Abstract
All customary parametric survival models show a high probability of event occurance at early times. This is not appropriate in most cases where arrival time is age. A way to overcome this problem is the decomposition of the observed period into (not observed) sub‐periods with different duration distributions or different exposures to risk. In the search model presented in this paper a first sub‐period until marriageability is followed by subsequent searching (dating) periods. Parameter estimates from an Austrian female birth cohort sample show a very good coincidence with the observed age distribution. The combination with population heterogeneity is sketched in a final section.