Residence Location of Drivers Involved in Fatal Crashes


Jesse Blatt, Ph.D.
Office of Program Development and Evaluation
Susanne M. Furman, Ph.D.
Office of Budget and Policy Development

National Highway Traffic Safety Administration

ABSTRACT

Traffic safety researchers have long known that the majority of fatal crashes occur on rural roads, but it is not clear if these crashes involve people who live in rural areas or residents of urban areas traveling on rural roads. "Geodemographic" market-research tools permit determining the urbanization of drivers' residence locations from their postal "zip code." Using data from the 1988 through 1992 files of the Fatal Accident Reporting System (FARS) maintained by the National Highway Traffic Safety Administration (NHTSA), this study determined the residence location of several sub-groups of drivers involved in fatal crashes. Not only did the majority of fatal crashes occur in rural areas, the majority of fatal crashes involved rural residents, and the majority of the rural residents involved in fatal crashes were traveling on rural roads.

INTRODUCTION

Traffic safety researchers have long known that the majority of fatal crashes occur on rural roads (NHTSA, 1988a, 1988b, 1989, 1991a, 1991b, 1993, 1994a, 1994b). Although recent analyses have illuminated many of the differences between rural and urban crashes (NHTSA, 1996), it has never been clear if rural crashes involve people who live in rural areas or residents of urban areas traveling on rural roads. Having better information about people involved in crashes would foster the development of more effective safety countermeasures.

Data from the Fatality Analysis Reporting System (FARS), maintained by the National Highway Traffic Safety Administration (NHTSA), have shown a consistent relationship between rural and urban crashes, with approximately 55 percent of fatal crashes on rural roads each year. Figure 1 shows the number of rural and urban fatal crashes reported in FARS from 1986 through 1993.

Figure 1
Figure 1.

The advent of market research tools that combine geographic, demographic, and life-style information (now termed "geodemographics") makes it possible to determine the relative involvement of rural and urban dwellers in fatal crashes at the national level. Geodemographic software permits researchers to use individuals' postal zip codes to characterize the urbanization of their residence location. In 1988, FARS began recording drivers' residence zip codes obtained from the drivers' license recorded on the police accident report.

NHTSA recently conducted several independent studies of different population subgroups using geodemographic methods to facilitate development of safety programs tailored specially for each subgroup (Bradbard, S.L. & Lisboa-Farrow, E., 1995; Graham, J.D., Isaac, N.E., Kennedy, B., & Winsten, J., 1995; Bradbard, S.L., Panlener, J.C., & Lisboa-Farrow, E.,1996). These analyses revealed that, in each subgroup examined, residents of rural areas and small towns were consistently over-represented in fatal crashes relative to their numbers in the population.

This report augments and presents the data from these analyses and investigates the extent to which rural and urban residents were involved in crashes on rural or urban roadways.

METHODS

We created a simple database for analysis by copying selected fields from driver records from the 1988 through 1992 FARS data files. The resulting database contained driver's zip code, age, gender, blood alcohol concentration (BAC), and whether a child of 5 years of age or less died in the crash.

We used the 1995 update of Compass Prizm geodemographic software by Claritas Corporation to place each driver's residence location into one of five levels of population density: Rural, Town, 2nd City, Suburban, Urban. Rural and Town represent the most sparsely-populated areas. Second City locations include places larger than towns but not major metropolitan areas, such as Youngstown, OH, or Fresno, CA. They might also be "edge" cities, located on the exurban fringe of a major metropolitan area, such as Reston, VA or Frederick, MD, on the outskirts of Washington, DC. Suburban and Urban zip codes represent the most densely populated areas (Claritas, 1995a, 1995b, 1996).

In geodemographic analyses, the primary measure of a population's level of involvement is the "index of concentration." This index is obtained by dividing the percentage of a population sub-group involved in an activity by the percentage of the sub-group population in the whole population and multiplying by 100. If the characteristic of interest is distributed evenly throughout the population, then the index of concentration would be 100.

For example, if 20 percent of the male drivers involved in fatal crashes lived in small towns and 20 percent of the male population lived in small towns, the index of concentration would be 100 (100 x 20% involved 20% resident). If 40 percent of male drivers involved in fatal crashes lived in small towns, the index would be 200 (100 x 40% involved 20% resident), indicating that twice as many were involved as would be expected if everyone in the whole population had the same tendency toward involvement in fatal crashes. Values below 100 indicate lower- than-expected involvement.

