Abstract
Smoking behavior has been shown to be influenced by individuals’ perceptions of social norms about smoking. This study examines whether local regulations regarding clean indoor air and youth access to tobacco are associated with residents’ subsequent perceptions of smoking norms. Data came from Massachusetts surveys of adults and youths and from records of local tobacco control policies. Indices of perceived smoking norms were based on perceived smoking prevalence and perceived community acceptance of smoking. Multilevel models tested the association between perceived norms and the presence of strong local regulations in four policy domains (restaurant smoking bans, smoking restrictions in other venues, enforcement of laws prohibiting sales to youths and youth-oriented marketing restrictions). The model controlled for town voting results on a tobacco tax referendum, which served as a measure of antismoking sentiment pre-dating the regulations. Results showed that youths perceived community norms to be significantly more ‘antismoking’ if they lived in a town that had strong regulations in at least three of the four domains. For adults, having strong regulations in as few as one to two domains was associated with perceiving community norms to be significantly more antismoking. Implementing and publicizing local regulations may help shape perceptions of community smoking norms.
Introduction
Four decades after the first Surgeon General's report on smoking and health [1], smoking has become much less prevalent but remains the leading preventable cause of death and disease in the United States [2]. In response to the recognized danger of tobacco use and second-hand tobacco smoke, state and local regulations increasingly restrict where smoking is permitted and how tobacco is marketed. There is good evidence that tobacco control policies are associated with reductions in smoking prevalence among both youth and adults [3–7]. The question addressed by this research is whether one of the mechanisms underlying that association is the impact of tobacco control policies on social norms regarding smoking.
Local tobacco control regulations are hypothesized to communicate messages about community smoking norms as well as constraining specified behaviors. For example, restaurant smoking bans may influence the way individuals perceive the community norm because they can no longer smoke in restaurants, because they observe fewer people smoking in restaurants and/or because they see the restaurant's ‘No Smoking’ sign as indicating community disapproval. Regulations may therefore influence perceptions of both descriptive norms (what proportion of people smoke) and injunctive norms (to what extent is smoking disapproved). Further, regulations may influence people's perceptions when their own behavior is not regulated, such as a non-smoking adult who sees a store clerk check the identification of a young person attempting to buy cigarettes. A logical extension of the hypothesis is that multiple regulations, by presenting concordant antismoking messages, will influence perceived norms more strongly than will a single regulation.
Perceived norms and smoking behavior
Research examining the impact of perceived smoking norms has shown both descriptive and injunctive norms to be important influencers of youth and adult smoking behavior [8, 9]. Social unacceptability has been found as important as tobacco taxes in influencing tobacco consumption [10].
Teenagers’ overestimates of smoking prevalence (i.e. descriptive norms) have frequently been shown to relate to their own tendency to smoke [11, 12]. Injunctive norms—youths’ belief that adults in their community disapprove of teen smoking—have been found related to youth smoking [13]. A link between descriptive and injunctive norms has been shown: youths who observed smoking in various community locations perceived smoking to be socially acceptable in those locations [14].
Among adults, perceived norms have been shown to influence smoking [15, 16] as well as such diverse behaviors as sun protection [17] and cancer screening behaviors [18]. More broadly, it has been argued that one of the most potent forces responsible for the reduction in tobacco use has been the change in social norms regarding smoking [19].
Local tobacco control policies and smoking norms
Although research has linked perceived norms to smoking, there is little empirical evidence that tobacco control policies influence perceived norms. Theoretical work indicates that law and regulation in general, and especially law originating at the local level, can have such effects [20, 21]. Local tobacco control regulations are often asserted to serve as statements of smoking acceptability in the community [22, 23]. Empirically, a few studies have indicated that restrictions on smoking in locations such as workplaces and restaurants can increase community residents’ support for the restrictions themselves [24–27]. Increasing public support for tobacco regulations has been interpreted as indicating a change in social norms [28–30]. Direct evidence that regulations affect individuals’ perceptions of community norms, however, is even more limited.
Analysis of a portion of the data presented here [31] found that among those who frequently ate at restaurants in their own town, strong local restrictions on restaurant smoking were associated with youths’ and adults’ perceptions of adults’ approval of restricting restaurant smoking and with youths’ perceptions of lower adult smoking prevalence and greater adult disapproval of smoking. Aside from that study, no research, as best we know, has examined the effect of local tobacco control policies on broader measures of perceived social norms.
Prior to July 2004 when a statewide smoke-free worksite policy was implemented, clean indoor air policies and those designed to affect youth access to tobacco varied among the 351 cities and towns of Massachusetts. The relationship between the strength and breadth of those policies and the perception that one's community possessed an antismoking norm is the focus of this paper.
Methods
Overview
This analysis used data from population surveys of Massachusetts adults and youths conducted in 2001–02. Indices of respondents’ perceptions of their community smoking norms were created from survey items concerning perceived smoking prevalence, perceived acceptability of smoking in restaurants and, additionally for youth, perceived adult disapproval of smoking. Individuals’ survey responses were linked with the policy and demographic characteristics of their towns. To determine the breadth of a town's tobacco control regulation, we examined four policy domains (two types of clean indoor air policy and two types of restrictions on sales to youth) and counted the number of domains of strong regulation. Multilevel statistical models examined the breadth of local regulation as a predictor of perceived smoking norms, controlling for community smoking norms that predated the local regulations and for other town and individual characteristics.
