Why are Depressive Symptoms More Prevalent Among The Less Educated? The Relevance of Low Cultural Capital and Cultural Entitlement

ABSTRACT

Analyzing nationally representative survey data collected in the United States in 2014 (n = 1,932), we formulate and test a novel explanation for the educational gradient in depressive symptoms. We theorize that status as cultural capital drives this gradient in addition to well-established economic and social factors, via the feelings of cultural entitlement it inspires. Therefore, we use structural equation modeling to decompose the education effect. We demonstrate that in addition to economic (job security and income) and social factors (embeddedness in the neighborhood), cultural capital indeed accounts for the educational gradient in depressive symptoms via cultural entitlement. We conclude that for understanding social gradients in mental health it is vital to be sensitive for the cultural mechanisms that status as cultural capital can inspire. Based on our findings we propose suggestions for further research.

Introduction

It is clear that the study of depression does not exclusively belong to the domain of psychology: Because of the pronounced social stratification of depressive symptoms, it is a popular theme in sociology as well (see, e.g., Petersen 2011). The level of education especially plays a significant role in processes that affect mental health (Bjelland et al. 2008; Bracke, Pattyn, and Von dem Knesebeck 2013). Several studies have established that the more educated experience fewer depressive symptoms than the less educated (Bauldry 2015; Bjelland et al. 2008; for a meta-analysis, see Lorant et al. 2003), and a majority of longitudinal studies (e.g., Johnson et al. 1999) indicates that more education leads to less depressive symptoms, rather than vice versa. Scholars have thus acknowledged education as a “fundamental cause” of (mental) health disparities (Link and Phelan 1995; Pampel 2009). While, evidently, there has been a lot of attention on educational differences in health outcomes, the association still requires further explication (e.g., Bauldry 2015; Van Kippersluis et al. 2010).

As indicated by several authors, existing explanations traditionally focus on the economic privileges of the higher educated (e.g., Mirowsky and Ross 1998; Pampel, Krueger, and Denney 2010). In addition to the well-established economic causes for the mental health advantage of the more educated, it is argued that social factors also play an important role. Kawachi et al. (1997), for instance, found that higher levels of educational attainment are positively associated with being embedded in cohesive social structures, and the mental health benefits of this embeddedness are widely acknowledged (see, e.g., Ferlander 2007). For a sociological understanding of the educational gradient in depressive symptoms it is therefore important to consider social factors next to economic ones. However, a sociological approach could make an additional contribution by focusing on the role of status as cultural capital (henceforth, cultural capital). Cultural capital indicates one’s affinity with elite culture, and level of education is one of its primary indicators (Bourdieu, 1984; Bourdieu and Passeron, 1977).

Building on earlier suggestions of the relevance of cultural capital for mental health issues (Abel, 2008; Pinxten and Lievens, 2014), we formulated and empirically tested a specific mechanism through which cultural capital could contribute to a better understanding of the relationship between education and depression. In short, we argue that an individual’s cultural capital shapes a sense of “cultural entitlement,” which entails a sense of being a relevant and legitimate citizen who matters in society. This is, for instance, illustrated by research by Spruyt, Van Droogenbroeck and Kavadias (2015), who showed that individuals with lower levels of education more often experience a sense of futility. They feel, moreover, less entitled to participate in legitimate culture and institutions (Lamont, Beljean and Clair 2014; Laurison 2015), disenfranchised, and excluded (Myles 2008). As we explain in this article, such a low sense of cultural entitlement not only is shaped by educational level as an indicator for cultural capital but also could have important consequences for individual mental health.

By proposing a mechanism based on cultural entitlement, we aim to contribute to existing knowledge on the relationship between education and depression and provide an empirical basis for a novel, sociological interpretation of how cultural capital is relevant to depression. After explicating how this interpretation adds to existing notions of the role of economic and social factors, we empirically test hypotheses about the role of economic position, embeddedness in a cohesive social structure, and status as cultural capital using nationally representative survey data collected in the United States in 2014.

Positioning the role of status as cultural capital next to economic and social factors

Economic position

The positive relationship between education level and economic indicators such as income and employment is both well known and well documented (e.g., Psacharopoulos 2014). The resulting variations in economic circumstances can, in turn, be related to health inequalities, as Wilkinson and Pickett (2010) pointed out.

