�� Behavioral reactions to labels (forgoing of cigarettes and avoidance) were assessed by asking: ��In the last month, have the health warnings stopped you from having a cigarette when you were about to smoke one?�� with response options ��Never,�� ��Once,�� ��A few times,�� and ��Many times�� and ��In the last month, have you made any effort to avoid looking at or thinking customer review about the health warnings?�� (Yes/No). Data Analysis In order to test whether the introduction of pictorial warning labels in Thailand increased salience of the labels (noticing and reading) and psychological reactions to the labels (thinking about the risks, avoiding labels, increasing the likelihood of quitting, and forgoing a cigarette), the proportion of respondents responding in the affirmative for each measure was estimated for each of the three waves.

Significant increases were expected in Thailand between Waves 1 and 2, due to the introduction of pictorial warning labels while no changes were expected in Malaysia. Logistic regression, estimated using generalized estimated equations (GEEs), was used to test whether outcome measures changed significantly over time and whether the changes differed by country. In other words, these models tested the country �� time interaction effect, where a statistically significant interaction effect would indicate that the change in one country over time differed from the change in the other country. All models controlled both time-invariant covariates and time-varying covariates.

Time-invariant covariates were sex, age group, urban/rural residence, income, education, ethnicity, and cohort/wave of recruitment, whereas time-varying covariates were daily/nondaily smoking status, cigarettes smoked per day, and exclusive use of RYO cigarettes. Additional analyses were conducted using only the Thai data to explore whether exclusive use of RYO cigarettes moderated the effects found. The analysis was conducted using SUDAAN version 10.0.1 in order to account for both the multistage sampling design used in the ITC-SEA Project and for the longitudinal nature of the data. Analyses were conducted using weighted and unweighted data for all models, with no significant differences observed between weighted and unweighted analyses. Results are presented for weighted analyses, with standard errors and model coefficients adjusted accordingly. RESULTS Sample Characteristics Entinostat As seen in Table 1, there were more female, older, and rural respondents in the Thai sample than in the Malaysian sample.