The sample
The sample was a statewide list-assisted random digit dialed sample. The survey was conducted by professional interviewers at the Center for Survey Research, University of Massachusetts, Boston. An initial screening survey with an adult informant established household composition, demographics and smoking status of all adult members. All resident youths between 12 and 17 years were selected for extended interview. In most households, an adult (age ≥18 years and not institutionalized, in the military or living in group quarters) was also selected for extended interview, oversampling smokers, recent quitters, and young adults 18–30 years to improve sample sizes of these particularly interesting groups. Interviewers successfully screened 66% of households. Extended interviews were completed with 70% of the eligible adults for a final sample size of 6739 and an overall response rate of 46%. In households with age-eligible youth, parental permission was sought to speak with the youth and was granted in 78% of the cases. Interviews were completed with 84% of those, resulting in a sample of 3862 adolescents and an overall response rate of 43%. Design weights were applied during analysis to correct for the probability of selection.
Breadth of local regulation
To describe the comprehensiveness of regulations within a town, we derived a composite variable based on two domains of clean air regulation and two domains of youth access regulation. To the extent possible with the available data, we focused on ‘strong’ policies, such as a complete ban on smoking rather than a restriction to designated smoking areas. The data available for some domains described only the existence of regulations rather than their details or stringency, in which case our criterion for ‘strength’ was the presence of two related regulations.
Data on regulations came from the Massachusetts Tobacco Control Program (MTCP), which maintains data on tobacco-related ordinances and regulations in the state's 351 cities and towns (we use ‘town’ to refer to cities and towns inclusively), as well as enforcement activities carried out by town boards of health that receive MTCP funding. Additional data on restaurant regulations were obtained from several sources, as described elsewhere [31, 32].
The two measures of clean air regulations were as follows:
(i) Restrictions on smoking in restaurants: Strong regulation meant that towns had ordinances or regulations that prohibited smoking in restaurants, including all customer seating areas, and allowed no variances [31].
(ii) Other smoking restrictions: A town was defined as having strong regulation if it had some smoking restriction regarding (a) private workplaces in general and (b) municipal buildings.
The two measures of youth access regulation were as follows:
(i) Enforcement of prohibition on sales to youth <18 years of age: Most towns received state funding for the local board of health to conduct compliance checks, in which supervised youths attempted to buy cigarettes from retailers (>90% of respondents lived in towns that reported some compliance checks). If town ordinances or regulations required licenses for tobacco vendors and/or established fines or other penalties for sales to underage youth, compliance checks could result in license suspensions or fines. Strong regulation was (a) having a licensing requirement and/or a provision for fines and (b) conducting at least two annual compliance checks per vendor, averaged over 1999–2000.
(ii) Merchandising restrictions: A town was considered to have strong regulation if it both limited retailers’ use of free-standing cigarette displays and limited vending machine marketing (a complete ban, a ban except in adult-only establishments or required lockout devices).
The coding of regulations is believed to have good reliability. Interrater reliability testing of the restaurant data, which required assessment of the content of the regulation, showed high reliability (Cronbach = 0.98) [32]. The coding of other regulations was not tested but would be expected to be comparably accurate because only the presence or absence of the regulation was coded. The compliance check measure was based on reports filed monthly by each town's board of health and is therefore more vulnerable to missing or inaccurate reports, but no formal tests were possible. Data on compliance checks were available for 1999–2000, the 2-year period prior to the beginning of the survey. The measures regarding marketing restrictions and smoking restrictions for locations other than restaurants were defined to indicate the existence of a strong policy throughout the same 2-year period. Restaurant policy was measured as of the end of the period (31 December 2000) on the grounds that restaurant smoking bans tend to be controversial and spark public discussion for many months before actual implementation. An alternate specification, in which restaurant and all other regulations were measured at 2 years before the interview, yielded very similar results to those presented here.
The four indicators of strong regulation were positively correlated at moderate levels, with bivariate correlation coefficients ranging from 0.20 to 0.45.
Breadth of strong regulation, the predictor in the multivariate analyses, is the count of the number of domains with strong regulation. We use a three-level measure, where the categories are low breadth of regulation (0 domains of strong regulation), medium breadth (1–2 domains) and high breadth (3–4 domains).
Perceived smoking norms
Indices were constructed to include perceptions of both descriptive and injunctive norms. To represent descriptive norms, we used two survey questions about perceived smoking prevalence. Adults were asked about adult prevalence and youth prevalence: ‘About how many of the adults (teenagers) in [RESPONDENT'S TOWN] smoke cigarettes? Would you say very few, less than half, about half, more than half, or almost all?’ Youth respondents were asked a similar question about adult prevalence; regarding youth prevalence, they were asked about ‘kids your age at your school’. The composite variable for youth excludes the ‘kids your age’ question for respondents who did not attend school in their town of residence (23.7% of the sample) in order to focus on perceptions of norms within the town of residence.
To represent perceived injunctive norms, the adult index used one item: ‘How do most [TOWN] adults that you know feel about smoking in restaurants? Do you think the majority would prefer that smoking be allowed throughout the restaurant, only in special smoking areas, or not at all?’ A similar item was used for youth injunctive norms: ‘How do most adults that you know in [TOWN] feel about other people smoking in restaurants? Would you say most of them don't mind, disapprove a little, or disapprove a lot?’ In addition, the youth survey contained two questions about general approval of smoking by adults and youths, and these were included in the youth norm index: ‘How do most [TOWN] adults that you know feel about teenagers (other adults) smoking? Would you say most of them don't mind, disapprove a little, or disapprove a lot?’
To construct the indices of perceived smoking norms, we coded each item to range from 0.0 to 1.0, with equal intervals between response categories and the value of 1.0 representing the most ‘antismoking’ perception. The indices were computed as the simple average of the five youth items and the three adult items, respectively. If a respondent did not have valid responses for all items, we averaged those items with valid responses. Both indices were scaled by their standard deviation (SD) and centered at zero. Coefficients can therefore be interpreted as the number of SDs of movement on the outcome variable associated with a one-unit change in the predictor variable.
