An Exploratory Study of the Relationship Between Socioeconomic Status and Motor Vehicle Safety Features

Abstract

Objective: The purpose of this study was to assess the association between motor vehicle owners' socioeconomic status (SES) and the safety of their motor vehicles.

Methods: Truncated vehicle identification numbers (VINs) were obtained from the Maryland Motor Vehicle Administration office. ZIP code–level income and educational data were assigned to each VIN. Software was used to identify safety-related vehicle characteristics including crash test rating, availability of electronic stability control and side impact air bags, age, and weight. Correlations and analyses of variance were performed to assess whether a ZIP code's median household income and educational level were associated with its proportion of registered vehicles with safety features.

Results: For 13 of the 16 correlations performed, SES was significantly associated with the availability of vehicle safety features in a direction that favored upper-income individuals. Vehicle weight was not associated with income or education. When ZIP codes were divided into median household income quintiles, their mean proportions of safety features also differed significantly, in the same direction, for availability of electronic stability control, side impact air bags, vehicle age, and crash test ratings.

Conclusions: Safer motor vehicles appear to be distributed along socioeconomic lines, with lower income groups experiencing more risk. This previously unidentified mechanism of disparity merits further study and the attention of policy makers.

Keywords:

• Motor vehicle safety features
• Socioeconomic status

INTRODUCTION

Motor vehicle safety improvements have been recognized by the U.S. Centers for Disease Control and Prevention as one of the ten most notable public health achievements of the 20th Century (Centers for Disease Control and Prevention 1997). In awarding this designation, experts cited a 90 percent decrease in the annual death rate per vehicle miles traveled between 1925 and 1997 (Centers for Disease Control and Prevention 1999). This dramatic decline is the result of a multifaceted approach to prevention, but in recent years vehicle design improvements have driven fatality rate reductions (Farmer and Lund 2006). It is estimated that vehicle safety technologies saved 328,551 lives from 1960 to 2002 (Kahane 2004). In fact, Robertson (2007) has concluded that “a substantial majority” of today's motor vehicle–related fatalities could be prevented if vehicle characteristics with demonstrated effectiveness were universally adopted.

Typically, however, safety advances are not introduced into our nation's fleet of passenger vehicles in a uniform fashion. They may be first offered as an option by one manufacturer or included as a standard feature on luxury vehicles. Once enough data have accumulated to suggest that the equipment indeed reduces deaths or injuries, the National Highway Traffic and Safety Administration (NHTSA) may consider mandating its inclusion on all new vehicles. They do so over the course of a long process, however, that includes federally mandated rule-making procedures and a phase-in period that is coordinated with the auto industry's design cycles. Not surprisingly then, it takes years for mandated safety equipment to be featured in all vehicles coming off the assembly line. People who buy used vehicles do not benefit from such advances until these safer vehicles “trickle down” to the used car market. According to NHTSA, it takes a median of 12.5 years for cars to age out of circulation in the United States (Lu, 2006).

This system would seem to favor consumers who come from the upper socioeconomic strata. Roberts and Power (1996) have speculated whether the much more modest decline in motor vehicle death rates they observed among children from “manual” social classes might be due to “differential exposure to …health promoting physical and social environments” (p. 786), such as cars equipped with modern safety equipment. U.S. data do suggest that promising national trends in motor vehicle safety can mask declines experienced by some population subgroups, such as those who drive older vehicles (Farmer and Lund 2006). Cubbin and Smith (2002) cited access to crashworthy vehicles as a possible explanation for the inverse relationship between socioeconomic status (SES) and motor vehicle injuries that has been reported in rural areas. Whitlock et al. (2008) suggested that drivers in their lower SES groups might be more likely to “use old vehicles …that have fewer modern safety features such as air bags” (p. 516). Such comments were purely speculative, however; attempts to interpret epidemiologic trends that had been observed. The authors in question did not study the distribution of safer vehicles directly.

We are not aware of any prior attempts to assess whether an association exists between vehicle owners’ socioeconomic status and the safety of their motor vehicles. In a study of which vehicles parents chose for novice drivers, however, Hellinga et al. (2007) reported that teenagers from low-income households were more likely to drive older vehicles.

METHODS

This project utilized a secondary data analysis design, combining data sets from three sources. The unit of analysis for this study was the ZIP code tabulation area (ZCTA), as designated by the U.S. Census Bureau. ZCTAs were created by the U.S. Census Bureau to approximate the mail delivery area for a U.S. Postal Service ZIP Code. They differ from ZIP codes, however, in that they are defined by true geographic boundaries rather than by less stable mail delivery routes (U.S. Census Bureau, 2007). ZCTAs offer a convenient scale of population analysis for variables relating to mortality and SES, and recent studies have determined them to be superior to ZIP codes when used for this purpose (Kreiger et al. 2002; Thomas et al. 2006). In keeping with the terminology used by the Census Bureau, ZCTAs will be referred to simply as ZIP codes in the remainder of this article.

