A Look at Various Estimators in Logistic Models in the Presence of Missing Values
Two commonly used procedures for estimating the parameters of a logistic regression function are the maximum likelihood estimators and the discriminant function estimators. When data are missing, researchers may not be willing to base their estimates on the subset of complete cases. This paper describes four modifications of these procedures for handling the missing-values case. One modification of the discriminant function estimators involves estimating the sample means and correlations from the complete pairs of observed values. An alternative procedure involves replacing the missing entries by zeros and augmenting the logistic regression model with indicator variables for the missing values. Two other modifications require replacing the missing entries by some fitted values based on all other available information. The resulting approximation errors are then accounted for in the covariance structure of the observations.