Abstract
Despite extensive evidence of the importance of marriage and marital processes for mental health, little is known about the interpersonal processes around depression within marriage and the extent to which these processes are gendered. We use a mixed-methods approach to explore the importance of gender in shaping processes around depression within marriage. We approach this in two ways: First, using quantitative longitudinal analysis of 2,601 couples from the Health and Retirement Study (HRS), we address whether depressive symptoms in one spouse shape the other spouse’s depressive symptoms and whether men or women are more influential in this process. We find that a wife’s depressive symptoms influence her husband’s future depressive symptoms, but a husband’s depressive symptoms do not influence his wife’s future symptoms. Second, we conduct a qualitative analysis of in-depth interviews with 29 couples wherein one or both spouses experienced depression to provide additional insight into how gender impacts depression and reactions to depression within marriage. Our study points to the importance of cultural scripts of masculinity and femininity in shaping depression and emotional processes within marriage and highlights the importance of applying a gendered couple-level approach to better understand the mental health effects of marital processes.
Keywords
For the past several decades, empirical studies have concluded that married people report better mental health than those who are unmarried (Frech & Williams 2007; Gove, Hughes, & Style 1983; Mirowsky & Ross 2003; Waite & Gallagher 2000). Recent research demonstrates that the psychological benefits of marriage depend on interpersonal processes within marriage (Hawkins & Booth 2005; Williams 2003). Implicit in these studies is the idea that the mental health of one spouse is shaped, for better or worse, by the other spouse. Yet because these studies are based on individual-level analysis, comparing individuals across couples, little is known about the within-couple processes that influence mental health in the context of marital relationships. Specifically, we do not know (1) how depressive symptoms in one spouse influence depressive symptoms in the other spouse over time or (2) what interpersonal and emotional processes underlie experiences of depression within marriage. Considering how these patterns unfold over time is important because most theoretical frameworks suggest that the spousal transmission of depressive symptoms within marriage is a reciprocal and dynamic process, requiring longitudinal dyadic data for analysis.
Further, because depression, emotions, and many marital experiences are gendered (Gove et al. 1983; Ridgeway 2011; Rosenfield, Vertefuille, & McAlpine 2000; Simon 2002), we expect gender differences in the direction and nature of interspousal depressive symptom transmission and in the emotional and interpersonal processes around depression. Societal norms and structures around gender constrain and facilitate psychosocial processes around depression (Bird & Rieker 2008; Link & Phelan 1995; Schnittker & McLeod 2005). For example, scholars theorize that women’s higher incidence of depression is linked to women’s internalization, rather than externalization, of emotions and psychological distress (Rosenfield et al. 2000; Rosenfield, Lennon, & White 2005; Simon 2002). This internalization of distress may be especially salient within marriage, a key site for the production and reproduction of gender (Ferree 2010; Reczek & Umberson 2012); thus many past studies find gender differences in the psychological benefits and costs of marriage (Gove et al. 1983; Horwitz, McLaughlin, & White 1998; Marks & Lambert 1998; Simon & Marcussen 1999). Beyond gender shaping spousal depressive symptom transmission within marriage, we also expect gender to influence interpersonal emotional processes within marriage during periods of depression. A key characteristic of these interpersonal processes involves gender differences in emotion work which is tied to depression (Erickson 2005; Lois 2010; Zapf 2002). Understanding how gender shapes these interpersonal processes around depression sheds light on how gender influences depression experiences within marriage, including transmission of depressive symptoms.
In this study, we use a mixed-methods approach to address the gendered dynamics around depression in marriage. We examine this in two ways. First, using couple-level longitudinal quantitative methods, we explore unfolding linkages between the depressive symptoms of husbands and wives, describing the general patterns of interspousal depressive symptom transmission and examining the extent to which this is gendered. We ask the following:
(1) Does a married person’s depressive symptoms at one point in time influence his or her spouse’s depressive symptoms at a later point in time?
(2) Is this influence an asymmetric process moderated by gender or a reciprocal process in which both partners are equally influential?
Second, our dyadic quantitative models are complemented with qualitative analysis of in-depth couple-linked interviews with spouses from the Marital Quality over the Life Course (MQ) project to examine how people react to their own depression and/or their spouse’s depression. This qualitative analysis allows us to evaluate the subjective meanings of depression and dynamic processes around depression, providing more insight into how gender shapes processes of depression within marriage, including interspousal depressive symptom transmission. In our analysis of the in-depth couple-linked interviews with 58 adults in long-term marriages, we ask the following:
(3) What are the interpersonal and emotional processes that occur when one or both spouses are depressed?
(4) (How) do these processes differ when women are depressed compared to when men are depressed? How are these processes described differently by women and men?
A mixed-methods approach allows us to address multiple gaps in our understanding of gender, depression, and marriage. Using quantitative and qualitative methods, we move beyond considering general patterns of depressive symptom transmission to also examining meanings and processes around depression in marriage. We consider how depression within a marital relationship is not an individual experience; rather, depression reverberates between partners in gendered ways. Using quantitative methods and nationally representative data enables us to test the extent to which these reverberations are gendered, an empirical question largely unexamined by past studies. Moreover, this quantitative assessment is complemented by a qualitative analysis of in-depth interviews in ways that extend our knowledge of gendered processes that undergird depression within marriage.
Background
Depression within Marriage
Depressive symptoms are not experienced in isolation with consequences only for the depressed individual. Rather, a depressed person affects and is affected by others (Coyne 1976b; Frech & Williams 2007; Joiner & Katz 1999). Scholars from a wide range of disciplines have applied this interpersonal model of depression to a multitude of relationships, most commonly, marriage (Butterworth & Rodgers 2006; Coyne 1976a, 1976b; Holahan et al. 2007; Kouros & Cummings 2010; Sandberg, Miller, & Harper 2002; Siegel et al. 2004; Townsend, Miller, & Guo 2001). These studies provide insight into how spouses experience and react to one another’s depressive symptoms. These studies, however, leave many questions unanswered, particularly related to the importance of gender in shaping the spousal transmission of depressive symptoms within marriage and the interpersonal processes around depression within couples.
