Outcome Informed Mental Health Care
Equating mental health scores from different organizations.
1. Develop statistical programming to compare mental health improvement scores of clients' from different mental healthcare organizations.
2. Use single parameter (Rasch) item response theory psychometrics
3. Equate items across organizations by putting a broad sample of organization's clients into a single two-facet model
4. Estimate consistent item measure scores and person measure scores from as many organizations as possible.
5. Rasch person measure scores and item measure scores would be comparable at all sites in the psychometric model.
6. Client improvement rates can now be compared across organizations.
7. Submit a report (with executable computer code): "Are improvement rates in nationally scattered mental health treatment roughly constant, or do some organizations get better results?"
Evaluate mental health outcome using modern longitudinal models.
1. All repeated measurements on a given patient are modeled, not just the first and the last measurement
2. Write statistical in SAS V9 so that it is fully compatible with older CCI code.
3. Develop longitudinal models that are fully up-to-date with current longitudinal practice, e.g.(Singer & Willett, 2003).
4. High quality graphics will be generated by software without user intervention
5. The new programming will not interfere with existing code, e.g. the calculation of residual endpoint model scores currently used to evaluation outcome.
6. Evaluate both linear and non-linear models of client change over time will be evaluated.
7. Develop a Cohen's (Cohen, 1992) d = (X1 - X2) / SDpooled estimate or substitute from the longitudinal model. This will be equivalent to past outcome effect sizes so they can be used interchangeably.
8. Provide a report with annotated SAS code that runs in CCI SAS programs: "Client improvement rates: Comparing two-wave residual endpoints and multi-wave slopes of client improvement."