Background:

Existing research shows that residents exposed to neighborhood socioeconomic disadvantage (NSD) have a relatively high risk of some types of mental illness. This study investigated whether this pattern is present for depressive symptoms among persons in late middle age to extreme old age, and if so why. Investigated factors include: stressors (neighborhood physical disorder and financial strain) and psychosocial resources (mastery, social support, and religious service attendance). Age differences were examined.

Methods:

This was a secondary analysis of data from the 2006 and 2008 Health and Retirement Study Psychosocial Supplemental Assessment. The analytic sample included 8,623 adults aged 52 to 104, residing in 3216 urban neighborhoods (census tracts). NSD was operationalized with a principal component of Census tract indicators (education, poverty, public assistance, and unemployment). Depressive symptoms were measured by an 8-item version of the Center for Epidemiologic Studies Depression Scale. Hierarchical linear regression estimated multilevel models.

Results:

NSD was positively associated with depressive symptoms, net of individual-level demographic characteristics used to control for selection. Exposure to stressors partially explained the association; while psychosocial resources suppressed it. A significant, curvilinear cross-level interaction with age was found: NSD is positively associated with symptoms among those under 64, but has little effect among persons 65 to 74 year, and is negative at older ages.

Conclusion:

The mental health disparity for depressive symptoms in NSD is partially due to greater exposure to stressors and fewer resources to counteract this exposure. The counter-intuitive inverse association among the oldest adults warrants further investigation.

Depression is a major public health challenge affecting millions of people worldwide, particularly women. Intra-individual explanations of gender differences in depression include biology (e.g., neurotransmitters, genes, hormones) and psychological factors (e.g., self-concept and esteem, mastery). Social explanations focusing on exposure to stressors (e.g., low socioeconomic status) and social role occupancy (e.g., marital and employment status) also have been considered. The recognition that environmental factors may influence mental health has given rise to studies examining the relationship between neighborhood conditions and risk for depression and psychological distress, but gender differences in neighborhood effects have yet to receive the needed research attention.

This dissertation sought an understanding of: (a) gender differences in neighborhood effects on depressive symptoms, and (b) neighborhood influences on variation in depressive symptoms among women. The research is guided by the neighborhood stress process framework focusing on stressors and psychosocial resources as mediators and moderators of the relationships among neighborhood conditions and depressive symptoms. Individual-level cross-sectional data come from the Health and Retirement Survey (HRS: baseline, 2006/2008 interviews and psychosocial questionnaire supplement). The HRS is a U.S. national probability sample of adults over the age of 50. Analyses are performed within a multilevel framework and urban neighborhood data come from the 2000 U.S. Census.

Among eight indicators of neighborhood disadvantage, including neighborhood socioeconomic disadvantage (NSD), and five measures of neighborhood advantage examined for gender differences in their effects on depressive symptoms, two were statistically significant but not in the expected direction. Neighborhood proportion non-family households was associated with fewer depressive symptoms among women and it had no effect among men. Neighborhood proportion married-couple households was not significantly related to depressive symptoms among women, but among men, living in a neighborhood with more married-couple households with children was associated with fewer symptoms. Overall, the impact on depressive symptoms of neighborhood characteristics do not differ for men and women.

Gender differences in neighborhood effects on three individual-level stressors and three individual-level psychosocial resources also were examined. Nine interactions were statistically significant. Consistent with expectations, people who reside in neighborhoods with more vacant housing units perceived more disorder and less social cohesion in their neighborhoods, and the effects were larger for women than men. Relative to men, women's perceptions of neighborhood social cohesion and social support are more sensitive to neighborhood economic conditions. In general, with a few notable exceptions, neighborhood effects on stressors and psychosocial resources do not vary by gender.

In analyses that only included women, NSD was positive and significantly associated with depressive symptoms and neighborhood proportion adults aged 65 and older was negative and significantly associated with symptoms. Perceived neighborhood social cohesion fully mediated the effect of NSD- and partially mediated the effect of neighborhood proportion older adults- on depressive symptoms. The effect on depressive symptoms of neighborhood disadvantage did not vary significantly by levels of stressors and psychosocial resources except for three significant cross-level interactions. Living in a neighborhood with more vacant housing units was associated with more depressive symptoms, and the effect was greater among women who perceived high levels of disorder in the neighborhood than those who perceived less disorder. Also as hypothesized, NSD had the largest positive effect on depressive symptoms among women with less social support than women with more support. However, mastery did not function as a stress-buffer.

