We averaged Y-27632 2HCL responses to these seven items to create a measure of retrospective recall of exposure to protobacco marketing and media (�� = .69). Smoking Intentions At the end-of-study survey, participants indicated their future intentions to smoke by completing a three-item scale adapted from Choi, Gilpin, Farkas, and Pierce (2001) and shown to be predictive of smoking initiation. The items are, ��Do you think you will try a cigarette anytime soon,�� ��Do you think you will smoke a cigarette anytime in the next year,�� and ��If one of your best friends offered you a cigarette, would you smoke it?�� Responses were made on a 1 (Definitely not) to 10 (Definitely yes) scale and averaged to create a measure of intention to smoke (�� = .94).

Following standard practice, we dichotomized this measure such that youth who had anything but a firm commitment not to smoke (i.e., a score >1.00) were considered at risk for future smoking (Pierce, Choi, Gilpin, Farkas, & Merritt, 1996). Results Sample Characteristics Table 1 shows the distribution of the sample on several demographic characteristics and on current smoking status. Ever-smokers comprised 61% of our sample; of those, 85% (n = 70) were experimental smokers who had not smoked in the past month and 15% (n = 13) were nonestablished regular smokers who smoked on 6 days (SD = 4.4) in the past month and smoked an average of 2.2 cigarettes (SD = 1.3) on the days that they smoked. Validity of EMA as a Measure of Exposure to Tobacco Advertising and Media The zero-order correlation between participants�� retrospective recall and EMA-based measure of exposure was .

37. Thus, less than 14% of the variation in the EMA-based measure of exposure was explained by the retrospective recall measure. To further characterize the association between retrospective recall and EMA, we grouped participants into low, medium, and high tertiles on the retrospective recall and EMA-based measures of exposure and then cross-tabulated these two categorical (low�Cmedium�Chigh) indices. Fifty-eight percent of participants were classified in different groups by the two measures; among those, 26% were classified into extreme opposite categories (e.g., participants classified as low exposure on the basis of retrospective recall were classified as high exposure based on EMA). Thus, Batimastat retrospective recall and EMA appeared to measure different aspects of exposure to tobacco advertising and media. To compare the predictive validity of the EMA and retrospective recall measures of exposure to protobacco marketing and media, we used separate logistic regression models to estimate the association between amount of exposure and participants�� intention to smoke at the end of study.