Education may, for instance, influence depression through income. As Marmot and Wilkinson (2001) highlighted, individual income can influence general health in two ways. First, income is directly related to health through the material conditions and the resources it provides to secure basic needs and, ultimately, health (Pinxten and Lievens 2014). Income also influences mental health through the social and emotional meaning it implies: Income shapes evaluations of relative social status, and this may influence factors like life satisfaction and depression (Diener et al. 1993:196). A related aspect of the economic privilege of the more educated is their relatively secure labor market position (Näswall and De Witte 2003). In general, insecurity about whether one will be able to maintain a job in the future leads to stress reactions and an overall reduction in psychological well-being (De Witte 1999), which may result in anxiety and depression (Orpen 1993).

An economic explanation of the relationship between education level and depression could thus be summarized as follows: The less educated are more likely to report depressive symptoms because they have a lower income (Hypothesis 1a) and experience more job insecurity (Hypothesis 1b).

Embeddedness in cohesive social structures

Next to a stronger economic position, social characteristics of the more educated could also play a role in their better mental health. The more prevalent embeddedness in heterogeneous networks among the higher educated (Ferlander 2007) has, for instance, been linked to assets like employment opportunities (Granovetter 1983; Small 2006). This aspect of social integration has already been discussed in the previous section, as such opportunities are associated with higher incomes and lower job insecurity and their potential role in the association between education and depression. In this section we focus on the psychosocial benefits of social integration instead.

In general, the more educated are more embedded in cohesive social structures than less educated individuals (Kawachi et al. 1997). In line with work of classical sociologists such as Durkheim on the beneficial effects of social integration, current studies show that the strong and intimate relationships that constitute social structures like the family are crucial to mental health (Ferlander 2007). When it comes to the family domain, the more educated more often have a stable relationship with a significant other (Ono 2009), which serves as a rich source of social support (Ferlander 2007) and as such mitigates depressive symptoms (Dahl and Malmberg-Heimonen 2010). Other facets of family life, like having children, are argued to play a role as well (Helbig et al. 2006), but the relationship between education and parenthood is not straightforward (see, e.g., Liefbroer and Corijn 1999).

The more educated also have access to a rich potential of social support in their local community, namely, the neighborhood. Several studies have found more educated individuals to be more embedded in a cohesive neighborhood community, which is considered to reduce individual depression through feelings of security and mutual trust (Ross 2000).

In short, there are two ways in which more embeddedness in cohesive social structures could underlie greater prevalence of depressive symptoms among the less educated: less access to social support from a significant other and weaker integration in a cohesive neighborhood. We thus anticipate the following: The less educated are more likely to report depressive symptoms, because they less often have a partner (Hypothesis 2a) and are less strongly embedded in a cohesive neighborhood (Hypothesis 2b).

Status as cultural capital and cultural entitlement

Next to the discussed economic and social factors, Bourdieusian cultural capital could also play an important role in explaining why the better educated experience less depressive symptoms. Cultural capital denotes the extent to which one appreciates, demonstrates, and masters the lifestyle of the cultural elite (shorthand: one’s affinity with elite culture, or affinity with legitimate culture; Bourdieu 1984; Bourdieu and Passeron 1977) and indicates one’s status: one’s standing on the cultural ladder. It results from socialization within the family and educational institutions.

We are certainly not the first to claim that cultural capital is relevant for understanding social gradients in mental health (see, e.g., Abel 2008; Pinxten and Lievens 2014), nor do we claim that the mechanism we formulate is the only way in which cultural capital might be relevant. After all, affinity with elite or legitimate culture is more encompassing than what its standard indicators in stratification research—level of education (often referred to as “institutionalized cultural capital”) and highbrow cultural consumption (indicating one’s “embodied cultural capital”; DiMaggio 1982; Jæger and Breen 2016)—account for. Cultural capital might affect mental health in various ways. (We elaborate on this in the Discussion section.) However, our main goal is to formulate, and subsequently test, a specific mechanism by means of which cultural capital might account for the educational gradient in mental health, in addition to economic and social factors.

