Electronic statistics textbook banner

Statistical Advisor, How To Calculate Descriptive Statistics for Survival/Failure Times

Use SURVIVAL ANALYSIS. This chapter describes standard life tables as well as Kaplan-Meier product-limit estimates. Methods for fitting specialized distributions (Weibull, Gompertz, linear hazard) to observed survival/failure times are also described.

In the standard life table analysis, the survival or failure times are divided into a user-specified number of intervals. Typically, software programs will then compute various statistics for each interval, for example, the proportion surviving up to the respective interval, the proportion failing in the interval, the hazard rate for the interval (probability of death/failure per unit of time), etc. In addition, the percentiles of the cumulative survival function can be computed.

The Kaplan-Meier product-limit method will compute statistics similar to those in the life table, however, they are computed 'continuously,' that is, for each distinct failure time in the data set. Therefore, results using this method are not affected by the (arbitrary) choice of the number of time intervals to be used for the analysis.