Background: In longitudinal smoking prevention studies, a difficulty in evaluating treatment effects is understanding whether bias is associated with those who do not complete the study. This study presents the significant predictors of attrition and suggests how to reduce attrition bias in evaluating program effects.
Methods: Survival analysis methods were used to assess factors associated with attrition at different time points of the study.
Results: Results of the analysis indicate that those who drop out tend to be of lower academic achievement, have lower tobacco and health knowledge, and have lower social influence/resistance skills knowledge, and are more likely to be smokers and to be marijuana users. Blacks are more likely to drop out than the other ethnic groups. Gender is not a significant predictor for dropout. The dropout rates among the treatment conditions are significantly different.
Conclusions: The findings of this study demonstrate that attritions in longitudinal smoking prevention studies are not at random. By considering the characteristics of dropouts, one can reduce attrition bias using available procedures and can take appropriate strategies for reducing dropout rates in future smoking prevention studies.