|Home | About | Journals | Submit | Contact Us | Français|
Public support for tobacco control policies is one catalyst advocates use to facilitate policy enactment. We examined whether tobacco-control policy support varied by demographic group, including by nativity (immigrants compared to US-born).
We used multiple logistic regression and the 1995–2002 Current Population Survey Tobacco Use Supplement (n=543,951). The outcome was a summary attitudinal measure for supporting smoking bans in 4 of 6 venues.
Populations who were US-born, smokers, male, unmarried, Native American, White, lower SES, in non-smokefree contexts, were less likely to support smoking bans. Immigrants exhibited stronger support for banning smoking in every venue, with a gradient by generation, including erosion of support with greater assimilation to the US. Immigrants had over twice the support for smoking bans than the US-born 3rd generation (OR=2.16, 95% CI:2.12–2.19). Naturalized citizens displayed higher support than US-born citizens, which may be relevant for electorate mobilization. Differential population composition and contexts (such as smokefree workplaces) only partially accounted for immigrants’ stronger support.
Immigrants and their children may be valuable tobacco control allies given their supportive attitudes for smokefree policy.
A substantial body of evidence suggests that Environmental Tobacco Smoke (ETS) causes numerous diseases1–3. Over 50 compounds in secondhand smoke are known human carcinogens 4, and there is no risk-free level of exposure 5. Policies restricting smoking in public places and worksites are effective for reducing population exposure to secondhand smoke,5,6 reducing cigarette consumption, and increasing cessation and quit attempts among smokers7. The coverage of smokefree policies in the US has increased substantially over the past two decades, covering only 3% of US indoor employees in 1986, to 77% of employees in 20038. Since 1993, 337 localities and 30 US states have enacted 100% smokefree policies9,10.
Although support for smokefree policies varies across demographic groups11–13, no prior analyses to our knowledge have examined smokefree policy support by nativity. Understanding tobacco control attitudes among immigrants is important for strategic, ethical, and demographic reasons.
The population’s opinions about where smoking should be allowed are a general indicator of support for tobacco control policy14. Such support is one important catalyst for tobacco control policy enactment15,16. Public opinion influences election outcomes and establishes the policy agenda of elected officials via voter selection of candidates17. Public opinion is also important for gauging the appropriateness of state ballot initiatives or referenda which offer the most direct political participation of citizens to change tobacco policies, because they bypass state legislative processes dominated by tobacco industry; indeed, ballot initiatives/referenda have facilitated passage of 28 state level tobacco measures from 1988–200618.
Knowledge about public support for smoking bans using publicly funded data may aid advocates to identify voter constituencies for coalition building in election strategies and may help them target media messages, given that advocates have vastly fewer resources than industry for public opinion polling.
Understanding the tobacco policy-related attitudes of different demographic groups, including immigrants, may help to protect against tobacco industry manipulation. The tobacco industry has a sophisticated understanding of the immigrant market segment19, and a long history of targeting immigrants, e.g. by investing in relationships with minority community leaders and organizations20. The industry in turn expects recipients of its funds to advocate for its legislative agenda20,21, which may counter the interests of a group’s constituents. Notably, many immigrants are not voters because they are not naturalized citizens, and may also be illegal aliens. Because non-citizen immigrants lack rights and privileges and fear retribution22, they may be less vocal to advocate for workplace protections like smokefree workplaces. Lastly, the growing size of the foreign born population and their children – now comprising over 22% of the US population – necessitates a demographic imperative for examining their policy-related attitudes23–26.
In this study we present descriptive patterns of immigrant support for smokefree policy, and test whether US immigrants have different levels of support for smoking bans compared to the native born, and explore explanations for these patterns. We hypothesize that immigrants hold stronger attitudes in support of smoking bans.
Immigrants may be more likely to support smoking bans because they are comprised of demographic subgroups that are more supportive of smoking bans (compositional explanations). Immigrants are much less likely to be smokers27 and much more likely to be Hispanic or Asian28 than the US-born. These demographic subgroups are also more likely to support smoking bans14.
