Do attitudes predict uptake of smoking in teenagers? Case not proven

A recent paper by Charlton and Blair (Soc. Sci. Med. 29, 813-818, 1989) reported a large prospective study on smoking in 12 and 13 year-old schoolchildren. The main new findings were that attitudes, particularly among girls, were important predictors of the uptake of smoking in never-smokers. The present paper presents a critique of this study and argues that the findings may not reflect true predictive relationships. Data are presented from OPCS-conducted national surveys which indicate that a substantial minority of children who initially classify themselves as never-smokers admit to having tried a cigarette when prompted. These undisclosed triers differ from confirmed never-smokers on relevant predictor variables, and are significantly more likely to agree that they will try a cigarette again before leaving school (12% vs 4%, P less than 0.005). A similar group would be expected in Charlton and Blair's study and might wholly or partially account for the apparent predictive relationships found. This is first because such undisclosed triers might be more accurate in their self-reports when questioned a second time, but also because this group would contribute disproportionately to those trying smoking in the follow-up period. Other issues raised concern the homogeneity of the attitude variables employed, and the difficulty of attributing any specific importance to particular attitude items.