Table 2.
Linear Mixed Models to Determine the Variables Associated With the Five Ad Outcomes for Each Type of Message Frame, With Gender Weights
| Variable | Recall | Like | Action | Think | Emotion | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Loss ( n = 1393) | Gain ( n = 915) | Loss ( n = 655) | Gain ( n = 506) | Loss ( n = 671) | Gain ( n = 507) | Loss ( n = 655) | Gain ( n = 504) | Loss ( n = 668) | Gain ( n = 508) | |
| Age | −0.04 (NS) | −0.17 (.020) | 0.08 (NS) | 0.06 (NS) | <0.01 (NS) | −0.06 (NS) | 0.02 (NS) | 0.06 (NS) | 0.02 (NS) | <0.01 (NS) |
| Gender | ||||||||||
| Male | −4.51 (.001) | 1.88 (NS) | −5.86 (.004) | −3.77 (NS) | −7.08 (.034) | −9.79 (.011) | −2.41 (NS) | −7.63 (.004) | −3.27 (NS) | −13.66 (<.001) |
| Female | ref | ref | ref | ref | ref | ref | ref | ref | ref | ref |
| Education | ||||||||||
| <High school | 0.40 (NS) | −11.11 (<.001) | 3.58 (NS) | 1.66 (NS) | 16.06 (.002) | 9.60 (NS) | 8.53 (.023) | −0.79 (NS) | 12.57 (.010) | 11.33 (NS) |
| ≥High school | ref | ref | ref | ref | ref | ref | ref | ref | ref | ref |
| Smoking status | ||||||||||
| Smoker | −1.56 (NS) | −3.97 (NS) | −3.53 (NS) | −11.11 (<.001) | −3.12 (NS) | −9.29 (NS) | −10.40 (.005) | −8.10 (.024) | −3.76 (NS) | −4.60 (NS) |
| Nonsmoker | ref | ref | ref | ref | ref | ref | ref | ref | ref | ref |
| Knowledge of SHS exposure risks | 2.13 (.002 ) | 1.50 (NS) | 8.43 (<.001) | 8.32 (<.001) | 3.57 (.030)* | 4.38 (.018) | 10.82 (<.001) | 9.32 (<.001) | 8.13 (<.001) | 6.17 (<.001) |
| Support for SF workplace policy | 0.87 (NS) | 1.66 (NS) | 11.30 (<.001) | 12.16 (<.001) | 4.09 (.028) | 1.26 (.56) | 6.35 (<.001) | 9.22 (<.001) | 8.23 (<.001) | 8.86 (<.001) |
SHS = secondhand smoke; SF = smoke-free; NS = non-significant. Only those who recalled the ad were asked to give their opinions about it. The full sample ( N = 1518) were asked about recall of loss-framed ads, but only 1006 were asked about recall of gain-framed due to ad selection in County C. Although there were relatively few missing values for any single variable, only participants with complete information on all variables were included in the mixed models.