Background: Longitudinal studies suggest that socioeconomic status (SES) and mental health have a bidirectional relationship, such that declines in SES lead to a deterioration of mental health (social causation), while worsening mental health leads to declines in SES (social drift). Nevertheless, existing research has important substantive and methodological gaps. Most notably, studies often employ one from a diverse range of SES indicators and arrive at different conclusions, with labor market indicators (e.g., earnings) providing more consistent evidence of bidirectional effects and non-labor market indicators (e.g., family income) generally offering only support for social causation dynamics.
Studies frequently estimate “average effects” failing to examine differences in social causation and social drift effects across populations. From a methodological standpoint, studies often have limited ability to draw causal inferences. For instance, studies examine either social causation or social drift effects independently without controlling for reverse causation. Other studies fail to control for time-invariant differences across individuals that could significantly bias estimates. Furthermore, studies on the association between material hardship and mental health often rely on measures of material hardship with unknown validity and reliability. This three-paper dissertation seeks to tackle several shortcomings in existing research, with the goal of improving and advancing our understanding how SES and mental health affect each other over time and how these dynamics vary across populations.
Methods: This dissertation employs data from a five-wave representative panel (n=3,103) of working-age (18-64) New York City adults with yearly measures of individual earnings, family income (income-to-needs), material hardship, and psychological distress. Paper 1 examines bidirectional effects between income types (individual earnings and family income) and distress by relying on cross-lagged panel models with unit fixed effects (FE-CLPM). Subgroup analyses are conducted by examining effects by age, gender, education, and racial/ethnic identification. Paper 2 develops measurement models for material hardship and examines the relationships longitudinal trajectories of income, material hardship, and distress. To identify dimensions underlying the seven observed material hardship indicators, Exploratory Factor Analyses (EFA) were performed on a randomly selected training sample (n=1,542). Subsequently, cross-sectional Confirmatory Factor Analyses (CFA) and longitudinal invariance tests were conducted on the holdout sample (n=1,561) to further examine the factor structure extracted via EFA and test its measurement equivalence across time. A latent state-trait model examined the extent to which indicators vary or persist over time. Additional CFA models were specified to examine the association between material hardship and income types and psychological distress. Lastly, utilizing factor scores calculated based on CFA models, parallel linear growth curve models were estimated to examine the association between the longitudinal trajectories of income types, material hardship, and psychological distress.
Paper 3 examines the bidirectional effects between material hardship, psychological distress, earnings, and family income. Material hardship is measured via a single scale and two subscales for unmet needs (e.g., food insufficiency, housing instability, medical needs, cash hardship) and billpaying hardship (e.g., difficulty paying for rent/mortgage and utilities, utilities disconnection). Factor scores for material hardship measures were estimated based on measurement models developed in paper 2 of this dissertation. I utilize FE-CLPMs to examine social causation and social drift effects between material hardship and psychological distress. An initial model examines effects between material hardship and distress only controlling for partnership status and number of children as time-varying covariates. Subsequently, three-variable FE-CLPMs examine effects between income (earnings or family income), material hardship, and distress. Total, direct, and indirect effects are estimated to examine the effect of income on distress via material hardship, and the effect of distress on material hardship through income. Follow-up models examine the simultaneous effects of unmet needs and billpaying hardship. Finally, subgroup analyses examine bidirectional effects between the material hardship subscales and distress by age, gender, race/ethnicity, education, and permanent family income.
Conclusions: The findings of this dissertation provide new evidence about the bidirectional effects between SES and psychological distress. Nonetheless, this study also reveals important differences in the magnitude and direction of effects depending on the SES indicator employed and the population studied. Across income types, individual earnings may be stronger determinants of mental health than family income. Additionally, social causation and social drift effects between income and distress vary by age, education, gender, and racial/ethnic identities. In paper 2, two distinct, although highly correlated, dimensions of material hardship were identified, namely, unmet needs and billpaying hardship. Consistent with prior research, the rate of change in material hardship mediated the association between the rates of change in income and distress. However, the mediating role of material hardship seems to be driven by the unmet needs factor and not billpaying hardship. Unmet needs (e.g., food, housing, medical care) may be more important social determinants of mental health than difficulties paying for bills (e.g., rent, utilities). The findings of paper 3 offer evidence supporting the reciprocal relationship between material hardship and psychological distress, particularly highlighting the significance of unmet needs as a social determinant of mental health. Difficulties in paying bills seem to be especially important among individuals facing economic disadvantage and those nearing retirement age. From a methodological perspective, the findings of this three-paper dissertation make a case for employing rigorous methods to improve the causal inference of studies about the relationship between SES and mental health.
Particularly, this study underscores the importance of methods that can control for unobserved differences between individuals and examining social causation and social drift effects simultaneously. From a substantive perspective, this dissertation also underscores the importance of moving beyond ‘average effects’ and examining potential disparities in the way that subpopulations experience the effects of SES and mental health. From a social policy standpoint, this study highlights the importance of providing support to mitigate the impact of material hardship and income shocks, particularly earnings losses, as these factors have independent effects on distress. Moreover, future research ought to prioritize the development of interventions aimed at alleviating the economic and mental health consequences arising from bidirectional effects between SES and psychological distress.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/ch33-6k71 |
| Date | January 2024 |
| Creators | Jimenez-Solomon, Oscar |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0023 seconds