Substance abuse in states: A methodological report on model based estimates from the 1994-1996 national household surveys on drub abuse
Folsom, R. E., Shah, B., & Vaish, A. K. (1999, January). Substance abuse in states: A methodological report on model based estimates from the 1994-1996 national household surveys on drub abuse. Presented at American Statistical Association, Section on Survey Research Methods, .
This paper summarizes the 1994-1996 small area estimation project completed recently for the Substance Abuse and Mental Health Services Administration. The main purpose of this project was to correct deficiencies in the previous NHSDA small area estimation exercise which used the penalized quasi-likelihood (PQL) estimation approach. In the current project, hierarchical Bayes estimation methods employing Gibbs samplingwere used. A secondary purpose was to test the new methodology on pooled 1994-1996 data. This methodologicaltest was not designed to published small area estimates but to begin preparation for the 1999NHSDA. Beginning in 1999 SAMHSA will require that small area prevalence estimates be produced on a tight schedule for states and age groups for up to 20 drug use related binary outcomes.