Table 2

Association between campaign exposure and campaign-targeted attitudes towards cigarette gifts

Attitude measure	Campaign exposure	n	Per cent	Logistic regression ORs (95% CI)
Attitude measure	Campaign exposure	n	Per cent	Unadjusted	Adjusted
Disagreeing that cigarettes are good gifts for friends and family	Model 1
	Control cities	965	55	1	1
	Intervention cities	2333	60	1.22 (0.86 to 1.73)	1.59 (1.11 to 2.27)*
	Model 2
	Not recalled	2043	58	1	1
	Recalled the campaign	282	71	1.75 (1.28 to 2.41)**	1.63 (1.18 to 2.25)**
	Model 3
	Not recalled	2043	58	1	1
	Recalled† 1 channel	173	75	2.09 (1.25 to 3.50)**	2.00 (1.22 to 3.29)**
	Recalled 2 and more channels	58	60	1.05 (0.43 to 2.55)	0.84 (0.34 to 2.09)
	Model 4
	Not recalled‡	2043	58	1	1
	Recalled TV	190	70	1.62 (1.02 to 2.55)*	1.51 (0.96 to 2.37)
	Recalled poster	54	62	0.84 (0.49 to 1.42)	0.80 (0.44 to 1.47)
	Recalled mobile media	67	71	1.32 (0.67 to 2.60)	1.19 (0.62 to 2.26)

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 logistic 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.