Deep Blue (University of Michigan) · 1984 · 14 citations · 0 references
Open access
Public HealthBiostatisticsStatistical ModelingStatisticsMissing Data ComputationMissing Data HandlingMissing Data ImputationMissing Data MethodsModel FittingRegressionMissing Completely At RandomMissing DataMissing Data AnalysisMissing Data InferenceMissingness Mechanisms
In this paper it is argued that maximizing the complete data (log) likelihood function with respect to the missing data and the unknown parameters will not improve the efficiency of the estimators but may affect consistency instead. If no auxiliary relations are available or additional assumptions are made, the maximum likelihood estimator based on the observed data is (asymptotically) the most efficient estimator.