The effect on depressive symptoms of neighborhood advantage varied significantly by psychosocial factors. Living in a neighborhood with higher proportions of older adults was associated with fewer depressive symptoms more so for women who report low levels of perceived neighborhood physical disorder than women who report average levels of disorder. Also consistent with expectations, higher neighborhood proportion of affluent households and owner-occupied housing units were associated with fewer depressive symptoms, and the effects were larger for women with high levels of mastery than women with low mastery. However, these neighborhood characteristics were less beneficial to the mental health of women with high than low levels of social support.

The findings from this dissertation largely indicate that relationships among components of the neighborhood stress process model do not differ by gender or by levels of stressors and psychosocial resources. However, the significant results that emerged make a valuable contribution to the research literature by identifying urban neighborhood conditions that are consequential to the mental health of middle-aged and older adults and that should be the target of interventions.

Violence is a serious public health problem that is a threat to society as a whole. Even when it does not result in death, it takes a toll on victims, families, and communities. The processes by which people become violent, and then continue or discontinue being violent, are not well understood. Although it is one of the leading causes of death for people 1-44 years of age, much of the focus of violence research has focused on adolescents and the factors that contribute to the onset of violence at that stage of the life course. Less is known about patterns of violence over time and what factors contribute to its persistence and cessation. Furthermore, little is known about female violence, its patterns over time, and how it differs from male violence.

The overall goal of this dissertation is to elucidate the impact of relationships with parents, friends, and romantic partners over time on trajectories of violent behaviors from adolescence into early adulthood through adulthood. This study makes a significant contribution to the knowledge base on violence by: (1) identifying four trajectories of violent behaviors in adolescence through adulthood in the full sample and in gender-stratified subsamples, (2) explicating the effects of social relationships at different life stages on these trajectories, and (3) ascertaining the extent to which these processes differ for males and females. It is guided by important concepts from the life course perspective, attachment theory, and social control theory to explore the ways in which social relationships over time are associated with violent behaviors across these life stages. Secondary analysis of Waves 1-4 (1994 to 2008) from the National Longitudinal Study of Adolescent Health (N = 11,197) was conducted to achieve the goals of this dissertation.

There are a number of key findings from this study. First, four distinct trajectories of violence from adolescence through adulthood were identified in this dissertation in the full sample and for females and males. They are: low desister, high desister, chronic perpetrator, and late escalator. Although similar groups were found in the gender subsamples, the specific shapes and compositions of these groups vary somewhat for males and females. Second, child abuse victimization by a parent and associating with delinquent friends during adolescence increase the risk of violence and significantly distinguish violence trajectories. These negative aspects of parent and friend relationships have proximal, distal, and persistent effects on violence perpetration and were found in the full sample in males and females. Third, romantic partnerships can be both deterrents to and provocations of violence. The direction of influence of romantic partnership depends on age, or life stage, the type of violence trajectory group a person is most likely to be assigned to, and gender. Fourth, violence victimization begets violence perpetration. Experiencing violence victimization is a very strong predictor of membership in a group that has increased or prolonged levels of violence for the full sample and for females and males. Five, the influence of certain social relationships on trajectories of violence vary by gender, emphasizing the importance of understanding how these processes vary by gender.

It is important that factors that put people at increased risk of violence be diminished, while simultaneously stimulating those that deter violence in order to decrease its pernicious and destructive reach. By elucidating the processes by which social relationships act as encouragements to or deterrents of violence at varying time points between adolescence and adulthood, this dissertation highlights opportunities for the reduction and/or prevention of violence couched within an understanding of what period on the life course might be more sensitive to intervention.

Despite decades of research and public health intervention, disparities in cardiovascular diseases among social groups persist and are only partially explained by parallel differences in socioeconomic status or conventional risk factors, such as smoking. For example, relative to non-Hispanic whites, racial and ethnic minorities are particularly at risk of developing and dying from cardiovascular diseases (Rogers et al., 2011) and, in terms of gender, women experience higher mortality rates than men despite their lower prevalence rate. Understanding these systematic differences in health necessitates acknowledging that the social patterning of cardiovascular diseases is not a chance occurrence; racial disparities in health are rooted in a system of social stratification that consistently and systematically disadvantages certain subgroups of the population (Adler & Rehkopf, 2008) through the unequal distribution of fiscal and material resources, on the one hand (Williams & Mohammed, 2009), and through the unequal distribution of stressors and coping resources, on the other hand (Pearlin, 1989; Williams & Mohammed, 2009).

