Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Aug 10.
Published in final edited form as: J Health Soc Behav. 2014 May 12;55(2):181–195. doi: 10.1177/0022146514531986

Early-Life Socioeconomic Status and Mortality at Three Life-Course Stages: A Growing Within-Cohort Inequality

Tetyana Pudrovska 1,*
PMCID: PMC7416735  NIHMSID: NIHMS625116  PMID: 24818953

Abstract

Using the 1957–2011 data from 10,317 participants in the Wisconsin Longitudinal Study, I examine how socioeconomic status (SES) at age 18 affects all-cause mortality between ages 18 and 72. Integrating the fundamental cause theory, the gender relations theory, and a life-course perspective, I evaluate the cumulative advantage and age-as-leveler processes as well as gender differences in these processes. Findings indicate that higher early-life SES at age 18 is related to lower mortality over the life course, and the effect of early-life SES is not explained by socioeconomic achievement and health behaviors in adulthood. Consistent with the cumulative advantage (CA) model, early-life SES generates growing within-cohort inequality with age, and this CA process is stronger for women than men. Results also show that unequal selection by SES obscures the CA process and creates an illusion of the age-as-leveler process. This study calls for a lifelong gendered approach to socioeconomic health disparities.

Major causes of death in the United States are chronic illnesses that develop gradually over decades and have a long latency period. Biological, social, behavioral, and psychological factors implicated in the development of chronic diseases operate across the life course and across generations (Ben-Shlomo and Kuh 2002; Pearlin, Schieman, Fazio, and Meersman 2005). Therefore, understanding how the processes of social inequality affect mortality requires a life-course perspective, starting with early-life exposures (Ben-Shlomo and Kuh 2002).

Studies exploring early-life influences on mortality disparities reveal the importance of socioeconomic status (SES) of the family of origin (Galobardes, Lynch, and Davey Smith 2004; Hayward and Gorman 2004). Research indicates that all-cause and cause-specific mortality is higher among men and women who experienced poorer socioeconomic conditions during childhood (Galobardes et al. 2004; Hayward and Gorman 2004). Although the importance of early-life socioeconomic resources for health disparities in adulthood is well-documented, there is a paucity of studies that conducted an evaluation of theoretically derived mechanisms linking early-life SES to mortality and embedded their findings in the context of theoretical models of social inequality and life-course development. Moreover, previous studies have not explored how the effect of early-life SES on mortality changes with age within cohorts and whether SES inequality launched in early life decreases or increases as individuals grow older. Finally, relatively little attention has been devoted to a systematic theoretically-informed analysis of gender differences in the focal association and life-course mechanisms.

I use the 1957–2011 data from 10,317 participants in the Wisconsin Longitudinal Study (WLS) to examine how SES at age 18 (in 1957) affects all-cause mortality between 1957 and 2011. I apply the cumulative advantage (CA) and age-as-leveler conceptual models to explore how the effect of early-life SES changes with age. I further evaluate the role of unequal selection by SES in the CA and age-as-leveler processes. Moreover, I analyze gender differences by comparing the effect of early-life SES between men and women. This study is based on a theoretical framework that integrates the fundamental cause theory (Phelan et al. 2010), the gender relations theory (Ferree 2010), and a life-course perspective (Ben-Shlomo and Kuh 2002; Pearlin et al. 2005) to guide an empirical evaluation of life-course processes linking early-life socioeconomic environment to mortality.

Gendered Mechanisms Linking Early-Life SES and Mortality

According to the fundamental cause theory (FCT), SES is a fundamental cause of population health disparities because it affects multiple diseases through a multiplicity of risk factors (Phelan et al. 2010). SES facilitates or blocks access to a wide range of flexible resources that can be used strategically to maximize health under widely divergent conditions (Phelan et al. 2010). Thus, the effect of SES cannot be reduced to proximal individual-level risks because socioeconomic inequalities in health reproduce themselves even when intervening mechanisms change (Phelan et al. 2010). Research within the fundamental cause framework documents pronounced advantages of higher-SES individuals with respect to mortality (Masters, Hummer, and Powers 2012).

Recent studies suggest two important directions for elaborating the FCT. First, because the notion of fundamental causality focuses primarily on between-SES differences and generic SES processes, little attention has been devoted to within-SES heterogeneity (Miech et al. 2011). Yet, SES disparities in mortality vary substantially across gender and race/ethnicity, thus, presenting a provocative challenge to the FCT (Miech et al. 2011). Therefore, it would be beneficial to integrate the FCT with research on intersectionality and multiple social dimensions of health inequality (Miech et al. 2011; Springer and Mouzon 2011). Second, the FCT can be enhanced by integrating a life-course approach, including distinctive historical experiences of birth cohorts (Masters et al. 2012) and a dynamic view of SES as evolving throughout the life course (Willson, Shuey, and Elder 2007). In this study, I elaborate the FCT by considering SES and gender as intersecting contexts and by adopting a life-course view of SES and health.

The Intersection of SES and Gender

I integrate the fundamental cause perspective with the gender relations theory to consider within-SES heterogeneity due to the gendered nature of the relationship between social class and mortality. According to the gender relations theory, gender is not a static role but a relational, ongoing, and negotiated process characterized by micro-macro dynamics (Connell 2005; Ferree 2010; Yancey Martin 2003). In addition to its emergent and shifting nature, gender is institutionalized and widely recognized as a system of practices (Yancey Martin 2003). The system of gender stratification subsumes power asymmetries and contradictions, shapes the interplay of agency and structure within and across social locations, affects access to resources and opportunities, and influences exposure to social risks and constraints (Ferree 2010; Springer, Hankivsky, and Bates 2012).

To explore the complexities of the relationship between SES and health, I adopt an intersectionality perspective that emphasizes how gender is inseparable from other structural inequalities, including social class (Ferree 2010). Gender and SES are intersecting contexts that define each other and synergistically affect individual outcomes. The health implications of socioeconomic advantage should be considered through the prism of gender relations because the cultural meanings and social practices of masculinity and femininity can attenuate or amplify health-promoting resources of higher SES (Pudrovska 2010; Springer and Mouzon 2011). Moreover, research on SES, gender, and health should incorporate biology and take into account an inextricable link between biological processes and gendered social practices and cultural norms (Springer et al. 2012). Thus, because of social and biological factors, gender can be a potentially important source of heterogeneity in the relationship between early-life socioeconomic resources and mortality.

Life Course: Historical Context and Long-Term Influences

Individual lives unfold within historical and sociocultural contexts reflected in the experiences of birth cohorts (Gee, Pavalko, and Long 2007). Cohort effects encompass cultural and social processes through which individuals sharing a birth year move together at a particular life-course stage (Elder 1994). Participants in this study were born in 1939 and entered adulthood in the 1950s and early 1960s – the period of early marriage, high fertility, and the traditional demarcation of public and private spheres (Carr 2004). Men of the WLS cohort were primary providers and focused on the achievement in the public sphere, whereas women were nurturers and caregivers who focused on childrearing and domestic responsibilities and relied on their husbands for economic security (Carr 2004; Pudrovska 2010). Women in this study had limited opportunities for educational attainment and professional advancement compared to recent cohorts of U. S. women (Pudrovska et al. 2013). Thus, the life course of men and women of the WLS cohort has been unfolding in pervasively gendered social contexts.