Several different levels of context may also impact an immigrant’s support for smoking bans (contextual explanations), including the immediate work and home context with respect to smokefree policies14, their state of residence (e.g. California)14,28, their family socialization context, or their country of origin. For example, immigrants from different countries of origin may hold different attitudes and norms towards the acceptability of smoking which may have been cultivated in that country in the past (e.g. by smokefree policies in the country of origin), and/or may be maintained by the current US-based family and community context29.
We used Tobacco Use Supplement (TUS)-Current Population Survey (CPS) data for years 1995/96, 1998/99, and 2001/0230. The CPS TUS is a multistage probability sample, representative of U.S. non-institutionalized, civilian population aged 15+.
Our outcome assesses support for smoking bans in 6 venues (restaurants, hospitals, indoor work areas, bars and cocktail lounges, indoor sports venues, and indoor shopping malls), based on the TUS questions about whether smoking should be allowed in “all areas, some areas, or not allowed at all”. We added the number of venues in which each person supported complete smoking bans (“smoking should not be allowed at all”) and created a dichotomous variable equal to one if a person supported smoking bans in 4+ venues, zero otherwise. We chose this summary measure because it is straightforward, has been applied in prior studies14, and is policy relevant (e.g. WHO smokefree policy tracking)31. This summary measure weights each venue equally. This was justified in our data based on factor analysis, showing that the 6 attitudes loaded approximately equally on the first factor. We excluded observations missing responses for any attitudinal measure, as well as proxy respondents because they were not asked the smokefree attitudinal questions14. The final sample size was 543,951.
Our predictor of interest was nativity, categorized as: 1st generation (respondent was foreign born), 2nd generation (US-born respondent with at least one foreign-born parent), or 3rd generation or higher (US-born of US born parents). We hypothesize that if family context is one vehicle by which norms and expectations are transmitted about smokefree attitudes, then we will observe gradient effects – e.g. that the magnitude of 2nd generation support will fall between 1st and 3rd generation support. Since voting may be an important pathway by which public opinion translates to policy enactment, we adjusted for citizenship (naturalized citizen immigrants, noncitizen immigrants, and US-born (who are all citizens)). To test country/region of origin context, we modeled country of birth (for countries with unweighted sample sizes over 200), or region of birth (for countries with less representation) to ensure we had large enough categories with sufficient power to meaningfully examine differences. We modeled immigrant length of stay in 5-year categories.
Smoking status was categorized as current smoker, former smoker, or never smoker based on whether respondents had smoked 100 lifetime cigarettes, and whether they currently smoked everyday, some days, or not at all14,32. We excluded those of indeterminant smoking status (0.29% of the original sample). A smokefree home indicator was based on self report that “no one is allowed to smoke anywhere” at home. Respondents who were employed and who worked indoors were asked whether their workplace had an official policy restricting smoking in any way. Those in smokefree workplaces were modeled against non-smokefree workplaces (non-smokefree workplaces included those lacking workplace tobacco policy and those who were not in universe for receiving the survey question)14. The time period of the survey was modeled as the TUS year (1995/96, 1998/99, or 2001/02). Demographic covariates included gender; age (divided by 10) and age-squared (both centered at age 45); race/ethnicity; educational attainment; marital status; annual income; occupation; employment; and residence in California versus elsewhere in the US. See Table 1 for further detail.
We first tested bivariate associations with policy support using cross tabulations and chi-square tests in SAS 9.1. We then conducted multiple logistic regressions using SUDAAN 9.02 with the summary smoking ban measure as the outcome. Model 1 tested unadjusted bivariates including only nativity/generation. Models 1a-1n added one variable at a time to Model 1, to test how the immigrant generation associations with smoking ban support were affected. Model 2 added all demographic variables simultaneously. Model 3 added smoking status to Model 2. Model 4 added workplace and home smoking bans to Model 3. Model 5 added California residence to Model 4. Models 6–8 built on Model 4 to explore whether the patterns of support vary among immigrants, including by length of stay, country of origin, and citizenship by substituting these variables in lieu of the first generation variable. For all the immigrant variables, the reference group (omitted) is the 3rd generation, and the 2nd generation is included as an indicator variable. We tested heterogeneity of % increase in support across time for generation subgroups with logistic regression and time-generation interactions.
We applied the self-response weights and self-response replicate weights (created using a Balanced Repeated Replication method) to adjust the variance induced by the CPS multistage complex survey design.