Discrimination is a chronic stressor that disproportionately affects racial minorities (Kessler, Mickelson & Williams, 1999). It manifests as discrete life events (i.e., major lifetime discrimination) that significantly hinder opportunities for status attainment and as day-to-day experiences of unfair treatment (i.e., everyday discrimination), such as being treated with less respect than others or being threatened or harassed (Williams, Yu, Jackson, & Anderson, 1997). Prior research has linked exposure to discrimination to multiple adverse health outcomes including low infant birth weight, depressive symptoms, all-cause mortality, and risk factors for cardiovascular diseases, such as high blood pressure (for a review of this literature see: Mays, Cochran & Barnes, 2007; Pascoe & Richman, 2009; Williams & Mohammed, 2009). Therefore, racial disparities in cardiovascular diseases may be attributed to differences in exposure to discrimination.

The current study sheds light on the mechanisms through which group differences in exposure to and the impact of discrimination generate disparities in cardiovascular diseases. The study specifically looks at its effects on C-reactive protein (i.e., CRP), a protein produced in response to exposure to stressors. It is a marker of systemic inflammation that is positively associated with cardiovascular diseases such as stroke, atherosclerosis, and myocardial infarction (Ridker, Cushman, Stampfer, Tracy, & Hennekens, 1997; Ridker, Hennekens, Buring, & Riafai, 2000). The current study goes beyond a basic description of the population-level distribution of CRP by focusing on psychosocial factors that mediate and/or moderate these relationships. The specific aims of the dissertation are to investigate: (1) racial/ethnic and gender differences in CRP levels; (2) differential exposure to discrimination as an explanation for the social patterning of CRP; (3) differential vulnerability to discrimination; and (4) the extent to which coping resources dampens these effects.

The dissertation uses data from the Health and Retirement Study (HRS), a multi-cohort longitudinal survey of a large and nationally representative sample of adults age 51 years and older. The HRS aims to identify and better understand the social, economic, psychosocial, and physical factors that influence and result from retirement. Since its inception in 1992, data have been collected every two years on the original HRS cohort and on subsequent cohorts added to the study. The "core" HRS interview is conducted face-to-face at baseline and collects data on demographic characteristics; housing and family structure; employment, income, assets, and insurance; health care, health status, cognition, and disability; and life expectations. Beginning in 2004, HRS respondents were asked to complete a psychosocial questionnaire that included measures of everyday and lifetime discrimination, among other factors, and in 2006 biomarkers--including CRP--were collected from one half of the study sample; biomarkers were collected from the other half in 2008. To maintain sufficient statistical power to detect differences in CRP, multivariate analyses are conducted with data from a hybrid 2006/2008 sample. The analysis involves statistical methods needed to appropriately adjust for the complex sampling design of HRS and to test the intricate causal pathways leading to race differences in CRP.

In line with our hypotheses, African Americans have higher levels of CRP than non-Hispanic whites and reported greater exposure to everyday and lifetime discrimination. Hispanics do not significantly differ from non-Hispanic whites in CRP levels or in their reports or everyday and lifetime discrimination. In unadjusted models, everyday and major lifetime discrimination are positively associated with CRP. This association remains for lifetime discrimination in fully adjusted models but not for everyday discrimination. Both everyday and lifetime discrimination mediate racial/ethnic and gender differences in inflammation. Lifetime discrimination mediates these differences in and of itself, and in conjunction with other factors, particularly waist circumference. Everyday discrimination also mediates group differences in inflammation in conjunction with waist circumference. Vulnerability to discrimination does not differ by race/ethnicity or by gender, and the coping resources evaluated in this study do not buffer the effects of discrimination on inflammation.

This study supports the presence of racial differences in CRP--an indicator of systemic inflammation and a clinical risk factor for cardiovascular diseases--on a population-level. The findings also suggest that exposure to discrimination, a chronic stressor that disproportionately affects racial minorities, is associated with higher levels of CRP and partially explains racial and gender differences in inflammation. In all, these findings provides further support for the significant contributions of the social environment on health and health disparities.

The organization of work in the United States has changed dramatically over the last four decades as a result of globalization, industrial shifts, and technological innovation (Burgard et al., 2009; Seigrist & Marmot, 2005; Sverke & Hellgren, 2002). Long-term, stable employment relationships have increasingly been replaced with work arrangements characterized by shorter job tenure and fewer worker protections, leading to an increase in perceived job insecurity (Standing, 2011). Perceived job insecurity is defined as an anticipatory stressor related to the threat of losing a job, or important features of a job, and a sense of powerlessness to overcome this threat (Greenhalgh & Rosenblatt, 2010; De Witte, 2005).

The Employment Conditions Knowledge Network of the World Health Organization (WHO) Commission on Social Determinants of Health (Benach et al., 2007) suggests that the effects of perceived job insecurity on health may be even more harmful than job loss itself. Although evidence for a causal relationship between perceived job insecurity and health is growing, little is known about mechanisms and group differences in this relationship.