In addition to historical contexts, a life-course perspective focuses on long-term trajectories of individual development and enduring influences of past experiences. I adopt a dynamic view of SES and emphasize a lifelong approach to the gendered processes underlying socioeconomic disparities in mortality. In this study, SES is considered as a trajectory characterized by long-term patterns of stability and change (Pearlin et al. 2005). Moreover, the life-course approach underscores that health outcomes at older ages reflect life-course processes of accumulation of opportunities, constraints, resources, and adversities launched earlier in life (Pearlin et al. 2005). Therefore, health in later life cannot be explained solely by temporally proximate conditions because earlier conditions and characteristics have long-term implications for later well-being (Pearlin et al. 2005). In life-course sociology, two major perspectives have been applied to explore how within-cohort socioeconomic health disparities change with age: the cumulative advantage model and the age-as-leveler model (Ben-Shlomo and Kuh 2002; Dupre 2008; Lynch 2003).

The cumulative advantage (CA) model posits that the advantage of some individuals or groups grows over time, which leads to increasing within-cohort inequality with advancing age (DiPrete and Eirich 2006). Among several types of CA described by DiPrete and Eirich (2006), the status-dependent CA is the most relevant to this study. In a status-dependent CA process, the growth rate in outcome Y is affected by a social status, and the effect of this status on Y becomes stronger over time (DiPrete and Eirich 2006). The status-dependent CA model suggests that socioeconomic health disparities initiated early in life are magnified as cohorts grow older, with the health advantage of higher-SES individuals steadily increasing compared to their lower-SES peers (Willson et al. 2007). Research provides empirical support for the CA process by revealing that the effect of SES on health and mortality strengthens with age, generating the divergent health trajectories of low- and high-SES cohort members (Lynch 2003; Willson et al. 2007). Based on the CA model, I hypothesize that the effect of early-life SES becomes stronger with age, with the growth rate of mortality increasing faster for persons from low-SES family background.

The age-as-leveler model.

Contrary to CA predictions, some studies show that the effect of SES on health decreases with age, and the SES difference in mortality may disappear or even reverse itself at older ages (House et al. 1994; Sautter et al. 2012). This pattern is consistent with the age-as-leveler mechanism. One explanation for narrowing health disparities in old age reflects the forces of senescence and finitude of human life (Willson et al. 2007). Another explanation is selection bias due to higher mortality of socioeconomically disadvantaged individuals at younger ages, which leaves a non-representative group of robust low-SES survivors among high-SES individuals (Dupre 2008; Lynch 2003; Willson et al. 2007). Unequal selection by SES can attenuate the CA process and create an illusion of the age-as-leveler process; therefore, it is important to incorporate attrition into the relationship between early-life SES and mortality for a more accurate understanding of the CA and age-as-leveler processes. The longitudinal nature of the WLS allows to model explicitly selective survival and attrition. Specifically, I estimate the effect of early-life SES on mortality after age 18 and then explore how this effect changes conditional on survival into adulthood and midlife.

Gender Differences

This study’s theoretical framework emphasizes the importance of within-SES gendered processes. Yet, because most previous studies have not explicitly focused on gender differences (Dupre 2008; Lynch 2003; Willson et al. 2007), little is known about the ways that the CA process can operate differently for men and women with respect to SES and mortality. Given the paucity of existing evidence, the direction of the gender difference is not clear because some mechanisms suggest that the effect of early-life SES may be stronger among women whereas other mechanisms suggest that this effect may be stronger among men. First, it is plausible to speculate that growing inequality based on early-life SES is stronger among women than men because childhood SES may be overall more important for women than men in this traditional cohort. Men and women in our study came of age in 1950s and 1960s, experiencing a traditional gender-typed division of labor. Most women were primary caretakers of their husbands and children, whereas men assumed the breadwinning responsibility in the public sphere. For women these socioeconomic achievement avenues were limited. Even women from socioeconomically advantaged families had fewer educational and professional opportunities than men (Carr 2004), whereas for men childhood SES was a starting point for launching trajectories of own status attainment (Hamil-Luker and O’Rand 2007). Another reason why early-life SES can be more important for women’s than men’s adult health is that health behaviors are partly learned in the family of origin (Giskes et al. 2008), and higher-SES parents are more likely to instill healthy habits in their children than low-SES parents (Saint Onge and Krueger 2011). Health of family members and competence in health-related matters are commonly viewed as women’s domain (Reczek and Umberson 2012). Thus, higher-SES parents may have emphasized health management more in the upbringing of daughters than sons because girls as future gatekeepers of their families’ health are socialized early in life to be attentive to health matters (Reczek and Umberson 2012).

Conversely, it is also plausible to speculate that inequality in mortality launched by early-life SES is stronger among men than women. Some studies suggest that low SES is more detrimental to men’s than women’s health (Catalano et al. 2005; Montez et al. 2009). Males are more susceptible than females to deleterious exposures, including economic stress, during gestation, infancy, and childhood (Catalano et al. 2005; Catalano and Bruckner 2006). Some researchers even advance the “fragile male” thesis and argue that at every life-course stage men are more vulnerable than women to environmental stressors, including low SES, due to genetic, physiological, and socio-cultural influences (Catalano and Bruckner 2006; Kraemer 2000). For example, children growing up in low-SES families have a greater infection burden than socioeconomically advantaged children (Dowd, Zajacova, and Aiello 2009), and male babies are more vulnerable to infections than female babies (Drevenstedt et al. 2008). Because of males’ less vigorous immune system compared to females, early-life infections are more likely to inflict long-lasting damage among men than women (Catalano and Bruckner 2006; Read, Troendle, and Klebanoff 1997).

Therefore, my purpose is to examine how the CA process is modified by gender. I will explore whether the effect of early-life SES on mortality becomes stronger with age and whether this growing inequality launched by early-life SES is more pronounced among women or men.

Adult SES and Health Behaviors as Potential Links

I analyze to which extent the effect of early-life SES and the CA process can be explained by socioeconomic attainment in adulthood and health behaviors over the life course. Research indicates that the association between parents’ SES and offspring’s mortality is mediated by SES in adulthood (Hayward and Gorman 2004). Parents’ SES is positively associated with individuals’ own SES (Sewell and Hauser 1975). In turn, socioeconomic conditions in adulthood are strongly related to mortality, with low-SES individuals having a greater risk than their high-SES counterparts (Zajacova 2006). The gendered process of status attainment can lead to differences between men and women. In the traditional WLS cohort, women had limited opportunities for establishing their occupation and income (Carr 2004). Therefore, husbands’ SES, especially during women’s childrearing years, may be a more adequate predictor of women’s mortality than women’s own SES. I compare women’s own SES and husbands’ SES in adulthood and midlife as mechanisms linking early-life SES and women’s mortality. In addition to adult SES, an important pathway linking socioeconomic family background and mortality is health behaviors (Hayward and Gorman 2004; Power and Matthews 1997). Higher childhood SES is related to lower risk of smoking, obesity, and physical inactivity (Kuh and Cooper 1992; Power and Matthews 1997). In turn, low-risk lifestyle (not smoking, regular exercise, healthy weight, and moderate alcohol consumption) reduces mortality (Ford et al. 2011). Thus, I explore whether the effect of early-life SES persists after adjustment for adult SES and health behaviors.

