with a 1339% increase in the odds of being in the women’s moderate, men’s low class. Men’s higher neuroticism was associated with a 5241% increase in the odds of being in the men’s moderate, women’s low class.

Table 3.

Unstandardized Parameter Estimates, Odds Ratios, and Significance Levels for Dyadic Growth Mixture Models (N = 1215 couples).

	Men’s moderate, women’s low class (23%)					Women’s moderate, men’s low class (21%)
	b	SE	OR	95% CI	p	b	SE	OR	95% CI	p
Women’s relationship satisfaction	−0.28	.16	.76	.56, 1.03	.08	−0.53*	.15	.59	.44, .79	.00
Men’s relationship satisfaction	−0.30	.16	.74	.55, 1.01	.06	−0.38*	.12	.69	.54, .88	.00
German native-born	−1.03	.59	.36	.11, 1.13	.08	−1.36*	.63	.26	.07, .89	.03
Women’s age	−0.04	.03	.96	.87, 1.07	.45	−0.01	.05	.99	.89, 1.10	.89
Men’s age	0.01	.05	1.01	.92, 1.10	.92	0.04	.05	1.04	.94, 1.15	.45
Women’ education	0.03	.17	1.03	.74, 1.43	.85	−0.12	.13	.88	.69, 1.14	.34
Men’s education	0.10	.14	1.11	.84, 1.47	.49	−0.32*	.14	.73	.55, .95	.02
Women’s health	−0.08	.26	.92	.56, 1.53	.76	−0.91*	.24	.40	.25, .65	.00
Men’s health	−0.41	.27	.66	.40, 1.12	.12	−0.51*	.26	.60	.31, .99	.04
Married	−0.15	1.19	.87	.09, 8.84	.90	−1.15	.91	.32	.05, 1.88	.21
Cohabitating	−0.63	1.09	.53	.06, 4.51	.57	−1.03	.88	.36	.06, 1.99	.24
Number of children	0.27	.23	1.31	.84, 2.03	.23	0.16	.21	1.17	.77, 1.78	.46
Household income	−0.04	.05	.96	.90, 1.02	.22	−0.03	.02	.97	.93, .101	.23
Women’s neuroticism	−0.02	.47	.98	.39, 2.47	.97	2.67*	.47	14.39	5.79, 35.78	.00
Men’s neuroticism	3.98*	.64	53.41	15.25, 187.02	.00	0.37	.56	1.45	.48, 4.37	.51

Note. Reference class is men’s and women’s low class (66%). All predictors were measured at Wave 2. *p < .05.

Discussion

This study extends our previous work in which we investigated dyadic classes of both partners’ depressive symptoms at one time point (Morgan et al., 2020). This is the first known study to examine dyadic classes of both partners’ depressive symptoms over time. Previous studies have tested depression classes for either one person (e.g., Musliner et al., 2016) or both partners in separate models (e.g., Lavner & Bradbury, 2010), which offers a limited understanding of dyadic processes. We sought to examine couple classes of depressive symptom trajectories and investigate the degree to which relationship satisfaction predicted class membership. In response to our first aim, we identified three classes of couples’ trajectories of depressive symptoms: women’s moderate and men’s low class, men’s moderate and women’s low class, and men’s and women’s low stable class. The identification of these couple classes indicated that couples vary in regard to whether one or both partners experienced depressive symptoms and the severity of symptoms experienced over time. Findings related to the second aim highlighted nuance in the way relationship satisfaction associated with these classes. Findings relevant to each class will be discussed further below.

Men’s and Women’s Low Stable Class

The majority of the sample in this study was in the men’s and women’s low stable depressive symptom class (66%). This was to be expected since studies of community samples typically report lower rates of depressive symptoms (e.g., Fredriksen et al., 2019; Kouros & Cummings, 2011) compared to clinical samples. Unlike the other classes, both partners in this class reported similar trajectories of depressive symptoms, suggesting that they had similar experiences regarding the severity of depressive symptoms over time. Although women reported slightly higher depressive symptoms than men in this class, they both had low initial levels of symptoms that slightly increased over time.

