In 2016, 48% of passenger vehicle occupants (PVOs) killed in crashes in the United States were unrestrained drivers and passengers (National Highway Traffic Safety Administration, 2017). With overall levels of seat belt use at 90% in 2017, this means that the remaining 10% of the population accounts for almost half of all passenger vehicle occupant deaths in the United States (National Highway Traffic Safety Administration, 2016a).

Given the life-saving potential of seat belts, public health and transportation professionals have sought to identify strategies to improve seat belt use among drivers and passengers. Some of the most effective population-based interventions have been the implementation of seat belt laws and the enhanced enforcement of such laws (Dinh-Zarr et al., 2001; Goodwin et al., 2015; Lee et al., 2015). These interventions have been shown to increase seat belt use as well as decrease crash-related injuries and deaths. Primary enforcement seat belt laws, which allow police officers to stop vehicles and issue tickets when lack of seat belt use is observed, are more effective than secondary enforcement seat belt laws, which only allow police officers to issue tickets after the vehicle has been stopped for another reason (Beck, Shults, Mack, & Ryan, 2007; Dinh-Zarr et al., 2001; Sunshine, Dwyer-Lindgren, Chen, & Mokdad, 2017). Other factors, such as the amount of the fine and whether the seat belt law covers all seating positions (front and rear seats) or only the front seats, have been shown to affect seat belt use as well (Bhat, Beck, Bergen, & Kresnow, 2015; Goodwin et al., 2015; Houston & Richardson, 2005; Nichols et al., 2010). Similarly, enhanced enforcement of these laws, which involves a period of increased levels of enforcement accompanied by communications and outreach (in the form of both paid advertising and earned media), is associated with higher seat belt use (Dinh-Zarr et al., 2001; Goodwin et al., 2015; Nichols & Ledingham, 2008).

While seat belt use has reached record levels overall with the implementation of strategies such as those mentioned above, key populations continue to travel unrestrained. Groups with lower levels of seat belt use include men, young adults (18–34 years of age), obese people, rear seat passengers, and rural residents (Beck, Downs, Stevens, & Sauber-Schatz, 2017; Bhat et al., 2015; National Highway Traffic Safety Administration, 2017; Strine et al., 2010). There remains a critical need to identify approaches that can improve seat belt use among these at-risk populations.

The purpose of the current study was to investigate the association between the belief that seat belt use is important and the behavior of always wearing a seat belt among adults in the United States. Belief and behavior were investigated separately for front and rear passenger seats.

We used data from Porter Novelli Public Services (2012), the most recent year for which data were available. Knowledge Networks: A GfK Company collected the data for Porter Novelli, randomly recruiting participants through probability-based sampling using random-digit dial and address-based sampling methods. Surveys were completed electronically, and households without existing Internet access were provided with laptop computers and access to the Internet. The Summer ConsumerStyles survey was fielded from June 19 to July 3, 2012 to 6,402 adults (18 years or older) who had previously participated in the Spring ConsumerStyles survey. A total of 4,170 surveys were completed, for a response rate of 65%. Data were weighted to match the U.S. Current Population Survey proportions for sex, age, household income, race/ethnicity, household size, education level, census region, metropolitan status, and whether the respondent had internet access prior to joining the panel (Porter Novelli Public Services, 2012). CDC licensed the 2012 Summer ConsumerStyles survey data file (without personal identifiers) from Porter Novelli. Because CDC licensed previously collected data for secondary analysis, the project was exempt from institutional review board approval.

Survey respondents were asked two questions for each of three seating positions (driver, front seat passenger, and back seat passenger). They were asked both about their belief in the importance of seat belt use and about their seat belt wearing behavior. All analyses were stratified by seating position. Preliminary analyses indicated a strong correlation between seat belt use among drivers and front seat passengers (Spearman correlation: 86%) and between belief in importance of seat belt use among drivers and front seat passengers (Spearman correlation: 97%). Further, seat belt use and belief among front seat passengers was close to 100% predictive of the same in drivers (Gamma statistic). For these reasons, subsequent analyses focused only on front and rear seat passengers.