The mechanism we propose revolves around what we label “cultural entitlement”—a “sense of ease” (Laurison 2015:926) because of the feeling that one matters, is a relevant, valuable, and legitimate part of society, and is entitled to participate in its dominant institutions such as education and politics. The low affinity with elite or legitimate culture and dominant institutions among the less educated is likely to be accompanied with a low sense of cultural entitlement. Research indeed demonstrates that the low status accompanying lower levels of educational attainment decreases people’s sense of cultural entitlement (see, e.g., Lareau, 2015; Spruyt et al. 2015). The less educated do not perceive themselves as legitimate actors in society, leading to withdrawal from dominant institutions (Kuppens et al. 2015; Lamont et al. 2014; Laurison 2015). In addition, the less educated are excluded from dominant discourse because their dispositions are not considered legitimate or appropriate (see, e.g., Myles 2008).

When it comes to its relevance to mental health, cultural entitlement could influence depressive symptoms in at least three ways. Determining through which of these three pathways cultural entitlement influences depressive symptoms goes beyond the scope of this study and is for future research to decide. First, because of the negative identity that is tied to their low-status position (Kuppens et al. 2015; Spruyt and Kuppens 2015), groups with less education experience a sense of being looked down on (also called “perceived contempt”; Spruyt et al. 2015). As is commonly acknowledged, perceived stigmatization is negatively related to mental health outcomes (Kessler, Mickelson, and Williams 1999). A lower sense of cultural entitlement could also influence depression through internalization of this negative image: Low levels of self-esteem are related to higher chances on depression (Orth, Robins, and Roberts 2008). Finally, not feeling entitled to participate in legitimate culture and institutions relates to low feelings of control and efficacy (Laurison 2015:928). Such feelings play an important role in mental health because they are crucial to active problem solving and coping (Mirowsky and Ross 1998), which aids prevention and solving of health issues (Pudrovska et al. 2005).

While we do not aim to disentangle the specific pathways through which low cultural entitlement associates with depressive symptoms in the present study, the aforementioned clearly indicates that lower cultural entitlement is a potential mechanism through which a lower level of education and higher prevalence of depressive symptoms are connected. In short, we expect that their smaller amount of cultural capital explains why the less educated experience more depressive symptoms, because it inspires a weaker sense of cultural entitlement (Hypothesis 3).

Data and measurements

We use data gathered on our behalf by the GfK Group (formerly Knowledge Networks) in 2014 (Van der Waal, Achterberg, and De Koster, 2014). Contrary to other surveys used to scrutinize social gradients in mental health, this one allows us to incorporate both the standard measure for embodied cultural capital and an indicator for cultural entitlement. There were 3,966 people sampled from a probability-based panel carefully designed to be representative of noninstitutionalized adults age 18 or older residing in the United States. There were 2,062 respondents who completed the survey (response rate = 52%).¹ Respondents who completed the survey in 10 minutes or less, which is not enough time to provide serious answers, were excluded from our analyses. This leaves a data set of 2,006 respondents.

To measure depressive symptoms, we used the Center for Epidemiological Studies Depression Scale (CES-D; Radloff 1977), commonly used in research on education and depression (see, e.g., Miech and Shanahan 2000). Following studies demonstrating that shorter, four- or five-item versions of the CES-D are as sensitive and specific as the full version (see, e.g., Fendrich et al. 1989; Lewinsohn et al. 1997), a five-item version of the CES-D was used. We follow Van de Velde, Bracke, and Levecque (2010) and Radloff (1977) by using the CES-D as an interval-level measure indicating the frequency and severity of depressive symptoms and not as a diagnostic tool with a specific cutoff point. For uniformity with the rest of the survey, answer categories for the items range from 0 (rarely or none of the time) to 10 (most or all of the time). Respondents were asked how often they experienced the following during the previous week: “I felt depressed,” “My sleep was restless,” “I felt lonely,” “I had a crying spell,” and “I could not get going.” A factor analysis produced a first factor with an eigenvalue of 3.25, explaining 65% of the variance. We constructed a scale (Cronbach’s α = 0.86) by calculating the mean score for respondents with valid scores on at least four of five items. Higher scores indicate greater prevalence and persistence of depressive symptoms.

Education level was measured using the highest degree received, ranging from 1 (less than high school) to 4 (bachelor’s degree or higher).

To measure income, the respondents were asked to indicate which of 19 categories their annual household income fell into, ranging from 1 (less than $5,000) to 19 ($175,000 or more).

To measure job insecurity, we combined two items. Employed respondents were asked how worried they were about losing their job, ranging from 0 (not much at all) to 10 (a lot). Using the same answer categories, all other respondents were asked how worried they were about not being able to find a job in the near future. Combining the two, we have created a variable where higher scores indicate higher levels of job insecurity.