Studies examining depressive symptom transmission within marriage that rely on large, representative, longitudinal samples find that having a spouse with high levels of depressive symptoms is associated with significantly more depressive symptoms in the respondent at later points in time (Butterworth & Rodgers 2006; Holahan et al. 2007; Siegel et al. 2004). These studies, however, do not examine the significance of gender in shaping the transmission of depressive symptoms between spouses. The few studies that do address gender rely on small, nonrepresentative samples and yield inconsistent findings, partly reflecting the divergent samples considered in these studies. A cross-sectional study of 176 Finnish couples and an 11-year longitudinal study of 316 Australian couples found that a woman’s depressive symptoms positively influence her husband’s depressive symptoms but a man’s depressive symptoms do not as strongly shape his wife’s symptoms (Kivela et al. 1998; Walker et al. 2011). Yet, a cross-sectional study of 553 Mexican-American couples and a 3-year longitudinal study of 296 midwestern couples with adolescent children concluded the opposite—that the husband’s depressive symptoms influence his wife’s depressive symptoms but not vice versa (Kouros & Cummings 2010; Peek et al. 2006). A study that examined changes in depressive symptoms among depressed adults upon entering marriage concluded that gender does not moderate depression outcomes (Frech & Williams 2007). In sum, past representative longitudinal studies of depressive symptom transmission within marriage largely ignore gender, and studies that acknowledge the potentially gendered nature of the transmission process are mostly nonrepresentative and inconclusive.
Gender, Depression, and Marriage
We expect the experiences of depression within marriage to be gendered because heterosexual marriage—particularly around emotions, caregiving, and mental health, specifically, depression—is gendered. According to gender theory, gender is not a static role but a relational, ongoing, and negotiated process characterized by micro-macro dynamics (Connell 2005; Ferree 2010). In addition to its emergent and shifting nature, gender is institutionalized and widely recognized as a system of practices (Yancey Martin 2004). Gender relations and practices are internalized by individuals and affect the larger social order, including family processes, emotions, and mental health (Connell 2005; Ferree 1990; Reczek & Umberson 2012; West & Zimmerman 1987).
Masculinities and femininities are often enacted within marriage, and the home is an important site for the production and reproduction of gender (Reczek & Umberson 2012; Ridgeway 2011; Simon 1995). Heterosexual marriages are organized around gender, and the social constructions of men and women as fundamentally different, particularly in regard to nurturing abilities and emotions, are the central driver of inequality in the household around unpaid work (Ridgeway 2011). Due to marriage’s gender frame, women are primarily responsible for emotion work, which are efforts done to foster one’s own or others’ positive emotional states (Erickson 2005; Hochschild 1979). Women are socially constructed as emotionally responsible for the well-being of their husbands, and this position of “emotion expert” is viewed as a natural feature of feminine traits and responsibilities related to women’s experiences as mothers (Gray & Smith 2009; Hochschild 1979; Risman & Johnson-Sumerford 1998). As gender scholars emphasize, masculinities are constructed in opposition to femininities (Connell 2005; Connell & Messerschmidt 2005; Ridgeway 2011), so while women are seen as emotion experts, men are often viewed as emotionally deficient and incompetent. This dynamic of difference plays out in the performance of emotion work by women and the general lack of emotion work by men (Duncombe & Marsden 1993; Erickson 2005; Pfeffer 2010). So while women may be willing to provide emotion work for their husband during his periods of depression, men may not reciprocate this work if their wife is depressed. We expect these gender divergent processes around emotion work to be important in understanding processes of depression within marriage.
Not only are marriage and emotion work structured by gender, but so is depression itself. Women’s positions of lower social power, greater responsibility for the private sphere of family, and social construction as nurturers all serve to promote self-schemas that emphasize the collective and an other-orientation (Rosenfield et al. 2000). Self-schemas that prioritize others above the self are associated with the development of internalizing symptoms, including those of depression (Rosenfield et al., 2000, 2005; Rosenfield, Phillips, & White 2006). Because the cultural prescription for femininity involves privileging the needs of others over one’s own self (Rosenfield et al. 2006) and because femininity is less valued by society than masculinity (Barrett & White 2002; Ridgeway 2011), higher levels of femininity and lower levels of masculinity are associated with a greater risk of depression (Barrett & White 2002). Men’s psychological distress may be more likely than women’s to manifest in externalizing ways, including substance abuse (Rosenfield et al. 2005; Simon 2002); however, this is not to say that men do not experience depression. Men’s depressive symptoms often manifest differently than women’s, such as through being less likely to seek professional help, and these differences may be particularly salient within marriage (Addis 2008).
In addition to experiencing depression less than women due to prevailing cultural scripts of masculine stoicism, independence, and invulnerability (Courtenay 2000), men may be less likely than women to divulge feelings of depression to their spouse (Addis 2008; Addis & Mahalik 2003; Möller-Leimkühler 2002), potentially reducing the likelihood that wives will have the opportunity to perform emotion work dedicated to alleviating husbands’ depressive symptoms. Further, men may be resistant to emotion work provided by their spouse, particularly during times of depression when men may deny that they have any mental health problems to protect their masculine identity of strength and independence (Addis & Mahalik 2003).
Other gender theories, however, suggest that these processes of hiding depressive symptoms and emotions from a spouse are not unique to men. According to the nurturant role hypothesis, women, as part of their feminine identities, are more likely to be nurturers (e.g., provide emotion work) for their family members, both daily and during periods of illness, and they are less willing than men to receive care and adopt the “sick role” (Gove 1984). Whether this extends to mental health symptoms, like depression, has not been examined. If so, this would mean that depressed women are resistant to accepting emotion work from their husbands. Emotion work involves not only working to improve the emotions of others, as discussed above, but also acting to disguise certain emotions, including repressing or masking feelings of depression (Hochschild 2003). If depressed women disguise their depressed feelings from their husband as a form of emotion work and as a way of doing femininity (for example, to protect the emotions of others), this would reduce a husband’s opportunities to provide emotion work for his wife. For these reasons, we expect gender to shape interactions between depressed men and women and their spouse.