To measure belief, respondents were asked “how important is it to wear seat belts in the driver seat/front passenger seat/back seat of a car, truck, van, or sport utility vehicle (SUV).” Responses were assessed on a 5-point scale ranging from “not at all important” to “very important.” Responses were dichotomized (very important vs. less than very important) for analysis. To measure seat belt use, respondents were asked “how often do you wear seat belts when you drive/ride in the front passenger seat/ride in the back seat of a car, truck, van, or SUV.” Response options included always, nearly always, sometimes, seldom, never, or never ride in driver/front passenger seat/back seat. Those who reported that they never rode in the seat were excluded from analyses of that seating position (n=21 for front passenger seat, n=198 for rear passenger seat). Responses were then dichotomized (always vs. less than always) for analysis.

We examined the bivariable association between seat belt use (defined as always wears) and belief about seat belt use (very important or less than very important), type of state seat belt law, and a number of respondent- and household-level characteristics identified in the literature as being associated with seat belt use. These included sex, age group (18–24 years, 25–44 years, 45–64 years, 65+ years), racial/ethnic group (white non-Hispanic, black non-Hispanic, Hispanic, other non-Hispanic [American Indian/Alaska Native, Native Hawaiian/Pacific Islander, and multiracial]), highest level of education completed (high school graduate or less, some college, college graduate), household income group (less than $25,000, $25,000 to less than $50,000, $50,000 to less than $75,000, $75,000 or more), and region of residence (North-east, Midwest, South, West) along with several dichotomous yes/no variables (currently married, currently employed, living in a metropolitan statistical area [MSA]). State seat belt laws for 2012 were identified with information from Insurance Institute for Highway Safety, which maintains a list of traffic safety law characteristics by state (Insurance Institute for Highway Safety, 2018). For analyses of front seat passenger belt use, state seat belt law was grouped into two categories (primary or secondary enforcement for adult use). Because only one state (New Hampshire) had no law for adult use of seat belts, it was grouped with the secondary enforcement states for front seat passenger analyses (Fig. 1). For analyses of rear seat passenger belt use, state seat belt law was grouped into three categories (primary, secondary, or none for adult use; Fig. 2). Weighted percentages and 95% confidence intervals were computed for seat belt use along with chi-square tests of association for categorical variables. Linear trends were assessed where appropriate using the Cochran-Armitage test for linear trend.

Multivariable regression was conducted using the log-binomial model with a log link function to assess the association between belief and behavior regarding seat belt use, adjusting for state law and other demographic variables. Respondent sex, racial/ethnic group, and age group were forced into each model. Variables that were significant in bivariable analysis were also included in each initial model. Two-way interactions between belief and demographic characteristics found to be significant in preliminary analyses were also included in the initial models (front seat passengers: belief by state law; rear seat passengers: belief by sex, belief by state law, belief by region). Age group was subsequently removed from the initial model for front seat passengers due to model convergence issues. Interactions were assessed first using a backward stepwise approach followed by main effects, with non-significant predictors being removed in a backward stepwise manner. In addition to those variables forced into the model, variables with p-values < 0.05, Wald chi-square test, were retained in the final model as were any variables involved in two-way interactions, regardless of their significance, in order to keep the models hierarchical in nature. Results are presented in the form of adjusted prevalence ratios (adjPRs) and 95% confidence intervals (CIs). All analyses were conducted using Statistical Analysis Software (SAS) version 9.3 (SAS Institute, Inc., Cary, North Carolina).

Overall, the weighted sample was comprised of a slightly higher proportion of females (51.8%) than males (48.2%), and approximately 70% of the sample was aged 25–64 years (Table 1). The sample was predominately white/non-Hispanic (67.2%), married (53.8%), employed (55.5%), and living in an MSA (83.9%). More than one-third (37.2%) lived in the South, 18.1% lived in the Northeast, 21.6% lived in the Midwest, and 23.1% lived in the West. A total of 42.8% had at most a high school education, and 38.6% reported a household income of $75,000 or higher. Overall, 75.0% lived in states with a primary enforcement law for seat belt use in front seats. In contrast, only 38.5% lived in states with a primary enforcement seat belt law for the rear seats, 12.7% lived in states with a secondary enforcement law for rear seats, and almost half (48.7%) lived in states without laws that covered seat belt use in the rear seating position (Table 1).

