Application of the Survival Trees for Estimation of the Propensity to Accepting a Job and Resignation from the Labour Office Mediation by the Long-Term Unemployed People

Abstract

The obstacles in finding a job by the long-term unemployed people are their behaviours resulting from cognitive and emotional mistakes. Long-term unemployment results in depreciation of the human capital and discouragement to further job searching. In order to lead the effective social policy, identification of threatened group is essential. The goal of the research was estimation of the influence of gender, age and education on the probability of exit from the long-term registered unemployment and resignation from the labour office mediation. Due to the fact that there were censored observations, survival analysis methods were used. Survival trees were built by means of the Kaplan–Meier estimators, and the statistics of the log-rank test were used as splitting criteria. They are the example of methods of recursive binary partitioning, which aim in creation of homogeneous subsets with respect to the analysed response variables. In the analysis, the conditional inference trees were used.