METHODS

Sample

We analysed data from the 2001 Harvard School of Public Health College Alcohol Study (CAS), which surveyed a random sample of students enrolled in 120 four-year US colleges and universities.¹⁶ For the survey, each school provided a list of 215 randomly selected students from all full time undergraduate students enrolled during the 2000–01 school year. One college was excluded from analysis because its response rate was substantially lower than the others, leaving 119 schools.

The schools, located in 38 states and the District of Columbia, were selected to create a nationally representative sample of four-year US colleges and universities.¹⁷ Sixty nine per cent of respondents attended public colleges and 31% attended private colleges; the corresponding US distribution is 68% and 32%, respectively.¹⁸ Forty seven per cent of respondents attended large colleges (> 10 000 students), 23% attended medium sized colleges (5001–10 000 students), and 29% attended small colleges (< 5000 students); the national distribution is 37%, 24%, and 40%, respectively.¹⁸ Large colleges are over represented in the CAS sample because they were selected with probability proportional to size. Thirteen per cent of students attended schools with a religious affiliation, compared with 16% nationwide.¹⁸

Surveys were mailed to 21 055 students in February 2001. Three separate mailings were sent within three weeks: a questionnaire, a reminder postcard, and a second questionnaire. Responses were anonymous. Cash prizes were offered to encourage response. The response rate was 52% (n = 10 904).

Questionnaire and measures

The questionnaire assessed students’ demographic characteristics, tobacco use, opinion of proposed tobacco control policies, awareness of their school’s policy about smoking in student residences, and the current smoking policy of their residence. Demographic factors assessed were age, sex, ethnicity, year in school, marital status, and residence (on-campus or off-campus). Students were asked if they lived in housing designated as smoke-free, and if not, whether they would like to do so. Students who reported smoking a cigarette in the past 30 days were defined as current smokers. Cigarette smokers were asked to specify daily cigarette consumption and frequency of smoking (“On how many of the past 30 days did you smoke cigarettes?”).

To assess attitudes regarding tobacco control policies on campus, students were asked, “To what extent do you support or oppose the following possible school policies about smoking?” Response options were strongly support, support, oppose, and strongly oppose. Students were asked about three types of tobacco control policies: clean indoor air (four items), tobacco marketing restrictions (two items), and tobacco sales restrictions (one item). Clean indoor air policies tested were smoking prohibitions in: (1) all campus buildings; (2) all parts of residence halls including student sleeping quarters; (3) on-campus restaurants or dining areas; and (4) on-campus bars or pubs. Marketing restrictions assessed were prohibitions on tobacco industry sponsorship of school parties or events and tobacco advertisements in student newspapers. Students were also asked their opinion of prohibiting the sale of tobacco products on campus.

A separate written questionnaire was sent to administrators at each school in the sample. In 2001, two questions asked about the school’s smoking policies. Administrators were asked whether smoking was prohibited everywhere on campus, and if not, whether smoking was prohibited in residence hall private rooms and common areas. Responses were obtained from all schools.

Analysis

Statistical analyses were carried out using STATA.¹⁹ Data were weighted to account for colleges’ different sampling fractions and to match each college’s distribution of sex, race (white/non-white), and age (< 18 years, 18–22 years, > 22 years). Analyses were conducted to examine the potential bias introduced by non-respondents. Response rates did not differ between colleges that did and did not (1) prohibit smoking everywhere on campus (p=0.85) or (2) prohibit smoking in student residences (p = 0.60). The Spearman correlation coefficient between a college’s current smoking rate and its response rate was 0.15 (p = 0.10). There was no significant difference in current smoking rates between students who responded before and after the second mailing (24.8% v 26.0%, p = 0.19). College response rate was included as a continuous covariate in all multivariate regression models to control for non-response bias.

Attitudes toward policies were dichotomised into two categories: support or oppose. Bivariate analyses identified student level and college level characteristics associated with attitudes toward tobacco control policies. Significance was assessed with χ² tests and χ² tests of trend for categorical variables and t tests for continuous variables. Multivariate models were created to identify student level and college level factors that were independently associated with each proposed policy. All models adjusted for each schools’ survey response rate and included student age, sex, ethnicity, year in school, marital status, and current cigarette smoking. College level factors included in models were size of enrollment (< 5000, 5000–10 000, or > 10 000 students), geographic region, public versus private status, competitiveness for admission (based on percentage of applicants accepted), school smoking policy (for college residences), and school smoking rate (divided into tertiles). We used the generalised estimating equations (GEE) approach to fitting the logistic regression models to account appropriately for clustered outcomes arising in our sampling scheme.^20,21 Adjusted odds ratios with 95% confidence limits are reported.

