Full Information Maximum Likelihood

Another advanced missing data method is Full Information Maximum Likelihood. In this method, missing values are not replaced or imputed, but the missing data is handled within the analysis model. The model is estimated by a full information maximum likelihood method, that way all available information is used to estimate the model. In full information maximum likelihood the population parameters are estimated that would most likely produce the estimates from the sample data that is analyzed. 

Examples of models that are often estimated by full information maximum likelihood are structural equation models and multi-level models or growth models. Multiple imputation and full information maximum likelihood will come to similar results when outcome data are missing and the same information is incorporated in a multiple imputation model as in a full information maximum likelihood estimation (Collins, Shafer & Kam, 2001).