Safety Belt Use and the Switch to Primary Enforcement, 1991–2003

Abstract

State seat belt laws have increased use rates and have reduced traffic fatalities, but tremendous variation exists in the laws. New Hampshire does not have a law, and 30 states have only secondary enforcement laws. Whereas primary enforcement allows an officer to issue a citation for any infraction, secondary enforcement permits a citation only if a motorist is stopped for another infraction first.

We performed a cross-sectional time-series analysis of the impact of upgrading to primary enforcement on belt use rates for 47 states and the District of Columbia from 1991 to 2003. Our results suggest that states with secondary enforcement laws could increase belt use by 10 percentage points and improve public safety considerably by upgrading to primary enforcement.

SEAT BELTS ARE WIDELY regarded as an important tool for enhancing motor vehicle safety. It has been estimated that seat belts reduce the risk for a fatality by 45% in passenger cars and 60% in light trucks.1 The National Highway Traffic Safety Administration (NHTSA) estimated more than 14 903 lives were saved by seat belts in 2003, and more than 6803 additional lives would have been saved if all passenger vehicle occupants wore safety belts.2

Seat belt use rates have risen substantially in the United States since the adoption of the first adult mandatory belt use law in 1984. During the 21-year period between 1983 and 2003, the nationwide belt use rate increased from about 17% to 79%.3^,4 Even so, states vary widely in their use rates. For example, in 2003, Washington State had a 95% use rate, but New Hampshire had only a 50% use rate.4

Although safety belt use has increased substantially, it still falls short of the national goal of 90% by 2005.5 Toward this end, states with secondary enforcement laws have been encouraged to adopt primary enforcement laws.5^,6 Primary enforcement authorizes a law enforcement official to issue a citation if an occupant is observed to be traveling unbelted in a motor vehicle. By contrast, secondary enforcement authorizes the issuance of a citation only if another traffic violation has occurred.

Switching to primary enforcement enhances the deterrent effect of the law by increasing the certainty with which individuals will be sanctioned for not wearing a safety belt. In addition, the switch from secondary enforcement to primary enforcement changes the perception of the statute from a good suggestion to a “real” law.7 Secondary enforcement sends motorists the signal that the seat belt law is not all that important, because motorists can violate the law in the presence of a law enforcement officer and know that they will not be cited if no other infractions are committed. Even police officers take a secondary enforcement statute less seriously than a primary enforcement statute.8^–10

Although a substantial amount of research has examined the differential effects of primary and secondary enforcement, much less research has focused on the gains in belt use rates that accompany upgrading to a primary enforcement provision. Research on this topic generally examines the experience of states individually and makes simple pre- and post-upgrade comparisons of belt use rates. We examined the experience of all states and controlled for other covariates of belt use with a cross-sectional time-series regression of annual belt use rates for 47 states and the District of Columbia from 1991 to 2003.

UPGRADING TO PRIMARY ENFORCEMENT

In 1991, 40 states and the District of Columbia had mandatory use laws, 8 of which were originally adopted with primary enforcement provisions. By 1995, an additional 9 states implemented seat belt laws, all of which stipulated secondary enforcement. New Hampshire is the only state still without an adult mandatory use law.

Research that has examined the association between mandatory belt use laws and use rates has concluded that these policies are effective at inducing motor vehicle occupants to wear safety belts, and primary enforcement statutes are more effective than secondary enforcement statutes at doing so.11^–13 Studies that have used econometric techniques, which permit control for other correlates of belt use, have estimated a 9 to 12 percentage point increase with secondary enforcement and a 16 to 22 percentage point increase with primary enforcement.14^–16

In 1993, California became the first state to upgrade the enforcement provision of a mandatory belt use law from secondary to primary. By 2003, 11 other states and the District of Columbia followed California’s lead, and Tennessee upgraded in 2004.17 After implementation of primary enforcement, these states typically experienced an increase in belt use compared with use rates under secondary enforcement.7^,18^–20 These comparisons of rates before and after implementation of primary enforcement show a belt use increase that ranges from 12 to 25 percentage points. It is important to point out, however, that these evaluations of the switch from secondary enforcement to primary enforcement compare changes for individual states and do not control for other covariates of belt use.