In sum, using longitudinal data spanning over 50 years, this study integrates the fundamental cause theory, the gender relations theory, and a life-course perspective to explore the effect of early-life SES on mortality and how this effect changes with age as the cohort grows older. I further explore whether these age-related patterns of mortality reflect unequal selection by SES and analyze how the effect of early-life SES differs by gender.

DATA AND METHODS

The Wisconsin Longitudinal Study (WLS) is a long-term cohort study of 10,317 men and women who graduated from Wisconsin high schools in 1957. Participants were interviewed at ages 18 (in 1957), 36 (in 1975), 54 (in 1993), and 65 (in 2004). In 2000, 2006, and 2011 deceased participants were matched to the National Death Index (NDI) to ascertain the cause of death and age at death. This study is based on three analytic subsamples. The first subsample comprises 5,326 women (875 deaths) and 4,991 men (1,194 deaths) who participated in the original 1957 sample. The second subsample contains 4,808 women (739 deaths) and 4,330 men (910 deaths) who participated in 1957 and 1975 including persons who died after 1993 and persons who survived to participate in the 1993 wave. The third subsample includes 4,513 women (553 deaths) and 3,980 men (655 deaths) who participated in 1957, 1975, and 1993.

Measures

A binary indicator of mortality reflects being deceased (coded 1) or alive (coded 0) at each age between 18 and 72 years. Age at death is coded in years.

SES in 1957 includes father’s and mother’s education measured in years, family income measured in $100’s, father’s occupation (unskilled worker, farmer, skilled worker, white-collar worker, and professional/executive), father’s occupational education reflecting the percentage of persons in the 1970 Census in a given occupation who completed one or more years of college, and father’s occupational income representing the percentage of persons in the 1970 Census in a given occupation who earned ≥ $10,000. Family background characteristics were reported on several occasions: in 1957 and 1975 by participants and in 1964 by parents. Parental income data were collected from Wisconsin state income tax records for 1957–1960. In most cases, there are at least two measurements of each characteristic, which allowed to document their reliability (Hauser, Sheridan, and Warren 1999).

Health behaviors in 1957.

Body mass in adolescence is measured via the relative body mass (RBM) scale, coded from facial adiposity in high school yearbook pictures in 1957 (Reither, Hauser, and Swallen 2009). Six coders assigned RBM scores separately for boys and girls on scale ranging from 1 (the lowest) to 11 (the highest). Previous research has shown that the RBM measure is a valid and reliable indicator of body mass (Reither et al. 2009). Physical activity in adolescence was obtained from participants’ high school yearbooks. The measure of sports participation in high school reflects the sum of three types of activities: the number of varsity sports, the number of club sports, and the number of intramural sports.

SES in 1975 and 1993.

Education was assessed as the total completed years of schooling. Occupation of the current or last job is represented with the following categories: professional/managerial, clerical/sales/service, and crafts/operatives/laborers. I include a natural log of household income (1975) and a natural log of household wealth (1993). Occupational education reflects the percentage of persons in the 1970 Census (1975) and in the 1990 Census (1993) in a given occupation who completed one year of college or more. Occupational income in 1975 represents the percentage of persons in the 1970 Census in a given occupations who earned ≥ $10,000 in 1969. Occupational income in 1993 represents the percentage of persons in the 1990 Census who earned ≥ $14.30 per hour in 1989.

Health behaviors in 1993.

Physical activity was assessed with two measures: the frequency of light exercise and vigorous exercise. Response categories range from 1 = less than once per month to 4 = three or more times per week. Not smoking is coded 1 for persons who did not smoke in 1993 and 0 for those who were smokers. Moderate drinking is defined based on the Federal Dietary Guidelines for Americans (2010) as one daily drink for women and 1–2 daily drinks for men. Body mass index (BMI) was measured as weight in kilograms divided by the square of height in meters. A binary indicator of healthy weight was coded 1 for BMI < 25.

Control variables.

I include early-life variables that were shown to be related both to parents’ SES and health in adulthood (Hayward and Gorman 2004; Riva, Curtis, and Norman 2011): rural residence in childhood and intact family structure while growing up. I control for marital status in 1975 and 1993 coded 1 for the married and 0 for the unmarried and parental status assessed as the number of children in 1975 and 1993. Although being married and having children is associated with better health (Umberson, Crosnoe, and Reczek 2010), in the WLS individuals from higher-SES background were less likely to be married and had fewer children than their lower-SES peers. Thus, models that do not include these variables can underestimate the protective effect of advantaged family background. Finally, as a proxy for overall robustness or frailty, I adjust for parental death (coded 1 if both parents were deceased as of 1993).

Statistical Analysis

Selection.

All original WLS participants whose vital status is known as of 2011 are included in this study. The problem of attrition bias arises from individuals who dropped out of the survey and were not matched to the NDI. Among participants whose vital status is unknown, 368 people did not participate in 1975 and 963 people did not participate in 1993. An analysis of sample attrition reveals that those who were lost to follow-up had lower SES in 1957, were less likely to live in rural areas and less likely to grow up with both parents. Following the approach of Willson and colleagues (2007), I apply propensity score matching to correct for this selection bias. Using logistic regression, each person’s observed early-life characteristics (parents’ SES, family structure, and rural residence) are summarized into a gender-specific composite propensity score reflecting a predicted probability to be lost to follow-up.

Measurement models.

SES in 1957, 1975, and 1993 and health behaviors in midlife are measured as latent factors. The rationale for this approach and detailed methodology are described in Appendix A. Measurement models and summary statistics for all indicators are shown in Appendix B.

Survival analysis.

Using a continuous-time parametric survival model, I estimate the effect of early-life SES on mortality. The hazard function for person i at time j is modeled as:

h(tij)=h0(tj)exp(Xij'β+Zi'ω), (1)

where h(tij) is the hazard of mortality evaluated at exact age t, h0 is a baseline hazard of mortality, β and ω are vectors of parameters, Xij is a vector of predicted factor scores, and Zi is a vector of control variables. The baseline hazard h0 is represented with the Gompertz distribution chosen over other functional forms based on the model fit indices:

h0(tj)=exp(γt)exp(β0), (2)

where γ is a shape parameter and β0 is a level parameter. I also calculate predicted cumulative hazard to plot the effect of early-life SES:

H(tij)=γ−1exp(β0+Xij'β+Zi'ω)[exp(γt)−1] (3)

The Gompertz model is a proportional hazard model, which means that the difference in log hazard corresponding to the unit difference in the value of the predictor should be identical at every point in time (Singer and Willett 2003). Yet, a violation of the proportionality assumption is important substantively if the predictor’s effect varies over time, as in the CA and age-as-leveler models. Therefore, I use what Singer and Willett (2003) call a “nonproportional proportional hazard model,” which is a regular survival model that includes an interaction term between SES in 1957 and time.

All models were estimated using Stata 12.1. To evaluate the fit of hazard models, I use Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC).