We conducted separate analyses of several sub-populations, including (a) drivers involved in crashes resulting in the death of a child; (b) drivers involved in crashes in which alcohol was detected in the drivers' blood (at three levels of BAC); and (c) young drivers in two age groups according to legal drinking age (15-20 and 21-25). In addition, we analyzed the whole population of drivers involved in fatal crashes by gender.

The analyses of each subpopulation compared the percentage of drivers in fatal crashes in each social cluster to the percentage of the base population of that social cluster. In general, the base population represents the number and percent of the U.S. population age 15 and older residing in the areas included in each social cluster. Because the age categories used by Compass PRIZM did not correspond exactly to age categories for the youth crash analyses, we used the PRIZM breakdown of 18-24 as a base for both.

RESULTS

The analyses of the subpopulations by zip code demonstrated consistent results: In each of the populations, residential zip codes of drivers involved in fatal crashes fell predominantly in rural areas and small towns.

Distribution of crashes involving child fatalities.

Table 1 shows that Rural residents were over involved in crashes in which a child 5 years of age and under died, regardless of child restraint use or nonuse. Thirty-four percent of all child- fatal crashes involved Rural drivers, who represent only 17 percent of the population, yielding an index of concentration of 199. The majority (62 percent) of drivers live in Suburban, Urban, and 2nd City cluster locations but they were involved in only 42 percent of crashes involving child fatalities.

Table 1. Drivers in Fatal Crashes Involving Children Under 5 Years of Age*

Group Total Population 15+ Drivers in Crashes involving Child Fatalities Drivers in Crashes where Restraint Used Drivers in Crashes where No Restraint Used
N % N % Index N % Index N % Index
Urban 35,822,926 17.8 314 10.0 56 75 7.7 43 211 10.8 61
Suburb 49,996,921 24.9 454 14.5 58 160 16.3 65 266 13.6 55
2nd City 39,589,771 19.7 557 17.8 90 162 16.5 84 367 18.8 95
Small Town 40,852,227 20.4 727 23.2 114 257 26.3 130 418 21.4 105
Rural 34,842,714 17.3 1,075 34.4 199 325 33.2 192 689 35.3 204
TOTAL 201,104,559 3,127 979 1,951
* Table 1 does not include 197 child fatalities for which restraint use was unknown.

Distribution of crashes involving alcohol.

Table 2 shows fatalities for three levels of blood alcohol concentrations: BAC = .08 to .09, BAC = .10 to .14, and BAC = .15 and above. Drivers from the Rural clusters were involved in the majority of fatal crashes for each of the BAC levels, with 35 percent, 34 percent, and 37 percent, respectively. When considering all fatal crashes with alcohol present, 36 percent involved Rural drivers and 59 percent involved those drivers living in Rural and Small Town clusters.

Table 2. Drivers in Alcohol-Related Crashes

Group Population 15+ Drivers in Fatal Crashes where BAC = .08 to .09 Drivers in Fatal Crashes where BAC = .10 to .14 Drivers in Fatal Crashes where BAC = .15 & above
N % N % Index N % Index N % Index
Urban 35,822,926 17.8 260 9.5 53 907 9.9 56 2,430 8.4 47
Suburb 49,996,921 24.9 429 15.7 63 1,477 16.2 65 4,392 15.2 61
2nd City 39,589,771 19.7 501 18.4 93 1,620 17.8 90 4,894 16.9 86
Small Town 40,852,227 20.3 596 21.9 108 2,044 22.4 110 6,610 22.9 112
Rural 34,842,714 17.3 940 34.5 199 3,068 33.7 195 10,633 36.7 212
TOTAL 201,104,559 2,276 9,116 28,929

Figure 2 presents graphically the indexes of concentration for drivers involved in alcohol-related fatalities for each cluster. For all BAC levels, the Rural residents' index of concentration was about 200, twice the expected ratio of 100.

Figure 2
Figure 2.

Distribution of crashes involving young drivers.

