Zzoooom through counseling? No data for you.

The simple way of explaining the goal of Nisha C. Gottfredson, Daniel J. Bauer, Scott A. Baldwin and John C. Okiishi’s experiment is to say, it involves a Shared Perimeter Mixture Model (SPMM) and missing data in a naturalistic study of psychotherapy treatment. The goal is to get accurate estimates of change over time from enrolled psychotherapy patients and then to see expected rates of improvement vary depending on patient’s diagnosis. Missing data does not always mean the data of the patient is literally lost, it simply means the patient graduated through treatment a little too fast and the analyst does not have enough data in the patient’s recovery to foresee to how the patient will do outside of therapy. This could happen if a patient zooms through therapy and could possibly pretend to be at a better state in recovery then they actually are, or even if they are just a quick learner. The data could be missing at random or not at random, for example one of these patients who could be a quick learner and improve quickly through treatment is missing data because they left the program much quicker then a patient who improved more slowly. Therefore, this could lead to a bias. This leads us to use a tool called a Shared Perimeter Mixture Model (SPMM) that is a semiparametric model that can be used to find results approximated during estimation. This model takes the association between the missing data patterns of the patient with the growth trajectory using latent classes. After using the Shared Perimeter Mixture Model (SPMM), they concluded that the average patient recovers more quickly from negative psychological symptoms than they would the normal suggestion by standard growth models.In the end, the data proves the longer a patient stays in therapy, the easier it is to collect data and predict how they will do in the recovery process, or the psychological functioning over time. 

Abstract

Gottfredson, N., Bauer, D., Baldwin, S., & Okiishi, J. (n.d.). Using a Shared Parameter Mixture Model to Estimate Change During Treatment When Termination Is Related to Recovery Speed. Journal of Consulting and Clinical Psychology, 82(5), 813-827.