The measurement of respondents’ sense of cultural entitlement was informed by research on expressions of not feeling entitled among the lower educated. According to Kuppens et al. (2015), a potential reaction of the less educated to having a negative group identity is to “withdraw from areas associated with the low-status attribute” (p. 1261). Laurison (2015) similarly discussed how the less educated do not see themselves as sufficiently competent, and therefore do not feel entitled to participate in legitimate culture. Laurison (2015) indicated that this is illustrated by the greater frequency of answering “don’t know” in a political survey among the less educated, who withdraw from the arena of politics because they do not feel entitled to participate.

Following relevant research on this topic (e.g., Berinsky 2004; Bourdieu 1984; Laurison 2015), we measure low cultural entitlement by counting the “don’t know” and nonanswers (missings) provided by respondents on questions on five political issues (derived from Delli Carpini and Keeter 1996:304–306): “Do you happen to know what job or political office is now held by Joe Biden?” “Whose responsibility is it to determine if a law is constitutional or not?” “How much of a majority is required for the U.S. Senate and House to override a presidential veto?” “Do you happen to know which party had the most members in the House of Representatives in Washington before the election in 2012?” and “Would you say that one of the parties below is more conservative than the other at the national level?” A factor analysis produced a first factor with an eigenvalue of 2.53, explaining 51% of the variance. A scale (Cronbach’s α = 0.75) was constructed by summating the don’t knows and missing values on the five items.

We measure embodied cultural capital by means of the standard indicator in the status attainment literature: interest and participation in highbrow cultural activities (e.g., DiMaggio 1982; Jæger and Breen 2016). Respondents were asked how frequently they go to a live opera, ballet, or classical music performance; visit an art museum or gallery, historic park or monument, or neighborhoods for their historic or design value; and read books (with the exception of books required for school or work). They were also asked about the extent to which they consider themselves to be a lover of the arts and culture. All items had response categories varying from 0 (never or not at all) to 10 (very often or very much). A factor analysis produced a first factor with an eigenvalue of 2.41, explaining 60% of the variance. The mean score was calculated for the respondents with valid scores on at least three out of four items, producing a reliable scale (Cronbach’s α = 0.75). Higher scores represent more embodied cultural capital.

Embeddedness in cohesive social structures was measured with two indicators. First, inspired by research on the health effects of social relationships (see, e.g., Frech and Williams 2007), we include a measure for having a partner (people who are married or cohabiting coded as 1, others coded as 0). Second, inspired by research on the mental health effects of local communities (e.g., Latkin and Curry 2003), we include a measure of embeddedness in a cohesive neighborhood. We used the following items: “People around here are willing to help their neighbors,” “This is a close-knit neighborhood,” and “People in this neighborhood can be trusted” (taken from Oh and Kim 2009; Sampson and Raudenbush 1999). The response categories ranged from 0 (strongly disagree) to 10 (strongly agree). A factor analysis confirmed that these three items represent one underlying dimension of embeddedness in a cohesive neighborhood, with an eigenvalue of 2.47, explaining 82% of the variance. A scale (Cronbach’s α = 0.89) was constructed by calculating the mean for respondents with valid scores on all three items. Higher scores indicate more embeddedness in a cohesive neighborhood.

Age (in years), gender (coded 0 for male and 1 for female), and parenthood (coded 0 for no children and 1 for one or more children) are included as controls, as they are related to depressive symptoms (e.g., Helbig et al. 2006; Nolen-Hoeksema 2001; Wade and Cairney 1999), as is race/ethnicity (e.g., Turner 2013), which was measured using five dichotomous variables, indicating whether the respondents considered themselves to be non-Hispanic white (reference category), non-Hispanic black, Hispanic, non-Hispanic multiracial, or non-Hispanic of another race. We also control for the region where the respondents reside (Northeast, Midwest, South, and West) and whether they live in a metropolitan area (No = 0, Yes = 1). Finally, as the panel utilized includes both online and offline populations, we include a variable indicating whether (No = 0, Yes = 1) respondents had household Internet access at the time of recruitment. (Those who had not were provided with the digital means and assistance necessary for completing online questionnaires.)

