NCJ Number
157451
Date Published
1995
Length
8 pages
Annotation
This paper presents an overview of a prediction approach for drug prevention called Squashing Theory (Buscema, 1985, 1990, 1994a, and 1994b).
Abstract
The model is an analytical approach to predict behavior. In order to predict a behavioral trend by using this approach, users must measure multiple characteristics and incorporate variables that range from those that are more relevant to those that are supplementary. Characteristics that are relevant for one subject have the possibility of emerging as characteristics across subjects through network interactions. Following a summary of the Squashing Theory and methodology, this paper lists the 14 variables included in the analysis. They are scholastic characteristics, employment, parental and sibling characteristics, living conditions, sexual and partner characteristics, religious and behavioral characteristics, economic condition, use of alcohol and tobacco, legal involvement, friendships, use of free time, psychological characteristics, perceptions of the family and partner, and drug use. The Artificial Feed Forward Neural Network, a recently developed computer architecture inspired by the brain's structure (Dayhoff, 1990), is the framework for Squashing Theory. The Network was computer programmed by the Semeion Research Center in Rome, Italy. The model was able to predict drug behavior at the 92-percent level on prototypical cases and at the 80-percent level on uncertain cases based on self-reported drug use from two norming samples and the prediction sample. 10 references