Table 3 shows that young Rural adults were similarly over involved in fatal crashes. Only 16 percent of the base population 18-24 year olds resided in the Rural cluster, whereas 32 percent of the 15-20 year olds and 27 percent of 21-25 year olds fatal crashes involved Rural residents. Together, Small Town and Rural clusters accounted for 35 percent of the base population, but fatal crashes involved 57 percent of 16-20 year olds and 50 percent of 21-24 year olds. Rural 16-20 year old drivers and rural 21-24 year old drivers earned indexes of concentration of 202 and 170, respectively. Young drivers residing in Small Towns obtained indexes of 127 and 116 for the two age groups.

Table 3. Young Drivers

Group 18-24 YO Population 15-20 YO Fatalities 21-25 YO Fatalities
N % N % INDEX N % INDEX
Urban 4,932,636 18.7 4,031 9.5 51 5,344 12.7 68
Suburb 5,557,352 21.0 7,570 17.8 85 8,192 19.4 92
2nd City 6,595,589 25.0 6,721 15.8 63 7,704 18.2 73
Small Town 5,127,973 19.4 10,476 24.6 127 9,507 22.5 116
Rural 4,221,230 16.0 13,720 32.3 202 11,492 27.2 170
TOTAL 26,434,780 42,518 42,239

Distribution of crashes involving all drivers.

Table 4 shows that the majority (62 percent) of the total population 15 years of age and older resided in the Suburban, Urban, and 2nd City cluster locations. However, the majority (61 percent) of fatal crashes involved Rural and Small Town residents. Forty percent of all fatal crashes involved Rural residents, who represented only 17 percent of the population, yielding an index of concentration of 230.

Table 4. Drivers 15 Years Old and Older

GROUP Total Population 15+ Drivers 15+ Involved in Fatal Crashes INDEX
N % N %
Urban 35,822,926 17.8 16,197 9.0 51
Suburban 49,996,921 24.9 28,905 16.1 65
2nd City 39,589,771 19.7 24,508 13.6 69
Small Town 40,852,227 20.4 38,684 21.5 106
Rural 34,842,714 17.3 71,572 39.8 230
TOTAL 201,104,559 179,866

Table 5 shows the over-involvement in fatal crashes of Rural male drivers. The majority of males (63 percent) resided in Urban, Suburban, and 2nd City cluster locations but were involved in only 36 percent of male-driver fatal crashes (index=57). The majority of fatal crashes (44 percent) involved Rural males. However, only 18 percent of the male population resided in the Rural cluster, resulting in an index of concentration of 252.

Table 5. Male Drivers

GROUP Males in Population Male Drivers involved in Fatal Crashes INDEX
N % N %
Urban 16,979,794 17.5 9,980 8.4 48
Suburban 23,995,321 24.8 17,353 14.6 59
2nd City 18,846,030 19.5 14,288 12.0 62
Small Town 19,960,312 20.6 24,292 20.5 100
Rural 17,051,541 17.6 52,679 44.4 252
TOTAL 96,832,469 118,592

Table 6 shows that 31 percent of female-driver fatal crashes involved Rural females with only 17 percent of the female population residing in that cluster location, yielding an index of concentration of 180. Although 54 percent of female-driver fatal crashes involved Rural and Small Town residents, these females made up only 37 percent of the female population. Sixty- three percent lived in Suburban, Urban, and Second City locations but these females were involved in only 46% of the fatal crashes.

Table 6. Female Drivers

GROUP Females in Population Female Drivers Involved in Fatal Crashes INDEX
N % N %
Urban 18,843,132 18.1 6,217 10.1 56
Suburban 26,001,600 24.9 11,552 18.9 76
2nd City 20,743,741 19.9 10,226 16.7 84
Small Town 20,891,915 20.0 14,392 23.5 118
Rural 17,791,702 17.1 18,893 30.8 180
TOTAL 104,272,090 61,280

Figure 3 summarizes the results of the preceding analyses, showing the index of concentration for each population subgroup by the residence location of drivers involved in fatal crashes. These results show that rural residents were heavily over- involved in fatal crashes, while residents of small towns were slightly over-involved. Drivers residing in more urbanized areas were generally under-involved in fatal crashes.

Figure 3
Figure 3.

Effect of using licensed drivers instead of population as a base.