In sum, while past studies show that spouses influence one another’s depression, it is still unclear whether men or women are more likely to be affected by their spouse’s depressive symptoms and what interpersonal processes are involved during periods of depression within marriage. We use a mixture of qualitative and quantitative methods to inform our understanding of gender and depression within marriage.
Data and Methods
Our mixed-methods approach integrates quantitative and qualitative components. We use quantitative methods to analyze national couple-linked longitudinal data; these data allow us to assess population-level patterns of gender and depressive symptoms within marriage and between spouses. We utilize qualitative methods to analyze couple-linked in-depth interview data; these data allow us to examine the meanings, interactions between processes, and experiences of depression within marriage from the point of view of both spouses, complementing the quantitative analysis by providing additional insight into depression, gender, and marriage.
Quantitative Methods and Data
Data
In the first part of this study, we analyze four waves of data from the Health and Retirement Study (HRS), a nationally representative sample of respondents ages 51 to 61 years in 1992 and their spouses, any age, if married. Respondents were reinterviewed approximately every two years (Juster & Suzman 1995). The HRS oversampled African Americans, Latino Americans, and married couples. We limit our analytic sample to married couples in which both spouses were interviewed in 1994, 1996, 1998, and 2000. We restrict the sample to the earlier years of the HRS to minimize mortality selection. We ran supplementary analyses looking at additional waves, but the patterns remain similar across later years, so we use only these four waves to maximize the number of married couples in the sample. We begin in 1994, rather than 1992, because the Center for Epidemiologic Studies–Depression Scale (CES-D) questions in 1992 differ from later years. The sample size is 2,601 couples. For analysis, we use the RAND HRS data, provided by the RAND Center for the Study of Aging (RAND HRS Data 2010).
The HRS is large, nationally representative, and longitudinal, which is an improvement on previous studies on depression within marriage that were based on small, geographically limited, or cross-sectional samples (Kivela et al. 1998; Kouros & Cummings 2010; Peek et al. 2006; Walker et al. 2011). Using a representative sample allows us to gain insight into how these processes occur at the population level. Additionally, because past studies indicate that marital processes and depression unfold over time (Barrett 2000; Umberson et al. 2006), using longitudinal data collected over a number of years is critical. Finally, the HRS is one of the few large data sets in which both respondents and their spouses are interviewed, making dyadic data analysis possible. Dyadic data are critical, as this study hinges on examining the lived experiences of husbands and wives within marriage, requiring data that include both partners’ perspective.
Measures
Depressive symptoms
The main outcome variable is depressive symptoms. The depressive symptom index provided by the HRS uses eight items from the CES-D scale (Radloff 1977). These items measure whether the respondent experienced the following depressive symptoms all or most of the time: feels depressed, feels everything is an effort, has restless sleep, feels alone, feels sad, cannot get going, feels happy, and enjoys life. All items are coded so that higher values reflect more depressive symptoms. We sum the number of depressive symptoms the respondent had experienced all or most of the time, and this final scale ranges from 0 to 8. This short form of the CES-D scale has predictive accuracy when compared to the full-length version and strong correlation with poor mental health (Andreson et al. 1994; Grzywacz et al. 2006). All measures are collected from each spouse.
Covariates
Covariates include length of current marital duration (in years), number of years of education, self-rated health (0 = good, very good, or excellent, 1 = poor or fair), race/ethnicity (dummy variables with four categories: non-Hispanic white, non-Hispanic black, Hispanic, and other), number of living children, log of household income, and employment status. Marital duration, number of years of education, self-rated health, race, number of living children, log of household income, and employment status are all included as covariates because past research shows that each is associated with depressive symptoms and marital processes (Mirowsky & Ross 2003; Moen, Kim, & Hofmeister 2001). Current marital duration, number of years of education, and log of household income are treated as continuous variables. We use the log of household income to reflect income’s curvilinear association with well-being (Ecob & Davey Smith 1999). Current marital duration, log of household income, and number of living children are consistent within couples, and years of education, age, self-rated health, race/ethnicity, and employment status are specific to each respondent. Employment status consists of three mutually exclusive categories, treated as dummy variables: currently working for pay, currently retired, and currently not working for pay and not retired. Length of current marital duration (in years), number of years of education, number of living children, and race/ethnicity are all measured at baseline, and this baseline value is controlled for at every wave. This is because these values remain relatively consistent across waves. Variables that fluctuate across waves, namely, self-rated health, log of household income, and employment status, are measured at each wave, and the value specific to each wave is controlled for at each wave.
Analytic Strategy
The analysis begins with a comparison of summary statistics of men and women. Means, proportions, and standard deviations are calculated, and chi-square difference tests are conducted to compare the men and women in the sample. Next, we use a series of autoregressive models to test alternative assumptions about the direction of influence between husband’s and wife’s depressive symptoms. Specifically, we fit three models:
Model 1: a model assuming a bidirectional influence between husband’s and wife’s depressive symptoms across time;
Model 2: a model assuming a unidirectional influence from wife’s depressive symptoms to husband’s depressive symptoms across time; and
Model 3: a model assuming a unidirectional influence from husband’s depressive symptoms to wife’s depressive symptoms across time.
Each model also controls for correlation between husband’s and wife’s depressive symptoms at the same time point, husband’s depressive symptoms across time, and wife’s depressive symptoms across time (see Figure 1 for illustration of full reciprocal model). We use autoregressive models, where past values of one variable predict future values of the same variable, because this allows for estimation of simultaneously linear relationships among various combinations of observed variables (Kline 2011). Additionally, we are able to assess the directionality of the relationship between variables, testing whether husbands or wives are more influenced by their spouse’s depressive symptoms.
Illustration of full reciprocal autoregressive model.
All models are estimated using Mplus 6.1 (Munthen & Munthen 2010). Mplus uses Full Information Maximum Likelihood (FIML) procedure to deal with missing data (Arbuckle 1996). We use maximum likelihood parameter estimates with conventional standard errors and chi-square test statistics. We evaluate the fit of the models using three goodness of fit measures: adjusted χ2, the Akaike information criterion (AIC), and the Bayesian information criterion (BIC).