A cross-sectional time-series analysis found that the 6 states that upgraded between 1993 and 1997 experienced an average 13.5 percentage point increase in safety belt use with primary enforcement, which supports results from the more simplistic studies. 14 More importantly, this study concluded that the 6 states had postupgrade belt use rates that were comparable to those of the original primary enforcement states. However, of the 6 upgrades during this time period, 3 states had had primary enforcement for no more than 3 months, and a fourth state had had it for only 18 months.

Our analysis provides a more comprehensive evaluation of the association between safety restraint use and the switch from secondary enforcement to primary enforcement. We examined annual observed seat belt use rates in 47 states and the District of Columbia from 1991 to 2003, which covered a larger number of state upgrades and more post-upgrade observations. In addition, the estimation of fixed effects models controlled for other state factors that are correlated with use rates. To better isolate the benefits of this policy change, we estimated separate models to directly compare the experience of states that upgraded with those that have maintained secondary enforcement.

METHODS

Data

We analyzed annual data for 47 states and the District of Columbia from 1991 to 2003. The dependent variable was the observed annual state seat belt use rate reported by NHTSA.1^,4^,21 These rates were determined on the basis of observed daytime shoulder belt use of drivers and front-seat occupants in all passenger motor vehicles. Beginning in 1998, uniform criteria stipulated that site selection must use a probability-based sampling procedure and produce estimates with a relative precision of ± 5%.22 Although general guidelines had been issued for previous years, there was less uniformity across the states in the procedures used to collect observed belt use data.24 The lack of uniformity in use rates obtained before 1998 suggests greater variability that may be unassociated with the seat belt laws, and the variation likely suppresses the observed correlation between state use rates and the enforcement provisions. Because of missing data on safety belt use rates for some years, we did not include Maine, New Hampshire, and Wyoming in the analysis.

Covariates of Belt Use

We created 3 binary variables to measure the enforcement provision of mandatory belt use laws. First, a secondary enforcement variable was assigned a value of 1 for each year in which a state had a secondary enforcement statute in effect and 0 for all other years. Second, for the 13 states that upgraded a secondary enforcement statute by 2003, a value of 1 was assigned for each year that primary enforcement was in effect, and all other observations were coded 0.

A third binary variable showed the presence of a statute that was originally adopted with primary enforcement. Because the 8 original primary enforcement states all implemented laws before 1991, this variable had a value of 1 for each year, and the 31 other states and the District of Columbia had a value of 0 for all years. We included the statutory fine (in dollars) imposed for violating a state seat belt law as a continuous variable. For policy changes that occurred midyear, we coded these variables as the proportion of days in the year for which a law was in effect. Information on safety belt use laws were gathered from the Insurance Institute for Highway Safety and NHTSA’s Traffic Safety Facts.17^,25

Several sociodemographic characteristics have been correlated with belt use, including race/ethnicity, gender, and age.26^–29 We used US Census Bureau data to measure these state characteristics in our models. State racial/ethnic composition was measured as the percentage of the population that was White. The female percentage of the population and the percentage that were aged 15 to 24 years also were included. Previous studies have suggested that higher socioeconomic status is predictive of belt use,27^,28 and we measured this concept with the percentage of a state’s population that had at least 12 years of education and disposable income per capita in thousands of constant dollars (1982–1984 dollars), both of which were obtained from US Census Bureau data. Table 1 ▶ shows the descriptive statistics of the variables we included in the models.

