Piecewise constant exponential models
Hans‐Peter Blossfeld, Götz Rohwer, Thorsten Schneider, Brendan Halpin · 2019
This chapter examines the piecewise constant exponential model. This is a simple generalization of the standard exponential model, but it is extremely useful in many practical research situations. The basic idea is to split the time axis into time periods and to assume that transition rates are constant in each of the intervals but can change between them. If there are L time periods, the piecewise constant transition rate is defined by L parameters. The first is to assume that only a baseline rate, given by period-specific constants, can vary across time periods, but that the covariates have the same effects in each period. The second option allows for period-specific effects of covariates. The problem with the analysis without covariates is that important heterogeneity between the individuals has not been taken into account. The piecewise constant exponential model with period-specific effects is particularly interesting because it provides an opportunity to assess hypotheses based on filter or signaling theory.