RESULTS

The Cumulative Advantage and Age-As-Leveler Mechanisms

Model 1 of Tables 1 and 2 shows the effect of early-life SES in 1957 on mortality between ages 18 and 72.

Table 1.

Gompertz Survival Models Estimating the Effect of Socioeconomic Status (SES) in 1957 on Mortality: Women

Variables Mortality [18, 72] Mortality [36, 72] Mortality [54, 72]
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9
SES in 1957 −.198*** (.039) −.195*** (.037) −.139*** (.042) −.131** (.042) −.118** .042 −.116** (.039) −.074* (.037) −.079* (.039) −.099* (.038)
SES in 1957 × Time a .096*** (.009) .104*** (.003) .101*** (.005) .101*** (.005) .101*** .005 .127*** (.016) . 127*** (.015) . 127*** (.015) . 127*** (.016)
SES in 1957 × Time2 −.001 (.002)
Mediators:
RBM in 1957 .199*** (.044)
High school sports −.073* (.043)
SES 1975 (Own) −.093* ( .041)
SES 1975 (Spouse) −.113*** (.035)
SES 1993 (Own) −.256*** (.049)
SES 1993 (Spouse) −.198*** (.044)
Healthy lifestyle 1993 −.227*** (.046)
Attrition propensity .509* (.261) .414* (.264) .653* (.358) .627* (.359) .629* (.358) .829* (.409) .818* (.407) .852* (.409) .853* (.411)
Level (β0) −11.515 −11.520 −12.567 −12.357 −12.432 −14.797 −14.624 −14.823 −14.683
Shape (γ) 1.107 1.107 1.108 1.106 1.107 1.163 1.162 1.162 1.163
LL (df) b −1655 (7) −1644 (8) −1608 (8) −1607 (9) −1603 (8) −1086 (9) −1074 (8) −1078 (7) −1076 (8)
AIC c 3326 3307 3237 3236 3229 2193 2168 2176 2172
BIC d 3379 3366 3303 3308 3301 2259 2233 2241 2237
Open in a new tab

Note: Predictors’ effects are given as regression coefficients (logged hazard ratios) with standard errors in parentheses.

*

p < .05.

**

p < .01.

***

p < .001. All models control for the area of residence and family structure in adolescence. Models 3–10 control for marital and parental statuses. Models 7–10 also control for parental vital status.

a

Centered at the median age at death 62 years.

b

Log likelihood (degrees of freedom).

c

Akaike information criterion.

d

Bayesian information criterion.

Table 2.

Gompertz Survival Models Estimating the Effect of Socioeconomic Status (SES) in 1957 on Mortality: Men

Variables Mortality [18, 72] Mortality [36, 72] Mortality [54, 72]
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
SES in 1957 −.182*** (.032) −.175*** (.032) −.113*** (.037) −.106** (.037) −.112** (.036) −.075* (.038) −.103** (.037)
SES in 1957 × Time a .084*** (.002) .085*** (.002) .084*** (.004) .084*** (.004) .119*** (.014) .119*** (.014) .118*** (.014)
SES in 1957 × Time2 −.001 (.002)
Mediators:
RBM in 1957 .144*** (.037)
High school sports −.058** (.022)
SES in 1975 −.075* (.035)
SES in 1993 −.141*** (.039)
Healthy lifestyle in 1993 −.178*** (.041)
Attrition propensity .606* (.261) .611* (.264) .717* (.358) .724* (.359) .853* (.409) .852* (.410) .844* (.423)
Level (β0) −10.544 −10.495 −11.862 −11.185 −13.824 −13.759 −13.738
Shape (γ) 1.103 1.103 1.104 1.105 1.158 1.158 1.158
LL (df) b −1856 (6) −1846 (8) −1765 (8) −1762 (9) −1167 (8) −1161 (8) −1158 (8)
AIC c 3727 3711 3549 3547 2354 2342 2337
BIC d 3773 3770 3614 3618 2419 2407 2402
Open in a new tab

Note: Predictors’ effects are given as regression coefficients (logged hazard ratios) with standard errors in parentheses.

*

p < .05.

**

p < .01.

***

p < .001. All models control for the area of residence and family structure in adolescence. Models 3–7 control for marital and parental statuses. Models 5–7 also control for parental vital status.

a

Centered at the median age at death 62 years.

b

Log likelihood (degrees of freedom).

c

Akaike information criterion.

d

Bayesian information criterion.

A standard-deviation increase in early-life SES is associated with an 18% (e−.198 = .820, p < .001) lower risk of mortality among women and a 16.6% (e −.182 = .834, p < .001) lower risk among men. This effect of early-life SES does not differ significantly by gender (bSES in 1957 × female = .016, SE = .044, p > .05), as shown in Table 3.

Table 3.

Gompertz Survival Models Estimating Interactive Effects of Gender and Socioeconomic Status (SES) in 1957 on Mortality

Variables Model 1 Model 2 Model 3
[18, 72] [36, 72] [54, 72]
SES in 1957 −.201*** (.031) −.119*** (.033) −.194*** (.045)
SES in 1957 × Time .086*** (.002) .086*** (.004) .116*** (.014)
Female = 1 −.410*** (.046) −.039 (.074) −.036 (.085)
SES in 1957 × Female .016 (.044) −.005 (.045) .019 (.086)
SES in 1957 × Time a × Female .016*** (.003) .013* (.005) .079* (.038)
Level (β0) −10.759 −12.238 −15.845
Shape (γ) 1.104 1.106 1.159
LL (df) b −3515 (9) −3377 (11) −2245 (12)
AIC c 7051 6778 4517
BIC d 7124 6865 4611
Open in a new tab

Note: Each cell contains regression coefficients (logged hazard ratios) with standard errors in parentheses. All models control for area of residence and family structure in adolescence, marital and parental status in 1993, and the attrition propensity score. Bolded cells denote significant gender differences.

*

p < .05.

**

p < .01.

***

p < .001.

a

Centered at the median age at death 62 years.

b

Log likelihood (degrees of freedom).

c

Akaike information criterion.

d

Bayesian information criterion.

Model 1 estimates the effect of SES in 1957 taking into account selection bias. Sample attrition is disproportionately concentrated at the lowest levels of early-life SES, as shown by a plot of the attrition propensity score against SES in 1957 (Figure 1). When the propensity score is not included in the model, the effect of parents’ SES on offspring’s mortality becomes even stronger: b = −.216, SE = .034, p < .001 among women and b = −.221, SE = .030, p < .001 among men (not shown in the table).

Figure 1.

Figure 1.

Open in a new tab

Sample Attrition between 1957 and 1975

Further, a significant and positive interaction term between SES in 1957 and time suggests that the effect of early-life SES becomes stronger with age. Specifically, each year the SES gap in mortality increases by 10.1% (e.096 = 1.101, p < .001) among women and by 8.8% (e.084 = 1.088, p < .001) among men. For example, the difference in the cumulative hazard of mortality between men with high (+1 SD) and low (−1 SD) early-life SES is .0012 at age 20 and .0566 at age 70. Among women, this difference equals .0004 at age 20 and .0232 at age 70. A test of interactive effects (Table 3) indicates that this mechanism is significantly stronger among women than men (bSES in 1957 × time × female = .016, SE = .003, p < .001). I also tested an interaction between SES in 1957 and a quadratic transformation of time. Because this interaction term was not significant among men and women as shown in Model 1 of Tables 1 and 2 (b = −.001, SE = .002), there is no evidence of convergence of mortality trajectories of high- and low-SES individuals, at least up to age 72.