Given the low levels of depressive symptoms, it is unsurprising that couples in this class were more likely to be satisfied with their relationship when compared to the women’s moderate, men’s low class. This is consistent with the extant literature, which demonstrated that non-depressed couples report higher relationship satisfaction and less relationship stress (e.g., Byrne & Carr, 2000; Gabriel et al., 2016). These couples were also more likely to be in better health than the other two classes, which is consistent with literature that those in better health are less likely to be depressed (Musliner et al., 2016).

Women’s Moderate, Men’s Low Class

Unlike the men’s and women’s low stable class, the other two classes included couples composed of partners who differed in their severity of depressive symptoms. Particularly, women experienced higher severity of depressive symptoms compared to their partners in the women’s moderate, men’s low depressive symptom class. More specifically, women in this class experienced moderate levels of depressive symptoms that varied over time, slightly increasing initially and then decreasing in later years. Their partners on the other hand, had low initial levels that increased over time. That women’s symptomatology improves and worsens over time suggests that there could be both competing efforts to decrease and increase their symptoms. Furthermore, given that men’s depressive symptoms increased over time suggests that they may be negatively affected by women’s symptoms as time progresses. This finding aligns with the stress generation theory (Hammen, 1991) and Coyne’s interpersonal theory (Coyne, 1976; Lemay & Cannon, 2012), which suggest that one partner’s symptomatology can negatively affect the nondepressed partner and increase their odds of developing depression.

Contrary to these theoretical assumptions, both women and men in this class had lower odds of higher relationship satisfaction at Wave 2. This poses an interesting dynamic for these couples in that only women experience greater depressive symptoms, but both partners had lower odds of higher relationship satisfaction. This sex difference in depressive symptoms has been widely debated and has many possible explanations (Hilt & Nolen-Hoeskema, 2014). From a stress generation lens, one possible explanation could be that women experience greater relationship distress and when depressed are more susceptible to distress in their relationship, which corroborates with one previous study (Gabriel et al., 2016). Furthermore, partners with depression can experience a depressogenic bias where they perceive greater negativity about their relationship than their partners (Cowden et al., 2011). For men, it is possible that they feel less satisfied in their relationship because their partner’s depressive symptoms strain their interactions with their partner (Sharabi et al., 2016).

In addition to relationship satisfaction, neuroticism was a significant predictor in of couple classes. In fact, women’s higher neuroticism had a larger association with membership in this class than relationship satisfaction. This is consistent with prior research in which wives’ neuroticism was associated with depression (Cao et al., 2017). Membership in this class also appeared to be related to health stressors since men and women in better health were less likely to be in this class. German native-born individuals were also less likely to be in this class, which suggests that there are more immigrant couples in this class; this aligns with prior research indicating that immigrant couples in Germany are likely to experience depression (Tibubos et al., 2018). From a stress generation perspective, it is possible that for immigrant women, living in a foreign country brings more stress such as economic and family stress, that leads to depression. Many immigrant couples in our sample reported living in Germany for more than a decade and it is possible that some stressors may have been present for years and could have affected women’s depression for years. Additionally, one study showed that immigrants in Germany from Turkey and Southern Europe had small, but greater declines in their mental health trajectories over time (Nesterko et al., 2019). Due to small sample sizes of couples from Turkey and other countries in this sample, we were unable to further examine differences in our study among participants from different countries. In light of our findings, however, we encourage future research to continue to study the course of depression among immigrant couples.

Men’s Moderate, Women’s Low Class

Finally, in the men’s moderate, women’s low class, men experienced moderate levels of depressive symptoms that remained stable over time while their partners experienced low levels of depressive symptoms that increased over time. There were fewer couples in this class, which is also consistent with the literature that suggests more women report depression than men (for an overview see Hilt & Nolen-Hoeskema, 2014; Salk et al., 2017). Despite these sex differences, this class illustrates that among some couples, men experience moderate levels of depression that persist over time, while women experience low levels of symptoms. In contrast to the women’s moderate, men’s low class, men’s and women’s relationship satisfaction was not associated with being in this class. Furthermore, the association of men’s and women’s relationship satisfaction did not differ between this class and men’s and women’s low stable class. This is surprising because we would expect partners in this class to experience less relationship satisfaction. This finding is in contrast to the marital discord model of depression (Beach, et al., 1990) and recent research on European couples showing that men’s and women’s greater marital discord is associated with higher depressive symptoms in themselves and in their partners (Salinger et al., 2020). This finding may be attributed to men’s neuroticism, which was the only predictor of this class. This finding adds to the continuing discussion on confounding variables in the association between relationship satisfaction and depressive symptoms (e.g., Cao et al., 2017; Whisman et al., 2006). Previously, neuroticism was either not associated with depression or was associated with depression, but only for women (Cao et al., 2017; Whisman et al., 2006). However, these studies used samples with mild averages of depression. Our findings add to this work by suggesting that couples in which men experience moderate depressive symptoms and women experience no symptoms, men are likely to also experience higher neuroticism. Overall, neuroticism appears to be a notable factor in couples in which one partner experiences moderate depressive symptoms.