Differences in survey content resulted in an imbalance between the adult and youth indices in the number of injunctive items (one and three items, respectively), which may limit comparability of the measures. We therefore constructed alternative formulations, one that omitted the two ‘extra’ items from the youth measure and another set that weighted the items to give equal weight to descriptive (perceived prevalence) and injunctive norms (perceived disapproval) for both youth and adults. Substituting the alternative indices in the models yielded results that did not differ importantly from those presented below.
Individual-level covariates
We controlled for adults’ smoking status, reasoning that pre-existing smoking behavior may condition respondents’ reactions to policy and that most adults’ smoking behavior was established before local policies. Excluding smoking status from the model made no material difference. Similarly, the youth model included a measure of whether the youths’ parents smoked. Youths’ own smoking status and peer smoking were not included in the youth model because these may have been influenced by policy.
We controlled for five individual-level characteristics that are commonly related to smoking prevalence and might therefore be associated with perceptions of community smoking norms: age (continuous variable), gender, education (no more than high school education versus post-high school), household income (>$50 000 versus income up to $50 000 or unreported) and race (non-Hispanic white versus other). For youths, the value for education was that of the adult informant in the household.
The youth models also included two items designed to measure rebelliousness and sensation seeking. The rebelliousness index [33] averaged six Likert-scaled items concerning conflict at home and willingness to get into trouble outside the home and was scaled 0–1, with 1.0 indicating the most rebellious response. Similarly, the sensation-seeking index averaged four items related to the proclivity for exciting or scary experiences. Both of these constructs have been associated with youth smoking initiation [34, 35].
Town-level covariates
Although the hypothesis being tested is that local policies influence perceived norms, we envisioned a cyclical process in which community norms influence the adoption of policy, and the adopted policy leads in turn to adjustments to the community norms. To represent the norms that predated the policies of interest, we used the ‘percentage of the town's voters supporting Question 1’. Question 1 was a 1992 Massachusetts referendum calling for an increase in tobacco excise taxes. The percentage of town voters supporting Question 1 has been found strongly related to individual residents’ support for a variety of tobacco control policies [36] as well as the towns’ subsequent adoption of local restaurant smoking restrictions [37, 38]. By including the vote on Question 1 as a covariate, we estimated the association between breadth of town tobacco control regulations and perceived smoking norms in 2001/02 over and above the association that can be attributed to pre-existing antismoking norms. Question 1 voting results for Massachusetts’ 351 cities and towns were obtained from the Massachusetts Office of the Secretary of State [39].
Additional covariates were town population (scaled as the natural log to reduce the outlier effect of the largest cities), percentage of population that is white and percentage of population <18 years of age. Previous research has shown these factors to be related to the adoption of regulations and/or to smoking attitudes and behaviors[6, 31, 32, 37, 38]. Data for these variables came from the 2000 US Census.
Analysis
Because the study used data gathered at both town and individual levels, variation both between and within towns must be accounted for. Failure to account for variation at both levels can lead to model misspecification and aggregation bias. To avoid these problems, we have employed a multilevel data analysis strategy, also referred to as hierarchical linear modeling [40]. Further, our analyses must accommodate predictor variables at both the town (e.g. tobacco control policies) and individual (e.g. smoking status) levels, another feature of multilevel models.
Two-level models were estimated, with individual and household characteristics at the first level and town variables at the second. Although the youth survey structure suggests three levels (individuals, households and towns), only 26% of households had more than one youth respondent, providing only minimal data to separately estimate individual and household levels with reliability and precision.
The analysis began with the estimation of a ‘null’, or ‘fully unconditional’, model, which contained no predictor variables. This model was used to partition total variance into its within- and between-town components. This model also provided baselines for both of these variance components so that we could compute how much variance was explained by each of our subsequent models. To the model, we first added all the predictors associated with individuals, then the non-policy variables at the town level and finally the town-level policy variables. We tested the incremental improvement to model fit using a deviance test. We excluded covariates for which the deviance test yielded a P-value >0.05. Lastly, we tested each individual-level covariate to determine whether it should be included as a fixed or random effect; fixed effects were used unless the test indicated that a random effect significantly improved model fit (P < 0.05).
Results
Sample characteristics
Of the 6398 adults included in the analysis, Table I shows that 65% lived in towns with strong regulation in one or two policy domains, while 20% lived in towns with no strong regulations and 15% in towns with strong regulation in three or all four of the domains considered. Relative to adults in other towns, adults living in towns with high breadth of regulation had higher average education and income and lower smoking rates, and their towns more strongly supported Question 1 in 1992. Adults living in towns with medium breadth were more likely to be racial/ethnic minorities, and their towns had larger minority populations. Table II shows similar distributions for the 3670 youths in the sample.
Table I.