Model 2 of Tables 1 and 2 includes relative body mass in adolescence and high school sports participation as potential mechanisms that could explain the effect of early-life SES and the growing inequality with age. Although body mass and physical activity are related to mortality in the expected direction, the inclusion of these variables does not significantly attenuate the effect of early-life SES and the CA process.

Model 3 in Tables 1 and 2 shows the effect of SES in 1957 on mortality between 1975 and 2011 (ages 36 and 72). Conditional on surviving to age 36, a standard-deviation increase in early-life SES is associated with .870 (e−.139, p < .001) lower hazard of mortality among women and .893 (e−.113, p < .001) among men. The effect of early-life SES does not differ significantly by gender as shown in Table 3 (bSES in 1957 × female = −.005, SE =.045, p = .865). Further, the association between early-life SES and mortality is attenuated among people who survive until age 36, which reflects a disproportionate selection of low-SES individuals out of the sample. Yet, the interaction between early-life SES and time remains large in magnitude and statistically significant, pointing to the increasing SES gap in mortality between 36 and 72 years. Table 3 indicates that this increase is significantly greater among women than men (bSES in 1957 × time × gender = .013, SE = .005, p < .05).

Models 4 and 5 of Table 1 explore whether the effect of early-life SES changes with the inclusion of SES in adulthood. Both women’s own and husband’s SES at age 36 are associated negatively with mortality between the ages of 36 and 72. Women’s own SES explains about 6% of the effect of early-life SES on mortality ([.139 – .131]/.139), and husband’s SES explains about 15% of this effect ([.139 – .118]/.139). Yet the early-life SES coefficient remains large in magnitude and statistically significant. Similarly, Model 4 in Table 2 shows that after adjustment for men’s SES at age 36 the effect of parents’ SES diminishes only slightly and remains large in magnitude and statistically significant. Importantly, the CA process is unaffected by adult SES as indicated by the unchanging interaction term between SES and time.

Model 6 in Table 1 and Model 5 in Table 2 present the effect of early-life SES on mortality between 1993 and 2011 conditional on surviving to age 54. A standard-deviation increase in parents’ SES is associated with a 10.9% reduction in the hazard of mortality (e−.116 = .891, p < .01) among women and a 10.6% reduction (e−.112 = .894, p < .01) among men. With respect to the CA process, the gap between individuals from low- and high-SES families of origin increases by 12.6% with each year between ages 54 and 72 among men, as indicated by the significant interaction between early-life SES and time in Model 5 of Table 2 (bSES in 1957 × Time = .119, e.119 = 1.126, p < .001). Among women this increase is even steeper and equals 13.5% per year, as shown in Model 6 of Table 1 (bSES in 1957 × Time = .127, e.119 = 1.135, p < .001). The gender difference in the CA process is statistically significant as shown by the three-way-interaction term in Model 3 of Table 3 (bSES in 1957 × Time × Female = .079, SE = .038, p < .05).

Finally, Models 7–9 in Table 1 and Models 6–7 in Table 2 adjust for SES and health behaviors in 1993. Women’s own SES explains 36% ([.116-.074]/.116) and her husband’s SES explains 32% ([.116-.079]/.116) of the effect of early-life SES on post-1993 mortality. Health behaviors explain about 15% ([.116-.099]/.116). Yet the coefficient for early-life SES remains significant and large in magnitude. Among men the change in the effect of early-life SES after adjustment for midlife SES and health behaviors is very similar to patterns among women.

DISCUSSION

Using the 1957–2011 data from the Wisconsin Longitudinal Study, I examine the link between parents’ SES and offspring’s life-course mortality as well as the extent to which this link reflects the cumulative advantage and the age-as-leveler processes. This study highlights within-SES heterogeneity and the complexity of gendered processes underlying the enduring effect of early-life SES on mortality over the life course.

Early-life SES at age 18 is significantly related to all-cause mortality up to age 72. Specifically, a standard-deviation increase in parents’ SES reduces the risk of mortality over the life course by 18% among women and 16.6% among men. Although the fundamental cause theory typically focuses on social health disparities in adulthood, the present study suggests that, in addition to own socioeconomic resources, parental SES is another fundamental cause of health disparities operating through multiple mechanisms over considerable periods of time. Unequal trajectories of high- and low-SES individuals originate in early life and are “anchored in conditions that long antecede the decades in which these trajectories take shape” (Pearlin et al. 2005:207). This finding underscores the importance of a dynamic life-course approach to SES.

My findings also emphasize the importance of socioeconomic differentials in sample attrition for the substantive conclusions about the effect of early-life SES. The effect of parents’ SES on mortality becomes weaker if offspring survive until mid-30s and, especially, mid-50s. This attenuation of the effect of early-life socioeconomic resources is consistent with the age-as-leveler model and reflects the disproportionate selection of low-SES individuals out of the sample, resulting in a non-representative group of hardy survivors from disadvantaged family background (Crimmins, Kim, and Seeman 2009; Dupre 2008; Willson et al. 2007). Thus, selection bias can obscure the CA process and create the age-as-leveler process, particularly, in studies that do not follow individuals since childhood but begin with samples of robust survivors in midlife and old age (Crimmins et al. 2009). Moreover, I find that the effect of early-life SES on mortality is diminished with the inclusion of attrition propensity score that captures selection bias. Taken together, these findings suggest that simply “controlling” for selection can be misleading because of the underestimation of the true effect of family background. It is important to understand substantive implications of the selection mechanism and complex ways in which selection bias affects conclusions about socioeconomic disparities in health and mortality (Willson et al. 2007).

Further, this study indicates that the effect of early-life SES on mortality becomes stronger with age. Each year the gap in mortality reflecting unequal socioeconomic resources of the family of origin increases linearly by about 10% among women and 8.8% among men. This pattern is consistent with the status-dependent CA model (DiPrete and Eirich 2006) and not with the age-as-leveler model. A possible reason is that the WLS participants were 72 years old in 2011 and have not yet reached the age at which convergence typically occurs. Because of the rectangularization of mortality, the point at which superior resources of the socially advantaged are superseded by biological senescence is postponed until very old age, with the survival advantage of higher-SES persons disappearing around 85 years old (Phelan et al. 2004).

Gender Differences

With respect to within-SES heterogeneity by gender, the CA process reflecting growing within-cohort inequality based on early-life SES is significantly stronger for women than men. In other words, early-life socioeconomic disadvantage generates greater inequality in mortality decades later among women than men. This gender difference may reflect the fact that in this traditional cohort women were expected to focus on familial obligations; thus, low-SES women had fewer opportunities for their own status attainment than low-SES men (Carr 2004; Hamil-Luker and O’Rand 2007). It was difficult for women to overcome initial socioeconomic disadvantage and narrow the gap with their higher-SES counterparts because of their limited mobility through educational and professional achievement. Moreover, low-SES women had limited opportunities to improve their socioeconomic standing by marrying higher-SES men. Educational homogamy increased in the 1960s and 1970s, sharply reducing the odds of women with low levels of education to marry up (Schwartz and Mare 2005). Thus, the CA effect is more pronounced among women because low-SES women of this cohort had more social constraints than low-SES men that hindered their ability to break the chain of accumulation of disadvantages.

