To calculate the final maximum likelihood estimator.

To calculate the expected value of the log-likelihood function with respect to the missing data, given the current parameter estimates.

To calculate the variance of the observed data.

To maximize the observed log-likelihood directly.