Survival Analysis

Survival analysis is a collection of statistical methods that are used to describe, explain, or predict the occurrence and timing of events. The name survival analysis stems from the fact that these methods were originally developed by biostatisticians to analyze the occurrence of deaths. However, these same methods are perfectly appropriate for a vast array of social phenomena including births, marriages, divorces, job terminations, promotions, arrests, migrations, and revolutions. Other names for survival analysis include event history analysis, failure time analysis, hazard analysis, transition analysis, and duration analysis. Although some methods of survival analysis are purely descriptive (e.g., Kaplan-Meier estimation of survival functions), most applications involve estimation of regression models, which come in a wide variety of forms. These models are typically very similar to linear or logistic regression models, except that the dependent variable is a measure of the timing or rate of event occurrence. A key feature of all methods of survival analysis is the ability to handle right censoring, a phenomenon that is almost always present in longitudinal data. Right censoring occurs when some individuals do not experience any events, implying that an event time cannot be measured. Introductory treatments of survival analysis for social scientists can be found in Teachman (1983), Allison (1984, 1995), Tuma and Hannan (1984), Kiefer (1988), Blossfeld and Rohwer (2001), and Box-Steffensmeier and Jones (2004). For a biostatistical point of view, see Collett (2003), Hosmer and Lemeshow (2003), Kleinbaum and Klein (2005), or Klein and Moeschberger (2003). Specific desiderata for applied studies that use survival analysis are presented in Table 31.1 and later explained in detail.