Limitations

This study extended prior research by examining longitudinal data using novel methods; however, there are a number of limitations that are important to acknowledge. Our sample was composed of a community population of German adults and may not reflect the experiences of clinical populations. Future research should explore these associations within clinical populations using more clinically relevant assessments of depression. Participants’ depressive symptom trajectories were all relatively stable, with slight variation over time for some partners, which limited our understanding of change over time. Our ability to understand the experiences of partners who were less satisfied with their relationship was limited because we excluded participants who were divorced at Wave 2, partners in this sample reported being generally satisfied with their relationship, and few couples divorced in later waves of the study. Further, this study used existing data that assessed biological sex, but not gender identity. Future research is needed to explore the course of depression among divorced and separated couples, as well as partners with non-binary gender identities.

Future research can also expand upon these findings by examining the documented bidirectional association between relationship satisfaction and depression (Morgan et al., 2018; Najman et al., 2014; Whisman & Uebelacker, 2009). This was beyond the scope of the study, which focused on identifying couple classes of depressive symptom trajectories. Further, prior literature noted self-esteem and stressful life events as possible confounding factors in examinations of couples with depressive symptoms (Cao, et al., 2017), and these variables were not assessed in the waves of Pairfam data used in this analysis. We encourage future research to further explore these possible confounding factors among couples with depressive symptoms.

Conclusion

This is the first known study to investigate couple classes of both partners’ depressive symptom trajectories. We identified three classes of couples’ depressive symptom trajectories: women’s moderate and men’s low class, men’s moderate and women’s low class, and men’s and women’s low stable class. Findings indicate that the course of symptoms varies by each partner’s severity, but generally remains stable over 4 years, except for couples with moderately depressed men in which their depressive symptoms worsened over time. Furthermore, moderately depressed partners are less satisfied with their relationships, which aligns with the three theories on depression among couples and prior empirical research. These findings suggest that partners with low symptoms, even among couples with men who report moderate symptoms, were just as satisfied with their relationship as partners in the low stable class. In conclusion, the course of depressive symptoms varies by partner and severity and couples with low depressive symptoms experience higher relationship satisfaction.

Footnotes

Author’s note

This paper uses data from the German Family Panel (pairfam), coordinated by Josef Brüderl, Sonja Drobnič, Karsten Hank, Franz Neyer, and Sabine Walper. Pairfam is funded as long-term project by the German Research Foundation. The current study was not funded.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Preston C Morgan

Open research statement

As part of IARR’s encouragement of open research practices, the author(s) have provided the following information: This research was not pre-registered.

The data used in the research are available. The data can be obtained at: .

References

Acock

A. C

(2005). Working with missing values. Journal of Marriage and Family, 67(4), 1012–1028. https://doi.org/10.1111/j.1741-3737.2005.00191.x

American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders: DSM-5 (5th ed.). American Psychiatric Association. https://doi.org/10.1176/appi.books.9780890425596

Beach

S. R. H.

Sandeen

E. E.

O’Leary

K. D

(1990). Depression in marriage: A model for etiology and treatment. Guilford Press.

Beck

A. T.

Steer

R. A

(1987). Manual of the revised beck depression inventory. Psychological Corporation.

