To impute missing data using a random generator.

To update the parameter estimate by maximizing the expected log-likelihood calculated in the E-step.

To calculate the gradient of the Fisher information matrix.

To perform a significance test for the model parameters.