This dissertation investigates the relationship between perceived job insecurity and quality of life. Quality of life is defined in this study as a general state of health, well-being, and satisfaction across multiple dimensions of life (Drotar, 2014; Kobau et al., 2010; Bowling, 1991). Although the WHO defines health very broadly as a “complete state of physical, mental, and social well-being,” health is usually assessed in studies of perceived job insecurity in very narrow terms based on the presence or absence of disease (CDC, 2011). In contrast to this deficit perspective on health, quality of life is an overall state of wellness.

This dissertation makes an important contribution to the literature on perceived job insecurity and quality of life by achieving three aims: 1) identifying determinants of perceived job insecurity among adults in the United States; 2) testing causal pathways linking perceived job insecurity and perceived inequality at work to quality of life; and 3) estimating group differences in the effect of perceived job insecurity and perceived inequality at work on quality of life based on social status, social resources, and demographic characteristics.

Data for this research were drawn from a U.S. sample of non-institutionalized, English-speaking adults in the 1995-1996 and 2004-2006 waves of the National Survey of Midlife Development in the United States (MIDUS). Multivariate generalized ordinal structural equation modeling was used to test hypotheses for the first aim (N=5,424). Multivariate linear structural equation modeling with full information maximum likelihood estimation was used to test hypotheses for the second and third aims (N=5,113). This study was informed by Pearlin and colleagues’ (1981) stress process model, which posits that unequal exposure to stressors and access to resources for coping with stressors contributes to health disparities.

Key findings from this study add to existing knowledge on job stressors and health. First, for determinants, perceived job insecurity is influenced by objective job insecurity over and above the effects of psychosocial job stressors and demographic characteristics. Effort-reward imbalance also significantly increases perceived job insecurity, but job strain does not. High effort may improve security, but not in the absence of control. People of color report significantly higher perceived job insecurity than non-Hispanic white people, but there is no association between gender and perceived job insecurity, all other factors held constant. Perceived inequality at work is a strong determinant of perceived job insecurity.

Second, for consequences, perceived job insecurity is associated with quality of life net of controls for demographic and health characteristics, but not when controlling for other job stressors. Perceived inequality at work confounds the relationship between perceived job insecurity and quality of life. Negative work to family spillover of stress and social support outside of work from family, friends, and spouse/partner significantly mediate the relationship between perceived inequality at work and quality of life.

Third, for conditioning factors, the effect of perceived inequality at work on quality of life is conditional on household income, social support at work, age, and wave of interview. No significant group differences were found by education, gender, or race.

The findings demonstrate that perceived job insecurity is associated with other psychosocial stressors from established job stress models but that inequality may be even more threatening to health and well-being than insecurity. Stress proliferation from work to family life partially explains the relationship between perceived inequality at work and quality of life. Programs to enhance social support at work and home may help to contain the negative health effects of inequality and insecurity on quality of life.

Co-occurring disorder (COD) refers to concurrent psychiatric and substance use disorders (SUD). Compared to those with a single disorder, individuals with COD often require more complex treatment, have poorer health outcomes, and incur higher treatment costs. Researchers have extensively studied both the high lifetime prevalence and age of onset for psychiatric disorders and SUD independently, but little is known about the social antecedents of COD, especially how these antecedents vary by race/ethnicity and gender.

I expect the antecedents do not behave universally, though they are currently treated that way. Guided by the Stress Process Model, the Theory of Fundamental Causes, and the Life Course Perspective, this dissertation aims to better understand the role of childhood poverty and childhood adversity in the occurrence of COD for males and females, and for different racial/ethnic groups. This dissertation employs a secondary analysis of existing community-based survey data recorded in the National Epidemiologic Survey of Alcohol and Related Conditions III.

Using multinomial logistic regression with a four-category variable for disorder (categories: COD, SUD only, psychiatric disorder only, no disorder), on a bivariate level, childhood poverty is associated with COD, however, with the addition of all other covariates there is no longer an association between poverty and COD. Childhood adversities are strongly associated with COD, net of other factors, in all of the models estimated.

There are clear race/ethnicity differences in prevalence of disorder when COD is studied in the whole population. For COD relative to no disorder, Blacks, Asian Americans, and Hispanics, are all approximately half as likely as Whites to have COD, net of other factors. There are no conditional race/ethnicity relationships for COD. There are, however, gender differences in both disorder prevalence and the associations between childhood poverty and COD as well as childhood adversity and COD. Childhood poverty is associated with COD in opposite directions for males and females: for males it increases the relative risk ratio of COD compared to SUD, and for females it decreases the relative risk for this same comparison. This study found no moderation of the childhood poverty and COD relationship by number of adversities in the regressions conducted.

