Goodness-of-Fit Testing in Survival Analysis: Imputation vs Adaptation for Censored Data

Abstract

In survival analysis, the focus is typically on the time until a certain event occurs, such as survival time after surgery or another medical treatment. Since studies are often time-limited and participants may leave the study for various reasons, the issue of randomly right-censored data becomes a significant challenge. When classical complete-case testing procedures are applied in such situations, their stability and reliability are compromised—they may or may not yield accurate results, and the factors affecting their performance are not known. Therefore, modifications are necessary: either the testing procedure itself must be adapted to account for censoring information or missing values must be imputed before applying the standard procedure. In this talk, we will present the latest advancements in both of these approaches for different tests and discuss the pros and cons of each method.