Table 1 shows that public support for smoking bans has increased across time among all Americans. Support for banning smoking in 4 of 6 venues rose from 54.8% in 1995/96, to 68.5% in 2001/02. The prevalence of the US population covered by workplace smoking bans also increased over this period (results not shown). On average, 75.7% of immigrants support banning smoking in 4 of 6 venues, compared with 65.7% of the 2nd generation, and 59.1% of the 3rd generation. Populations who were smokers, male, unmarried, Native American, White, lower SES, in non-smokefree contexts, were less likely to support smoking bans.
Smoking bans in hospitals garner the greatest support and smoking bans in bars garner the least, across all groups (Figure 1). Within each nativity group, attitudes became more supportive of smoking bans across time, for all venues, and there is a gradient in support across all venues by immigrant generation. The 1st generation is the most supportive of bans, and the 3rd generation express weakest support. However the 3rd generation outpaced the 1st generation on a relative basis for 6 of the 7 venues/outcomes regarding increasing support indicated by the 1st generation’s significantly shallower slope of support across time compared to the 3rd generation, as noted by the % increase numbers and asterisks above the figure bars.
Multivariate models demonstrated that foreign born individuals were more likely than their 3rd generation counterparts to support smoking bans in 4+ venues both before (Model 1, Table 2, OR=2.16, 95% CI (2.12–2.19)), and after adjusting for demographic factors (Model 2, Table 3, OR=1.67, (1.64–1.70)). When each variable was added to the model in turn (trivariate models 1a-1n Table 2), the 1st generation Odds Ratio for support for smoking bans decreased the greatest with the addition of race-ethnicity (−41% when compared with the bivariate model), home smoking ban (−35%), smoking status (−26%), and California residence (−25%). All the demographic variables had significant associations with smoking ban support, similar to bivariates. As anticipated, differences in smoking status also attenuated the coefficient for immigrant status with smoking ban support (Model 3, OR=1.50, CI 1.48–1.53), compared to Models 1 & 2, but immigrants still held more supportive attitudes vs. the US-born. Throughout the models, second generation immigrants exhibited less support for smoking bans than the first generation, but were still significantly more supportive than the 3rd generation.
We next tested whether differential smoking policy context affected nativity associations with attitudes. Adjusting for the smoking policy of the workplace and the home did further attenuate the 1st generation coefficient for support to OR=1.39(1.36–1.41)(Model 4), as did California state of residence (Model 5). Immigrants continue to be more likely than the US-born to support smoking bans, OR=1.31 (1.28–1.33)(Model 5), although the 2nd generation now differed little from the 3rd generation (OR=1.03, 1.01–1.04).
We lastly tested whether we observed patterns of support within immigrant groups. Model 6 shows that immigrants who were not US citizens (OR=1.47 (1.44,1.50)) and immigrant naturalized citizens (OR=1.27 (1.24,1.31)) had significantly stronger support for smoking bans than the US-born. Recent immigrants (within the last 4 years) were 48% more likely to support smoking bans than the US born (OR=1.48, 1.42–1.53), and more likely to support bans than immigrants who have arrived more than 20 years prior (OR=1.28), and compared to the US-born, with a gradient in between, even after adjusting for covariates (Model 7). The heterogeneity in immigrants’ support can also be discriminated by their country or region of origin; but most immigrants displayed stronger support than the US born regardless of their area of origin. For example, immigrants from Africa (OR=1.86), the Dominican Republic (OR=1.73), and Mexico (OR=1.70) had the greatest support compared to the US-born in adjusted models (Model 8).
This analysis has 4 principal findings: (1) American immigrants support smoking bans more strongly than their US-born counterparts, with a clear gradient of support across immigrant generation and assimilation measures (e.g. length of stay, citizenship); (2) demographic composition and smoking-policy related context each explain part of the higher immigrant support, although residual higher support among immigrants remained; (3) there was some initial support for the family socialization theory of smoking ban support (since 2nd generation immigrants had higher support than 3rd generation), but it was largely accounted for by differential composition and smoking policy context; (4) Lastly, the US has experienced a substantial secular increase in support for smoking bans over the seven years spanning this analysis among all groups. We address these findings below.