Socioeconomic Status

Socioeconomic status data by ZIP code were obtained from the U.S. Census Website (www.census.gov). The Census Bureau parameters selected to represent SES in this investigation included median annual household income and the percentage of residents aged 25+ with a high school diploma. High school completion was chosen as our education-related measure of SES because that indicator has been strongly associated with motor vehicle occupant death rates (Braver 2003). These data were taken from the latest available dicentennial census (1999). Histograms showed them both to be approximately normal in their distribution.

Vehicle Safety Information

To create our vehicle sampling frame, 150 of the 483 ZIP codes listed on the Census Bureau's Maryland Web page were randomly selected. This number was chosen for reasons of feasibility. A power analysis carried out in the study's planning phase had indicated that it would provide adequate (i.e., 80%) power to detect a moderate effect size (r= .227), based upon a two-sided p value of .05. To explore whether any unintentional bias was introduced by this step, the mean median household incomes of Maryland's sampled and unselected ZIP codes were examined and found to be comparable.

Sampled ZIP codes were submitted to the Maryland Motor Vehicle Administration (MVA). The MVA randomly selected vehicles registered within those areas by eliminating all vehicles that had any number other than three as the 17th digit of their vehicle manufacturer's serial number. The MVA provided the study team with truncated vehicle identification numbers (VINs) corresponding to the vehicles they had sampled. (Note that by truncating VINs to 12 digits the database that was stripped of personally identifying information but would still allow for decoding of each vehicle's make, model, year and body style.) A total of 8418 truncated VINs, drawn from our range of ZIP codes, was then forwarded to the Insurance Institute for Highway Safety (IIHS). The IIHS is an independent, nonprofit organization funded by automobile insurance companies. The IIHS had agreed to use their VINDICATOR software to link vehicle makes and models to crash ratings and specific safety feature information.

The safety features selected for examination in this study were restricted to those with established injury prevention effects that were not universally, or nearly universally, available. Seat belts and frontal-impact air bags, for example, were excluded—despite the fact that they reduced motor vehicle fatality risk immensely (Kahane 2004)—because they have been mandated in new cars in the United States for many years. Our analysis did include side-impact air bags and electronic stability control (ESC). A driver's risk of death in a nearside impact collision is reduced by 37 percent if his or her vehicle is equipped with side air bags (with head protection) (McCartt and Kyrychenko 2007). ESC has been characterized as “perhaps the single most effective means of reducing motor vehicle–related mortalilty” since the invention of seat belts (Robertson 2007, p. 309). A recent review article summarizing the world's experience with ESC concluded that it reduces fatal vehicle crashes by 30 to 50 percent and fatal rollovers by 70–90 percent (Ferguson 2007). In the United States alone, it is estimated that 10,000 fatal crashes could have been avoided in 2004 if all vehicles had been equipped with ESC (Farmer 2006). (Note that NHTSA will begin requiring ESC in all new cars in September of 2012 [NHTSA 2007b], and vehicles must be equipped with side-impact protection by 2013 [NHTSA 2007a]).

Vehicle weight was included in this investigation because occupant death rates are strongly (and inversely) related to the weight of their vehicles (IIHS 1998). To illustrate, if two cars are involved in a crash and one is twice as heavy as the other, the driver of the lighter car is 12 times more likely to die than the other driver (Evans 2004). Vehicle age was included because older vehicles have also been demonstrated to pose greater risks to occupants. A New Zealand study team found that cars that were older than 15 years were three times more likely to be involved in a crash that resulted in hospitalization or death (Blows et al. 2003). Other investigators have reported that driver fatality risk increases even in the early years of ownership, although they acknowledge that vehicle use patterns may underlie such findings (Farmer and Lund 2006).

In selecting a global indicator of vehicle safety, consideration was given to the National Highway Traffic Administration's star system, which utilizes sophisticated dummies to estimate the severity of injuries that a vehicle occupant would suffer in several different types of crashes. Ultimately, however, IIHS's ratings of “Good,” “Acceptable,” “Marginal,” and “Poor,” were chosen because they have been shown to be better correlated with real-world crash outcomes (Farmer 2005; Nirula et al. 2004).

Data Analyses

Safety feature information was obtained for 7451 vehicles. These data were used to calculate parameter estimates for each of the safety features under study, which were then assigned to our 150 ZIP codes. For vehicle weight and age, this was a straightforward process; mean values were calculated for each ZIP code. These data were available for 99 and 100 percent of our vehicles, respectively.

