ABSTRACT
In practical research, the analysis of event history data with nonparametric estimation methods is associated with several disadvantages. First, as discussed in the previous chapter, with an increasing number of subgroups normally a point is rapidly reached, at which it is no longer sensible to estimate and compare survivor functions due to the small number of cases left in the various subgroups. Second, even in the case where it is feasible to estimate a rising number of survivor functions for important subgroups, comparisons of these functions quickly become complex and interpretation difficult. Third, in the case of quantitative characteristics (e.g. income, age, etc.), variables must be grouped (e.g. “high income group” vs. “low income group,” etc.), with a loss of information, to be able to estimate and compare survivor functions. Finally, multi-episode processes can hardly be analyzed with nonparametric methods. Over the last 20 years, transition rate models have therefore increasingly been used in practical research for the analysis of event history data instead of nonparametric methods.