Qualitative Methods and Data
Data Collection and Recruitment
The second part of our analysis addresses the interpersonal processes around depression within a marriage and the ways in which these processes might be gendered. We conduct qualitative analysis of in-depth interviews from the Marital Quality over the Life Course (MQ) project, which entailed 60 individual in-depth couple-linked interviews conducted between 2003 and 2006 with both spouses in long-term (more than 7 years) heterosexual marriages. The goal of the MQ project was to gather marital life course narratives to explore how spouses influence one another’s health and how this changes over time. Respondents were asked specifically about their own and their spouse’s mental health, making this an ideal qualitative sample for our study. Spouses were interviewed separately and privately to preserve each person’s perspective and to allow the respondents to comfortably and honestly discuss topics that they felt might upset or offend their spouse.
Couples were recruited in a southwestern city in the United States. Most respondents (40) were recruited through a local newspaper article, and the rest were recruited through newspaper advertisements and referrals by other respondents. Interviews lasted 1.5 hours on average and typically occurred in the respondent’s home. The interviews were tape-recorded and transcribed verbatim with pseudonyms assigned to protect confidentiality. The interviews were semistructured and consisted of questions on a number of topics related to how their health experiences shaped their relationship with their spouse and vice versa. The interviews were retrospective and traced the entire course of the relationship, from when the couple began dating to the time of the interview. Respondents were specifically asked, “Have you or your spouse ever had a significant period of depression or other mental health problem?” In 29 couples, one or both spouses responded yes and specifically discussed episodes of depression, and these couples composed the subsample used for this present analysis. These couples were asked to describe what happened during the episode(s) of depression, specifically, how it affected themselves, their spouse, and their marriage; how their spouse responded; what treatments they pursued and/or were encouraged to pursue; and pertinent follow-up questions.
Sample
The final subsample consists of 29 married couples (58 individual interviews). The majority of respondents are white (48), six are African American, one is Asian American, two are Latino American, and one identifies as multiracial. Average marital duration is 25 years, ranging from 8 years to 52 years, and the average age of the sample is 53 years. Most respondents have at least some college education. The average household income is $60,000. In 10 couples, both spouses mention at least one episode of depression; for 8, only the husband reports being depressed at some point; and in the remaining 11 couples, only the wife reports an episode of depression. This reflects national patterns showing that, on average, women are more likely than men to experience depression (Kessler et al. 2005).
Compared to the quantitative sample from the HRS, the MQ sample is slightly younger (average age of 53 years compared to 61 years in the HRS data), more educated (most of the qualitative sample are college graduates compared to less than half of the HRS sample), and married for fewer years (25 years married compared to 32 years). Both groups, however, are composed of couples married for at least 7 years, due to methodological design, with most respondents in both groups married 20 years or more. Further, though the qualitative sample is younger, both groups are primarily composed of mid- to later-life adults. Both samples are similar on race/ethnicity, as they are majority white while still including African American and Latino American respondents. The respondents in the MQ interviews are not participants in the HRS survey. Indeed, it is rare for a nationally representative survey to include in-depth interviews. However, the two samples are complementary and unusual in that both data sources include dyadic data and address the topic of depression. Thus, taken together, the qualitative and quantitative data analyzed for this study provide a rich and in-depth examination of gender, depression, and marriage.
Analytic Strategy
Our goal in the qualitative analysis is to examine how a depressed spouse shapes interspousal dynamics with a particular emphasis on how these processes are shaped by the gender of the depressed spouse. We use Charmaz’s (2006) qualitative analysis approach, which builds upon Glaser and Strauss’s (1967) grounded theory. This approach emphasizes coding major themes that emerge from participants’ interviews instead of being predetermined by researchers (Charmaz 2006). We began this analysis by carefully reading through the transcripts and field notes multiple times. Then, we did open line-by-line coding to identify repeated concepts. We used NVIVO qualitative software to aid with the housing of the data and codes. We focused on interview excerpts regarding depression within a couple. Themes and subthemes were identified and extracted for analysis. We then read and analyzed these sections multiple times to identify key subthemes.
Results: HRS Data
Table 1 presents descriptive statistics of variables used in the analysis. First we compare men in the sample with women in terms of the study variables. On average, compared to men, women in the sample report more depressive symptoms at every time point, report similar health, are younger, are less educated, and are less likely to be working for pay as well as less likely to be retired. These differences are significant at the p < .001 level. Depressive symptoms increase across time for both men and women.
Means and Standard Deviations of Variables (Health and Retirement Study, N = 5,202).
Note: CES-D = Center for Epidemiologic Studies–Depression Scale.
Autoregressive Models
Table 2 shows the estimated effects of husband’s and wife’s depressive symptoms on each other across time and associated goodness-of-fit statistics from the three models. In Table 2, we report only the correlations between husband’s and wife’s depressive symptoms from time t to time t+1 as these are the key variables of interest. Model 1 tests the bidirectionality of the relationship between men’s and women’s depressive symptoms from time t to time t+1. In this model, respondents’ own depressive symptom levels are significantly correlated longitudinally (i.e., over time) (p < .001). All cross-spousal effects are observed net of controls. Cross-sectionally (i.e., within each time t), husbands’ and wives’ depressive symptoms are positively associated with each other (p < .001). Longitudinally, the husband’s depressive symptoms at time t are not significantly associated with his wife’s depressive symptoms at time t+1, with the exception of time 2 to time 3 when it is marginally significant (p < .05). Yet the wife’s depressive symptoms at time t are positively associated with her husband’s depressive symptoms at time t+1, with this significant at the p < .001 level at almost all time points. This suggests that the wife’s depressive symptoms are more consistently predictive of the husband’s future depressive symptoms than vice versa.
Summary of Fitted Model Coefficients for the Associations between Husband’s and Wife’s Depressive Symptoms and Model Goodness-of-Fit Indices.
