Socioeconomic Status and Health in Middle and Late Life: A New Look at an Old Issue

Restricted (Penn State Only)
Acciai, Francesco
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
July 14, 2016
Committee Members:
  • Melissa Hardy, Dissertation Advisor
  • Melissa Hardy, Committee Chair
  • Glenn A Firebaugh, Committee Member
  • Steven Andrew Haas, Committee Member
  • Leif Jensen, Outside Member
  • Social inequalities in health
  • Health disparities
  • Socioeconomic status
  • SES
  • Chronic conditions
  • Chronic diseases
  • Elderly
  • Late life
  • Europe
  • Aging
  • Health trajectories
  • Longitudinal analysis
  • Attrition
  • Sample selection
One of the most consistent findings in social science research is the existence of a health gradient based on socioeconomic status (SES). Despite the large evidence linking SES and health from birth and infancy throughout all adulthood, there are contrasting results on whether the health advantages associated with higher SES persist at older ages. In other words, it is not clear whether at older ages, socioeconomic factors keep exerting a protective effect on health, whether such effect gradually fades away, or whether such effect keeps increasing. Three hypotheses have been proposed to describe how social inequalities in health can evolve with age: (1) The persistent inequality hypothesis; (2) The age-as-leveler hypothesis; and (3) The cumulative advantage hypothesis. The current study provides new evidence relevant to this debate. The overarching question is whether socioeconomic status continues to exert a protective effect on health at older ages. This theme is developed throughout three chapters. In each chapter I examine the relationship between socioeconomic status and health in middle and late life from a different angle. Chapter 1 focuses on the measurement of SES for elderly individuals. The rationale is that inconsistent findings relative to the SES-based health gap among the elderly may derive from the use of SES indicators that are not particularly suited for this population. In particular, I hypothesize that wealth is a more appropriate indicator of SES at older ages than the more commonly used measures of SES, such as income and education. I test this hypothesis by examining the age pattern of social inequalities in prevalence of conditions, by using simultaneously wealth, income, and education. Chapter 2 analyzes individuals’ trajectories of chronic conditions and self-rated health. The research question I address is whether at the individual level aging is associated with a diminution of the protective effect of SES on health. Chapter 3 focuses on the problem of attrition (or sample selection). While no longitudinal studies is immune to this problem, studies that focus on health, especially at older ages, must pay particular attention because the process of sample selection due to health deterioration is closely linked to the outcome under investigation. In this chapter I describe four different methods that can be used to adjust for sample selection: Propensity score; Heckman’s model; Inverse probability weighting; and Multiple imputations. Then I apply each method to the study of incidence of new chronic conditions between two points in time and compare how they perform.