Table 4

 Analysis of robustness of main study findings: adjusted odds ratios* for effect of strong† local restaurant smoking regulations on progression to established smoking‡

OR* (95% CI)
*Odds ratio for progression to established smoking. Odds ratio is derived from GEE logistic regression model that accounts for clustering of responses within towns. Data are weighted to account for baseline probability of respondent selection. Odds ratios are adjusted for all other variables in the analysis.
†Strength of local restaurant smoking regulation was defined as strong if it banned smoking completely in all restaurants with no variances.
‡Progression to established smoking was defined as smoking 100 or more cigarettes in one’s life.
Primary analysis0.39 (0.24 to 0.66)
Results when all youths who moved (from baseline to follow up) (n = 103) are excluded0.40 (0.24 to 0.68)
Results when sample weights are excluded0.39 (0.24 to 0.64)
Results when all observations with missing data (n = 539) are deleted (stepwise elimination)0.34 (0.19 to 0.60)
Results when all ever smokers (at baseline), including puffers, (n = 589) are excluded0.26 (0.10 to 0.68)
Results when all ever smokers and susceptible non-smokers (n = 1042) are excluded0.07 (0.01 to 0.52)