Note: Data are from Waves 2 through 5 of the Health and Retirement Study, N = 5,202. Adjusted for marital duration, educational attainment, self-rated health, race/ethnicity, number of living children, log of household income, and employment status. SE = standard error; AIC = Akaike information criterion; BIC = Bayesian information criterion.
We also estimate unidirectional path models, Models 2 and 3 in Table 2, to examine whether these models improve fit compared to the bidirectional model. Model 2 specifies that wives’ depressive symptoms affect husbands’ depressive symptoms but not vice versa. In contrast, Model 3 specifies that depressive symptoms are transmitted only from husband to wife. The difference in chi-square statistics between Models 1 and 2, Δχ2(3) = 8.7, p = .03, is marginally statistically significant, indicating that the model in which husbands’ depressive symptoms do not affect wives’ depressive symptoms is not a much better fit than Model 1 with reciprocal cross-spousal transmissions. This lack of substantial improvement in model fit is consistent with the pattern that most paths from husbands’ to wives’ depressive symptoms are not significant in Model 1 or Model 3. Taken together, these findings support the conclusion that the husband’s depressive symptoms do not affect wives’ depressive symptoms longitudinally. The difference in model fit of Model 3 compared to Model 1, Δχ2(3) = 33.0, p < .001, is large in magnitude and highly significant, indicating that Model 3 is a worse fit and supporting that the direction of influence on depressive symptoms from wives to husbands fits the data the best. Moreover, the AIC and BIC for Model 2 is lower than for Models 1 and 3, indicating that this model has a better fit. These patterns are consistent with the finding that all paths from wives’ to husbands’ depressive symptoms are significant in Models 1 and 2.
These results demonstrate that the wife’s depressive symptoms at one point in time are positively associated with her husband’s depressive symptoms at a later point in time, but the husband’s depressive symptoms do not influence his wife’s later depressive symptoms. These results raise questions about interpersonal dynamics around depression within a marriage that cannot be well addressed quantitatively. In particular, what are the interpersonal processes introduced by depression within a marriage? How do these differ when the husband is depressed compared to when the wife is depressed? What meanings do men and women assign to their own and their spouse’s depressive symptoms? We now turn to a qualitative analysis of our in-depth interview data to address these questions.
Results: In-Depth Interviews
In the qualitative analysis, we examine relationship processes around depression. The qualitative analysis suggests several consistent themes involving emotion work, including subthemes of negative emotions, conflict over concealed emotions, and absence of emotion work. We present quotes that best illustrate the predominant themes and subthemes revealed in the analysis, and we indicate where these themes are most predominant for men and women. Several important patterns emerge, and these patterns are highly gendered. These themes characterize the majority of couples where one or both partners experienced depression, though there are a few exceptions. In couples where the wife (but not the husband) reported depression (n = 11 couples), the wife and husband typically described negative emotions, hostility, and conflict regarding the wife’s depression and/or a lack of emotion work by the husband to his wife, either because the wife concealed her depression from her husband or the husband felt incapable of emotionally supporting his wife. In contrast, in couples where the husband (but not the wife) reported depression (n = 8 couples), the husband and wife discussed his depression frequently, they worked together to cope with his depression, and the wife provided emotion work for her husband. In couples where both husband and wife were depressed (n = 10 couples), women prioritized their husband’s well-being over their own well-being. In turn, husbands received emotion work from their wives, and women felt more negative emotion due to this unequal exchange, though they typically concealed this from their husband. The themes for these couples are similar to themes for couples in which only the wife was depressed. These findings, elaborated on below, shed light on interpersonal dynamics around depression within marriage.
Couples with a Depressed Wife
In general, themes that emerged from interviews with couples that included a depressed wife involve three interrelated processes: negative emotions and conflict, lack of emotion work by husbands, and concealed emotions.
Negative Emotions and Conflict
In several couples, communications regarding the wife’s depression were characterized by negativity and conflict. In five couples, husbands told their depressed wives to “get over it,” and both husbands and wives described these interactions as stressful. Irene said that when she was in “a dark state,” her husband Brian would “get furious with me and tell me to snap out of it.” This dismissal of the wife’s depression sometimes occurred because the spouses disagreed over whether the wife’s depression was legitimate. Chantelle, who was eventually hospitalized due to her postpartum depression, said that her husband, Anthony, “denied it” because “it was something he was not knowledgeable about.” He confirmed this when he said during his interview that “I wasn’t convinced that there was anything was really, really wrong beyond just feeling blue.” He said they would fight over whether she was depressed or not, and he would tell her, “I understand on the one hand, but life goes on.” For other couples, the husband’s denial of his wife’s depression was not as hostile but was still a source of negative emotions for the wife. Angie described the conflict in her marriage created by her husband’s denial of her depression:
Before I was diagnosed with the depression, but I felt like there was something wrong, [my husband] kept saying, “There is nothing wrong with you. It’s just fine . . . you’re just stressed out or tired or sad,” or things like that. So, there was a lot of discord because I felt like he was discounting . . . I knew there was something really big that was wrong.
It was not until a friend convinced Angie to see a doctor and she received a depression diagnosis that her husband began to accept her assessment of her situation.
Sometimes these communications created negative emotions and conflict, not because the husband denied the wife’s depression but because the husband did not respond to the depression in ways that the wife preferred. When Sasha was depressed, Joel said he would tell her, “What would you like me to do? Tell me what to do, and I will do it for you,” emphasizing actions he could take. Sasha said in her interview that she just wanted Joel to talk to her, not do anything. Phil described this issue with his wife, Christine, when they would discuss her depression: “If she can get past the emotional reaction to what I am saying . . . she has a habit of wanting to go back and dredge up old problems and not focus on the subject we are talking about at the moment.” Christine, like Sasha, emphasized in her interview how Phil’s approach was not helpful. Overall, the interview data suggest that a disconnect between a husband’s and wife’s expectations about her depression often led to hostility, negative emotions, and conflict.
