A family.In Survival analysis , data deficiency often

A common problem faced in most research area nowadays is the problem of missing data.As such there is an increasingly awareness of theses problems and biases which can because by it. With the presence of missing data, an element of ambiguity is introducedinto data analysis. They can affect the properties of statistical estimators such as themean, variance or percentages which may result to misleading conclusions as well representloss of statistical power (Peter Schmitt et al, 2015). The underlying goal of anystatistical analysis to make valid inferences regarding the population of interest which;with the advent of missing data is been threaten since it create s a biased sample,thatmakes the sample different from the population from which it was drawn (Wayman,2003). There are three major approaches in handling missing data namely the deletionbased approach, simple replacement methods where missing observations are replacedwith plausible values based on available data and model based methods; where missingobservation are replaced by values estimated from models that makes use of data forall variables in the dataset (A M G Ali et al, 2011). The most commonly used methodfor handling missing data is one which edits the missing data to generate a completedataset popularly known as complete case analysis and usually the default in most statisticalpackage. However, this method is subjected to severe drawbacks such as theintroduction of bias in analysis and loss of substantial information. This paved way forthe development of many imputation methods. Hence it serves as alternative approach toderive complete dataset on which complete data methods can be applied. It operates byfilling in missing values rather than deleting incomplete sequences (Molenberghs, 2005).Multiple imputation was proposed by Rubin(1987) which is a commonly used methodto impute missing data. The underlying concept with multiple imputation is to replacemissing values with a number of plausible values based on the distribution of observeddata. This study consider and compare three different imputation method along withsurvival analysis modeling with parametric distributions of the location and scale family.In Survival analysis , data deficiency often occurs in the covariates since the survivaloutcome always consist of easy to measure data such as death or time to event beingpresent in registeries (Huo Zhao, 2015). As such, this study considers missing valueson covariates in the dataset for patients with cancer of the larynx which has as survivaloutcome, the time to death .