Adult sample characteristics, by breadth of strong tobacco control regulation in town of residence
| All respondentsa | Breadth of strong regulation: number of strongly regulated domains | ||||
| 0 | 1–2 | 3–4 | P-valueb | ||
| Unweighted n | 6398 | 1339 | 4168 | 891 | |
| Weighted distribution | 20.16% | 64.57% | 15.27% | ||
| Components of outcome variable: antismoking norms | |||||
| About how many adults in [TOWN] are smokers? | |||||
| Very few or less than half | 57.23% | 55.17% | 53.50% | 76.04% | <0.001 |
| About half, more than half, almost all | 42.77% | 44.83% | 46.50% | 23.96% | |
| About how many teenagers in [TOWN] are smokers? | |||||
| Very few or less than half | 50.84% | 48.41% | 48.04% | 66.02% | <0.001 |
| About half, more than half, almost all | 49.16% | 51.59% | 51.96% | 33.98% | |
| How do most [TOWN] adults that you know feel about smoking in restaurants? | |||||
| Allow anywhere or in special areas | 46.61% | 51.28% | 48.34% | 33.10% | <0.001 |
| Not allow at all | 53.39% | 48.72% | 51.66% | 66.90% | |
| Index of perceived smoking norms | |||||
| Mean | 0.00 | −0.06 | −0.05 | 0.31 | <0.001 |
| SE | 0.01 | 0.03 | 0.02 | 0.03 | |
| Individual-level characteristics | |||||
| Age | |||||
| 18–39 | 43.75% | 38.93% | 46.17% | 39.93% | 0.090 |
| 40+ | 56.25% | 61.07% | 53.83% | 60.07% | |
| Sex | |||||
| Female | 57.47% | 58.03% | 57.30% | 57.44% | 0.100 |
| Male | 42.53% | 41.97% | 42.70% | 42.56% | |
| Education | |||||
| High school graduate or less | 34.06% | 38.91% | 34.60% | 25.38% | 0.003 |
| Some education post-high school | 65.94% | 61.09% | 65.40% | 74.62% | |
| Race | |||||
| White non-Hispanic | 83.97% | 91.88% | 79.86% | 90.96% | 0.004 |
| Other | 16.03% | 8.12% | 20.14% | 9.04% | |
| Income | |||||
| <$50 000 or unreported | 51.55% | 49.99% | 53.16% | 46.83% | 0.179 |
| ≥$50 000 | 48.45% | 50.01% | 46.84% | 53.17% | |
| Smoking status | |||||
| Smoker | 18.51% | 19.85% | 19.13% | 14.08% | 0.001 |
| Non-smoker | 81.49% | 80.15% | 80.87% | 85.92% | |
| Town-level characteristics | |||||
| % Voting for Question 1 | |||||
| Mean | 49.99 | 46.98 | 48.79 | 59.08 | <0.001 |
| SE | 0.17 | 0.35 | 0.19 | 0.35 | |
| Town populationc | |||||
| Mean | 91 482 | 33 085 | 123 517 | 31 730 | 0.360 |
| SE | 2944 | 1232 | 4361 | 864 | |
| % White | |||||
| Mean | 82.21 | 89.62 | 78.02 | 90.32 | 0.026 |
| SE | 0.31 | 0.37 | 0.44 | 0.25 | |
| % <18 years of age | |||||
| Mean | 23.57 | 24.27 | 23.60 | 22.51 | 0.291 |
| SE | 0.08 | 0.15 | 0.10 | 0.25 | |
Omits respondents with missing data on any variable included in the multivariate analysis. Full sample = 6739.
Bivariate test, weighted data, adjusted for clustering.
Modeled as the log of town population.
Table II.
Youth sample characteristics by breadth of strong tobacco control regulation in town of residence
| All respondentsa | Breadth of strong regulation: number of strongly regulated domains | ||||
| 0 | 1–2 | 3–4 | P-valueb | ||
| Unweighted n | 3670 | 796 | 2292 | 582 | |
| Weighted distribution | 21.60% | 63.07% | 15.33% | ||
| Components of outcome variable: antismoking norms | |||||
| About how many of the [TOWN] adults that you know smoke cigarettes? | |||||
| Very few or less than half | 57.05% | 56.64% | 53.56% | 71.97% | <0.001 |
| About half, more than half, almost all | 42.95% | 43.36% | 46.44% | 28.03% | |
| How many kids your age at your school smoke cigarettes? | |||||
| Very few or less than half | 57.91% | 54.12% | 56.72% | 67.56% | 0.002 |
| About half, more than half, almost all | 42.09% | 45.88% | 43.28% | 32.44% | |
| How do most adults that you know in [TOWN] feel about other people smoking in restaurants? | |||||
| Don't mind or disapprove a little | 70.40% | 73.70% | 70.43% | 65.61% | 0.013 |
| Disapprove a lot | 29.60% | 26.30% | 29.57% | 34.39% | |
| How do most [TOWN] adults that you know feel about other adults smoking? | |||||
| Don't mind or disapprove a little | 83.56% | 85.49% | 83.87% | 79.54% | 0.003 |
| Disapprove a lot | 16.44% | 14.51% | 16.13% | 20.46% | |
| How do most [TOWN] adults that you know feel about teenagers smoking? | |||||
| Don't mind or disapprove a little | 30.55% | 31.69% | 31.88% | 23.49% | 0.033 |
| Disapprove a lot | 69.45% | 68.31% | 68.12% | 76.51% | |
| Index of perceived smoking norms | |||||
| Mean | 0.00 | −0.06 | −0.05 | 0.28 | <0.001 |
| SE | 0.02 | 0.04 | 0.02 | 0.04 | |
| Individual-level characteristics | |||||
| Age | |||||
| 12–14 | 50.16% | 51.81% | 49.33% | 51.23% | 0.287 |
| 15–17 | 49.84% | 48.19% | 50.67% | 48.77% | |
| Sex | |||||
| Female | 48.41% | 49.01% | 48.55% | 46.98% | 0.814 |
| Male | 51.59% | 50.99% | 51.45% | 53.02% | |
| Race | |||||
| White non-Hispanic | 78.15% | 87.11% | 72.89% | 87.12% | 0.004 |
| Other | 21.85% | 12.89% | 27.11% | 12.88% | |
| Rebelliousness | |||||
| Below mean | 47.13% | 46.59% | 47.69% | 45.55% | 0.501 |
| Above mean | 52.87% | 53.41% | 52.31% | 54.45% | |
| Sensation seeking | |||||
| Below mean | 44.14% | 44.22% | 44.74% | 41.54% | 0.430 |
| Above mean | 55.86% | 55.78% | 55.26% | 58.46% | |
| Education of adult informant | |||||
| High school graduate or less | 33.18% | 32.20% | 36.87% | 19.40% | <0.001 |
| Some education post-high school | 66.82% | 67.80% | 63.13% | 80.60% | |
| Household income | |||||
| <$50 000 or unreported | 46.59% | 44.54% | 49.60% | 37.14% | 0.006 |
| ≥ $50 000 or more | 53.41% | 55.46% | 50.40% | 62.86% | |
| At least one parent smokes | |||||
| Yes | 32.02% | 33.24% | 33.35% | 24.84% | 0.017 |
| No | 67.98% | 66.76% | 66.65% | 75.16% | |
| Town-level characteristics | |||||
| % Voting for Question 1 | |||||
| Mean | 50.19 | 48.40 | 48.85 | 58.19 | <0.001 |
| SE | 0.15 | 0.32 | 0.17 | 0.35 | |
| Town populationc | |||||
| Mean | 73 051 | 27 381 | 98 657 | 32 061 | 0.060 |
| SE | 2350 | 1053 | 3590 | 850 | |
| % White | |||||
| Mean | 84.45 | 90.63 | 80.83 | 90.62 | 0.037 |
| SE | 0.27 | 0.33 | 0.40 | 0.22 | |
| % <18 years of age | |||||
| Mean | 24.44 | 24.77 | 24.52 | 23.66 | 0.008 |
| SE | 0.07 | 0.13 | 0.08 | 0.21 | |
Omits respondents with missing data on any variable included in the multivariate analysis. Full sample = 3862.
