Attitudes Toward Secondhand Smoke, Smoking, and Quitting Among Young People

METHODS

After approval of the protocol by the institutional review board at the University of Pennsylvania, interviewers from Princeton Survey Research Associates used random-digit telephone dialing to contact 9301 households in the continental United States from May 27, 1999 through July 13, 1999. Within these households, 76% provided the needed screening information to identify a person 14 to 22 years of age. Approximately 15% of these cooperating households contained an eligible English-speaking respondent. In households with more than one eligible respondent, the person with the most recent birthday was selected for interviewing. Parental consent was obtained before interviewing respondents under 16 years of age. Respondents who reported smoking any cigarettes in the past 30 days were defined as smokers. The final sample contained 300 smokers and 300 nonsmokers with a total response rate of 51%.

The survey data were weighted to adjust for both sample design decisions (overrepresenting smokers) and for demographic nonresponse. This weighting was accomplished in 2 stages. In the first stage the relative proportions of smokers and nonsmokers were adjusted to correct for the fact that smokers were oversampled to increase their number in the final sample for analysis. The oversampling of smokers was accomplished by screening a large nationally representative sample of 14- to 22-year-olds on smoking status and then conducting the full interview on all smokers identified (n = 300) but only a portion of the nonsmokers. The weighting reflects the fact that smokers were oversampled 2.5 times relative to their prevalence in the large nationally representative sample of 14- to 22-year-olds screened on smoking status.

The second-stage weight involved bringing the sample of completed full interviews (after it had been adjusted to correct the smoker/nonsmoker proportions) into alignment with national parameters for sex by age, race, and region for Americans 14 through 22 years of age living in telephone households in the continental United States. These demographic weighting parameters were derived from a special analysis of the most recently available Census Bureau Annual Demographic File (from the March 1998 Current Population Survey). The weights are derived using an iterative technique that simultaneously balances the distributions of all weighting parameters.

In addition to age and gender, we included variables for ethnicity (Hispanic and black). We included an educational variable coded as 1 for respondents who were still enrolled in school or high school graduates and 0 if they had left school before completing high school. We assessed whether they considered smoking dangerous with this question: “In your opinion, is smoking very risky for a person's health, somewhat risky, only a little risky, or not at all risky?” and coded the responses “somewhat risky” or “very risky” as 1 (yes) and “a little risky,” “not at all risky,” and “don't know” as 0 (no). We assessed whether respondents considered secondhand smoke dangerous with this question: “True or false: each year thousands of nonsmokers die from breathing other people's smoke” and coded “true” as 1 (yes) and “false,” “don't know,” and “refused to answer” as 0 (no). We defined a current smoker as one who answered “yes” to this question: “During the past 30 days have you smoked any cigarettes?” We assessed whether a respondent had stopped smoking or planned to stop smoking with this question: “Do you plan to quit smoking?” coding “yes” or “already quit” as 1 (yes) and “no” or “don't know” as 0 (no).

Data were analyzed with a single logistic regression including all the variables using SPSS, Version 9.0(SPSS, Chicago, IL).