Observation Schemes Giving Incomplete or Selective Data

Thus far we have assumed that the event history process can be observed continuously over some followup period for each individual, and it has been assumed that the selection of individuals and their followup periods are independent in the sense described in Sections 1.4 and 2.6, and thus ignorable. In some settings these conditions are violated; for example, an individual may be observed only intermittently, or loss-to-followup may not be independent in the sense needed for ordinary analysis. In addition, individuals may be selected for a study because their event history satisfies some condition. In the following sections we discuss how to deal with these issues.