TABLE 1—

Descriptive Statistics for State Time-Series Data: United States, 1991–2003 (N = 624)

Variable	Mean	Median	SD	Minimum	Maximum
Seat belt use rate	65.7	66.1	12.7	25	95
Seat belt law: original primary enforcement	0.17	0	0.37	0	1
Seat belt law: switched to primary enforcement	0.10	0	0.30	0	1
Seat belt law: secondary enforcement	0.68	1	0.46	0	1
Seat belt law: fine, $	21.07	20	13.10	0	95
White, %	82.57	86.22	13.69	21.18	98.76
Female, %	51.62	51.84	1.09	46.76	51.12
Aged ≥ 15 y, %	18.16	18.04	1.87	13.16	27.38
Education ≥ 12 y, %	78.82	79.15	6.70	64.3	92.2
Disposable income per capita, 1982–1984, $1000	13.53	13.38	2.07	9.20	22.87

Statistical Analysis

We estimated the cross-sectional time-series regression models with a fixed effects specification. This estimation approach takes advantage of change over time within states and permits control for covariates of belt use. The need to include state and year fixed effects parameters was determined with separate F tests.29 State fixed effects captured the differences in belt use among states that were not represented formally in the model. The year fixed effects captured changes in belt use that occurred over time aside from any policy or sociodemographic change that was modeled. Diagnostic tests showed the presence of heteroskedasticity, which meant the error variance was not constant across states.29 Therefore, we estimated a feasible generalized least squares routine to correct the results for presence of both heteroskedasticity across states and first-order serial correlation.

RESULTS

Trends in Use Rates

Trends in average state belt use rates are shown in Figure 1 ▶. States and the District of Columbia were grouped into 3 categories. The first group comprised the 8 states that initially adopted and maintained a primary enforcement belt law from 1991 to 2003. The second group comprised the 13 states that upgraded from secondary enforcement to primary enforcement by 2003. In 1991, all but 1 of these states had adopted a secondary enforcement provision, with the last state doing so in 1992. The first upgrade to primary enforcement was in 1993, which was followed by 1 in 1995, 1 in 1996, and 3 in 1997. The third group comprised the 27 states that had secondary enforcement laws in 2003 (excluding the 3 states that were missing data on belt use rates). Of the 27 secondary enforcement states, 20 had initially adopted a law by 1991, and the remaining 7 states had done so by 1995.

FIGURE 1—

Average annual observed safety belt use by belt law category, United States, 1991–2003.

Although all 3 groups experienced a steady increase in belt use throughout the 13-year period, there were differences in the average level of use (Figure 1 ▶). States that had primary enforcement throughout this period consistently experienced the highest average observed belt use rates among the 3 categories. Conversely, the group of secondary enforcement states experienced the lowest use rates.

Most importantly, Figure 1 ▶ shows states that initially adopted secondary enforcement but upgraded to primary enforcement experienced the greatest average increase in belt use. From 1991 to 2003, these upgrade states raised their average belt use from 55.5% to 82.7% compared with an increase from 68.4% to 86.1% among the original primary enforcement states, which reduced the gap between these categories from 12.9 percentage points to only 3.4 percentage points. The increase in belt use for the upgrade states was especially pronounced between 1996 and 2000, when 8 of the 13 states upgraded to primary enforcement. Although the group of secondary states that did not upgrade to primary enforcement also experienced a substantial increase in average belt use from 49.9% to 73.4%, the average still lagged behind the 2 categories of states that had primary enforcement provisions in 2003.

Effects of Belt Laws

To properly assess the effect of upgrading from secondary enforcement to primary enforcement, we estimated cross-sectional time-series regression models with state and year fixed effects. The first model in Table 2 ▶ includes all 47 states and the District of Columbia for the 13-year time period. The 8 states that initially adopted primary enforcement experienced average use rates that were 23.5 percentage points higher than the average rate for years when states had no safety belt law in effect (95% confidence interval [CI] = 13.2, 33.8). In comparison, the presence of secondary enforcement accounted for only an 11.7 percentage point increase in belt use (95% CI = 8.7, 14.6). However, when a state upgraded from secondary enforcement to primary enforcement, belt use was on average 21.4 percentage points higher compared with states that had no law at all (95% CI = 17.8, 24.9). Tests for the equality of 2 coefficients showed no statistically significant difference between the 2 primary enforcement coefficients (i.e., original and upgrade). Alternatively, the coefficients for primary enforcement upgrade states and secondary enforcement states were statistically different (P < .001).