Results

Before inspecting the role of status as cultural capital in the educational gradient in depressive symptoms, we provide the descriptive statistics and the bivariate relationship to education for all variables in Table 1. This table shows that there is a very modest number of missing values on some variables. After listwise deletion, 1,932 respondents (96.3%) remain, and are included in the structural equation models.

Table 1. Descriptive statistics and bivariate relationship to education.

	M	SD	Range	n	Association with education
					r	η
Depressive symptoms	2.03	2.04	0–10	1,980	−0.16***
Education	2.88	0.96	1–4	2,006	N/A	N/A
Household income	12.10	4.40	1–19	2,006	0.41***
Job insecurity	3.49	3.35	0–10	1,981	–0.16***
Low cultural entitlement	1.83	2.08	0–5	2,006	–0.35***
Embodied cultural capital	3.51	2.14	0–10	2,003	0.30***
Partner	0.64	0.48	0–1	2,006		0.09***
Embeddedness in cohesive neighborhood	6.00	2.45	0–10	1,966	0.09***
Age	51.00	16.65	18–92	2,006	–0.03 (ns)
Gender (female)	0.50	0.50	0–1	2,006		0.06**
Parenthood	0.28	0.45	0–1	2,006		0.03 (ns)
Non-Hispanic white (ref.)	0.77	0.42	0–1	2,006
Non-Hispanic black	0.08	0.27	0–1	2,006
Hispanic	0.09	0.28	0–1	2,006		0.15***
Non-Hispanic other	0.03	0.18	0–1	2,006
Non-Hispanic multiracial	0.03	0.38	0–1	2,006
Northeast (ref.)	0.18	0.38	0–1	2,006
Midwest	0.26	0.44	0–1	2,006
South	0.34	0.47	0–1	2,006		0.07*
West	0.23	0.42	0–1	2,006
Internet access	0.83	0.38	0–1	2,006		0.21***
Metropolitan area	0.83	0.37	0–1	2,006		0.16***

*p < .05. **p < .01. ***p < .001 (two-sided tests for significance).

Figure 1 provides more in-depth insight into the bivariate relationship between education and depression and shows that the relationship between our measures of education and depressive symptoms does not appear to show signs of nonlinearity. This is furthermore supported by the linearity check we performed, which produced insignificant results (p = .29), indicating linearity assumptions are not violated.

Figure 1. Bar chart of relationship between education and depressive symptoms.

Moving on to the central issue of this article, we test our hypotheses using structural equation modeling (SEM), which allows us to disentangle the different paths through which education is associated with depressive symptoms. To handle the non-normality of the dichotomous variables included in our model, we use quasi-maximum likelihood to fit the standard errors. This method relaxes the normality assumptions of the default maximum likelihood estimation and produces robust standard errors (Klein and Muthén, 2007). Because the standard chi-square test statistic does not follow a chi-square distribution when data are non-normal, we use the Satorra–Bentler scaled chi-square to test the fit of our model (Satorra and Bentler, 1994). The Satorra–Bentler scaled chi-square, root mean square error of approximation (RMSEA), and comparative fit index (CFI) indicate that our model fits the data well (Satorra–Bentler scaled χ² = 49.78, p = .14; RMSEA = 0.01, pclose = 1; CFI = 0.996). The estimated path model is presented in Figure 2.^2,3

Figure 2. Structural equation model for testing hypotheses on the relationship between educational level and depressive symptoms. Note. n = 1,932. Standardized coefficients shown, controlling for concomitants age, gender, parenthood, race, region, Internet access and metropolitan status. All values are significant at the .05 level unless indicated otherwise.

To check the robustness of our results, we estimated two additional models. First, to take into account that one of our mediators, having a partner, is a binary variable, we estimated our final model using generalized SEM (GSEM). GSEM allows for specification of various types of responses and link functions, including binary variables and logit links. Despite these advantages of GSEM, we chose the SEM model presented in Figure 2 because no goodness-of-fit indicators and postestimation commands to estimate direct and indirect effects are available when using GSEM. Since all paths that are significant in the SEM model presented in Figure 2 are also significant in the alternative GSEM model, our findings appear robust.

Second, we estimated a model that included a correlation between education and embodied cultural capital instead of a one-way path from the former to the latter. We did this because such a model fits well with the theoretical idea that education and embodied cultural capital are both forms of cultural capital. It is, however, not possible to estimate the proportion of the relation between education and depression that is mediated (by each mediating variable and in total) when using such a model, which is why we chose to specify our model as presented in Figure 2. Again, this additional analysis indicates that our results are robust: The same paths are significant as in the model depicted in Figure 2.

