A COMPETING RISKS ANALYSIS OF OLDER AMERICANS' POVERTY ENTRY AND EXIT PATTERNS IN THE HEALTH AND RETIREMENT STUDY.

We examine how older Americans' poverty entry and exit patterns are associated with survey sample attrition using a longitudinal data set for 2002-2018 from the Health and Retirement Study. We consider how sample attrition affects estimates of variables associated with poverty entry and exit, and we find that attrition bias is less apparent in models estimating poverty entry than in models of poverty exit. The effect of aging on poverty exit is smaller in competing risks models than in proportional risk models, and cross-model differences in race and ethnicity effects are not statistically significant. This indicates that long-term respondents were more likely to exit poverty than were those who attrited from the sample, implying a change in sample representativeness over time. Our research confirms the importance of understanding panel data attrition biases when examining older Americans' poverty vulnerability.