Table 3

Association between campaign exposure and campaign-targeted knowledge of smoking harms

Knowledge measureCampaign exposurePre-campaign meanPost-campaign meanPoisson regression b (SE)
Knowledge index of smoking harms including stroke, lung cancer in smokers and cardiovascular diseaseModel 1
 Control cities1.371.63refref
 Intervention cities1.381.670.013 (0.034)0.045 (0.038)
Model 2
 Not recalled1.361.62refref
 Recalled the campaign1.511.970.168 (0.043)***0.147 (0.043)**
Model 2
 Not recalled1.361.62refref
 Recalled† 1 channel1.511.890.124 (0.046)*0.093 (0.043)*
 Recalled 2 and more channels1.452.240.301 (0.069)***0.259 (0.066)***
Model 4
 Not recalled‡1.361.62refref
 Recalled TV1.512.020.131 (0.044)**0.106 (0.040)*
 Recalled poster1.302.100.152 (0.074)*0.0109 (0.082)
 Recalled mobile media1.462.140.089 (0.071)0.087 (0.079)
  • Adjusted for age group, sex, income, education, marital status, smoking status, intention to quit, exposure to general anti-smoking campaign and survey time from the campaign end until the time of the post-campaign survey.

  • Statistically significant levels for ordinal regression: *p<0.05; **p<0.01; ***p<0.001.

  • †Participants (n=51) who recalled the campaign but did not report any channel through which they saw the campaign were removed from the analysis. ‡This analysis is a single model in which dummy variables were created for each channel, with no recall of any exposure to the campaign as the reference group.

  • TV, television.