zur analyse von zeitintervallen unter beruecksichtigung unbeobachteter heterogenitaet: anwendungen auf rueckfallintervalle nach der haftentlassung und die dauer der arbeitslosigkeit

摘要

abstract: transition rate models are very appropriate for the analysis of social processes as demonstrated by the work of coleman , hannan, tuma and other authors. in recent years these models became increasingly popular in empirical social research concerned with time related data. in particular, there is a great need for techniques which are able to deal with the problem of observed and unobserved heterogeneity. in this paper some characteristics of a model with unobserved heterogeneity are analyzed. it is assumed that the hazard rate depends on a gamma distributed error term. the model is applied to recidivism and unemployment data and compared with alternative hazard rate models. estimation of parameters is performed with tuma's programrate. important implications of the model are derived which allow clear and unambigious interpretations of estimated parameters.;