All of the preceding analyses used population as the denominator in determining the base proportion. Ideally, the denominator would accurately reflect the distribution of drivers among the cluster groups. Since a smaller proportion of urban residents are licensed drivers, relative to suburban and rural residents, using the population as the base may introduce a slight bias towards increasing the index of concentrations for the more rural clusters.

To determine the approximate magnitude of this bias, we adjusted the population base by the proportion of licensed drivers in the place-of-residence categories reported in the 1990 Nationwide Personal Transportation Survey (NPTS)(Hu & Young, 1993). The 1990 NPTS uses only three categories for population density: MSA, central city; MSA, non-central city; and non-MSA. The licensing rates for these categories were 85.0%, 92.1%, and 90.5%, respectively. Table 7 shows the original and adjusted bases and the resulting concentration indexes.

Table 7. Drivers 15 Years Old and >Older
Adjusted for Proportion of Licensed Drivers in Population

GROUP Total Population 15+ Drivers 15+ Involved in

Fatal Crashes

INDEX
% Lic Drivers Estimated Licensed Drivers
N % N %
Urban 35,822,926 17.8 16,197 9.0 51
85.0 30,449,487 16.8 54
Suburban 49,996,921 24.9 28,905 16.1 65
92.1 46,047,164 25.4 63
2nd City 39,589,771 19.7 24,508 13.6 69
92.1 36,462,179 20.1 68
Small Town 40,852,227 20.3 38,684 21.5 106
90.5 36,971,265 20.4 105
Rural 34,842,714 17.3 71,572 39.8 230
90.5 31,532,656 17.4 229
TOTAL 201,104,559 179,866
181,462,751

Using the estimated distribution of licensed drivers as the comparison base has only a small effect on the index of concentration across the five groups. The bias introduced by using the population distribution was one to three points on the index of concentration. The different methods produced less than one percent difference for the small town and rural categories.

Effect of annual miles driven.

Exposure to risk is another possible influence on the index of concentration. If rural residents typically drive more miles than their urban counterparts, then they would be exposed to a greater risk of crash involvement and would evidence a greater level of involvement in crashes. Table 8 provides data from the 1990 NPTS (Hu & Young, 1993) showing that licensed drivers in Non-central-city MSAs and licensed drivers in Non-MSAs drive nearly the same average miles per year, while the Central-city MSA drivers drive somewhat fewer miles. Thus, differences in annual miles driven cannot account for the large index of concentrations observed among rural residents.

Table 8. Miles Driven by Residence Location

MSA* Central City MSA Non Central City Non MSA
Personal Miles Traveled (PMT) (millions) 705,454 1,072,689 537,130
% PMT by Personal Vehicle 84.1 87.9 93.8
Miles Traveled by Personal Vehicle (millions) 593,287 942,894 503,828
Number of Licensed Drivers (thousands) 56,180 70,103 36,742
Personal Vehicle Miles per Licensed Driver (thousands) 10.56 13.45 13.71
*MSA=Metropolitan Statistical Area

Relationship between place of residence and crash location.

Table 9 provides a cross-tabulation of driver's residence cluster by FARS crash location (i.e., rural vs urban road classifications). The majority of fatal crashes involved rural residents traveling on rural roads: Forty-one percent of all drivers involved in fatal crashes were Rural or Small-town residents traveling on rural roads. However, only 15 percent of all fatal crashes involved Suburban, Urban, or 2nd City residents traveling on rural roads. Thus, nearly 3 out of 4 fatal crashes occurring on rural roads involved Rural and Small-town residents.

While the number of crashes was smaller, the pattern of involvement of urban residents in urban crashes was the mirror image of the rural pattern: About 3 out of 4 fatal crashes on urban roads involved drivers residing in Suburban, Urban, or 2nd City clusters.

Table 9. Number (& percentage) of Drivers Involved in Fatal Crashes
by Residence Cluster & FARS Crash Location

Cluster Group
FARS Crash Location Rural & Small-Town Residents Suburban, Urban, 2nd City Residents TOTAL
Rural 114,709 (41.2) 43,070 (15.5) 157,779 (56.7)
Urban 33,071 (11.9) 87,273 (31.4) 120,344 (43.3)
TOTAL 147,780 (53.1) 130,343 (46.9) 278,123 (100.0)

DISCUSSION

These analyses clearly demonstrate the over involvement of rural residents in fatal crashes: The majority of fatal crashes involved Rural residents, and the majority of these Rural residents involved in fatal crashes were traveling on rural roads. This over-involvement is not limited to only one or two subpopulation groups (e.g., males or females) but across the board when considering all drivers, alcohol-involved drivers, adolescent drivers, young adult drivers, and drivers with young child passengers.