Bivariate test, weighted data, adjusted for clustering.
Modeled as the log of town population.
For both adults and youths, the tables show that mean values of the smoking norm indices were higher in towns with greater breadth of strong regulation. This pattern is visible for each of the items that make up the indices. All associations were statistically significant (P < 0.05) in bivariate tests.
Model construction results
Results of the model construction procedure indicated that 24.6% of all variation in adults’ perceived smoking norms was at the town rather than the individual level, while only 8.6% of variation in youths’ perceptions was between-town variation. This means that town-level factors played a more important role in adults’ than youths’ perceptions.
A substantial portion of the between-town variation was due to differences in the characteristics of the residents included in the survey. Entering all individual-level variables in the model accounted for 26.6% of the between-town variation for adults and 47.9% for youth (Table III). Town-level demographic characteristics and support for Question 1 in 1992 were also important. Adding these town-level variables explained an additional 38.3% of the between-town variation for adults and 42.0% for youth.
Table III.
Model performance in explaining between-town variance
| Percentage of between-town variance explained | ||||
| Estimated between-town variance | Relative to Model 1, null model (%) | Relative to Model 2, individual-level variables (%) | Relative to Model 3, individual plus non-policy town-level variables (%) | |
| Adult model | ||||
| Null model, no variables | 0.1013 | |||
| Individual-level variables | 0.0744 | 26.6 | ||
| Individual plus non-policy town-level variables | 0.0356 | 64.9 | 52.1 | |
| All variables, including policy | 0.0344 | 66.1 | 53.8 | 3.4 |
| Youth model | ||||
| Null model, no variables | 0.0823 | |||
| Individual-level variables | 0.0429 | 47.9 | ||
| Individual plus non-policy town-level variables | 0.0083 | 89.9 | 80.6 | |
| All variables, including policy | 0.0073 | 91.1 | 83.0 | 12.1 |
Finally, adding the policy variables (medium and high breadth of strong regulation) explained 3.4% of the remaining between-town variation for adults (1.2% of adults’ total between-town variation) and 12.1% of the remaining between-town variation for youths (1.2% of total between-town variation). The full models thus explained a considerably greater proportion of the between-town variation in youths’ than adults’ perceptions of smoking norms (91.1 versus 66.1%). The remaining variation represents the influence of omitted factors and measurement error.
Deviance tests showed that random coefficients added significantly (P < 0.05) to model fit for all individual-level variables in the adult model. For the youth model, the only random coefficient adding significantly to model fit was that for parental smoking. Thus, the adult model was estimated with random coefficients for all individual and household variables, while only the intercept and parental smoking coefficients were treated as random in the youth model.
Multivariate results
The final multilevel model for adults (Table IV) indicated a significant positive association between the presence of strong local tobacco control policies and adults’ perceptions of community smoking norms. Adults living in towns with either high or medium breadth of strong policy perceived their communities’ norms as more antismoking than adults with low breadth (P = 0.004 for high breadth, 0.025 for medium breadth). High breadth (three to four domains of strong regulation) relative to low breadth (zero domains) added 13% of a SD in perceived norms. This was somewhat less than the effect of having a high school education (19% of a SD) and about the same as the effect of being a non-smoker (12%).
Table IV.
Results of multilevel regressions on adult index of perceived antismoking norms
| Coefficient | SEa | P-value | |
| Predictor: breadth of tobacco control regulation in respondent's town (number of strongly regulated domains) | |||
| Medium breadth (one to two domains) | 0.078 | 0.035 | 0.025 |
| High breadth (three to four domains) | 0.133 | 0.046 | 0.004 |
| Individual characteristics | |||
| Age | 0.007 | 0.001 | 0.000 |
| Female | 0.064 | 0.023 | 0.005 |
| More than high school | 0.186 | 0.025 | 0.000 |
| White non-Hispanic | 0.127 | 0.042 | 0.003 |
| Income > $50 000 | 0.174 | 0.023 | 0.000 |
| Smoker | −0.120 | 0.021 | 0.000 |
| Town-level characteristics | |||
| % Voting for Question 1 | 0.021 | 0.002 | 0.000 |
| Town population (natural log) | −0.059 | 0.014 | 0.000 |
| % White | b | ||
| % <18 years of age | b | ||
| Unweighted n | 6398 | ||
Robust SEs are reported, as these are less sensitive to violation of model assumptions.
Did not meet criterion for inclusion (P < 0.05 for improvement in model fit).
