Generalized Time-Varying Effect Models for Binary and Count Outcomes

In this chapter, we discuss extensions of time-varying effect modeling (TVEM) for binary and count outcomes. Specifically, this chapter will present model specification and interpretation of (a) logistic TVEM for binary dependent variables and (b) Poisson TVEM for count dependent variables. Interpretation of logistic models will cover time-varying logistic regression coefficients, odds ratios, and predicted probabilities. We will guide the reader through an empirical example that estimates the age-varying prevalence of past-year hypertension and estimates the age-varying effects of sex, racial/ethnic group, and their interaction on past-year hypertension. Interpretation of Poisson models will cover time-varying regression coefficients, incidence rate ratios, and predicted counts. An empirical example is included that estimates the age-varying mean number of drinks during one’s typical drinking occasion during the past year. The age-varying effects of sex, racial/ethnic group, and their interaction are similarly examined.