The current study found a strong association between belief about the importance of seat belt use and seat belt wearing behavior and further demonstrated that this relationship existed for both front and rear seat passengers. Previous research has also found that positive beliefs about seat belts (such as believing that seat belts are important for one’s health) increased the likelihood of seat belt use (Boyle & Lampkin, 2008; Steptoe et al., 2002). In a 2016 survey of adults who did not always buckle up in the rear seat, common reasons for not doing so included beliefs that the rear seat was safer than the front, a crash was unlikely, or they were not needed because of the type of trip (e.g., short distance; Jermakian & Weast, 2018).

Belief in the importance of seat belt use was the strongest predictor of use for both seating positions in the current study, but the strength of that relationship varied by several important factors. In many cases, the belief-behavior relationship was strongest for groups with lower levels of belt use. For example, among front seat passengers, the association was stronger for residents of secondary lawstates (where those who reported belief that seat belt use is very important were almost 3 times more likely towear seat belts than thosewho did not report that belief) than for residents of primary law states (where thosewho reported belief that seat belt use is very important were 1.6 times more likely to wear seat belts than those who did not report that belief). The relationship was more complicated for rear seat passengers. In each of the four Census regions, both males and females who reported the belief that seat belt use is very important were more likely to buckle up than those who did not report that belief. However, the belief–behavior association was stronger for males and females in the Northeast (adjPR 5.52 and 8.55, respectively) than for those in the South (adjPR 2.68 and 4.14, respectively) or West (adjPR 2.24 and 3.46, respectively).

Despite the high overall level of seat belt use in the United States, certain groups are less likely to buckle up than others. The current study found that rear seat passengers were significantly less likely to always wear seat belts (62%) than were front seat passengers (86%), which is consistent with previous reports (Boyle & Lampkin, 2008; Jermakian & Weast, 2018). Similarly, only 71% of respondents believed that seat belt use is very important in the rear seating position, compared with 84% who believed the same for the front passenger seat. This is particularly concerning given recent research that finds gains in occupant safety for the front seat have outpaced those for the rear seat, and (depending on the occupant’s age) the current vehicle fleet may not offer added protection in the rear seat (Durbin et al., 2015). Perceptions that the rear seat is safer than the front may be based on data from older vehicle models. In addition to the changes in relative safety of rear versus front seats, the growing popularity of ride-hailing services (Clewlow & Mishra, 2017) may lead to an increased proportion of adults who ride in rear seats. A 2016 study found that, among adults who had ridden in the rear seat in the past six months, 12% primarily rode in a hired vehicle (i.e., taxi or ride-hailing vehicle; Jermakian & Weast, 2018). The same study also found that seat belt use in the rear seat was lower among those who primarily used hired vehicles, compared to those who primarily used private passenger vehicles (Jermakian & Weast, 2018). People who use taxis or ride-hailing services may be an important target population for messaging about the importance of buckling up in the rear seat.

The study findings, in concert with previous research, suggest that interventions designed to change beliefs about the importance of seat belt use may have potential to change behavior, including in states without primary enforcement laws. With careful attention to messaging, mass media campaigns have been shown to promote positive health behaviors, including seat belt use (Wakefield, Loken, & Hornik, 2010). Mass media campaigns are most likely to be successful in conjunction with supportive resources (Wakefield et al., 2010). Enforcement of seat belt or alcohol-impaired driving laws that is accompanied by well-designed media campaigns is associated with reduced crash-related injuries and fatalities as well as increased prevalence of safety behaviors (Bergen et al., 2014; Dinh-Zarr et al., 2001; Piontkowski et al., 2015). For example, a combined media and enforcement campaign for Click it or Ticket in Nevada significantly increased seat belt use along with the belief that it is important for police to enforce seat belt laws (Vasudevan, Nambisan, Singh, & Pearl, 2009).