For youths, high breadth of strong regulation showed a significant positive association with perceived norms (P = 0.043). The association was not significant for medium breadth (P = 0.276). The estimated effect of living in a town with high versus low breadth of regulation was 11% of a SD, which is at the low end of the range of effects of the individual-level covariates.
The hypothesis that multiple regulations will have a cumulative effect on perceived norms was tested with mixed results. A continuous version of the predictor variable (number of domains of strong regulation, ranging from zero to four) showed a significant effect for adults (P < 0.001) and an effect approaching statistical significance for youth (P = 0.065), suggesting a linear association. On the other hand, when using the categorical coding of the predictor, the estimated effect of high breadth of regulation was not significantly different from the effect for medium breadth for either adults or youth (P > 0.15).
The measures of policy breadth included clean indoor air regulation (two domains) and youth-oriented marketing regulation (two domains) without distinguishing between them. A supplementary analysis, not shown, separated clean air and youth access regulations. For each, we created a three-level ordinal variable indicating whether the respondent's town had strong regulations in none, one or both domains. We substituted these measures, together and separately, for the policy variables in Tables IV and V. Both the clean air and the marketing regulation coefficients were positive and statistically significant in the adult models, and both were positive but non-significant in the youth models. It thus appears that both types of regulation contributed to the overall association between breadth of regulation and perceived norms.
Table V.
Results of multilevel regressions on youth index of perceived antismoking norms
| Coefficient | SEa | P-value | |
| Predictor: breadth of tobacco control regulation in respondent's town (number of strongly regulated domains) | |||
| Medium breadth (one to two domains) | 0.041 | 0.038 | 0.276 |
| High breadth (three to four domains) | 0.106 | 0.052 | 0.043 |
| Individual characteristics | |||
| Age | −0.119 | 0.008 | 0.000 |
| Female | b | ||
| White non-Hispanic | b | ||
| Rebelliousness | −0.483 | 0.071 | 0.000 |
| Sensation seeking | −0.186 | 0.053 | 0.001 |
| Adult informant has above high school education | 0.132 | 0.033 | 0.000 |
| Household income > $50 000 | 0.111 | 0.028 | 0.000 |
| At least one parent smokes | −0.486 | 0.036 | 0.000 |
| Town-level characteristics | |||
| % Voting for Question 1 | 0.019 | 0.002 | 0.000 |
| Town population (natural log) | b | ||
| % White | b | ||
| % <18 years of age | 0.013 | 0.004 | 0.001 |
| Unweighted n | 3670 | ||
Robust SEs are reported, as these are less sensitive to violation of model assumptions.
Did not meet criterion for inclusion (P < 0.05 for improvement in model fit).
Several individual-level factors were related to the indices of perceived smoking norms. Community norms were perceived as significantly more antismoking by adults who were non-smokers, older people, more educated people, females and non-Hispanic whites. Among youth, norms were perceived to be more antismoking by those who were younger, less rebellious and less sensation seeking and by those whose parents were more educated, more affluent and non-smokers.
Town-level support for Question 1 in 1992 was a significant predictor of adults’ and youths’ perceptions of norms a decade later. In addition, norms were perceived as more antismoking by adults in towns with smaller populations and by youth in towns with high proportions of the population <18 years of age.
Discussion
These results are consistent with the hypothesis that local tobacco control regulations influence residents’ perceptions of community smoking norms. Massachusetts adults and youths living in towns that had strong tobacco regulations in at least three of the four measured policy domains perceived their towns’ norms to be more antismoking than those living in towns with no strongly regulated domains. Among adults, but not youth, this effect existed even when only one to two domains were strongly regulated. Although such effects have not previously been shown for tobacco regulations, they are consistent with empirical and theoretical literature in a number of policy fields, suggesting that government regulations act as statements of norms that influence perceptions and behaviors [20, 21, 41, 42].
Local tobacco regulations may have less influence on the perceptions of youths than adults, as suggested by the non-significant effect of medium breadth of regulation in the youth model. In the variance components analysis, between-town variation represented a smaller proportion of the variance in youths’ than adults’ perceptions of norms. Therefore, youths may be less sensitive than adults to the implications of local regulations.
Both clean air regulations and youth access regulations contribute to adults’ and youths’ perceptions of community smoking norms. The contribution of youth access regulations is somewhat surprising, because these regulations do not target adults and have not generally been found to affect youth smoking [3]. The results suggest a positive effect of youth access regulations but, compared with clean air policies, a more modest, indirect or slowly developing benefit.
The influence of local regulations on perceived norms may represent an opportunity for health promotion. Publicity about the adoption and enforcement of regulations, visible reminders such as ‘No Smoking’ signs and community education activities may increase the awareness of regulations and hence their effect on perceptions of community smoking norms. Because signals about community norms are relevant to the full population, communications should not be limited to those groups directly targeted by regulations.
Communicating these signals effectively may require more effort for youths than for adults. Extra effort may also be required for groups that tend to perceive fewer antismoking norms. This includes adult smokers and children whose parents smoke, adults and children in households with less education and income, younger adults and older youths, rebellious youths, adult males and minority adults.
This study has several limitations. First, the cross-sectional study design cannot exclude the possibility that the observed association between policy and perceived norms could stem from reverse causation (i.e. communities with stronger antismoking norms may have adopted stronger policies and respondents’ perceptions may have reflected pre-existing norms). Although this possibility cannot be completely ruled out, a strength of this analysis is that it controlled for a town-level measure of antismoking sentiment that predated most local tobacco regulations. Together with other individual- and town-level covariates in the model, this minimizes the chances that the regulations simply serve as proxies for pre-existing norms. Another limitation is the rather simplistic measure of the breadth of local policy. A more sensitive measure might include information on the content of a regulation (e.g. whether a licensing requirement included a provision for license suspensions), the extent of public debate surrounding adoption of the regulation and the stringency of enforcement, but systematic data on these points were not available. The length of time a regulation has been in effect may condition its influence on perceived norms, and this was not examined in the analysis.
