Validity of alternative self-report indices of smoking among adolescents.

The convergent validity of popularly used open-ended and closed-ended self-report measures of smoking was examined. Carbon monoxide (CO) samples were obtained from 11th-grade Canadian students as an independent method of assessing recent smoking. In addition to CO, 5 known psychosocial correlates of smoking (attitude, subjective norm, risk taking, best friend's smoking, and other friends' smoking) were used to estimate convergence with the self-report smoking indices. Results indicate that both simple closed-ended scales, with only a few response options, and more continuous, open-ended measures performed about equally as well as correlates of CO and the psychosocial measures, but only if the open-ended scales were subjected to a normalizing transformation to optimize their convergence. After this transformation was performed, convergence depended more on the time-span covered by the self-report indices than on the open-ended/closed-ended distinction. Implications of these results for different assessment goals were discussed.