Effective media campaigns use theory as a conceptual foundation to identify which components of behavior change to target with messages and ensure that messages will guide the audience through behavior change (Noar, 2006). Past studies have used theoretical frameworks, including the Theory of Planned Behavior, the Health Belief Model, and Social Norms, to understand seat belt use. Positive attitudes (e.g., perceptions of seat belts as effective, comfortable to wear) were associated with an increased use of seat belts (Budd, North, & Spencer, 1984; Fhaner & Hane, 1974; Jonah & Dawson, 1982; Şimşekoğlu & Lajunen, 2008; Stasson & Fishbein, 1990). Perceived behavioral norms – that is, an individual’s perception of whether his/her peers engage (or not) in a given behavior – were associated with seat belt use, both for adults (Jonah & Dawson, 1982) and adolescents (Dunlop & Romer, 2010; Litt, Lewis, Linkenbach, Lande, & Neighbors, 2014).

Many of these theory-based studies were conducted in the United States prior to widespread passage of seat belt laws (Stasson & Fishbein, 1990) or in other countries (Budd et al., 1984; Fhaner & Hane, 1974; Jonah & Dawson, 1982; Şimşekoğlu & Lajunen, 2008). In addition, the studies were primarily conducted with samples of teens or college students (Budd et al., 1984; Dunlop & Romer, 2010; Litt et al., 2014; Şimşekoğlu & Lajunen, 2008; Stasson & Fishbein, 1990). Since the time of the Stasson and Fishbein (1990) study, both seat belt laws and social norms about seat belt wearing have changed in the United States. New studies that incorporate a theoretical framework and a sample that is representative of all drivers are warranted in order to understand the effect of attitudes and beliefs on seat belt use within the current US population and how these attitudes, beliefs, and their effects may vary with different types of state laws that are now in place. Findings could inform development of interventions to promote seat belt use among those drivers and passengers who continue to ride unrestrained.

There were several limitations of the study. First, belief about importance of seat belt use was measured with a single question. Within relevant theoretical frameworks, the concept of beliefs – and their role in predicting behavior – is more complex and includes measures such as perceived benefits or harms of a given behavior and perceived likelihood and/or severity of a given health outcome (Ajzen, 2002; Rosenstock, Strecher, & Becker, 1988). Second, self-reporting of certain behaviors can be subject to social desirability bias. However, belt use reported by front seat passengers (86%) was similar to the 2012 observed belt use for front seat passengers (84%; National Highway Traffic Safety Administration, 2012), and a previous evaluation found minimal social desirability bias in self-reported seat belt use (Ibrahimova, Shults, & Beck, 2011). Third, the operationalization of seat belt use as “always” versus “nearly always, sometimes, seldom, or never” precluded the study of occupants who wear seat belts in some situations. Understanding partial seat belt use and the circumstances under which occupants choose to buckle up could also inform efforts to increase the proportion of occupants who always wear seat belts. Fourth, the survey response rate was 65%, which may limit generalizability to the US adult population. A strength of the study is that the sample was drawn using probability-based methods (random-digit dial and address-based sampling; Porter Novelli Public Services, 2012). In addition, the study was conducted well after state seat belt laws were implemented throughout the United States (in 49 states and DC), which allowed for the assessment of how the type of law (primary or secondary) affected the relationship between belief about seat belt use and seat belt wearing behavior. The data set also allowed for the assessment of this relationship by seating position.

The present study found a positive relationship between belief about the importance of seat belt use and the behavior of always wearing a seat belt, even in the context of overall high rates of seat belt use and widespread implementation of seat belt laws throughout the United States. Development and implementation of effective interventions that target at-risk populations may be successful in improving seat belt use among these high-risk groups and, ultimately, reducing crash-related injuries and deaths. The National Highway Traffic Safety Administration estimated that almost 2,500 additional lives could have been saved in 2016 if all occupants (aged 5 + years) in the United States had been wearing seat belts (National Highway Traffic Safety Administration, 2016b). Because much of the theory-based literature on seat belt use was developed prior to changes in social norms and policies in the US landscape (Budd et al., 1984; Fhaner & Hane, 1974; Jonah & Dawson, 1982; Stasson & Fishbein, 1990), efforts to increase seat belt use may benefit from new research that relies upon validated behavioral theories. Given the differences observed by seating position in seat belt use and beliefs about seat belt use, these strategies may require that messages be tailored to front versus rear seat passengers.