Table 4

Logistic regression predicting support for comprehensive indoor smoking bans (n=1 02 831)

Overall (no themes) modelThemes modelled separatelyThemes modelled together (Model 5 emerging as the most stable, parsimonious and preferred model)
Independent variableModel 1
B (SE)
Model 2
B (SE)
Model 3
B (SE)
Model 4
B (SE)
Model 5
B (SE)
Model 6
B (SE)
Market-level volume of exposure to state antismoking ads— adult/general targeted0.10***
(0.005)
-----
Market-level volume of exposure to state antismoking ads— adult/general themes
 Efficacy appeal-0.08***
(0.008)
−0.14***
(0.019)
---
 Health consequences to others-0.18***
(0.010)
0.11***
(0.019)
0.17***
(0.017)
0.11***
(0.013)
-
 Health consequences to self-0.12***
(0.010)
0.03
(0.023)
−0.09***
(0.017)
-0.02
(0.013)
 Anti-industry appeal-0.18***
(0.011)
0.19***
(0.018)
0.10***
(0.013)
0.09***
(0.013)
0.14***
(0.013)
 Irrationality/addiction appeal-0.23***
(0.016)
0.20***
(0.025)
0.08***
(0.019)
0.08***
(0.019)
0.12***
(0.018)
Market-level volume of exposure to state antismoking ads—youth targeted0.00
(0.013)
Full set of control variables included in all models
Market-level volume of exposure to
truth campaign antismoking ads
0.00
(0.001)
% of state GDP generated from tobacco farming−0.13
(0.155)
Excise tax ($) per pack0.10**
(0.034)
Index of laws restricting youth access to tobacco0.01**
(0.001)
Per capita tobacco control funding0.00
(0.004)
No of location bans on cigarette use0.02
(0.010)
Sociodemographic control variables included
Max VIF with multiple exposure measures
(theme with highest VIF)
1.61 (adult)-8.65
(efficacy)
4.12
(other)
2.31
(other)
1.85
(self)
  • *p<0.05; **p<0.01; *** p<0.001.

  • Model 1 presents coefficients for a model that includes the overall volume of adult/general-targeted state antismoking ad exposure; Model 2 presents coefficients for separate models in which each state antismoking ad theme variable was the only state ad exposure variable included; Models 3 through 6 present coefficients from models in which various combinations of state antismoking ad theme exposure measures were entered simultaneously in the same model.

  • † Demographic control variables include dummy variables for categories of age, gender, race, ethnicity, citizenship status, marital status, educational attainment, household income, employment status, student status, number of children living in the household, smoking status, previous (ever) quit attempts and recent (past year) quit attempts.

  • VIF, variance inflation factor.