Bos

E. H.

Bouhuys

A. L.

Geerts

van Os

T. W.

Ormel

(2007). Stressful life events as a link between problems in nonverbal communication and recurrence of depression. Journal of Affective Disorders, 97(1-3), 161–169. https://doi.org/10.1016/j.jad.2006.06.011

Brüderl

Drobnič

Hank

Huinink

Nauck

Neyer

F. J.

Walper

Alt

Borschel

Bozoyan

Buhr

Finn

Garrett

Greischel

Hajek

Herzig

Huyer-May

Lenke

Müller

Wilhelm

(2018). The German family Panel (pairfam). GESIS Data Archive, Cologne. ZA5678 Data file Version 9.1.0 https://doi.org/10.4232/pairfam.5678.9.1.0

Byrne

Carr

(2000). Depression and power in marriage. Journal of Family Therapy, 22(4), 408–427. https://doi.org/10.1111/1467-6427.00161

Cao

Zhou

Fang

Fine

(2017). Marital well-being and depression in Chinese marriage: Going beyond satisfaction and ruling out critical confounders. Journal of Family Psychology, 31(6), 775–784. https://doi.org/10.1037/fam0000312

Chisholm

Sweeny

Sheehan

Rasmussen

Smit

Cuijpers

Saxena

(2016). Scaling-up treatment of depression and anxiety: A global return on investment analysis. The Lancet Psychiatry, 3(5), 415–424. https://doi.org/10.1016/S2215-0366(16)30024-4

10.

Coyne

J. C

(1976). Toward an interactional description of depression. Psychiatry, 39(1), 28–40. https://doi.org/10.1080/00332747.1976.11023874

11.

Davila

Bradbury

T. N.

Cohan

C. L.

Tochluk

(1997). Marital functioning and depressive symptoms: Evidence for a stress generation model. Journal of Personality and Social Psychology, 73(4), 849–861. https://doi.org/10.1037/0022-3514.73.4.849

12.

Eaton

W. W.

Shao

Nestadt

Lee

B. H.

Bienvenu

O. J.

Zandi

(2008). Population-based study of first onset and chronicity in major depressive disorder. Archives of General Psychiatry, 65(5), 513–520. https://doi.org/10.1001/archpsyc.65.5.513

13.

Fredriksen

von Soest

Smith

Moe

(2019). Depressive symptom contagion in the transition to parenthood: Interparental processes and the role of partner-related attachment. Journal of Abnormal Psychology, 128(5), 397–403. https://doi.org/10.1037/abn0000429

14.

Fülöp

Bőthe

Gál

E. J. Y. A.

Demetrovics

Orosz

(2020). A two-study validation of a single-item measure of relationship satisfaction: RAS-1. Current Psychology, 41(4), 1–13. https://doi.org/10.1007/s12144-020-00727-y

15.

Gabriel

Beach

S. R.

Bodenmann

(2010). Depression, marital satisfaction and communication in couples: Investigating sex differences. Behavior Therapy, 41(3), 306–316. https://doi.org/10.1016/j.beth.2009.09.001

16.

Gabriel

Bodenmann

Beach

S. R

(2016). Sex differences in observed and perceived stress and coping in couples with a depressed individual. Open Journal of Depression, 5(2), 7–20. https://doi.org/10.4236/ojd.2016.52002

17.

Gotlib

I. H.

Whiffen

V. E.

Mount

J. H.

Milne

Cordy

N. I

(1989). Prevalence rates and demographic characteristics associated with depression in pregnancy and the postpartum. Journal of Consulting and Clinical Psychology, 57(2), 269–274. https://doi.org/10.1037//0022–006x.57.2.269

18.

Hammen

(1991). Generation of stress in the course of unipolar depression. Journal of Abnormal Psychology, 100(4), 555–561. https://doi.org/10.1037/0021-843X.100.4.555

19.

Hilt

L. M

Nolen-Hoeskema

(2014). Sex differences in depression. In Gotlib

I. H.

Hammen

C. L.

(Eds), Handbook of depression (3rd ed.). Guilford Press.

20.

Huinink

Brüderl

Nauck

Walper

Castiglioni

Feldhaus

(2011). Panel analysis of intimate relationships and family dynamics (pairfam): Conceptual framework and design. Zeitschrift für Familienforschung, 23(1), 77–101. https://doi.org/10.20377jfr.235

21.

Kendler

K. S.

Karkowski

L. M.

Prescott

C. A

(1999). Causal relationship between stressful life events and the onset of major depression. American Journal of Psychiatry, 156(6), 837–841. https://doi.org/10.1176/ajp.156.6837

22.