Figure 2 demonstrates that our expectations regarding economic and social factors are largely confirmed. It shows that both income and job insecurity play a significant role in explaining why the less educated experience more depressive complaints, as stated in Hypotheses 1a and b. Embeddedness in the neighborhood plays an additional significant role, corroborating Hypothesis 2b. Contrary to our expectations, having a partner does not affect the frequency and severity of experiencing depressive symptoms, refuting Hypothesis 2a. The more educated thus experience less depressive symptoms partly because they have higher incomes, have less job insecurity, and are more embedded in a cohesive neighborhood.

Turning to the main point of our article, Figure 2 shows that status as cultural capital and cultural entitlement are indeed of additional relevance next to economic and social factors for explaining the relationship between education and depression. In line with extant stratification research, institutionalized cultural capital (level of education) is substantially positively related to embodied cultural capital (β = 0.30), and both inspire a sense of cultural entitlement (β = −0.28 and −0.11, respectively). Finally, a weaker sense of cultural entitlement relates to more depressive symptoms (β = 0.07). These results support our expectation that a smaller amount of cultural capital explains why the less educated experience more depressive symptoms, because it inspires a weaker sense of cultural entitlement.

To examine the proportion of the association between education and depression that is mediated, we calculated the strength and significance of direct and indirect effects (see Table 2). As Table 2 shows, cultural capital plays a significant role next to the included economic and social factors. Income and job insecurity both account for sizeable amounts of the total effect of education, respectively, 21.5% and 23.5%, whereas embeddedness in a cohesive neighborhood accounts for 4.6% of the total effect. With 12.1% and 1.4%, cultural entitlement proves to be a significant way in which both forms of cultural capital are related to feelings of depression. The fact that the less educated experience more depressive symptoms is partly accounted for by their weaker sense of cultural entitlement inspired by their lower level of cultural capital.

Table 2. Extent of mediation.

	Depressive symptoms
Coefficients
Total effect	−0.16
Direct effect	−0.07
Mediated effect	−0.09
Size of mediation	% of total effect
… via embodied cultural capital	ns
… via cultural entitlement	12.1%
… via embodied cultural capital and cultural entitlement	1.4%
… via income	21.5%
… via job insecurity	23.5%
… via partner	ns
… via embeddedness in neighborhood	4.6%

Note. n = 1,932. Standardized coefficients shown, controlling for age, gender, parenthood, race, region, Internet access and metropolitan status. All values are significant at the .05 level unless indicated otherwise.

As both Figure 2 and Table 2 show, a direct effect of education on depression remains after taking the hypothesized mechanisms into account (β = −0.07). We offer an interpretation of this remaining education effect in the discussion.

Discussion

Our results show that status as cultural capital plays a significant role in the negative relationship between education and depressive symptoms. Furthermore, our analyses indicate that what we call “cultural entitlement” is central to understanding how it plays a role. As our analyses show, a higher level of education, indicative of a higher level of institutionalized cultural capital, relates to a sense of cultural entitlement. The same goes for one’s amount of embodied cultural capital. Our findings resonate with highly influential sociological literature stressing the intrinsic connection between educational attainment and cultural capital (see, e.g., Bourdieu 1984) and expands on studies suggesting that cultural capital might play an important role in mental health issues (see, e.g., Abel 2008; Pinxten and Lievens 2014) by offering a specific interpretation and empirical test of this role.

Our analyses show that cultural entitlement is more strongly related to institutionalized cultural capital than to embodied cultural capital. This resonates with arguments that education is nowadays crucially important for the cultural stratification ladder in Western societies (see Spruyt and Kuppens 2014; Stubager 2009): Not only does it affect how individuals see their place in society because it provides individuals with affinity with elite culture, it is also directly related to perceptions of entitlement and “mattering.”