Lack of Emotion Work by Husbands
Seven of the men with a depressed wife discussed how they were unable to emotionally support her during episodes of her depression. This was also described by eight of the depressed women about their husband. Many of these husbands took a completely hands-off approach because they did not perceive the problem as serious enough to warrant any response, they did not know what to do, they did not think they could help, or, as in 7 of the couples (discussed in the next section), they did not realize their wife was depressed. In 11 couples in which the wife was depressed, the husband, the wife, or both mentioned that the husband was unsure of what to do, so he remained passive. Some men felt that they were unable to help. Rick explained how he justified not trying to help his wife during her depression: “You can’t talk a person out of depression. You can’t convince them not to be depressed because you are basically talking on a cognitive level and they are dealing on an emotional level.” However, many women wanted their husband to discuss their depression with them. Susan complained that her husband, Benjamin, “doesn’t respond the way I wish he would . . . I would like for him to really talk to me about it and ask, you know, ‘What is going on and what can I do?’” Because he does not, she said, she felt “more anxious, just because it frustrates me.” Benjamin agreed that he avoided talking about her depression with her, and he said that he realized that his lack of communication “puts a strain between us.”
Many of the women felt that their husband could not help them, and these women said they accepted this as an unchangeable part of who their husband was. For instance, Karen said of her husband, Nathan, “Well, it’s hard for him. . . He doesn’t know really what to do to make me get out of it.” Even though she felt this was just her husband’s nature, Karen also noted that because he did not feel he could help, “he just gets frustrated.” Karen cried during this portion of the interview, indicating the negative emotions she felt from their situation. Similarly, Tracy said of her husband, “He can understand equipment, he can understand cause and effect with things that are not me, but with me he seems totally clueless. . . . He has a blockage about human emotional things and especially females.” Because of this, Tracy said that his approach to her depression was “more hands-off, ignoring, kind of not knowing what to do.” This lack of support promoted isolation, negative emotions, and conflict for both spouses within these couples.
Concealed Emotion by Wives
Our analysis suggests that for some couples, the lack of emotion work by men was because women concealed their depression from their husbands. For seven of the couples in which the wife was depressed, the wife discussed actively concealing her depression from her husband. Louise, when asked about how she and her husband coped with her depression, said that she did not think he ever knew about it. She said, “I hadn’t told him that because it was my problem.” In some cases, women concealed their depression from their husbands in an attempt to protect the husband’s emotions, in this way performing emotion work through masking their own depressed feelings. Angie said when she was rediagnosed with depression after a period of feeling better, “I did not tell my husband because I knew it would scare the heck out of him.” Later, when she told him what was happening, after she began taking antidepressants again, she recounted that he said, “I’m glad you didn’t tell me because I would have been scared and worried.” Concealing depression often led to women feeling isolated. Irene described her period of depression when she chose not to tell her husband of her diagnosis: “I was essentially by myself and he had no conception of what was going on. No clue.”
This was not a theme unique to depressed women, though it was more prominent among depressed women than depressed men. In all the instances when the wife chose to conceal her depression from her husband, her husband either remained unaware that anything was wrong or realized something was wrong but decided to not discuss it with his wife. Frank said that when he realized his wife was depressed, he chose to leave her alone:
Well, when she gets into a real low mood she tends to go into isolation. And the best thing that I can do in many of those cases is just leave her alone and let her go through it. Because usually the things that she’s depressed about are nothing I can do anything about anyhow.
In contrast, only one-tenth of the depressed men, compared to one-third of the women, discussed trying to conceal their depression from their spouses, but in the two cases where this occurred, the wife sensed something was wrong and worked to learn what this was. Stephen described this process with his wife:
So, when I would get down, [my wife] wanted to take blame for that. She always felt it was something that she had done. And of course my response when she would ask what was wrong, “Nothing. Nothing is wrong.” You know. So, I would just clam up and I created tension occasionally when I would get in those moods. And now, I have enough sense to just tell her, “I am just down. I am just feeling down, and I don’t know why.”
Because Stephen’s wife continued to ask how he was, even when he denied anything was wrong, they began discussing his depression together. This theme of men and women working together to cope with the husband’s depression is discussed below.
Couples with Depressed Husbands
In couples with a depressed wife, the primary theme involved women dealing with depression alone, either not communicating with their husband about the depression and treatment or experiencing hostility and conflict about it. These relationships tended to be characterized by negative emotions, lack of emotion work by men, and emotion work through concealing depressed feelings by women. In contrast, in interviews with couples with a depressed husband, the primary theme involved both partners communicating often about the husband’s depression with these communications rarely producing conflict and the wife providing emotion work for her husband during depression episodes. Thus couples with depressed husbands were more likely to be characterized by high levels of emotion work from wives to husbands and a general absence of conflict.
In many of these couples, women communicated with their husband regarding his depression to provide support and improve his mood. These women often described themselves (or were described) as acting as cheerleaders and optimists who encouraged positive emotions in their husband. As Jenn said, “I’ve been more of a cheerleader. This is what you need to do. Let’s get up. Let’s get moving . . . complimenting him.” This positive attitude by wives helped their husbands improve, as expressed by Joe, who said that when he was depressed, talking to his wife “helped me get through that, and accept that fact that okay, I can do it.”
In contrast to interviews in which the wife was depressed, in interviews where the husband was depressed, we found that most couples discussed talking often about his depression and treatment, making decisions together. Toni monitored her husband’s depression through constant conversation. When she thought he needed to do something about his depression, her husband, Joe, agreed to go on antidepressants. He said of Toni, “I guess that’s normal that she could verbalize it and explain it and in a lot of ways help me get through it . . . and in ways that help me normalize taking antidepressants.” Joe saw this as a “natural” act for Toni, as his wife, to do for him.
When women did not provide emotion work to their husbands during their husband’s depression, or at least the amount of support expected by the husband, this caused conflict and strain. Kinsey said that after a long period in which her husband was depressed, she told him, “You are going to have to pull yourself up here.” When she told him this, she said that he “loses it,” telling her, “‘Like all I am asking from you is a little sympathy. . . . You are not the person I want you to be.’” This was reported by only one couple with a depressed husband. In all the other couple interviews, the husband and wife said that the wife facilitated emotionally positive conversations that the husband found helpful. Even in the case of Kinsey, she said that she responded to his criticisms of how she supported him and began providing a positive outlook on her husband’s depression, which she and he agreed “is what he needs.”
