Last Observation Carried Forward Versus Mixed Models in the Analysis of Psychiatric Clinical Trials