TABLE 2—

Cross-Sectional Time-Series Feasible Generalized Least Squares Estimates of State Seat Belt Use: United States, 1991–2003

	Model 1: All States, 1991–2003	Model 2: Original Secondary States, 1991–2003	Model 3: Original Secondary States, 1995–2003
Seat belt law: original primary enforcement (SD)	23.502*** (5.27)
Seat belt law: switched to primary enforcement (SD)	21.383*** (1.81)	20.143*** (2.12)	10.013*** (0.76)
Seat belt law: secondary enforcement (SD)	11.650*** (1.50)	10.313*** (1.75)
Seat belt law: fine (SD)	0.135*** (0.03)	0.178*** (0.05)	0.129*** (0.03)
White, % (SD)	−0.211 (0.24)	−0.151 (0.34)	−0.197 (0.36)
Female, % (SD)	−4.417*** (0.98)	−5.281*** (1.27)	−4.271*** (1.17)
Aged ≥ 15 y, % (SD)	0.106 (0.30)	0.096 (0.32)	−0.072 (0.32)
Education ≥ 12 y, % (SD)	−0.089 (0.12)	−0.202 (0.15)	−0.032 (0.11)
Disposable income per capita, 1982–1984 ($1000) (SD)	1.147** (0.48)	1.047* (0.62)	0.549 (0.58)
N	624	507	351
Log likelihood	−1636.3	−1371.1	−858.7
Wald X2	4680.9***	2904.9***	3558.9***
Adjusted R2	0.847	0.811	0.814

Note: Cell entries are unstandardized coefficients. Numbers in parentheses are standard errors that have been corrected for heteroskedasticity across states and first-order serial correlation serial correlation. Intercepts for each state and each year were estimated (data not shown).

* P ≤ .10; **P ≤ .05; ***P ≤ .01

Three conclusions emerged from model 1. First, the presence of a primary enforcement belt law was correlated with greater belt use compared with a secondary enforcement law. Second, states that switched from secondary enforcement to primary enforcement experienced average gains in restraint use of 9.7 percentage points, which is an 83% relative increase in use compared with the effect of secondary enforcement. Third, the difference in belt use between original primary enforcement states and upgraded primary enforcement states was not statistically significant. States that upgraded to primary enforcement experienced belt use rates that were comparable to the original primary enforcement states. The initial adoption of secondary enforcement did not doom these states to lower belt use rates after upgrading to primary enforcement. Instead, primary enforcement had the same effect on belt use whether it was the original state statute or an upgraded statute.

Estimates of the effect that original primary enforcement had on belt use, however, must be made with caution. A high level of collinearity between this variable and a linear combination of the state fixed effects resulted in a large standard error for this coefficient and a parameter estimate whose magnitude was sensitive to the specific state that was treated as the base case for the state fixed effects. This condition existed because all 8 original primary enforcement states adopted an adult mandatory belt use statute before 1991. However, the parameter estimates for the primary enforcement upgrade variables and secondary enforcement variables were consistent across numerous alternative specifications of model 1.

The same model was estimated a second time with the 8 original primary enforcement states excluded (Table 2 ▶, model 2). We did this to ensure that the experience of the 8 original primary enforcement states did not confound the parameter estimate for the primary enforcement upgrade variable. Once again, it was clear that seat belt laws matter. A secondary enforcement statute accounted for a 10.3 percentage point increase in belt use compared with absence of an enforcement statute (95% CI = 6.9, 13.7). However, this figure increased to 20.1 percentage points once primary enforcement was adopted (95% CI = 16.0, 24.3). The difference of 9.8 percentage points between the secondary and primary coefficients in model 2 was statistically significant (P < .001).

The base group in models 1 and 2 comprised the years for which some states were without a belt law, however, which accounted for only 21 observations or 3.4% of those used to estimate model 1 and 4.1% of those in model 2. Furthermore, all of these observations occurred early in the 13-year period. This raised concerns about estimating the effect of switching to primary enforcement when most of the upgrades took place later during this time period.