Although several authors suggest that embodied cultural capital has a negative effect on depression (see, e.g., Abel 2008; Khawaja and Mowafi 2006), we find no significant direct effect while utilizing its standard indicator (i.e., highbrow cultural consumption). This suggests that different, cross-pressuring mechanisms may underlie the relationship between embodied cultural capital and depression. We conducted an additional analysis that also points in that direction. As depressed individuals might participate less in cultural activities, we performed a robustness check by also conducting the analyses measuring embodied cultural capital without actual participation (i.e., only using the item on the extent to which respondents consider themselves to be a lover of the arts and culture). The resulting model shows a modestly significant positive effect on depressive symptoms (β = 0.05, p = .03; n = 1,921). When accounting for the possibility of reverse causality in the relationship between depressive symptoms and the indicator for embodied cultural capital, a moderate positive association between embodied cultural capital and depression thus remains.

This indicates that the theorized negative effects of embodied cultural capital on depressive symptoms (Abel 2008), for instance, through stimulating active help-seeking (Mirowsky and Ross 1998), may be cross-pressured by mechanisms that link embodied cultural capital to higher levels of those symptoms. A potential explanation can be found in the tendency to ruminate. The high level of reflexivity that accompanies embodied cultural capital (e.g., Threadgold and Nilan 2009) could in its extreme guise lead to excessive self-monitoring and rumination, which in turn could lead to more depressive symptoms (Crowe 2002). Such a cross-pressuring mechanism could explain why there is no association between embodied cultural capital and depression (main analyses), or even a small positive one (robustness check), while that capital is broadly conceived as beneficial for mental health.

This illustrates an issue we touched on in a previous section: Next to the specific one we analyzed here, cultural capital could also associate with depressive symptoms through other mechanisms. Abel (2008) for instance, argued that education stimulates the active construction of a healthy lifestyle. For him, cultural capital is mainly related to health outcomes because it offers certain values and knowledge, which are incorporated in a healthy lifestyle. Others link cultural capital to a general “know-how” and institutional knowledge, providing the more educated with tools to find their way in complex institutions (Lareau 2015). More specifically, the more educated are argued to be more able to navigate medical institutions (Heimer and Staffen 1998) and have a higher ability to understand, judge, and act upon health care information (Missinne, Neels and Bracke 2014). Taking such mechanisms into account might help explain the remaining direct effect of education.

To determine whether and to what extent these additional mechanisms play a role, future research could simultaneously include various more detailed measurements. For instance, we recommend incorporating multiple knowledge indicators. Specific measures of health knowledge are useful for assessing to what extent the ability to understand and act upon health care information explains the remaining education effect. In addition, one could include indicators of general institutional knowledge, which would aid examining if the ability of the more educated to find their way in contemporary institutions also inspires less feelings of depression. Similarly, including measures for active help-seeking behavior alongside indicators for self-monitoring and rumination could be valuable for further examining the potentially cross-pressuring mechanisms through which cultural capital affects feelings of depression. In addition, it would be useful to include specific measurements that could help determining how exactly cultural entitlement is related to depressive symptoms. As we discussed, a sense of cultural entitlement could shape depression in different ways, ranging from having a sense of control and active problem solving, to the consequences of feeling stigmatized. Future studies using additional indicators could unravel which of these mechanisms are relevant to the association between cultural entitlement and depression.

Finally, note that with regards to cultural entitlement and the rationale behind it, the word “status” may bring to mind theories of “status anxiety” (see, e.g., Wilkinson and Pickett 2010). In line with the literature we build on, our approach is, however, not based on economic or material notions of status: Cultural entitlement does not refer to a sense of being more economically successful. It refers to perceptions of being entitled to participate in legitimate culture and institutions based on one’s standing on the cultural ladder of stratification. However, our findings do suggest that an economic form of “status anxiety” also plays a role in addition to the cultural one formulated and tested in this study, since job security substantially matters even when controlling for the influence of income. Our study thus shows that a lower level of education is related to both the perception of one’s economic position and one’s sense of cultural entitlement, and in turn both matter significantly to the experience of depressive symptoms.

Conclusion

We aimed to develop and test a specific interpretation of the role of cultural capital in the negative association between education and depression, next to well-established economic and social factors. Inspired by recent studies in other fields that address level of education and social status, we proposed and tested a novel mechanism based on the notion of “cultural entitlement.” Our analyses confirm our expectation that the lower level of cultural capital of the less educated underlies the greater prevalence of depressive symptoms among this group, because cultural capital provides a sense of cultural entitlement. This resonates well with studies showing that the less educated experience feelings of futility (Spruyt et al. 2015) and studies tying these feelings to participation in legitimate culture and institutions (see Laurison 2015; Myles 2008).