Smokefree policies are workplace protection measures that prevent disease and premature death1,2. However, certain groups are less likely to be protected by smokefree policies at work including Hispanic immigrants and workers in low-wage jobs33. Since illegal immigrants, citizen children of illegal immigrants, and refugees are vulnerable populations compared to their counterparts who are US citizens, they may be less willing to advocate for worker-protection policies like workplace smoking bans, for fear of retribution and/or fear of deportation22. Indeed in the last decade, states have passed hundreds of laws that restrict immigrant rights22. Public health advocates must therefore protect immigrant health by advocating for smokefree policy on their behalf.
The finding that immigrants to the US consistently are more supportive of banning smoking than their US-born 3rd generation counterparts suggests that they are potential allies in the tobacco control movement, as partners for coalition building and campaign mobilization. Publicizing their higher support may also protect them from industry manipulation within legislative battles12.
The tobacco industry has targeted and exploited vulnerable populations like immigrants and racial minority groups, not only to increase product sales, but also to obstruct passage of smokefree policies; indeed, the industry has a complex understanding of the immigrant market, and has strategized and targeted immigrants for sales and marketing since the 1970’s19. Profit is the industry’s motivation factor, as this internal document from Brown & Williamson (B&W) articulates:
“Clearly, the sole reason for B&W’s interest in the black and Hispanic communities is the actual and potential sales of B&W products within these communities and the profitability of these sales … this relatively small and often tightly knit [minority] community can work to B&W’s marketing advantage, if exploited properly”34(Bates No. 531000141–2).
The tobacco industry has moreover cultivated relationships with leaders in minority communities not only to increase their tobacco use, but also for legislative ends: “to advance and defend industry policy positions and to … obstruct tobacco control efforts”20 (p.342). For example, an internal memo enumerates the Hispanic and Asian constituent organizations to which Philip Morris donated $145,655 in 1988. Forty-three percent of these organizations “provided support on legislative issues” including on “restrictive smoking”, “excise tax”, and “Prop 99” (a 1988 California tobacco control ballot initiative) issues35. In other instances, the industry has compelled organizations to which it has donated to write letters to support its legislative agenda20,21. Industry has thus forced many groups to oppose smokefree legislative efforts that would likely have improved the health of their constituents had the efforts passed. The tobacco industry has also specifically targeted voters in black and Hispanic communities with mass media campaigns to oppose ETS legislation36,37. Therefore documenting higher smokefree policy support among immigrants (including among immigrants who are US citizens and therefore voters) provides some protection against tobacco industry manipulation, including making it more difficult for minority constituency groups to lobby on behalf of industry’s positions, against the interest of constituents.
We found that population composition, including race/ethnicity and smoking status, did account for a portion of the higher support that immigrants expressed for smoking bans. Prior discussions of the implications of demographic differences in support for smokefree policies have centered on compositional explanations, including insufficient control for smoking status13. In immigrant health research, nativity and race/ethnicity are often conflated38, despite that these constructs are distinct dimensions of inequality that may influence tobacco use and support for tobacco control policy differently. Dispelling population compositional explanations may strengthen deductions from observational studies about the contextual effect of clean-indoor air policies. For example, Gilpin et al. concluded that California exhibited higher support for smoking bans compared to the rest of the US from 1993–1999 after controlling for demographic variables, and attributed the association to the strong tobacco control policy environment cultivated there14.
The pattern of stronger policy support among immigrants holds for most countries of origin, and erodes with longer residence in the US. We found a gradient of support by immigrant generation, including that children of immigrants (2nd generation) were initially more supportive of smoking bans than the 3rd generation in unadjusted models, as well as by the number of parents who were foreign born. These 2nd generation associations may signal effects of family anti-tobacco socialization e.g. about the unacceptability of smoking39. Norms and expectations play an important role in conditioning behaviors and attitudes, including those related to smoking, and immigrant kin and community networks have been documented as reinforcing certain behavior-related norms and sanctioning those who defy them40. Although we found only small residual associations among the 2nd generation after covariate adjustment, these covariates may have overadjusted for mediators since socioeconomic advancement is an important aspect of immigrant assimilation41. We lastly do not have the most appropriate data to test family socialization pathways that may be proxied by generation.
