American Statistical Association
New York City
Metropolitan Area Chapter

New York State Psychiatric Institute
at Columbia University Medical Center
Biostatistics Seminar



NEW APPROACHES FOR MATCHING AND ANALYSIS
IN AN OBSERVATIONAL STUDY OF MENTAL HEALTH PARITY

by

Frank Yoon, Ph.D.
Department of Health Care Policy
Harvard University


Abstract

In this talk, I will discuss an ongoing observational study of mental health parity in the Federal Employees Health Benefits (FEHB) Plans and illustrate statistical techniques used in this study. These include (1) recent developments for matching techniques that incorporate the propensity score, namely entire matching and matching with fine balance, (2) how to use multiple control groups to address unobserved biases, and (3) how to analyze multiple outcomes in observational studies with joint models. The basic motivation for these techniques follows here. Matching has as its goal the creation of comparison samples whose observed covariate distributions are similar (stochastically) or the same (fine balance); I will show how one can achieve fine balance with optimal match ratios. The resulting matched design removes bias due to observed covariates and, subject this constraint, efficiently uses the available control pool. Despite this, observational studies are subject to the concern that an important covariate was not measured so there might be unobserved bias. To address this concern I will invoke an older idea of using multiple control groups and illustrate a testing procedure that decomposes a complex hypothesis into smaller ones by testing them in a logical order of priority. Finally, multiple outcomes are with increasing frequency included in observational studies of policy interventions; for example, in the FEHB study, outcomes for intensity of therapy, medication management, and prescriptions are included. Joint models can capitalize on their correlations and provide more powerful tests of intervention effects, while adjusting for multiple testing. I will show how to analyze multiple outcomes in the FEHB study with recently developed software in SAS.

Biographical Note

Dr. Frank Yoon is a postdoctoral fellow in statistics at the Department of Health Care Policy, Harvard Medical School. His current research interests include matching and the propensity score, the analysis of multiple outcomes in mental health, and the use of geographic variation in treatment rates for cancer. While not pursuing his research, he spends a lot of time playing music and cooking. Before postdoctoral training, he received his PhD in statistics in 2009 from the The Wharton School, University of Pennsylvania under the mentorship of Paul Rosenbaum.


Date: Tuesday, January 25, 2011
Time: 3:30 - 4:30 P.M. Please note new time.
Location: New York State Psychiatric Institute
1051 Riverside Drive
New PI 6th Floor Boardroom (6601)
New York, New York
(Directions)

RESERVATIONS ARE NOT REQUIRED

Please note the new times.
Coffee: 3:00 to 3:30 P.M.
Reception: 4:30 to 5:00 P.M.


Home Page | Chapter News | Chapter Officers | Chapter Events
Other Metro Area Events | ASA National Home Page | Links To Other Websites
NYC ASA Chapter Constitution | NYC ASA Chapter By-Laws

Page last modified on January 19, 2011

Copyright © 1998-2011 by New York City Metropolitan Area Chapter of the ASA
Designed and maintained by Cynthia Scherer
Send questions or comments to nycasa@mindspring.com