171 – Survival Analysis
Comparison of Survival Analysis and Logistic Regression for Correlated Data
Niloofar Ramezani
University of Northern Colorado
Time is often modeled using Survival Analysis. The ability to consider the time element of event occurrences by proportional hazards models has meant that the logistic regression has played a less important role in the analysis of survival data (Abbott, 1985). This paper, however, discusses the situations in which the censored indicator can be modeled using Correlated Logistic Regression based on the binary nature of it and compares it to the model using Survival analysis which is rarely compared. This paper presents a comparison between Survival Analysis models and Logistic regression models for both independent and correlated observations. Applying these methods, this paper presents an example of the length of stay in the transitional housing facility in the greater Rocky Mountain Region which is an autocorrelated longitudinal data that are subject to both left truncation and right censoring. The results are explained in terms of the comparisons of models mainly based on the significance of the independent variables.