engleski

Influence of Different Methods of Data Imputation on Parameter Estimation – A Monte Carlo Simulation

In this study, we examined the influence of missing data imputation methods on OLS regression analysis parameter estimates. We used two data imputation methods: Deterministic Regression Imputation and Multiple Imputation ; we also estimated parameter values using listwise deletion for comparison. Estimated parameters were compared with regard to precision and bias. Effects of several factors were examined: degree of missingness, average intercorrelation among predictors as well as proportion of missing data in a given set. R software was used to perform a series of simulations and each method was tested using the same correlation matrices. Different methods are compared and practical implications are discussed.

missing data; parameter estimation; OLS regression

nije evidentirano