The analysis shows, conversely, that urban residents are primarily involved in urban crashes. By and large, people have crashes where they live, and by implication, where they drive: Rural residents crash on rural roads and urban residents crash on urban roads.

Highway safety researchers suggest a number of reasons for increased risk of fatal crashes on rural roads. Design elements (e.g., two lane highways, narrow or nonexistent shoulders, and limited sight distance due to hills and curves) contribute to this risk. A majority of all fatal crashes occur on two-lane, two-way highways, with a large percentage of these highways located in rural areas. Higher speeds are also a contributing factor. For example, speed limit and land use data show that 40 percent of all 1992 fatal crashes occurred on rural roads posted at 55 mph or higher (NHTSA, 1994a).

Economic and behavioral factors also influence crash outcomes. Lower rates of seat belt and child safety seat use, both of which characterize rural residents, contribute to increasing seriousness of injuries (NHTSA, 1995). Delays in discovery and extended Emergency Medical Services (EMS) response times (often a 10 to 30 minute delay in ambulance response and 30 to 60 minute additional travel time to hospitals), and lack of nearby emergency and trauma care facilities decrease the likelihood of surviving serious crashes in remote areas.

In order to address the unique needs of rural residents, NHTSA has adopted the techniques of social marketing, using geodemographics to help locate potential target populations and focus-group interviews and surveys to assist in developing and testing thematic messages and communication channels.

NHTSA has developed an integrated injury-control approach involving partnerships among health care, government, businesses, and community groups. While the rural motor-vehicle injury problem is too complex and resources are too limited for any one group to solve alone, rural community partnerships involving government, police, fire, EMS, schools, businesses, community groups, health care providers, and public health agencies may hold the key to reducing the personal, emotional, and financial impacts on rural residents due to motor-vehicle crashes.

BIBLIOGRAPHY

Bradbard, S.L. & Lisboa-Farrow, E., (1995) Program Strategies for increasing car seat usage in rural areas, DOT-HS-808-274, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

Bradbard, S.L., Panlener, J.C., & Lisboa-Farrow, (1996) Program strategies for increasing safety belt usage in rural areas, DOT- HS-808-505, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

Claritas, Inc., (1995a) Cluster Snapshots: Getting to Know the 62 Clusters, Arlington, VA.

Claritas, Inc., (1995b) PRIZM Lifestyle Segmentation, Arlington, VA.

Claritas, Inc., (1996) Cluster Narratives: Summary Lifestyle Descriptions, Arlington, VA.

Graham, J.D., Isaac, N.E., Kennedy, B., & Winsten, J., (1996) Strategic advertising plans to deter drunk driving. DOT-HS-806- 611, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

Hu, P. S., and Young, J., (1993) National Personal Transportation Survey, 1990 NPTS Databook, Volume I, FHWA-PL-94-010A, U.S. Department of Transportation, Federal Highway Administration, Washington, D.C.

NHTSA, (1988a) Fatal Accident Reporting System 1986, DOT-HS-807- 245, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1988b) Fatal Accident Reporting System 1987, DOT-HS-807- 360, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1989) Fatal Accident Reporting System 1988, DOT-HS-807- 507, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1991a) Fatal Accident Reporting System 1989, DOT-HS-807- 693, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1991b) Fatal Accident Reporting System 1990, DOT-HS-807- 794, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA,(1993) Fatal Accident Reporting System 1991, DOT-HS-807- 954, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1994a) Traffic Safety Facts 1992 (Revised), DOT-HS-808- 022, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1994b) Traffic Safety Facts 1993, DOT-HS-808-169, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.

NHTSA, (1995) National Occupant Protection Use Survey: Controlled Intersection Study, Research Note, U.S. Department of Transportation, National Highway Traffic Safety Administration, National Center for Statistics and Analysis, Washington, D.C.

NHTSA (1996) Rural and Urban Crashes: a Comparative Analysis, DOT-HS-808-450, U.S. Department of Transportation, National Highway Traffic Safety Administration, Washington, D.C.