Stronger Health Through Abuse Reduction and Prevention (SHARP) Program: Predictive Ability of the History Opinion Inventory-Revised (HOI-R).
Final rept. for May 96,
ARMSTRONG LAB BROOKS AFB TX
Pagination or Media Count:
This study evaluated the History Opinion Inventory-Revised HOI-R, currently given to all incoming recruits in basic military training, for its ability to predict future substance abuse events. There were 63,074 valid data records in the HOI-R questionnaire files, all of which were used in the exploratory factor analysis. Of these 63,074 records, only 19,306 passed the inclusion criteria for further logistic regression analysis. Factor analysis confirmed a seven factor model similar to the original ten factor model upon which the HOI-R was based. Multivariate regression analysis indicated the most significant predictors in order of subsequent substance abuse events were extroversion, high school performance, marital status, gender and age. Although regression analysis suggested a good prediction model, a classification table for the obtained prediction model suggested it was of little practical use. The model had a very high false-positive rate and would identify too many recruits for treatment who would not require it, therefore wasting much time and money. Although the obtained prediction model was not of applied use, the information obtained in this preliminary study suggested several research directions, including the creation of a new, unique substance abuse prediction tool, to be validated with Air Force recruits.
- Sociology and Law
- Medicine and Medical Research
- Statistics and Probability