Data are from the International Tobacco Control (ITC) China Survey Wave 4. The ITC China Survey is a cohort survey of adult smokers and non-smokers. The ITC China Survey used a multistage cluster sampling design to recruit 800 smokers and 200 non-smokers in each survey city: Stage 1 Jie Dao (street district), Stage 2 Ju Wei Hui (residential blocks) and Stage 3 at the household level. The survey used quantitative methods and was conducted via face-to-face interviews in seven cities: Beijing, Shanghai, Changsha, Guangzhou, Yinchuan, Kunming, and Shenyang according to the geographical representation and levels of economic development difference . Overall, there were 1324 non-smokers in the Wave 4 data collection. Further details about the methodology and survey design are available elsewhere [15, 16].
Key outcome variables
Noticing HWLs. ‘In the last month, how often, if at all, have you noticed the health warnings on cigarette packages?’ Responses were dichotomized as “often” (‘often’) vs. “less than often” (‘once in a while,’ ‘never,’ ‘refused’, or ‘don’t know.’)
Support for more health information on HWLs. ‘Do you think that cigarette packages should have more health information than they do now, less, or about the same amount as they do now?’ Responses were dichotomized as “does not support” (‘less information’, ‘same information’ ‘refused’, ‘don’t know’) vs. “support” (‘more health information’).
Support for including pictures on HWLs. ‘Would you support or oppose the government including pictures as part of the health warning on cigarette packs?’ Responses were dichotomized as “does not support” (‘strongly oppose, ‘oppose’, ‘refused’, ‘don’t know’) vs. “support” (‘strongly support’, ‘support’).
Surveyed Before/After new HWLs implemented. Because the new HWLs were introduced while the survey was in the field, a variable was coded to indicate surveyed “before” vs. “after” the new HWLs were implemented.
Key independent variables
Partner and Friend Smoking
Smoking spouse/partner. Respondents with a partner were asked: ‘Does your partner or spouse smoke?’ Responses were coded as “no smoking spouse/partner” = ‘no’; “have smoking spouse/partner” = ‘yes’; “no spouse/partner” = ‘refused’, ‘don’t know’. Respondents who were not married or not living with their partners were coded as “no spouse/partner”.
Number of smoking friends. ‘Of the five closest friends or acquaintances (not including family members) that you spend time with on a regular basis, how many of them are smokers?’ Responses were coded as “0, 1, 2, 3, 4, 5”.
Demographic variables included: city, ethnicity (Han nationality, other nationality), age (18–39, 40–54 and 55+), gender (male, female), monthly household income (Low/Medium: <¥3000, High:≧¥3000), and education (Low/Medium = no education, elementary school, junior/senior high school, High = college, university or higher.
The data were analysed using Stata/MP 14.0 software for Mac. Descriptive and sample characteristics were examined. Bivariate and multiple logistic regression models (controlling for demographics) were used to examine the relation between the outcome, noticing the HWLs often vs. less often in the last month, and respondents’ number of smoking friends, whether respondents had a smoking spouse/partner, and whether they were surveyed before vs. after new HWLs were introduced. To examine differences in support for adding more information and including pictures on Chinese HWLs, the predicted outcome of a logistic regression analysis was set to “support for more information” versus “no support” and “support for including pictures” versus “no support”. These analyses were controlled for demographics, noticing HWLs in the last month, number of smoking friends, having a smoking partner and surveyed before versus after new HWLs were introduced. All analyses were conducted on weighted data.