RESULTS

Characteristics of the sample

Table 1 shows the demographic characteristics of the 10 904 respondents and compares their age, sex, and racial distribution with the distributions of these characteristics among all full time undergraduates at the 119 participating schools and among a national sample of full time undergraduates.¹⁸ Respondents’ age and race were comparable to the comparison populations, but a greater proportion of respondents were female than in the comparison populations. Of the 119 schools in the sample, 22 (18%) prohibited smoking everywhere on campus. An additional 31 schools (26%) prohibited smoking in private rooms in college residences but not everywhere on campus.

Attitudes about tobacco control policies

A majority of students supported each of the seven proposed policies (table 2). More than three quarters of students supported a ban on smoking in all campus buildings, including residence halls and dining areas. Nearly as many supported restrictions on tobacco marketing on campus. Sixty per cent supported banning tobacco sales on campus and half favoured banning tobacco in campus bars.

Support for all policies was significantly stronger among non-smokers than smokers (table 2). However, even among smokers, a majority favoured five of the seven policies, including smoke-free residence halls and dining areas, and bans on tobacco sponsorship of campus social events and tobacco advertising on campus. Table 2 also shows that among smokers, support for each policy was inversely related to tobacco consumption.

Additional support for smoke-free residences is found in students’ responses to questions about their preferences for living quarters. Of the 39% of respondents who lived in a campus residence, 59% reported that their housing was designated as smoke-free, and over half (58%) of those who did not live in smoke-free housing stated that they would prefer to do so.

Factors associated with support for tobacco control policies

Tables 3 and 4 display independent predictors of student support for the proposed tobacco control policies. Support for all policies was most strongly associated with student smoking status, even after adjustment for other student and college level factors.

When these multivariate analyses were repeated among current smokers only, support for all tobacco control policies was greater among smokers who planned to quit in the next 30 days than among smokers not planning to quit (for example, for prohibiting smoking in all campus buildings: adjusted odds ratio (AOR) 1.58, 95% confidence interval (CI) 1.23 to 2.03; p < 0.001). Support for smoking restrictions in student residences, dining areas, campus bars, and all campus buildings was greater among lighter smokers (< 10 cigarettes/day) than among heavier smokers (for example, for prohibiting smoking in all campus buildings: AOR 1.52, 95% CI 1.14 to 1.98; p < 0.01) and among occasional smokers compared with daily smokers (AOR 1.39, 95% CI 1.08 to 1.82; p < 0.05). Support for marketing restrictions and for banning tobacco sales on campus did not vary by intensity of tobacco use.

Tables 3 and 4 demonstrate that all policies were more popular among females and married students than among males or unmarried students. Student support for all policies was higher at colleges in western states, at colleges with a lower smoking prevalence, and at schools that had already banned smoking in student residences.

Smoke-free policies for student residences and campus buildings were also more popular among schools in the north central states. These policies were less popular among students who live in campus residences. Smoke-free bars had more support among students in higher grades and among students at smaller schools. Support for tobacco marketing and sale restrictions, unlike smoke-free policies, varied by race, being more popular among non-whites than whites. These policies were also more popular at smaller schools. Age was a factor only for one policy: tobacco sales restrictions. Younger students were less supportive of this policy than older students, independent of grade level.

DISCUSSION

To our knowledge, this is the first study to report college students’ opinions about the campus tobacco control policies recommended by the American College Health Association and American Cancer Society. The study found strong support for all proposed policies among a large, nationally representative sample of US college students. College administrators and policymakers may not have previously been aware of the breadth and depth of student support for tobacco control policies. Our findings should reassure officials who may have been concerned about student opposition to the policies.

As we hypothesised, support for tobacco control policies was stronger among non-smokers than smokers. Even so, most policies had substantial support among smokers, because support was inversely related to tobacco consumption, and 76% of the smokers in the sample were either occasional smokers or light daily smokers. The multivariate analyses indicated that opposition to tobacco control policies was concentrated in the small subgroup of heavy daily smokers who did not intend to quit. It is possible that the inverse relation between policy support and tobacco consumption may apply to populations beyond college students. If so, the finding would have special relevance for populations with few heavy smokers, such as young adults in general. In these populations, support for tobacco control policies may be higher than many policymakers assume if they use tobacco prevalence as a rough measure of opposition to tobacco control policies.