Another explanation for a stronger CA process among women than men can reflect the fact that high- and low-SES men are more similar to each other in terms of heath behaviors and outcomes than high- and low-SES women. This gender difference suggests that higher-SES men do not benefit as much as women from the resources of socioeconomic advantage. From a sociological perspective, these gender differences may reflect the intersection of gender and SES as fundamental causes. It is possible that women are more efficient that men in deploying the health-enhancing resources of higher SES because the cultural ideals of femininity embrace competence in health-related matters whereas hegemonic masculinity can attenuate health benefits of socioeconomic advantage (Courtenay 2000; Reczek and Umberson 2012). Within the gender relations theory, researchers have consistently drawn a link between men’s poor health behaviors and hegemonic masculinity ideals that emphasize risk-taking and the denial of weakness and vulnerability (Mahalik, Burns, and Syzdek 2007). From a social constructionist perspective, health behaviors serve as a means of constructing femininity and masculinity (Reczek and Umberson 2012). The enactment of masculinity is influenced by cultural beliefs that men are more powerful and resilient than women and that men’s bodies are structurally superior to women’s bodies (Courtenay 2000). For example, among men who strongly adhere to hegemonic masculinity ideals, higher SES does not increase and even undermines compliance with preventive health care guidelines (Springer and Mouzon 2011). Thus, it is important to consider within-SES heterogeneity between men and women because gender relations and practices as well as cultural scripts of masculinity and femininity can modify the ways in which individuals employ flexible resources of higher SES (Reczek and Umberson 2012; Springer and Mouzon 2011).

The Enduring Effect of Early-Life SES

To explore how the effect of early-life SES is affected by adult characteristics, I adjusted for adult SES and health behaviors over the life course – mechanisms that were shown particularly important in existing research (Hayward and Gorman 2004; Miller et al. 2011; Power and Matthews 1997; Umberson et al. 2010). For women both own and husband’s SES at two life stages were included. As hypothesized, adult socioeconomic resources and health behaviors partly attenuate the effect of early-life SES on mortality. Yet the strong association between socioeconomic family background and mortality persists and remains significant and large in magnitude net of the most plausible adult mechanisms.

What can explain the persistence and long-term consequences of parents’ SES for offspring’s mortality? According to the fundamental cause theory, SES affects health through a multiplicity of mechanisms. Although this study includes the most widely documented pathways, there are a myriad of potential mechanisms conveying the effect of early-life SES on mortality that no single study can incorporate. It is possible that, in addition to SES and healthy lifestyle, the effect of socioeconomic family background is conveyed through life-course psychosocial variables not included in this study. Constrained opportunities and socioeconomic disadvantage in early life and adulthood undermine personal mastery in low-SES women and men (Rosenfield 1999; Ross and Mirowsky 2002). Because personal mastery is an important link connecting SES and health over the life course (Pudrovska et al. 2005), individuals’ sense of powerlessness and a perceived lack of control over the outcomes of their lives can have snowballing effects on health and gradually lead to the divergence in physical health between low- and high-SES cohort members. Similarly, low childhood SES is related to a greater risk of depression in adulthood (Goosby 2013). In turn, depressive symptoms are associated with increased mortality (Houle 2013). In addition, life-course stressors can be an important pathway. Early-life poverty increases exposure to chronic strains and stressful life events (Pearlin et al. 2005), and social stressors lead to physiological dysregulation that increases mortality (Aldwin et al. 2011).

Yet, it is also possible that the effect of early-life SES is not fully conveyed by adult characteristics but reflects a direct enduring physiological imprint of early-life circumstances. The critical or sensitive period model (Ben-Shlomo and Kuh, 2002; Miller, Chen, and Parker 2011) reflects a biological imprinting mechanism and posits that early-life SES has long-lasting effects on biological and behavioral systems, and these effects can be irreversible and permanent (critical period) or potentially modifiable (sensitive period). The critical period models suggests that health implications of early-life SES are not fully absorbed by SES and health behaviors over the life course; rather, the lingering effects of earlier socioeconomic disadvantage operate in concert with later characteristics. Changing socioeconomic circumstances can amplify or attenuate past exposures but do not wipe them out completely (Hamil-Luker and O’Rand 2007; Pearlin et al. 2005).

The enduring effect of early-life socioeconomic environment can reflect unfavorable exposures among children from disadvantaged social background. Latent infections are often acquired early in life and are more prevalent among low-SES children (Dowd et al. 2009). Children born to parents with low levels of income and education have a greater risk of individual infections, such as Helicobacter pylori, cytomegalovirus, and herpes simplex virus-1, as well as the overall infection burden (Dowd et al. 2009). An early exposure to infections during critical periods can increase vulnerability to chronic diseases of aging partly through diverging physiological resources from physiological maturation to inflammatory responses (Dowd et al. 2009). Infection is a plausible mechanism of how early-life sociological disadvantage becomes embodied and drives forward chronic physiological repercussions (Miller et al. 2011). Research shows that an early exposure to infections is a central mechanism underlying the “cohort morbidity phenotype” defined as long-reaching consequences of early-life environment for life-course morbidity and mortality of a cohort (Finch and Crimmins 2004). Another mechanism of long-term biological programming can be inflammation. Childhood adversity, including low SES, promotes chronic proinflammatory tendencies that foster the development of chronic diseases in the long term (Miller et al. 2011). Life-course low-grade chronic inflammation is more common among individuals who experienced early-life adversity (Miller et al. 2011). In turn, chronic inflammatory responses are related to major diseases of aging, including heart disease, cancer, and Alzheimer’s disease (Miller et al. 2011). These mechanisms are plausible explanations for the enduring effect of early-life SES. Because the WLS does not have measures of inflammation and latent infections over the life course, I cannot test these mechanisms directly. Yet the strength of the effect of early-life socioeconomic resources points to the importance of exploring the critical period model empirically in future research, especially given that more and more social surveys are now collecting biomarkers.

Limitations and Future Research

The WLS sample comprises only White participants who completed at least a high school education, which makes it representative of two-thirds of this generation of Americans. Although there is sufficient variability in early-life SES because participants come from widely diverse socioeconomic family backgrounds (Appendix B), a potential limitation of this study is the absence of individuals who did not graduate from high school. Given that persons who are the most disadvantaged with respect to their own SES are not included, my findings are likely to be conservative and underestimate socioeconomic disparities in mortality. Further, I could not explore race and ethnic differences in the effect of early-life SES on mortality. This direction is important for future research because minority groups derive fewer health benefits from higher SES compared to white adults (Cummings and Jackson 2008). In addition, this study is based on one cohort born in 1939. Socioeconomic inequality in health is more pronounced in recent than older cohorts (Lynch 2003). Therefore, the patterns observed in the WLS may be even stronger for younger cohorts. An important avenue for future studies is to compare effects of SES on mortality in older and younger generations.