We found that immigrants who have remained in the US for a longer period of time held weaker policy support attitudes, after adjustment. This erosion of attitudes with time aligns with other evidence that immigrants adopt the health behaviors or health profile of the US-born with greater assimilation to the US27,42. Erosion of attitudes may reflect residual effects of tobacco advertising and marketing to immigrants, since tobacco industry has researched, tracked, and targeted immigrants by levels of assimilation19.
Americans increased their support for smoking bans from 1995 to 2002, regardless of nativity, subgroup, or venue. These findings align with prior research12,14,43, and seem to be due to the success among the tobacco control movement to pass local and state-wide laws prohibiting smoking in public places9,10, the voluntary passage among employers, and the promotion of social norms that smoking is unacceptable44. Both of these pathways (policies and norms) are important for reducing population rates of smoking. As with seat belt use and drunk driving prevention45, the passage of laws inhibiting smoking in public areas may increase the level of support for such policies43,46. However, even though more restrictive smoking policies might cause an increase in attitudes supporting such policies, a certain level of baseline support for policies needs to be present before such passage is politically feasible47.
The majority of the US population supports smokefree policy, yet the US does not have national-level laws mandating smokefree environments. Sixteen countries worldwide have passed comprehensive national smokefree policies, including Ireland, New Zealand, and Uruguay31. Despite current strong support for workplace smoking bans, and despite the documented, well-publicized harmful health effects of secondhand smoke1,2, inequalities of workplace smokefree policy coverage remain11–13,33. These inequalities are generated by the U.S. tobacco-control policy patchwork of smoking ban policies at local and state levels, and reliance on voluntary smoking bans that disproportionately benefit white collar workers48. Mandating smokefree workplaces must be a matter of federal policy to effectively protect the health of all workers and reduce health inequalities.
We used the Current Population Survey, one of the few surveys representative of the U.S. civilian noninstitutionalized population with the power to examine smoking patterns across different American subgroups, with valid and reliable measures on tobacco use, tobacco control, and demographics. We also adjusted for the complex survey design using replicate weights, to report corrected standard errors.
Since this was a repeat-cross sectional study, we cannot deduce that attitudes changed within individuals across time. However, since the CPS is rigorously designed and executed by the Census Bureau, the TUS is representative of the US and state populations, thus our results indicate changes in population attitudes across time. The measurement of the construct of support for tobacco control policies may be underestimated by our method, which measures support for smokefree policies in only 6 venues. Other instruments, such as the Smoking Policy Inventory (SPI) have been developed to capture a range of different domains of tobacco control policy49. The TUS should consider adopting this measure in future surveys. Lastly, our measure of workplace smokefree policy may be measured with error since it was self-reported. Although we did not account for enforcement, US smokefree policies are largely self-enforcing50.
In this study we found that attitudes in support of smokefree policies were stronger for immigrants than for the US-born 3rd generation, and that composition as well as context explained part, but not all, of these patterns. Increasing the support for smokefree laws among all subgroups, and passing comprehensive smokefree policies, are important goals for preventing tobacco-related disease and improving population health.
Financial support for this analysis was provided to Dr. Osypuk by the Robert Wood Johnson Foundation Health and Society Scholars Program at the University of Michigan, and by a 2-year Association of Schools of Public Health (ASPH) and American Legacy Foundation (Legacy) STEP-UP to Tobacco Control Dissertation Grant L2010-02. Dr. Acevedo-Garcia was supported by NCI grant 1 R03 CA093198-01 and a grant from the ASPH and Legacy (L4002-01/03; Dolores Acevedo-Garcia, PI).
Statement of Institutional Review Board Approval: This research study was deemed exempt from IRB review by Northeastern University since we conducted a secondary analysis using public-use data that had been de-identified.
Statement of Author Contributions: Dr. Osypuk conceived of the study, conducted the analysis and wrote the manuscript. Dr. Acevedo-Garcia advised on the analysis and edited the manuscript.
Theresa L. Osypuk, Northeastern University, Bouvé College of Health Sciences, Department of Health Sciences.
Dolores Acevedo-Garcia, Department of Society, Human Development, and Health at the Harvard School of Public Health, Department of Society, Human Development, and Health.