For the other safety features under study, we calculated the proportion of vehicles in that ZIP code that came with a safety feature as standard equipment, as well as the proportion of vehicles that did not even offer that safety feature as an option. Note that these two proportions do not sum to 100, because no information was available about the presence or absence of the safety features on vehicles that offered the equipment as an optional feature (i.e., that could have been purchased from the dealer).

ESC status was available for 72 percent of VINs. The database included driver side air bag status for 55 percent of vehicles. Crash test ratings were available for only 47 percent of VINs because the IIHS only tests vehicles that represent large portions of the new car market, and the Institute did not begin testing motor vehicles until 1995.

Pearson's product-moment correlations, using data weighted by ZIP code population, were used to explore possible relationships between SES indicators and the vehicle safety characteristics. For ease of interpretation a categorical version of median household income (by ZIP code) was also created by dividing those data into quintiles. Analyses of variance were then conducted to compare the mean proportions of safety features that corresponded with our five income groups.

RESULTS

Correlations Between SES and Safety Feature Availability

A total of 16 Pearson's product-moment correlation tests were performed between our two measures of SES and the eight vehicle safety indicators. Thirteen of these correlations were found to be significant (p .01), and all of those were in a direction that provided additional protection to residents of upper SES ZIP codes. No association was found between vehicle weight and our measures of income or education. The mean proportion of vehicles with poor or marginal crash test ratings was significantly associated with income but not education. The strongest correlations we observed were between median household income and the percentage of vehicles with ESC as standard equipment (r= .50) and the proportion of residents with a high school diploma and vehicle age (r=−.52). See Table I for more details, including descriptive sample information.

Table I Correlation coefficients (r) of SES and vehicle safety features

	Unweighted		Correlation ^a		Correlation ^a with %
Safety	median		with median		of residents with
feature	(range)		household income		a high school diploma
Vehicle weight	3551	(3025–4115)	.17	(p= .13)	.11	(p= .37)
Vehicle age	7.92	(5.04–10.82)	−.48	(p< .01)	−.52	(p< .01)
Percentage of vehicles rated good or acceptable	33	(11–62)	.36	(p< .01)	.43	(p< .01)
Percentage of vehicles rated marginal or poor	12	(2–27)	−.31	(p= .01)	−.23	(p= .12)
Percentage of vehicles with side air bags as standard equipment	14	(0–36)	.37	(p< .01)	.41	(p< .01)
Percentage of vehicles with side air bags unavailable	40	(14–62)	−.39	(p< .01)	−.45	(p< .01)
Percentage of vehicles with ESC as standard equipment	6	(0–21)	.50	(p< .01)	.44	(p< .01)
Percentage of vehicles with ESC unavailable	67	(48–82)	−.32	(p< .01)	−.30	(p= .01)

Safety Comparisons by Categories of Income

Analyses of variance were performed to determine whether the mean proportion of safety features available (as standard features) or unavailable (even as an option) differed significantly between income groups. Group means, which are displayed in Table II, were found to differ in seven of eight instances. As mean median household income of ZIP codes increased, the mean age of their registered vehicles went down (F= 11.92, df = 4, p< .001). The mean weight of vehicles did not differ significantly across income groups (F= 1.56, df = 4, p= .19). The mean proportion of vehicles that had received a Good or Acceptable safety rating from the IIHS generally increased as ZIP codes’ median household incomes increased (F= 12.31, df = 4, p< .001). A clear (and inverse) “dose-response” relationship is suggested by the proportions of vehicles that received a Poor or Marginal crash test rating by median household income (F= 4.52, df = 4, p= .002). The mean proportion of vehicles that came with side impact air bags as standard equipment was positively associated with median household income (F= 18.29, df = 4, p< .001). The proportion of vehicles that came with no option of purchasing side impact air bags decreased significantly as the median household incomes of ZIP codes rose (F= 7.47, d.f. = 4, p< .001). A pattern of upper SES advantage was also observed for the proportion of vehicles that came with ESC as a standard feature (F= 17.40, df = 4, p< .001) and the proportion that lacked ESC as a purchase option (F= 4.37, df = 4, p= .002).

