On the Impact of Socioeconomic Status on Mortality in the Presence of Interval Censored Status Change

Göran Broström, Umea University
Joakim Malmdin, Umea University

In demographic studies of the impact of socio-economic status change on adult mortality, one often encounters the problem that the exact timing of such a status change is not recorded. The status is often reported only at the same time as a vital event, like marriage, birth of a child, etc. This means that the exact time of status change is interval censored. Nevertheless, in most practical work, this problem is neglected, and the date of change is set equal to the first date with the new status recorded. In a study on data from the Demographic Data Base, Umeå University, Sweden, neglecting the interval censoring lead to severe bias in the estimation of the effect of status on mortality. Two methods to handle interval censored data are presented, an imputation method, and the EM algorithm. A simulation study evaluates the performances of the suggested methods.