Couples with Depressed Husbands and Wives
Interviews with couples in which the husband and wife were depressed at the same time (n = 10) were characterized by themes involving emotion work from the wife to the husband, both in women concealing their own depression and caring for their husband’s depression, but no emotion work from the husband to the wife. The primary theme involved women concealing their own depression from their husband to support their husband and not risk exacerbating his depression. Both Jason and his wife, Maria, experienced episodes of depression, and while Maria knows they were both depressed, Jason did not know about Maria, saying, when asked directly, “If she was depressed, she disguised it very well, so I didn’t know.” Doris and Malcolm both experienced depression and were on antidepressants, though even at the time of the interview, Malcolm was not aware that this had happened to Doris. She chose not to tell him so that it would not exacerbate his depression. She sent him to the doctor first and decided not to see a doctor until after he began treatment, prioritizing his care over her own. Bill and Katherine were in a similar situation to Doris and Malcolm, in that both received treatment but Bill was not even aware of Katherine’s depression. Katherine encouraged Bill to see a psychiatrist and scheduled his appointment before her own, because “I thought he was the sickest and he needed to go first.” Women’s acts of prioritizing their husband’s treatment and well-being over their own could have negative effects on their own mental and emotional health through not being able to rely on their husband for emotion work and through delayed treatment for themselves.
Discussion
Most past research on marriage and depression addresses how marital status and marital dynamics affect depression and what these dynamics mean for the depressed person (Lin, Ye, & Ensel 1999; Whisman & Beach 2001; Williams 2003). The present study extends our understanding of depression and marriage by considering how depression in one or both spouses impacts each partner. Our quantitative analysis suggests that husbands’ and wives’ depressive symptoms are highly correlated over time and that the link between spouses’ depressive symptoms primarily flows from the influence of wives on husbands. A wife’s depressive symptoms contribute to her husband’s future depressive symptoms, but a husband’s depressive symptoms do not have a significant effect on his wife’s future symptoms. We follow our quantitative assessment with a qualitative analysis of in-depth interviews with couples in which one or both spouses experienced depression to address questions about how marital dynamics around depression might produce these gendered patterns. The qualitative analysis suggests that the relationship dynamics of couples with a depressed wife are markedly different from those of couples with a depressed husband. These differences revolve around the presence or absence of emotion work and conflict. We find that relationships of couples with a depressed wife tend to be characterized by hostility, isolation, and negative emotions, whereas marital relationships in which the husband is depressed are more likely to be characterized by emotion work performed by the wife to the benefit of the husband. Understanding the gendered aspects of depression and the interpersonal processes around depression provides new insight into gendered relationship dynamics within marriage and the importance of these gendered dynamics for altering or contributing to well-being of husbands and wives.
Past studies conclude that marriage provides important resources for the depressed that reduce the likelihood of depression (Thoits 2011; Umberson, Wortman, & Kessler 1992; Waite & Gallagher 2000), but our quantitative and qualitative findings point to a caveat: Being married to a depressed person comes with considerable psychological cost, and these costs and the processes underlying these costs differ by gender. This is seen in the quantitative analysis, as we find that one spouse’s depressive symptoms lead to increases in the other spouse’s depressive symptoms over time. According to the stress process model, stress proliferates and accumulates throughout a person’s life, manifesting as psychological distress and depressive symptoms (Pearlin et al. 2005). An extension of this model supported by our quantitative analysis suggests that depressive symptoms proliferate not only within persons but also between persons, from spouse to spouse. This has been demonstrated in past studies (Coyne 1976a; Holahan et al. 2007; Sandberg et al. 2002; Siegel et al. 2004), though we extend this by considering the role of gender. The gendered nature of cross-spousal sharing of depressive symptoms is particularly noteworthy and fits with the significant theoretical and empirical literature on gender and depression. Depressive symptoms in the general population are unequally distributed by gender, with women experiencing more depressive symptoms than men (Kessler et al. 2005). Additionally, marriage is a gendered institution (Ferree 2010), suggesting that the transmission of depressive symptoms within marriage would be gendered. Current debates about depression, gender, and marriage tend to focus on whether men or women benefit more from marriage (Hawkins & Booth 2005; Williams 2003). But we ask a fundamentally different question: Whose depressive symptoms are more influential in contributing to depression in the other spouse? We find that the cross-spousal influence of depressive symptoms does depend on gender, such that women’s depressive symptoms have a stronger influence on their husband’s depressive symptoms than vice versa. This has important implications for policy and clinical practice, as it demonstrates that protecting against depression in married women may also lead to less depression in their husband. A social network study of over 12,000 adults from the Framingham Heart Study similarly found that women’s depressive symptoms are more consequential than are men’s for the depressive symptoms of those in their social networks (Rosenquist, Fowler, & Christakis 2010), suggesting that these gendered processes of depressive symptom transmission extend beyond marriage.
We complement this new understanding of the gendered transmission of depressive symptoms with an examination of gendered processes around depression within marriage. Our qualitative analysis demonstrates additional psychological costs involved in being married to a depressed person, particularly for women. While the quantitative analysis focuses on general patterns of gender and depressive symptoms at the population level, our qualitative analysis highlights meanings and processes that underlie the lived experience of gendered marital dynamics around depression. This qualitative assessment then provides unique insights into gender, marriage, and depression. The qualitative results suggest that marital processes around depression look markedly different depending on whether the husband, the wife, or both are depressed and that gendered differences in the management of emotions are central to understanding these processes. Gender differences in emotion work and management are produced within a relational context (Connell 2005; Connell & Messerschmidt 2005; Ridgeway 2011), and this is especially apparent within marriage. Emotion work within marriage is generally viewed by respondents in our study as being the responsibility of women and as falling under the expertise of women more than men, as other scholars theorize (Erickson 2005; Ferree 2010). The wives of depressed men are much more likely than the husbands of depressed women to offer extensive emotion work to support and help their depressed spouse. But, even more striking, depressed women also perform emotion work to protect their husband from the stress and distress of having a depressed wife—something that depressed men did not describe.