Klein

D.N.

Allmann

A.E

(2014). Course of depression: Persistence and recurrence. In Gotlib

I. H.

Hammen

C. L.

(Eds), Handbook of depression (3rd ed.). Guilford Press.

23.

Kline

R. B

(2016). Principles and practice of structural equation modeling. Guilford publications.

24.

Kouros

C. D.

Cummings

E. M

(2011). Transactional relations between marital functioning and depressive symptoms. American Journal of Orthopsychiatry, 81(1), 128–138. https://doi.org/10.1111/j.1939-0025.2010.01080.x

25.

Krohne

H. W.

Schmukle

S. C.

Spaderna

Spielberger

C. D.

(2002). The state-trait depression scales: an international comparison. Anxiety, Stress & Coping, 15(2), 105–122. https://doi.org/10.1080/10615800290028422

26.

Lantagne

Furman

(2017). Romantic relationship development: The interplay between age and relationship length. Developmental Psychology, 53(1738), 1738–1749. https://doi.org/10.1037/dev0000363

27.

Lavner

J. A.

Bradbury

T. N

(2010). Patterns of change in marital satisfaction over the newlywed years. Journal of Marriage and Family, 72(5), 1171–1187. https://doi.org/10.1111/j.1741-3737.2010.00757.x

28.

Lemay

E. P.

Jr Cannon

K. T

(2012). Dysphoric reassurance seeking breeds contempt: Experimental evidence. Journal of Social and Clinical Psychology, 31(10), 1023–1050. https://doi.org/10.1521/jscp.2012.31.10.1023

29.

Leung

B. M.Y.

Letourneau

N. L.

Giesbrecht

G. F.

Ntanda

Hart

(2017). Predictors of postpartum depression in individualed mothers and fathers from a longitudinal cohort. Community Mental Health Journal, 53(4), 420–431. https://doi.org/10.1007/s10597-016-0060-0

30.

Barrera

Hops

Fisher

K. J

(2002). The longitudinal influence of peers on the development of alcohol use in late adolescence: A growth mixture analysis. Journal of Behavioral Medicine, 25(3), 293–315. https://doi.org/10.1023/A:1015336929122

31.

Lim

G. Y.

Tam

W. W.

C. S.

Zhang

M. W.

R. C

(2018). Prevalence of depression in the community from 30 countries between 1994 and 2014. Scientific Reports, 8(1), 1–10. https://doi.org/10.1038/s41598-018-2143-x

32.

Morgan

Dell’lsola

Nicholson

Spencer

(2020). Stress generation theory in couples with depression: A latent profile analysis. Journal of Social and Personal Relationships, 37(7), 2205–2228. https://doi.org/10.1177/0265407520919

33.

Morgan

Love

H. A.

Durtschi

May

(2018). Dyadic causal sequencing of depressive symptoms and relationship satisfaction in romantic partners across four years. The American Journal of Family Therapy, 46(5), 486–504. https://doi.org/10.1080/01926187.2018.1563004

34.

Musliner

K. L.

Munk-Olsen

Eaton

W. W.

Zandi

P. P

(2016). Heterogeneity in long-term trajectories of depressive symptoms: Patterns, predictors and outcomes. Journal of Affective Disorders, 192, 199–211. https://doi.org/10.1016/j.jad.2015.12.030

35.

Muthén

L.K.

Muthén

B.O

(1998-2017). Mplus user’s guide. (8th ed.). Muthén & Muthén.

36.

Najman

J. M.

Khatun

Mamun

Clavarino

Williams

G. M.

Scott

Alati

Hayatbakhsh

(2014). Does depression experienced by mothers leads to a decline in marital quality: A 21-year longitudinal study. Social Psychiatry & Psychiatric Epidemiology, 49(1), 121–132. https://doi.org/10.1007/s00127-013-0749-0

37.

Nesterko

Turrión

C. M.

Friedrich

Glaesmer

(2019). Trajectories of health-related quality of life in immigrants and non-immigrants in Germany: A population-based longitudinal study. International Journal of Public Health, 64(1), 49–58. https://doi.org/10.1007/s00038-018-1113-7

38.

