Full Information Maximum Likelihood (FIML)

We now explore another approach to dealing with missing data, based on the maximum likelihood function and used in logistic regression.

The basic premise is that instead of imputing the values of missing data, we try to estimate the value of some population parameter by determining the value that maximizes the likelihood function (actually the natural log of this function) based on the sample data that we have.

Topics

References

Enders, C. K. (2001) The performance of the full information maximum likelihood estimator in multiple regression models with missing data. Educational and Psychological Measurement, Vol. 61 No. 5.
https://asu.elsevierpure.com/en/publications/the-performance-of-the-full-information-maximum-likelihood-estima

Allison, P. D. (2012) Handling missing data by maximum likelihood
https://statisticalhorizons.com/wp-content/uploads/MissingDataByML.pdf

5 thoughts on “Full Information Maximum Likelihood (FIML)”

  1. Can I use FIML in a panel random effect model with N = 13 and T= 19. I have missing data about 40%. So, how can I deal with missing data without pairwise/listwise deletion? I tried xtdpdml but it is not executable with my data. An online consultant told me it is better when t < 10 and large N, which is not my case at all.
    thanks in advance

    Reply

Leave a Comment