The Army STARRS: Statistical Designs And Methods For Predicting Non-fatal Suicidal Behaviors & Reducing Response Bias in Estimating Prevalence of Mental Disorders

• Activity Number: 296
• Type: Topic Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Mental Health Statistics Section
• Author(s): Tzu-Cheg Kao* and Steven Heeringa and Alan Zaslavsky and James Naifeh and Pablo Aliaga and Patti Vegella and Tsz Ng and Bailey Zhang and Christina Buckley and Carol Fullerton and Gary Wynn and James McCarroll and Nancy Sampson and Lisa Colpe and Michael Schoenbaum and Kenneth Cox and Ronald Kessler and Murray Stein and Robert Ursano
• Companies: Uniformed Services University of the Health Sciences and University of Michigan Institute for Social Research and Harvard Medical School and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and CSTS, USUHS and Harvard Medical School and NIH/NIMH and NIH/NIMH and U.S. Army Public Health Command and Harvard Medical School and UC San Diego/VA San Diego Healthcare System and CSTS, USUHS
• Keywords: retrospective case-control studies ; longitudinal study ; non-fatal suicidal behaviors ; weight adjustment ; mental health ; big data

Abstract

We will use data from some components of The Army STARRS to illustrate statistical designs and methods used to investigate non-fatal suicidal behaviors and reduce response bias if applicable. To investigate non-fatal suicidal behaviors, like suicide attempt (SA), we will present related design and statistical methods by using some elements of the Historical Administrative Data Study (HADS). Additionally, based on the early data from the All Army Study Survey (AAS), we will present the statistical methods used to reduce response bias for self-administered questionnaire (SAQ) in estimating prevalence of mental disorders. Some main findings will be presented. Challenges in managing data, selecting variables for statistical models, recognizing and reducing response bias will be discussed.