End of Award Report Multilevel Multiprocess Models for Partnership and Childbearing Event Histories

Event history data provide a longitudinal record of the timing of events such as partnership formation and dissolution, births, and changes in employment and housing. Typically outcomes of one process will influence the occurrence of events in another process. For example the presence of children, who constitute prior outcomes of the fertility process, is often found to be negatively associated with the risk of union dissolution. Previous researchers have explored the relationship between childbearing and dissolution by including the number of children as a covariate in a model for dissolution. This approach, however, ignores the possibility that decisions about childbearing and partnerships are subject to shared influences, some of which will be unobserved. In other words, fertility outcomes may be endogenous with respect to partnership transitions. Lillard and Waite (1993) developed a simultaneous equations model to allow for the joint determination of marital stability and fertility in the US. Using this approach they found that some women were more likely than others to have unstable marriages (due to unmeasured time-invariant characteristics), and that such women were less likely to have children during marriage. Failure to take into account the simultaneity between processes led to biased estimates of the effects of having children on partnership transitions.