Knowledge of the Health Effects of Smoking. Adolescents were asked whether they knew or believed that smoking causes lung cancer in smokers, lung cancer in nonsmokers, stained teeth, and premature ageing. For each one that they indicated, Tivantinib they were assigned a score of 1, and these scores were summed to give an overall score (0�C4) (Yang, Hammond, Driezen, Fong, & Jiang, 2010). The distribution of the index score was highly skewed and so the five-point index score was recoded into two levels by median split (median = 4), no/low knowledge (score <4), and high knowledge (score = 4), for use as an outcome variable. Perceived Health Risk of Smoking. Perceived health risk of smoking (Yang et al., 2010) was assessed using the following two statements: (1) Smoking is harmful to smokers; (2) Smoking is harmful to nonsmokers.

Each of these items had four response options: (1) definitely not, (2) probably not, (3) probably yes, and (4) definitely yes. Each item was recoded into a 0�C3 score and summed to give an overall index score (0�C6). The distribution of the index score was highly skewed, and so it was recoded into a dichotomous item by median split (median = 6), no/low (score <6), and high (score = 6) for use as an outcome variable. Susceptibility to Smoking. This item was created based on two questions asked of never smoked adolescents: (1) If one of your best friends offered you a cigarette, would you smoke it? (2) At any time during the next year, do you think you will smoke a cigarette? Both of these questions had four response options: (1) definitely not, (2) probably not, (3) probably yes, and (4) definitely yes.

Adolescents were considered susceptible if they gave any response other than ��definitely not�� to both (Pierce, Choi, Gilpin, Farkas, & Merritt, 1996). Weight Construction and Data Analysis. A complex weighting procedure was employed to correct the estimates for sampling bias so that we can make population inferences. This involved household weight being constructed for each household in the sample, within its ��pseudo-PSU,�� namely urban or rural part of the each state (Malaysia) or province (Thailand). Following this, individual weight for everyone within his or her household was constructed. Then, the product Drug_discovery of household weight and individual weight within the household was raised to the national level, and finally, the weights were rescaled to national sample size for pooled analysis (International Tobacco Control South-East Asia Survey, 2005). Overall, 2,008 adolescents (smokers and nonsmokers) were included from Malaysia and Thailand in the baseline descriptive analysis. Point estimates (e.g., frequency and means) were computed using weighted data.