Examining Effective Factors on Duration Time of Recommitment Using Cox's Proportional Hazard Model

Abstract. Recently, in most of scientific studies, the use of survival analysis is performed for examining duration time models.  One of the important applications of survival analysis is the study of recommitment crime in criminology which has not yet been considered in Iran.  So, with attention to the necessity and importance of predicting recommitment time and the analysis of duration model, in this study 209 individuals who are released from prison are followed and examined for 2 years. Chosen explanatory variables are, age, marriage status, job skill, education, kind of crime, duration of detention and the number of previous arrests. The dependent variable is the duration time till recommitment. The nonparametric methods of Kaplan-Meier and semiparametric model of Cox’s proportional hazard are applied for analysis of the data. Three different Cox’s models are used for predicting hazard of duration time till recommitment.  The results show that age, marriage status and the kind of crime are significant on the hazard of time to recommitment. In the third model we found a significant interaction effect of the kind of crime and marriage status.