Rammstedt

John

O. P

(2005). Kurzversion des Big Five Inventory (BFI-K): Entwicklung und Validierung eines ökonomischen Inventars zur Erfassung der fünf Faktoren der Persönlichkeit. Diagnostica, 51(4), 195–206. https://doi.org/10.1026/0012-1924.51.4.195

39.

Rehman

U. S.

Gollan

Mortimer

A. R

(2008). The marital context of depression: Research, limitations, and new directions. Clinical Psychology Review, 28(2), 179–198. https://doi.org/10.1016/j.cpr.2007.04.007

40.

Ruscher

S. M.

Gotlib

I. H

(1988). Marital interaction patterns of couples with and without a depressed individual. Behavior Therapy, 19(3), 455–470. https://doi.org/10.1016/S0005-7894(88)80016-9

41.

Sacco

W. P.

Dumont

C. P.

Dow

M. G

(1993). Attributional, perceptual, and affective responses to depressed and nondepressed marital individuals. Journal of Consulting and Clinical Psychology, 61(6), 1076–1082. https://doi.org/10.1037/0022-006X.61.6.1076

42.

Salinger

J. M.

Whisman

M. A.

Randall

A. K.

Hilpert

(2020). Associations between marital discord and depressive symptoms: A cross-cultural analysis. . Advance online publication. Family Process, 60(2), 493–506. https://doi.org/10.1111/famp.12563

43.

Salk

R. H.

Hyde

J. S.

Abramson

L. Y

(2017). Gender differences in depression in representative national samples: Meta-analyses of diagnoses and symptoms. Psychological Bulletin, 143(8), 783–822. https://doi.org/10.1037/bul0000102

44.

Sander

Böcker

(1993). Die deutsche form der Relationship assessment scale (RAS): Eine kurze messung der Zufriedenheit in der Partnerschaft. Diagnostica, 39(1), 55–62.

45.

Sharabi

L. L.

Delaney

A. L.

Knobloch

L. K

(2016). In their own words: How clinical depression affects romantic relationships. Journal of Social and Personal Relationships, 33(4), 421–448. https://doi.org/10.1177/0265407515578820

46.

Soloski

K. L.

Durtschi

J. A

(2020). Identifying different ways people change: A latent basis growth mixture model example identifying nonlinear trajectories of binge drinking. Journal of Marital and Family Therapy, 46(4), 638–660. https://doi.org/10.1111/jmft.12382

47.

Spaderna

Schmukle

S. C.

Krohne

H. W

(2002). Berichteuber die deutsche Adaptation der State-Trait Depression Scales (STDS). Diagnostica, 48(2), 80–89. https://doi.org/10.1026//0012-1924.48.2.80

48.

Tibubos

A. N.

Beutel

M. E.

Schulz

Klein

E. M.

Brähler

Michal

Münzel

Wild

P. S.

Lackner

König

Pfeiffer

Wiltink

(2018). Is assessment of depression equivalent for migrants of different cultural backgrounds? Results from the German population-based gutenberg health study (GHS). Depression and Anxiety, 35(12), 1178–1189. https://doi-org.proxy1.cl.msu.edu/10.1002/da.22831

49.

UNESCO (2006). International standard classification of education: ISCED-97. Technical report, Montreal: UNESCO Institute for Statistics.

50.

Whisman

M. A.

Uebelacker

L. A.

Tolejko

Chatav

McKelvie

(2006). Marital discord and well-being in older adults: Is the association confounded by personality? Psychology and Aging, 21(3), 626–631. https://doi.org/10.1037/0882-7974.21.3.626

51.

Whisman

M. A.

Sbarra

D. A.

Beach

S. R

(2021). Intimate relationships and depression: Searching for causation in the sea of association. Annual Review of Clinical Psychology, 17(1), 233–258. https://doi.org/10.1146/annurev-clinpsy-081219-103323

52.

Whisman

M. A.

Uebelacker

L. A

(2009). Prospective associations between marital discord and depressive symptoms in middle-aged and older adults. Psychology and Aging, 24(1), 184–189. https://doi.org/10.1037/a0014759

53.

World Health Organization (2017). Depression and other common mental disorders: Global health estimates. Author.

54.

Zung

W. W. K.

Sartorius

Ban

T. A

(1986). Assessment of depression. Zung self-rating depression scale and depression status inventory (pp. 221–231). Springer.