A risk model for addictive behaviors in adolescents: interactions between personality and learning

To determine whether transdiagnostic risk, represented as elevations in one high-risk personality trait, interacts with behavior-specific risk, represented as elevated expectancies for reinforcement from either drinking or smoking, to account partly for early adolescent drinking and smoking behavior.

Multiple regression analysis.

Twenty-three public schools in two school systems in the United States.

A sample of 1897 adolescents tested in the spring of 5th, 6th, 7th, 8th and 9th grades.

Transdiagnostic risk was measured as negative urgency, the tendency to act rashly when distressed, using the UPPS-P child version. Drinking-specific and smoking-specific risk were measured as expectancies for reinforcement from drinking and smoking, using the Memory Model-Based Expectancy Questionnaire (alcohol) and the Adolescent Smoking Consequences Questionnaire (smoking).

There was consistent concurrent prediction from the interactions of (a) negative urgency and alcohol reinforcement expectancies to early adolescent drinking and (b) negative urgency and smoking reinforcement expectancies to early adolescent smoking, above and beyond prediction from the main effects of those variables. In each case, expectancies were more predictive at higher levels of negative urgency. Incremental R² values for main effects ranged from 0.07 to 0.26, and for interactions ranged from 0.01 to 0.03. Prospectively, the main effects predicted subsequent behavior but the interaction effects did not, except in one case.

Among elementary and high school students in the United States, the joint effects of negative urgency and behavior-specific expectancies help to explain drinking and smoking behavior. Joint elevations on the trait and the learning variable account for drinking and smoking behavior beyond the main effects of each predictor. However, there is reason to doubt whether the joint effects predict subsequent increases in drinking and smoking beyond the main effects of those variables.