Moreover, whereas all-cause mortality used as an outcome in this study is a comprehensive and reliable indicator of health disparities (Galobardes et al. 2004), attention to specific causes of death in future research may help further elucidate disease-specific mechanisms through which SES affects health. In addition, the WLS has only limited information about health behaviors in adolescence and no measures of healthy lifestyle in childhood and young adulthood. Ideally, future studies of mortality should include information about health behaviors collected prospectively over the life course.

Despite these limitations, my findings expand current knowledge of long-term consequences of early-life SES as a fundamental cause operating across the life course. This study is based on a theoretical framework that facilitates comprehensive evaluation of life-course mechanisms linking early-life socioeconomic environment to mortality as well as gender differences in these processes. The results emphasize the growing within-cohort inequality with advancing age and document the complexity of gendered processes underlying the enduring effect of early-life SES on mortality over the life course.

Supplementary Material

Appendix A & B

REFERENCES

  1. Aldwin Carolyn M., Molitor Nuoo-Ting, Spiro Avron, Levenson Michael R., Molitor John, and Igarashi Heidi. 2011. “Do Stress Trajectories Predict Mortality in Older Men? Longitudinal Findings from the VA Normative Aging Study.” Journal of Aging Research. Volume 2011, Article ID 896109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ben-Shlomo Yoav and Kuh Diana. 2002. “A Life Course Approach to Chronic Disease Epidemiology.” International Journal of Epidemiology 31(2):285–293. [PubMed] [Google Scholar]
  3. Carr Deborah. 2004. “‘My Daughter Has a Career; I Just Raised Babies’: The Psychological Consequences of Women’s Intergenerational Social Comparisons.” Social Psychology Quarterly 67(2):132–154. [Google Scholar]
  4. Catalano Ralph, Bruckner Tim, Anderson Elizabeth, and Gould Jeffrey. 2005. “Fetal Death Sex Ratios: A Test of the Economic Stress Hypothesis.” International Journal of Epidemiology 34(4):944–948. [DOI] [PubMed] [Google Scholar]
  5. Catalano Ralph and Bruckner Tim. 2006. “Child Mortality and Cohort Lifespan: A Test of Diminished Entelechy.” International Journal of Epidemiology 35(5):1264–1269. [DOI] [PubMed] [Google Scholar]
  6. Connell Robert W. 1995. Masculinities. Cambridge: Polity Press. [Google Scholar]
  7. Courtenay Will H. 2000. “Constructions of Masculinity and their Influence on Men’s Well-Being: A Theory of Gender and Health.” Social Science & Medicine 50(10):1385–1401. [DOI] [PubMed] [Google Scholar]
  8. Crimmins Eileen M., Kim Jung, and Seeman Teresa. 2009. “Poverty and Biological Risk: The Earlier ‘Aging’ of the Poor.” Journal of Gerontology: Medical Sciences 64A(2):286–292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cummings Jason L. and Jackson Pamela B. 2008. “Race, Gender, and SES Disparities in Self-Assessed Health, 1974–2004.” Research on Aging 30(2):137–167. [Google Scholar]
  10. DiPrete Thomas and Eirich Greg. 2006. “Cumulative Advantage as a Mechanism for Inequality: A Review of Theory and Evidence.” Annual Review of Sociology 32:271–297. [Google Scholar]
  11. Dowd Jennifer B., Zajacova Anna, and Aiello Allison. 2009. “Predictors of Inflammation in U.S. Children Aged 3–16 Years.” American Journal of Preventive Medicine 39(4):314–320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dupre Matthew E. 2008. “Educational Differences in Health Risks and Illness over the Life Course: A Test of Cumulative Disadvantage Theory.” Social Science Research 37(4):1253–1266. [DOI] [PubMed] [Google Scholar]
  13. Drevenstedt Greg L., Crimmins Eileen M., Vasunilashorn Sarinnapha, and Finch Caleb E. 2008. “The Rise and Fall of Excess Male Infant Mortality.” PNAS 105(13):5016–5021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Elder Glen H. 1994. “Time, Human Agency, and Social Change: Perspectives on the Life Course.” Social Psychology Quarterly 57(1):4–15. [Google Scholar]
  15. Ferree Myra Marx. 2010. “Filling the Glass: Gender Perspectives on Families.” Journal of Marriage and Family 72(3): 420–39. [Google Scholar]
  16. Finch Caleb E. and Crimmins Eileen M. 2004. “Inflammatory Exposure and Historical Changes in Human Life-Spans.” Science 305(5691):1736–1739. [DOI] [PubMed] [Google Scholar]
  17. Ford Earl, Zhao Guixiang, Tsai James, and Li Chaoyang. 2011. “Low-Risk Lifestyle Behaviors and All-Cause Mortality.” American Journal of Public Health 101(10):1922–1929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Galobardes Bruna, Lynch John, and George Davey Smith. 2004. ” Childhood Socioeconomic Circumstances and Cause-Specific Mortality in Adulthood.” Epidemiologic Reviews 26(1):7–21. [DOI] [PubMed] [Google Scholar]
  19. Gee Gilbert C., Pavalko Eliza K. and Scott Long J 2007. “Age, Cohort and Perceived Age Discrimination: Using the Life Course to Assess Self-Reported Age Discrimination.” Social Forces 86(1):265–90. [Google Scholar]
  20. Giskes Katrina, Frank van Lenthe Gavin Turrell, Kamphuis Carlijn, Brug Johannes, and Mackenbach Johan P. 2008. “Socioeconomic Position at Different Stages of the Life Course and Its Influence on Body Weight and Weight Gain in Adulthood.” Obesity 16(6):1377–81. [DOI] [PubMed] [Google Scholar]
  21. Goosby Bridget J. 2013. “Early Life Course Pathways of Adult Depression and Chronic Pain.” Journal of Health and Social Behavior 54(1):75–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hamil-Luker Jenifer and O’Rand Angela M. 2007. “Gender Differences in the Link Between Childhood Socioeconomic Conditions and Heart Attack Risk in Adulthood.” Demography 44(1):137–158. [DOI] [PubMed] [Google Scholar]
  23. Hauser Robert M., Sheridan Jennifer, and Warren John Robert. 1999. “Socioeconomic Achievements of Siblings in the Life Course: New Findings From the Wisconsin Longitudinal Study.” Research on Aging 21(2):338–378. [Google Scholar]
  24. Hayward Mark D. and Gorman Bridget K. 2004. “The Long Arm of Childhood: The Influence of Early-Life Social Conditions on Men’s Mortality.” Demography 41(1):87–107. [DOI] [PubMed] [Google Scholar]
  25. House James S., Lepkowski James M., Kinney Ann M., Mero Richard P., Kessler Ronald C., and Regula Herzog A 1994. “The Social Stratification of Aging and Health.” Journal of Health and Social Behavior 35(3):213–234. [PubMed] [Google Scholar]
  26. Houle Jason N. 2013. “Depressive Symptoms and All-Cause Mortality in a Nationally Representative Longitudinal Study with Time-Varying Covariates.” Psychosomatic Medicine 75(3):297–304. [DOI] [PubMed] [Google Scholar]