A number of factors that might influence perceived smoking norms are not included in our models. Local factors could include unsuccessful efforts to adopt regulations, grassroots antismoking campaigns, tobacco industry lobbying and tobacco education in school. Systematic data were not available on these factors, and omitting them could lead to bias if they were correlated with local tobacco regulations. Omitted statewide or national factors—such as tobacco taxation, mass media antitobacco campaigns or tobacco marketing—would not generally be expected to influence the relative effect of differing local policies. However, Massachusetts had a strong statewide tobacco control program before and during the survey period, and the local regulations examined here might make a greater difference in a policy environment with weaker state-level programs.
Only a few items were available for use in the indices of perceived norms. More sensitive measures of policy breadth and/or perceived norms might reveal stronger associations than those observed here.
In sum, this study provides evidence supporting the hypothesis that local communities can influence adults’ and youths’ perceptions of community smoking norms by adopting a broad array of strong tobacco control regulations. The findings are consistent with the proposition that local tobacco control regulations influence smoking rates through the indirect mechanism of influencing perceived norms, but it remains for future research to determine more fully whether this accounts for some of the impact of policy on smoking behavior.
Funding
National Cancer Institute's Tobacco Research Initiative for State and Community Interventions (CA86257).
Conflict of interest statement
None declared.
References
- 1.US Department of Health Education and Welfare. Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service. Washington, DC: US Government Printing Office; 1964. [Google Scholar]
- 2.US Department of Health and Human Services. The Health Consequences of Smoking: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2004. [Google Scholar]
- 3.Levy DT, Chaloupka F, Gitchell J. The effects of tobacco control policies on smoking rates: a tobacco control scorecard. J Public Health Manag Pract. 2004;10:338–53. doi: 10.1097/00124784-200407000-00011. [DOI] [PubMed] [Google Scholar]
- 4.McMullen KM, Brownson RC, Luke D, et al. Strength of clean indoor air laws and smoking related outcomes in the USA. Tob Control. 2005;14:43–8. doi: 10.1136/tc.2004.007880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Siegel M. The effectiveness of state-level tobacco control interventions: a review of program implementation and behavioral outcomes. Annu Rev Public Health. 2002;23:45–71. doi: 10.1146/annurev.publhealth.23.092601.095916. [DOI] [PubMed] [Google Scholar]
- 6.Siegel M, Albers AB, Cheng DM, et al. Effect of local restaurant smoking regulations on progression to established smoking among youths. Tob Control. 2005;14:300–6. doi: 10.1136/tc.2005.012302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wakefield M, Chaloupka F, Kaufman N, et al. Effect of restrictions on smoking at home, at school, and in public places on teenage smoking: cross-sectional study. Br Med J. 2000;321:333–7. doi: 10.1136/bmj.321.7257.333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Wiium N, Torsheim T, Wold B. Normative processes and adolescents’ smoking behaviour in Norway: a multilevel analysis. Soc Sci Med. 2006;62:1810–8. doi: 10.1016/j.socscimed.2005.08.029. [DOI] [PubMed] [Google Scholar]
- 9.van den Putte B, Yzer MC, Brunsting SB. Social influences on smoking cessation: a comparison of the effect of six social influence variables. Prev Med. 2005;41:186–93. doi: 10.1016/j.ypmed.2004.09.040. [DOI] [PubMed] [Google Scholar]
- 10.Alamar B, Glantz SA. Effect of increased social unacceptability of cigarette smoking on reduction in cigarette consumption. Am J Public Health. 2006;96:1359–63. doi: 10.2105/AJPH.2005.069617. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sussman S, Dent CW, Mestel-Rauch J, et al. Adolescent nonsmokers, triers and regular smokers’ estimates of cigarette smoking prevalence: when do overestimations occur and by whom? J Appl Soc Psychol. 1988;18:537–51. [Google Scholar]
- 12.US Department of Health and Human Services. Preventing Tobacco Use among Young People: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1994. [Google Scholar]
- 13.Eisenberg ME, Forster JL. Adolescent smoking behavior: measures of social norms. Am J Prev Med. 2003;25:165–6. doi: 10.1016/s0749-3797(03)00116-8. [DOI] [PubMed] [Google Scholar]
- 14.Alesci NL, Forster JL, Blaine T. Smoking visibility, perceived acceptability, and frequency in various locations among youth and adults. Prev Med. 2003;36:272–81. doi: 10.1016/s0091-7435(02)00029-4. [DOI] [PubMed] [Google Scholar]
- 15.Kim S-H, Shanahan J. Stigmatizing smokers: public sentiment toward cigarette smoking and its relationship to smoking behaviors. J Health Commun. 2003;8:343–67. doi: 10.1080/10810730305723. [DOI] [PubMed] [Google Scholar]
- 16.Todd M, Chassin L, Presson CC, et al. Role stress, role socialization, and cigarette smoking: examining multiple roles and moderating variables. Psychol Addict Behav. 1996;10:211–21. [Google Scholar]
- 17.Jackson KM, Aiken LS. A psychosocial model of sun protection and sunbathing in young women: the impact of health beliefs, attitudes, norms, and self-efficacy for sun protection. Health Psychol. 2000;19:469–78. doi: 10.1037//0278-6133.19.5.469. [DOI] [PubMed] [Google Scholar]
- 18.Hill D, Rassaby J, Gardner G. Determinants of intentions to take precautions against skin cancer. Community Health Stud. 1984;8:33–44. doi: 10.1111/j.1753-6405.1984.tb00422.x. [DOI] [PubMed] [Google Scholar]
