How well does the index of receptivity to tobacco industry promotion discriminate between smoking and never smoking adolescents.

Tobacco advertising is often named as the culprit that causes children to start smoking (Lancaster & Lancaster, 2003). This belief can partly be attributed to the Index of Receptivity to Tobacco Industry Promotion (IRTIP) developed by Evans, Farkas, Gilpin, Berry, & Pierce (1995). IRTIP was later modified and used by Pierce, Choi, Gilpin, Farkas, & Berry (1998) in a longitudinal study that claimed to have found a causal link between advertising and adolescent cigarette trial. The model is advertised by the American National Cancer Institute (2004) as being able to measure the likelihood of an adolescent starting smoking. Because of Pierce’s causality claim and this endorsement, IRTIP has been widely adopted by tobacco-control researchers. Consequently, the results from IRTIP based surveys have played a central role in influencing tobacco control policy. Based on the logic that a model used to predict the chances of a non-smoker becoming a smoker should be able to distinguish between these two groups, discriminant analysis with dummy coded variables was used to validate IRTIP. The results show that while IRTIP classifies never-smokers well, it grossly misclassifies smokers. This leads to questions about the validity of the model and of studies using IRTIP.