Multivariate analyses further demonstrated that support for proposed tobacco control policies was broad and strong in all sociodemographic groups and types of colleges. Some colleges have voiced concern that the adoption of a smoke-free policy in student residences might discourage prospective students who smoke from choosing to attend the school. Our data suggest otherwise. Even among smokers, 45% preferred a smoke-free dorm policy, and the majority of students who currently lived in a residence that permitted smoking stated a preference for a smoke-free residence. We did observe that students who lived in dormitories were less in favour of smoke-free residence policies than students who lived in off-campus housing. Students who do not live in dormitories may find it easy to favour a policy that does not affect them, while students who live in a dormitory may be more reluctant to impose rules on classmates.

Support for tobacco control policies of all types was stronger among students attending colleges with a low smoking prevalence. The relative absence of smokers on these campuses may foster positive attitudes toward proposed tobacco control policies. Support for all types of tobacco control policies was also stronger among students who reported that housing on their campus was already smoke-free. One interpretation is that students who live in smoke-free residences like them and may find it easier to support additional tobacco restrictions. Alternately, colleges with smoke-free residences may attract students with more support for tobacco control policies. Because this is a cross-sectional survey, it is impossible to distinguish between alternate interpretations of the finding.

Smoking bans in campus bars and prohibiting tobacco sales on campus were the least popular tobacco control policies. This may reflect the fact that both policies directly impact smokers. Smoke-free bar policies make it difficult for students to smoke in one of the last places where smoking has been allowed, while banning tobacco sales on campus makes access to tobacco products less convenient. In this context, smokers’ strong support for bans on smoking in campus buildings, housing, and dining areas is remarkable. It probably reflects students’ perception of social norms about where smoking is now acceptable in US society. It also reflects larger public opinion; state and community smoke-free bar laws face stronger opposition than do smoking bans for public places, worksites, and restaurants.

Proposed restrictions on tobacco marketing, which have little direct effect on students, had little opposition, even among smokers. Tobacco company sponsorship of campus events is a new tobacco industry strategy for recruiting young adult smokers.^6–8 It occurs widely and students who attend these events have higher rates of tobacco use than students who do not.⁹ Our survey suggests that students will not oppose efforts to stop these promotional activities.

The prevalence of smoke-free dormitory policies in 2001 in our sample (44%) was substantially higher than the 27% prevalence observed in 1999 in a larger national sample of US colleges.⁵ This apparent discrepancy probably represents an increase in smoke-free policy adoption between 1999 and 2001. A survey of 50 US public universities conducted in 2001 found a doubling in the prevalence of smoke-free dormitories between 1999 and 2001.¹³

The results of this study are subject to several limitations. The survey response rate of 52% raises the potential for selection bias. However, the sociodemographic characteristics of the sample resembled those of a national sample, and response rates did not correlate with smoking prevalence or differ between schools with and without smoke-free policies. These facts suggest that selection bias is not a large concern. Second, our results are generalisable to students enrolled in US four-year colleges and universities but cannot be assumed to represent the opinions of students attending two-year colleges. Nonetheless, the results do apply to a large number of young adults—an estimated 5.3 million young adults attend four-year colleges.² Finally, because of survey length limitations, we did not assess all proposed campus tobacco control policies. We did not solicit student opinions of bans on smoking outside building entrances or at all campus events, about the provision of free smoking cessation treatment, or on free distribution of tobacco products on campus.

In summary, this study elicited student opinions about a range of policies and derived data from a large, random sample of students at a nationally representative sample of US colleges. These data provide strong support for tobacco control policymakers, advocates, and college administrators who are considering whether to adopt policies to discourage tobacco use among college students. Policy strategies are important for the college population because of their broad reach. They can discourage both the uptake and continuation of tobacco use and they have the potential to affect a much larger number of students than traditional smoking cessation interventions in this setting. Future research should include surveillance of the prevalence of college tobacco control policies, identification of factors that facilitate and impede policy adoption, and evaluation of the impact of specific tobacco control policies on tobacco use prevalence. In the meantime, this study provides support for adopting college tobacco control policies, which should be a high priority for college administrators, public health policymakers, and practitioners.