METODE MULTIPLE IMPUTATION UNTUK MENGATASI KOVARIAT TAK LENGKAP PADA DATA KEJADIAN BERULANG

Abstract

Multiple imputation is one of estimation method used to impute missing observations. This method imputes missing observation several times then it is more possible to get the right estimate than just one time imputation. In this research, the method will be applied to estimate missing observations in covariates of recurrent event data. Some multiple imputation methods will be considered including combination of the event indicator, the event times, the logarithm of event times, and the cumulative baseline hazard. To compare these methods, Monte Carlo simulation will be used based on relative bias and Mean Squared Error (MSE). The recurrent events will be modelled using Cox proportional hazard model. Furthermore, real data application will be presented.