Figure 3.

Who has the harshest home smoking restrictions? (A) shows the change in predicted probability of a complete home smoking ban with 95% CIs after entering all predictors into a single logistic regression equation. An overlapping CI with zero indicates statistical insignificance. (B) shows the predicted probability of a home smoking ban for high-risk and low-risk profile respondents using estimates from the logistic equation described in A. Low risk was indicated by 25% of familial alters smoking, no smoking friends, not smoking, and 16 years of education. High risk was indicated by 75% of familial alters smoking, most/all smoking friends, smoking, and 12 years of education. In both profiles, age was fixed at 30, gender as male, and they had children. All predictions were produced by simulation using 1,000 randomly drawn estimates from the coefficient covariance matrix.