Importantly, this emotion work may have negative consequences for women (Lois 2010; Zapf 2002), particularly those who are already depressed. An orientation toward the other has been shown to be detrimental to women’s well-being, promoting internalized forms of distress, such as depression (Rosenfield et al. 2006). To shield their husband from their depression, many women engage in self-silencing, an act described by feminists as the devaluation of personal experiences and suppression of emotions (Beauboeuf-Lafontant 2007; Jack 1993). As posited by Gove in his nurturant role hypothesis, women’s attempts to care for others conflict with the sick role (i.e., being a patient by resting and receiving care from others), and thus women do not fully adopt the sick role—or in this case, the depressed role—in the way men do (Gove 1984).
While women demonstrate acts of nurturing, other-orientation, and self-silencing as ways of enacting femininity, husbands of depressed women describe themselves, and are generally described by their wife, as behaving and interpreting their behavior as incapable of dealing with emotions—either their own or their wife’s. These descriptions offered by husbands and wives closely conform to cultural scripts of masculinity as described by gender scholars (Beauboeuf-Lafontant 2007; Connell 2005). Masculinity is often constructed as at odds with emotional expertise because being the “emotional expert” is a characteristic of femininity (Hochschild 1979; Risman & Johnson-Sumerford 1998). Rosenfield and others also describe how masculinity often involves a self-orientation (Rosenfield et al., 2000, 2005; Simon 1995), and this is seen in our qualitative analysis when most husbands do not provide emotion work directed at reducing their wife’s depression, even going so far as to deny that their wife is depressed when she says that she is. Our qualitative results suggest that these dynamics promote isolation and hostility within a couple, and as a result, depressed women do not receive emotion work and care from their spouse—resources frequently enjoyed by depressed husbands. Additionally, as past studies demonstrate that hostile marital environments, typified by isolation and conflict within a marriage, increase the transmission of depressive symptoms from person to person (Kouros & Cummings 2010; Yorgason et al. 2006), it may be that these gendered processes of hostility and isolation identified in the qualitative analysis provide an explanation for our quantitative finding that women’s depressive symptoms influence their husband’s depressive symptoms but not vice versa. Past empirical tests of the nurturant role hypothesis, self-silencing, and the other-orientation focus on the negative mental health effects these processes have for women (Rosenfield et al. 2006), but our study introduces the possibility that these processes also negatively impact men’s well-being.
Our conclusions might be markedly different if we had used an alternative measure of mental health rather than depressive symptoms and depression. Depressive symptoms are an internalized form of psychological distress that is more prevalent in women than men (Rosenfield et al. 2005). We find that men’s depressive symptoms do not influence their wife’s depressive symptoms. This may occur because depression is a relatively less common experience for men than for women, so depression in a man generates great concern and attention from his wife. In contrast, cultural scripts of femininity and greater prevalence of depression among women than among men may inspire less concern and reactivity from the spouse of a depressed woman. Or it may be that women’s depressive symptoms are more strongly influenced by their husband’s substance use than their husband’s depressive symptoms. Thus, if we were to conduct a similar analysis of gendered cross-spousal transmission of distress with a focus on alcohol consumption as the key outcome measure, it might yield a different and informative pattern of results. While a parallel analysis of alcohol consumption, in addition to depressive symptoms, is beyond the scope of this mixed-methods study, we conducted some preliminary analyses that suggest that alcohol use is correlated across time within a marriage, and husband’s alcohol use increases wife’s depressive symptoms, though wife’s alcohol use does not influence her husband’s depressive symptoms. But these are only exploratory results that await further investigation. We strongly urge sociologists of mental health to examine the cross-spousal transmission of alcohol consumption in addition to depressive symptoms, as this would expand our current understandings of gender and mental health in marriage.
Despite the unique contributions provided by using a mixed-methods approach to examine the gendered patterns and processes in the transfer of depressive symptoms within marriage, several limitations should be addressed. First, the HRS asks questions regarding marital quality in limited waves, and it asks these questions only to a subset of respondents. We chose to exclude these waves for a number of reasons, including concerns over mortality selection and inconsistent CES-D measures; thus we do not formally test whether or not marital quality moderates depressive symptoms transmission, as is suggested in our qualitative analysis and in previous research (Choi & Marks 2008; Kramer 1993). Second, these quantitative analyses include only those couples who remained married over the eight-year study period, and the qualitative analysis is limited to couples who have been married at least seven years. Therefore, couples who divorced, perhaps due to the depression of one or both partners, are not included, though they could provide important examples of how depression influences marital dynamics. The consequence of this limitation is likely an underestimation of the effects of depression within marriage. Third, the qualitative analysis is limited by the heterogeneity of the sample in terms of race, ethnicity, and income. Gendered marital and depression processes likely differ between groups, but we are unable to address this possibility. Fourth, our quantitative and qualitative analyses are based on different samples. The qualitative sample is younger as well as less representative of the general population. We acknowledge limitations in using in-depth interview data from one sample of people to understand quantitative processes for another sample, and we call on future research designs to include in-depth qualitative interviews within larger, nationally representative surveys.
Our study shifts the focus of research on marriage, mental health, and gender from status to process. Even though it is generally known that married people, on average, have better mental health than those who are unmarried (Frech & Williams 2007; Mirowsky & Ross 2003; Waite & Gallagher 2000), we do not know much about the dynamic processes through which mental health outcomes are produced within marriage. Because both marriage and depression are gendered processes, one needs to understand the profound role of gender at the dyadic level rather than simply cataloguing individual-level gender differences in mental health across marital status categories. Marital relationships and gendered emotion work within marriage are dynamic and fluid processes. Moreover, depression does not develop in a vacuum but is embedded throughout the course of a marriage. By carefully analyzing gender and mental health using blended qualitative and quantitative methods, we can better understand marital processes around depression and the mental health consequences of these processes for men and women.