- 19.Susser M. Editorial: the tribulations of trials—intervention in communities. Am J Public Health. 1995;85:156–8. doi: 10.2105/ajph.85.2.156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sunstein CR. On the expressive function of law. Univ PA Law Rev. 1996;144:2021–32. [Google Scholar]
- 21.McAdams RH. An attitudinal theory of expressive law. Oregon Law Rev. 2000;79:339–90. [Google Scholar]
- 22.Lynch BS, Bonnie RJ, editors. Committee on Preventing Nicotine Addiction in Children and Youths, Institute of Medicine. The publisher is listed as Washington DC: National Academy Press, 1994. [PubMed] [Google Scholar]
- 23.Brownson RC, Koffman DM, Novotny TE, et al. Environmental and policy interventions to control tobacco use and prevent cardiovascular disease. Health Educ Q. 1995;22:478–98. doi: 10.1177/109019819502200406. [DOI] [PubMed] [Google Scholar]
- 24.Gilpin EA, Lee L, Pierce JP. Changes in population attitudes about where smoking should not be allowed: California versus the rest of the USA. Tob Control. 2004;13:38–44. doi: 10.1136/tc.2003.004739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Heloma A, Jaakkola MS. Four-year follow-up of smoke exposure, attitudes and smoking behaviour following enactment of Finland's national smoke-free work-place law. Addiction. 2003;98:1111–7. doi: 10.1046/j.1360-0443.2003.00429.x. [DOI] [PubMed] [Google Scholar]
- 26.Tang H, Cowling DW, Lloyd JC, et al. Changes of attitudes and patronage behaviors in response to a smoke-free bar law. Am J Public Health. 2003;93:611–7. doi: 10.2105/ajph.93.4.611. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Fong GT, Hyland A, Borland R, et al. Reductions in tobacco smoke pollution and increases in support for smoke-free public places following the implementation of comprehensive smoke-free workplace legislation in the Republic of Ireland: findings from the ITC Ireland/UK survey. Tob Control. 2006;15(Suppl. 3):iii51–8. doi: 10.1136/tc.2005.013649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.National Cancer Institute. State and Local Legislative Action to Reduce Tobacco Use. Smoking and Tobacco Control Monograph No. 11. Bethesda, MD: US Department of Health and Human Services, National Institutes of Health, National Cancer Institute; 2000. NIH Publication No. 00–4804. [Google Scholar]
- 29.US Department of Health and Human Services. The Health Consequences of Involuntary Exposure to Tobacco Smoke: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, Coordinating Center for Health Promotion, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2006. [Google Scholar]
- 30.US Department of Health and Human Services. Reducing Tobacco Use: A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2000. [Google Scholar]
- 31.Albers AB, Siegel M, Cheng DM, et al. Relation between local restaurant smoking regulations and attitudes towards the prevalence and social acceptability of smoking: a study of youths and adults who eat out predominantly at restaurants in their town. Tob Control. 2004;13:347–55. doi: 10.1136/tc.2003.007336. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Skeer M, Siegel M. The descriptive epidemiology of local restaurant smoking regulations in Massachusetts: an analysis of the protection of restaurant customers and workers. Tob Control. 2003;12:221–6. doi: 10.1136/tc.12.2.221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Sussman S, Dent CW, Galaif ER. The correlates of substance abuse and dependence among adolescents at high risk for drug abuse. J Subst Abuse. 1997;9:241–55. doi: 10.1016/s0899-3289(97)90019-5. [DOI] [PubMed] [Google Scholar]
- 34.Stephenson MT, Hoyle RH, Palmgreen P, et al. Brief measures of sensation seeking for screening and large-scale surveys. Drug Alcohol Depend. 2003;72:279–86. doi: 10.1016/j.drugalcdep.2003.08.003. [DOI] [PubMed] [Google Scholar]
- 35.Zuckerman M, Ball S, Black J. Influences of sensation seeking, gender, risk appraisal, and situational motivation on smoking. Addict Behav. 1990;15:209–20. doi: 10.1016/0306-4603(90)90064-5. [DOI] [PubMed] [Google Scholar]
- 36.Hamilton WL, Biener L, Rodger CN. Who supports tobacco excise taxes? Factors associated with towns’ and individuals’ support in Massachusetts. J Public Health Manag Pract. 2005;11:333–40. doi: 10.1097/00124784-200507000-00012. [DOI] [PubMed] [Google Scholar]
- 37.Bartosch WJ, Pope GC. Local restaurant smoking policy enactment in Massachusetts. J Public Health Manag Pract. 1999;5:63–73. doi: 10.1097/00124784-199901000-00010. [DOI] [PubMed] [Google Scholar]
- 38.Skeer M, George S, Hamilton WL, et al. Town-level characteristics and smoking policy adoption in Massachusetts: are local restaurant smoking regulations fostering disparities in health protection? Am J Public Health. 2004;94:286–92. doi: 10.2105/ajph.94.2.286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Elections Division. Public Document No. 43. Massachusetts Elections Statistics 1992. Boston, MA: Office of Massachusetts Secretary of State; 1992. [Google Scholar]
- 40.Raudenbush SW, Bryk AS. Hierarchical Linear Models: Applications and Data Analysis Methods. 2nd edn. Thousand Oaks, CA: Sage Publications; 2002. [Google Scholar]
- 41.Carlson AE. Classifying social norms. In: Chen J, editor. The Jurisdynamics of Environmental Protection: Change and the Pragmatic Voice in Environmental Law. Washington, DC: Environmental Law Institute; 2003. [Google Scholar]
- 42.Opp K-D. When do norms emerge by human design and when by the unintended consequences of human action? The example of the no-smoking norm. Ration Soc. 2002;14:131–58. [Google Scholar]