Table II Comparison of vehicle safety features by ZIP code's median household income

Median			Mean % of	Mean % of	Mean % of	Mean % of	Mean % of
household	Mean	Mean	vehicles with	vehicles	vehicles	vehicles in which	vehicles	Mean % of vehicles
income of ZIP	vehicle	vehicle	good or	with marginal	equipped	side air	equipped	in which
codes (N= 150),	age	weight	acceptable crash	or poor crash	with side	bags were	with	ESC was
by quintile	(years)	(lbs)	test rating	test rating	air bags ^a	unavailable ^b	ESC^a	unavailable ^b
$20,636–$37,798	8.31	3497	.30	.15	.12	.42	.05	.69
$37,799–$45,692	8.56	3601	.28	.14	.12	.43	.04	.69
$45,693–$56,977	8.17	3562	.31	.13	.12	.40	.05	.68
$56,878–$69,825	7.39	3547	.37	.11	.18	.38	.08	.67
$69,826–$103,879	7.23	3501	.42	.10	.23	.34	.11	.62

DISCUSSION

These results suggest that a robust positive association exists between owners’ SES and the safety of their motor vehicles. In this sample, vehicles registered in ZIP codes with median household incomes in the lower portion of the range tended to share a similar disadvantage.

Vehicle weight did not prove to be significantly associated with either of our SES measures. This may be due to the fact that heavier vehicles are available in the new and used car markets at both ends of the price spectrum.

This exploratory study is subject to a number of limitations. Our SES indicators were drawn from ZIP code–level data, whereas our safety features were computed from individual vehicle specifications. Such a design is vulnerable to ecological fallacy. To elaborate, our reported associations should be interpreted with caution because they may not reflect individual level associations accurately. Also, the SES data that were obtained from the Census Bureau dates from 1999, whereas the VIN registration data provided by the Motor Vehicle Administration was from 2007. To mitigate these potential sources of error, two measures of SES were utilized and a relatively large number of individual vehicles were included per ZIP code (mean = 50). Future investigators could produce more precise estimates of the safety advantage that seems to accrue to wealthier vehicle owners by working with individual family incomes.

This study's conclusions related to safety equipment are restricted to vehicles in which such features either came standard or were unavailable. Though one might speculate that the presence of optional safety features—which increase a vehicle's purchase price—would be positively associated with a consumer's SES, this investigation did not explore that relationship. Finally, this study was conducted with data from just one state. Though we have no reason to believe that the associations we observed are unique to Maryland, it would be desirable to replicate our analyses using national data.

The results of this research suggest that individuals from lower SES neighborhoods are less likely to drive crash-avoidant and crashworthy vehicles. Vehicles registered in ZIP codes that fell within the highest income quintile were approximately twice as likely to come with electronic stability control and side impact air bags, when compared to vehicles registered in the lowest income quintile. This is not surprising when you consider that ESC came standard in all 2006 models of only five makes of vehicles: Audi, BMW, Infiniti, Mercedes, and Porsche. ESC was at least optional in all Cadillac, Jaguar, Land Rover, Lexus, Mini, Toyota, Volkswagen, and Volvo models, but it could not even be purchased in the majority of all Chevrolet, Dodge, Ford, Hummer, Mazda, Mitsubishi, Saturn, Subaru, and Suzuki models (IIHS 2006). According to the IIHS, all but 3 of the 15 vehicles with the lowest overall death rates from 2002 to 2005 came with ESC, usually as a standard feature. None of the 16 vehicles with the highest death rate during that period came with ESC, and it was only offered as an option for one such vehicle (IIHS 2007). Automatic crash response (e.g., OnStar), which is available to consumers who can afford to pay a monthly subscriber fee, extends this inequitable model.

An ideal system would, of course, make uniform levels of safety available to all road users. This is the standard we accept for air travel. Society allows airlines to offer premium passengers better food and more comfort but would not tolerate it if first-class tickets came with a crash survival advantage. The editor of Automotive News has stated that “all safety-related devices should become standard equipment on all [motor] vehicles” (p. 12) characterizing the decision as a moral rather than economic one (Crain 2004).

In reaction to the United States’ recent economic downturn and increased environmental consciousness, the president and congress acted quickly to approve a federal program that provided vouchers to consumers who replaced older, “gas guzzlers” with newer, more fuel-efficient vehicles. A similar mechanism, which extends to the used car market, might be explored for accelerating the diffusion of safety technologies that have proven to be “extraordinarily beneficial” (Ferguson 2007, p. 337), such as ESC. Needs-based tax incentives could also be considered. Our current system of vehicle safety evolution leaves low-income families vulnerable for much longer than their wealthy counterparts.

ACKNOWLEDGMENTS

This study would not have been possible without the very generous assistance of Jack Joyce of the Maryland Motor Vehicle Administration and Kim Hazelbaker of the Insurance Institute for Highway Safety. Dr. Cara Olsen, of the Uniformed Services University of the Health Sciences, also provided valuable biostatistical support for this work.

The views expressed are those of the authors, who are not representing any federal agency.

This article not subject to U.S. copyright law.

Notes

^a Correlations were calculated from data that were weighted to reflect ZIP code populations.

^a As standard equipment.

^b Even for purchase.