Retrospective analysis in smoking cessation research.

Abstract

The use of retrospective data in smoking research represents possible sources of bias resulting from: (1) baseline incomparability of current, former, and non-smokers, (2) instability of parameters under investigation, and (3) difficulties in recall. A self-report measure of smoking motives was employed in a longitudinal design to examine bias associated with retrospective data analysis. Thirty-six pairs of former smokers and recidivists (smokers who had tried to quit) were matched on age and amount smoked. Although initial responses for both groups were comparable, former smokers were significantly higher on smoking motive factors after quitting compared to recidivists who stayed essentially the same. The findings suggest that mean differences in smoking motives between recidivists and former smokers assessed cross-sectionally may not accurately reflect the magnitude of the differences that existed when both groups were smoking.