ABSTRACT
This chapter introduces several statistical models. Depending on the type of response variables, it discusses two popular models: a linear model for continuous Gaussian response of the event rate or measure, and a loglinear model for event frequency count out of population exposure. Based on the factors included for model covariates, the models can be single factor models, two factor models, or three factor models. Single factor models include age effect (A) model, period effect (P) model, and cohort effect (C) model. Two factor models include age-period (AP) model, age-cohort (AC) model, and period-cohort (PC) model. Three factor models include all age, period, and cohort effects. The chapter considers linear models for age-period-cohort (APC) data. It then provides R programs for fitting the linear models. The chapter also provides R functions and examines how to use these functions to conduct APC analysis. A loglinear model can serve the needs by incorporating the extra piece of information through a Poisson distribution.