  27. Kraemer Sebastian. 2000. “The Fragile Male.” British Medical Journal 321(7276):1609–1612. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kuh Diana and Cooper C 1992. “Physical Activity at 36 Years: Patterns and Childhood Predictors in a Longitudinal Study.” Journal of Epidemiology and Community Health 46(2):114–119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lynch Scott M. 2003. “Cohort and Life Course Patterns in the Relationship Between Education and Health: A Hierarchical Approach.” Demography 40(2):309–331. [DOI] [PubMed] [Google Scholar]
  30. Mahalik James R., Burns S, and Syzdek M 2007. “Masculinity and Perceived Normative Health Behaviors as Predictors of Men’s Health Behaviors.” Social Science and Medicine 64:2201–2209. [DOI] [PubMed] [Google Scholar]
  31. Masters Ryan K., Hummer Robert A., and Powers Daniel A. 2012. “Educational Differences in U.S. Adult Mortality: A Cohort Perspective.” American Sociological Review 77: 548–572. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Miech Richard, Pampel Fred, Kim Jinyoung, and Rogers Richard G. 2011. “The Enduring Association between Education and Mortality: The Role of Widening and Narrowing Disparities.” American Sociological Review 76:913–934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Miller Gregory, Chen Edith, and Parker Karen. 2011. “Psychological Stress in Childhood and Susceptibility to the Chronic Diseases of Aging.” Psychological Bulletin 137(6):959–997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Montez Jennifer K., Hayward Mark D., Brown Dustin C., and Hummer Robert A. 2009. “Why is the Educational Gradient of Mortality Steeper for Men?” Journals of Gerontology: Social Sciences 64(5):625–634. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Pearlin Leonard, Schieman Scott, Fazio Elena, and Meersman Stephen. 2005. “Stress, Health, and the Life Course: Some Conceptual Perspectives.” Journal of Health and Social Behavior 46(2):205–219. [DOI] [PubMed] [Google Scholar]
  36. Phelan Jo C., Link Bruce G., Anna Diez-Roux Ichiro Kawachi, and Levin Bruce. 2004. “Fundamental Causes of Social Inequalities in Mortality: A Test of the Theory.” Journal of Health and Social Behavior 45(3):265–285. [DOI] [PubMed] [Google Scholar]
  37. Phelan Jo C., Link Bruce G., and Tehranifar Parisa. 2010. “Social Conditions as Fundamental Causes of Health Inequalities: Theory, Evidence, and Policy Implications.” Journal of Health and Social Behavior 51(Suppl):S28–40. [DOI] [PubMed] [Google Scholar]
  38. Power Chris and Matthews Sharon. 1997. “Origins of Health Inequalities in a National Population Sample.” Lancet 350(9091):1584–1589. [DOI] [PubMed] [Google Scholar]
  39. Pudrovska Tetyana. 2010. “Why is Cancer More Depressing for Men than Women among White Older Adults?” Social Forces 89(2):535–558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Pudrovska Tetyana, Schieman Scott, Pearlin Leonard, and Nguyen Kim. 2005. “The Sense of Mastery as a Mediator and Moderator in the Association between Economic Hardship and Health in Late Life.” Journal of Aging and Health 17(5):634–660. [DOI] [PubMed] [Google Scholar]
  41. Pudrovska Tetyana, Carr Deborah, McFarland Michael, and Collins Caitlyn. 2013. “Higher-Status Occupations and Breast Cancer Incidence and Mortality: A Life-Course Stress Approach.” Social Science and Medicine 89:53–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Read Jennifer, Troendle James, and Klebanoff Mark. 1997. “Infectious Disease Mortality among Infants in the United States, 1983 through 1987.” American Journal of Public Health 87(2):192–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Reczek Corinne and Umberson Debra. 2012. “Gender, Health Behavior, and Intimate Relationships: Lesbian, Gay, and Straight Contexts.” Social Sciences and Medicine 74(11):1783–1790. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Reither Eric, Hauser Robert, and Swallen Karen. 2009. “Predicting Adult Health and Mortality from Adolescent Facial Characteristics in Yearbook Photographs.” Demography 46(1):27–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Riva M, Curtis S, and Norman P 2011. “Residential Mobility within England and Urban-Rural Inequalities in Mortality.” Social Science and Medicine 73(12):1698–1706. [DOI] [PubMed] [Google Scholar]
  46. Rosenfield Sarah. 1999. “Splitting the Difference: Gender, the Self, and Mental Health” Pp. 209–224 in Handbook of the Sociology of Mental Health, edited by Carol S. Aneshensel and Jo C. Phelan. New York: Springer. [Google Scholar]
  47. Ross Catherine and Mirowsky John. 2002. “Age and the Gender Gap in the Sense of Personal Control.” Social Psychology Quarterly 65(2):125–145. [Google Scholar]
  48. Saint Onge Jarron M. and Krueger Patrick M. 2011. “Education and Racial-Ethnic Differences in Types of Exercise in the United States.” Journal of Health and Social Behavior 52(2):197–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Sautter Jessica M., Thomas Patricia A., Dupre Matthew E., and George Linda K. 2012. “Socioeconomic Status and the Black-White Mortality Crossover.” American Journal of Public Health 102(8):1566–1571. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Schwartz Christine R. and Mare Robert D. 2005. “Trends in Educational Assortative Marriage From 1940 to 2003.” Demography 42(4):621–646. [DOI] [PubMed] [Google Scholar]
  51. Sewell William and Hauser Robert. 1975. Education, Occupation, and Earnings: Achievement in the Early Career. New York: Academic Press. [Google Scholar]
  52. Singer Judith and Willett John. 2003. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press. [Google Scholar]
  53. Springer Kristen W. and Mouzon Dawn. 2011. “‘Macho men’ and Preventive Health Care: Implications for Older Men in Different Social Classes.” Journal of Health and Social Behavior 52(2):212–227. [DOI] [PubMed] [Google Scholar]
  54. Springer Kristen W., Hankivsky Olena and Bates Lisa M. 2012. “Gender and Health: Relational, Intersectional, and Biosocial Approaches.” Social Science and Medicine 74(11):1661–6. [DOI] [PubMed] [Google Scholar]
  55. Umberson Debra, Crosnoe Robert, and Reczek Corinne. 2010. “Social Relationships and Health Behavior Across the Life Course.” Annual Review of Sociology 36:139–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Willson Andrea E., Shuey Kim M., and Elder Glen H. Jr. 2007. “Cumulative Advantage Processes as Mechanisms of Inequality in Life Course Health.” American Journal of Sociology 112(6):1886–1924. [Google Scholar]
  57. Yancey Martin Patricia. 2003. “‘Said and Done’ Versus ‘Saying and Doing’: Gendering Practices, Practicing Gender at Work.” Gender and Society 17(3): 342–66. [Google Scholar]
  58. Zajacova Anna. 2006. “Education, Gender, and Mortality: Does Schooling Have the Same Effect on Mortality for Men and Women in the US?” Social Science and Medicine 63(8):2176–2190. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Appendix A & B
NIHMS625116-supplement-Appendix_A___B.pdf (61.6KB, pdf)

RESOURCES