Júnior, Luiz Carlos SantosTsúnemi, Miriam Harumi [UNESP]2023-07-292023-07-292019-01-01Revista Brasileira de Biometria, v. 37, n. 2, p. 205-228, 2019.1983-0823http://hdl.handle.net/11449/248868The public servants' pensions need the development of research that contributes to a better understanding of the variables related to the receipts and payments of their Institutes. The present study aims to analyze the time of municipal server stays in the public service, based on information provided by Institute of Social Security of the Municipal Servants of Cabedelo (Paraíba / Brazil) and through the use of survival models that incorporate competitive risks. Thus, the functions of cumulative risk, survival and cumulative incidence were estimated, as well as the effects of four covariables on the time of permanence, considering the presence of five competing causes to remove the server from the public service. Among the main results, it has been shown that the proportional models were considered adequate, being that: the older the server, and the earlier the date of admission to the service, greater the chance of leaving the service for any of the causes; being a man in relation to women, as well as being a teacher, in relation to other categories, can increase or decrease the risk, depending on the specific cause considered. Once the biometric behavior of the analyzed group has been mapped, better predictability conditions for future commitments are provided.205-228porCompetitive risks under the multistate approachInstitute of Social Security of the Municipal Servants of CabedeloSurvival analysis for censored dataTime of stay in the public serviceArtigo10.28951/rbb.v37i2.3742-s2.0-85160041850