Therefore, model 3 examined the experience of original secondary enforcement states from 1995 to 2003 only. By 1995, all states included in our analysis had adopted a mandatory use law; thus, the base group for this model was the presence of secondary enforcement. The coefficient for the primary enforcement upgrade showed that on average states experienced a 10.0 percentage point increase in belt use above that experienced by states with secondary enforcement (95% CI = 8.5, 11.5). This gain is comparable to the estimates generated by models 1 and 2, which showed that the results were not an artifact of either the original primary enforcement states’ possible impact on the parameter estimate or the move from no belt law to secondary enforcement.

Beyond the enforcement provision, the minimum statutory fine for a seat belt law violation was positively correlated with belt use. The coefficient for the fine variable was statistically significant in all 3 models, yet its size varied somewhat from model to model. In accordance with the parameter estimate provided by model 1, which used all available observations, every $1 increase in the minimum statutory fine resulted in an average increase of 0.135 percentage points in belt use rates (95% CI = 0.07, 0.20). This suggests that a $10 increase in the statutory fine will increase belt use by 1.35 percentage points.

Other Covariates of Belt Use

Among the sociodemographic covariates of safety restraint use, the female percentage of the population and the personal income per capita were the only 2 that tended to be statistically significant across these models. Whereas income was positively correlated with belt use as expected, the negative association between females and belt use was contrary to previous research.25^–28 Although the covariates did not all perform as hypothesized, the results were partially a function of including state fixed effects. For example, when the state fixed effects were excluded from the models, the variables for the percentage of young drivers and education tended to be statistically significant in the expected negative and positive directions, respectively. The state fixed effects coefficients likely captured much of the differences between states that were reflected by these demographic variables.

CONCLUSIONS

Upgrading to a primary enforcement provision had significant positive implications for traffic safety. On average, states that made this switch experienced an increase in belt use of about 10 percentage points (Table 2 ▶, model 2). Furthermore, these states experienced belt use rates that were comparable to those of the original primary enforcement states. Thus, states that make the switch from secondary enforcement to primary enforcement can take full advantage of the safety benefits primary enforcement offers.

Perhaps the importance of this increase in belt use can best be understood in terms of number of motor vehicle fatalities that can be or have been averted because of an upgrade to primary enforcement. It is estimated that every 1 percentage point increase in the national safety belt use rate translates to 136 fewer occupant fatalities.14 On the basis of this figure, it is estimated that in 2003, 12 states and the District of Columbia saved the lives of 419 motor vehicle occupants by upgrading an adult mandatory belt use law from secondary enforcement to primary enforcement. Furthermore, the cumulative effect of all state upgrades since California upgraded in 1993 has been 3158 fatalities averted. Finally, the remaining 27 secondary states in our analysis could have averted an estimated 621 motor vehicle occupant fatalities in 2003 had these states upgraded to primary enforcement. In accordance with NHTSA estimates of the economic costs of crash fatalities, an upgrade to primary enforcement in all states would save more than $700 million a year in medical and emergency costs, lost productivity, insurance, rehabilitation costs, and legal costs.30

In addition, our results show that states can further enhance traffic safety by adopting higher minimum fines for not using a safety belt. The statutory maximum fine in 2003 ranged from a low of $5 to a high of $75 for a first offense, with the median fine being $25.17 It is estimated that for every additional $10 a fine is raised, belt use will increase by 1.35 percentage points (Table 2 ▶, model 1). The median fine in 2003 was $25, and an increase from $25 to $50 would be associated with an observed increase in belt use of 3.4 percentage points. In July 2005, Texas had a $200 maximum fine for a first offense, but there were still 17 states with fines less than $25. Traffic safety could be enhanced significantly with higher fines in these states.

Overall, our analysis suggests that state policymakers who seek to improve public safety should upgrade mandatory seat belt use laws from secondary enforcement to primary enforcement. Even after we controlled for sociodemographic characteristics, the results of our cross-sectional time-series analysis showed considerable improvements in seat belt use—of at least 10 percentage points—that were attributable to an upgrade from secondary enforcement to primary enforcement. Higher fines would further enhance this safety improvement. Finally, our results show states that have already adopted primary enforcement have gained the most benefits thus, far from the adoption of belt use laws, and states with secondary enforcement can catch up rather quickly in belt use rates if they upgrade to primary enforcement.