Statistics Glossary → Full information maximum likelihood (FIML)
Full information maximum likelihood (FIML)concept
An estimation method for data with missing values in which each case contributes to the likelihood only through its observed variables, so incomplete cases are not discarded.
Under listwise deletion, a case missing any variable is dropped from the analysis entirely, which loses information and introduces bias if the missingness is not random. With FIML, each case's likelihood is computed from the distribution the model implies for the variables observed in that case, and the contributions of all cases are summed. Missing values are not filled in; the estimation instead finds directly the parameters consistent with the observed information.
FIML gives consistent and efficient estimates under the assumption that missingness may depend on observed data but not on the unobserved values (MAR); under completely random missingness (MCAR) it is more efficient than listwise deletion. If missingness depends on the unobserved value itself (MNAR), bias can remain. Adding auxiliary variables that explain the missingness, even if they are not part of the model, makes the MAR assumption more plausible. In large samples FIML and a correctly specified multiple imputation give similar results.
An everyday example
In a survey, 120 of 500 people left the income question blank. Listwise deletion throws away these 120 people's answers on age, education and satisfaction too. FIML keeps using the answers they did give and simply leaves out the income information, since it is not there.
In MerQur
The 'Missing data' option (Listwise deletion, FIML) in Confirmatory Factor Analysis, Structural Equation Modeling and Path Analysis; the 'Missing indicators' option (Listwise deletion, FIML) in Latent Class Analysis.
Analyses where this concept appears
Latent Class Analysis (LCA)Confirmatory Factor Analysis (CFA)Structural Equation Modeling (SEM)Path Analysis
Related concepts
Listwise and pairwise deletionMissing-data mechanisms (MCAR, MAR, MNAR)