Because of the widespread prevalence of depressive symptoms, the findings of this study are relevant to a relatively large group. We are, however, cautious with claims regarding other mental health outcomes. Seeing as mental disorders like specific phobia or impulse-control disorders do not necessarily bear much resemblance to depression, and they are not categorized as mood disorders like depression (The WHO World Mental Health Survey Consortium 2004), we do not expect them to be related to cultural capital and cultural entitlement in exactly the same way as depressive symptoms. For more general measures of mental well-being like life satisfaction and happiness, on the other hand, we expect that comparable or even stronger effects of cultural capital could be found, as they are most likely more strongly socially shaped than a “mental illness” like depression (Keyes 2002). Future research that simultaneously takes into account various mental health outcomes could establish whether this is the case indeed.

Addressing an additional issue of generalizability, we would like to point out that this study has been conducted in a specific context, namely, the United States. Socioeconomic inequality, and therefore the salience of economic position, is relatively major in the United States as compared to other Western countries (Thewissen et al., 2015). It is therefore to be expected that economic factors will play a less prominent role in the education gap in the latter. On the other hand, we expect the role of cultural capital and cultural entitlement in the educational gradient in depression to be greater in contexts where the status order is more insistent, such as France. To unravel whether and how different country characteristics play a role in shaping how cultural capital is related to educational differences in depressive symptoms, future research could apply a cross-national approach.

Acknowledgments

This article was presented at the annual meeting of the Dutch and Flemish Sociological Assocations (Dag van de Sociologie) and the PhD Day of Erasmus University’s Department of Public Administration and Sociology in 2015. We thank all participants who provided feedback. In addition, we are especially grateful for very valuable constructive comments made by three anonymous reviewers.

Additional information

Notes on contributors

Josje ten Kate

Much of Josje ten Kate’s research aims to explain how social factors play a role in mental health disparities. In addition, she studies how social inequalities in mental health are shaped by cultural, economic, and institutional country characteristics.

Willem de Koster

Much of Willem de Koster’s research addresses the genesis, manifestations, and consequences of cultural conflict in Western countries. For more information, see www.willemdekoster.nl.

Jeroen van der Waal

Much of Jeroen van der Waal’s research aims to explain why social stratification is linked to value orientations, voting behavior, and health disparities in Western societies. In addition he studies the consequences of globalization and cultural change for inequalities, value orientations, and vote choice. For more information, see www.jeroenvanderwaal.com.

Notes

¹The response rate is not high, but comparable to recent waves of widely used large-scale survey programs such as ANES (2014:31) and ISSP (2014, Appendix A). Detailed information about the panel is available through http://www.knowledgenetworks.com/ganp/docs/KnowledgePanel(R)-Design-Summary.pdf and http://www.knowledgenetworks.com/ganp/docs/KnowledgePanelR-Statistical-Methods-Note.pdf.

²The results presented in Figure 2 are controlled for the influence of age, gender, ethnicity, parenthood, region, living in a metropolitan area, and Internet access. A full model containing paths between the control variables and all other variables was trimmed by stepwise removal of nonsignificant paths, each time checking whether the Akaike information criterion (AIC) and Bayesian information criterion (BIC) decreased. This resulted in a more elegant model (AIC and BIC of full model are 66,258.32 and 66,926.27, respectively, and AIC and BIC of trimmed model are 66,241.42 and 66,720.12). We removed the paths from gender to depression, job insecurity, having a partner, and embeddedness in a cohesive neighborhood; from ethnicity to depression, embeddedness in a cohesive neighborhood, and highbrow orientations; from region to depression, cultural entitlement, and job insecurity; from living in a metropolitan area to depression, cultural entitlement, job insecurity, and having a partner; from having Internet access to depression, job insecurity, and highbrow orientations; and, finally, the paths from parenthood to cultural entitlement, job insecurity, and depression.

³The estimated model also includes correlated error terms. More specifically, covariance between the errors of all mediator variables is included except for covariance between the errors of cultural entitlement and job insecurity, because it was not significant. Removing this covariance improved the model (AIC and BIC for the model including all correlated error terms are 66,241.97 and 66,726.24 respectively; AIC and BIC for the model without the covariance just discussed are 66,241.42 and 66,720.12). For a clear presentation, we do not display the correlated error terms in Figure 2.