Conducting a survival analysis with only respondents who have at least one disorder indicates that having psychiatric disorder compared to having SUD is associated with a 36% increase in the hazard ratio of subsequently developing COD overall. The significant conditional relationship between disorder sequence and gender shows that hazard of co-occurrence with a psychiatric disorder is higher for males than females. On the contrary, the hazard of co-occurrence when one has SUD is higher for females than it is for males.

This research has clear public health relevance: above and beyond the genetic risk incurred by having a parent with a disorder, experiencing adverse events in childhood is associated with COD. Efforts to help children and adolescents ameliorate the adversity they are exposed to are important and may be able to diminish the risk of COD associated with harmful early experiences.

The goal of this dissertation was to explore older adults’ perceptions of and experiences with quality of life technologies and the meanings they ascribe to a specific subset of quality of life technologies: personal emergency response systems (PERS). Specifically, the purpose was to analyze how older adults perceive the need for PERS and how they negotiate the uptake and use of PERS with themselves and with others.

In collaboration with three community-based organizations in western Los Angeles, California, 18 persons (age 71-97 years) were interviewed. The sample was comprised of eight PERS subscribers and ten non-subscribers. The interviews were conducted using a semi- structured interview guide that included questions about perceptions of PERS prior to subscription and experiences with the use of PERS. Data collection and analyses were guided by constructivist grounded theory methodology. Analytic techniques included initial coding, focused coding, theoretical sampling, and memoing.

The results are presented in a substantive theory that is grounded in the words and narratives of study participants. The theory situates participants’ pre-subscription experiences in the context of their efforts to counteract the impacts of aging. It is comprised of three processes: reclaiming control, protecting personhood, and walking the balance beam. Participants took action to reclaim control they had lost over their bodies due to aging-related changes. Interviewees also sought to protect their personhood from social forces that encroached on their sense of self. Participants appraised PERS with regards to the extent to which the technology could thwart or support these goals. In many cases, these appraisals stood in opposition to each other and participants’ repeatedly used phrases like “I’m not ready yet” to describe this conflict. Thiss internal conflict led participants to walk the balance beam, which entailed postponing their decision with regards to PERS while re-evaluating the meanings of PERS. Over time, participants edged closer to acquiring a PERS through imagined, vicarious, and actual experiences of emergency situations. Additionally, input from members of their social environment facilitated interviewees’ progression towards PERS.

This research is the first to provide crucial insights into the decision-making process specifically prior to PERS subscription. Future research and interventions should conceptualize PERS adoption can be productively conceptualized as behavior change in future research and intervention, which should take into account older adults’ level of readiness to adopt a PERS.

Across the life course, substance use is highest in adolescence and young adulthood in the U.S. Although substance use declines with age, a significant number of young adults go beyond normative to problematic use. Despite the strong relationship between low social status and poor health, substance use varies by social status in inconsistent ways. Smoking is higher among lower social status groups. Alcohol use is higher among higher social status groups. Substance use is often higher among Whites compared to racial/ethnic minorities. Problematic substance use is often higher among minorities compared to Whites. These inconsistent patterns may be due to drug type (e.g., smoking vs. alcohol) or difficulty in assessing social status during the transition to adulthood and across race/ethnicity.

The goal of this dissertation is to ascertain the effects of social status on substance use behaviors across race/ethnicity during the transition to adulthood using secondary data analysis of three waves of the National Longitudinal Study of Adolescent Health. Using social stratification theories and the life course perspective, this study conceptualized social status as a life-course construct from adolescence (ages 12-17) to young adulthood (ages 18-26) and adulthood (ages 24-32). Furthermore, social status was examined across the domains of economic, human, and social capitals. Cigarette smoking and heavy episodic alcohol use served as the key outcomes.

Through a person-oriented framework, latent class analysis captured the ebb and flow of social status advantages and disadvantages with four latent groups for the domains of economic and social capitals, and five groups for the domain of human capital. These latent classes were substantively similar across Whites, Blacks, and Latinos; however, variations within subgroups differed. Regardless of domain or race/ethnicity, persistently low social status had higher smoking prevalence when compared to other groups. For heavy episodic drinking, there was no clear pattern across domains and race/ethnicity, which suggest a complex picture that may be difficult to disaggregate. Overall, these findings highlight several dimensions of social status that present opportunities to reduce substance use disparities, and reveal life-course social status groupings from adolescence to adulthood that may provide important avenues to prevent problematic substance use.