National Academies Press: OpenBook

Highway Safety Behavioral Strategies for Rural Areas (2023)

Chapter: Data Analysis (Task 2)

« Previous: Define Rural Area Roads (Task 1)
Page 22
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 22
Page 23
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 23
Page 24
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 24
Page 25
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 25
Page 26
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 26
Page 27
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 27
Page 28
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 28
Page 29
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 29
Page 30
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 30
Page 31
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 31
Page 32
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 32
Page 33
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 33
Page 34
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 34
Page 35
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 35
Page 36
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 36
Page 37
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 37
Page 38
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 38
Page 39
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 39
Page 40
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 40
Page 41
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 41
Page 42
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 42
Page 43
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 43
Page 44
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 44
Page 45
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 45
Page 46
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 46
Page 47
Suggested Citation:"Data Analysis (Task 2)." National Academies of Sciences, Engineering, and Medicine. 2023. Highway Safety Behavioral Strategies for Rural Areas. Washington, DC: The National Academies Press. doi: 10.17226/27196.
×
Page 47

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

22 Data Analysis (Task 2) Building on the previous task, the research team used the combined classification system (i.e., rural areas and along rural roads), as well as descriptive and inferential statistical analysis to examine crashes from several perspectives. First, they used detailed crash data from five states to determine where crashes happen most often—Alabama, Alaska, Minnesota, North Carolina, and Washington. Second, they used data from Minnesota, North Carolina, and Washington to better understand design and environmental factors that best predict crash incidence. Third, the research team used the same data to analyze crash severity by rural road type and rural county type for five states. Fourth, they used national crash data to examine which rural county types have the highest rates of crashes involving drunk driving, speeding, restraint use, and distracted driving. Lastly, they again used national data to determine which rural counties have experienced the greatest road safety improvements in recent years. An important limitation of the analysis was the small sample size. Because the research team applied a novel approach that combines rural road types with rural county types, they required very detailed crash information. They ultimately chose to limit their analysis to the five states listed above because they (a) are located in several regions of the U.S., (b) have many rural counties, and (c) provide the most complete and detailed data. It should be noted that there were some additional limitations in the use of Alaska as one of the five states. As a rural state outside the contiguous U.S., Alaska had additional unique transportation and data characteristics to consider. Ultimately, it was determined that with the focus of the project being rural, Alaska data should remain. The research team realized that focusing on just five states limited the generalizability of the analysis, however, they intended to use the findings presented herein to inform later phases of this research. They also hoped the findings would inspire further quantitative and qualitative investigation into the unique safety characteristics of rural places. Purpose of the Safety Analysis The purpose of the safety analysis was three-fold: (a) to find patterns in safety outcomes in different rural county and rural road types, (b) to determine the influence of different factors on crash frequency and severity, and (c) to identify rural counties that experienced safety improvements in recent years. This analysis supported the identification of successful behavioral traffic safety countermeasures in later phases of the research project. It also informed the selection of case studies and the discussion guides used in conversations with local safety officials. Safety Data Sources There are many sources of safety data, each with its own advantages and limitations. The research team reviewed several publicly available data sources to determine which were most appropriate for the analysis. To be included, data sources were required to (a) report crashes by county, (b) use the Highway Functional Classification System to categorize roads, and (c) provide variables related to driver behavior and road design that are consistent with those in the research literature. The following sections describe each of the data sources the team considered for this study and whether they were used or not. Highway Safety Information System The FHWA Highway Safety Information System (HSIS) is an extensive dataset containing information about crashes, roads and roadway geometry, and traffic volumes across nine states for multiple years.

23 Maintained by FHWA’s Turner-Fairbank Highways Research Center, HSIS contains a wealth of safety data collected by states and translated into a standard format. Unlike the data sources described below, HSIS includes physical roadway characteristics in addition to details about the incidents themselves—two aspects that are key to safety analysis. Variables relevant to this study include local speed limits, lane and shoulder width, injury classification, and terrain. Two of the nine states included in HSIS have adequate coverage for this study. Other states contained too few roads in particular area types or data years that were decades in the past. The two states that meet the data requirements and are used in certain parts of the analysis are Minnesota and Washington. Highway Performance Monitoring System FHWA’s HPMS is a national level highway information system that includes data on the extent, condition, performance, use, and operating characteristics of the nation's highways. State DOTs are responsible for the collection and reporting of this data to FHWA. The HPMS provides the basis for FHWA’s Highway Statistics Series and Condition and Performance Report. HPMS provides certain information for all public roads in the United States. This includes roadway characteristics for higher functional classes that include pavement condition, traffic, and roadway geometry. However, HPMS provides limited information for lower functional systems, particularly local roads. While HPMS does not provide crash information or other data directly related to safety outcomes, it does provide the traffic data necessary to calculate fatality and injury rates. In the analysis of county-level crash reductions, the research team used HPMS data on average annual daily traffic (AADT) and highway segment length to calculate annual VMT by county, which they then used to calculate measures such as fatal crashes per 100 million VMT. Fatality Analysis Reporting System NHTSA’s Fatality Analysis Reporting System (FARS) is a census of fatal crashes in the U.S. To be included in FARS, a crash must (a) involve a motor vehicle traveling on a traffic way customarily open to the public, and (b) result in the death of a vehicle occupant or nonoccupant within 30 days of the crash. FARS data is collected through cooperative agreements between NHTSA and state agencies responsible for gathering data on qualifying fatal crashes. FARS data is obtained solely from state-level documents including police crash reports, vehicle registration files, state highway department data, death certificates, Emergency Medical Services (EMS) reports, and more. Building on these documents, FARS codes more than 140 data elements related to many aspects of fatal crashes including vehicles involved, driver characteristics, atmospheric conditions, EMS response, contributing factors, geographic context, and more. The research team used FARS data to calculate vehicle fatality rates by county. FARS provides the total number of vehicle fatalities by county for multiple time periods of interest. As described later, they divide this count of fatal crashes by VMT to calculate crash rates by county and smooth the number of fatalities (which vary from year to year) by averaging crash counts over a five-year period, which is a typical practice in safety analysis. This approach is particularly important for rural counties where crash counts tend to be low. Although there is overlap between the years 2014, 2015, and 2016, the midpoint of each range (2014 and 2016) creates a two-year trend for the analysis that utilized the most recently available data at the time.

24 National EMS Information System NHTSA’s National Emergency Medical Services Information System (NEMSIS) is a national database that provides a universal standard for patient care information related to emergency 911 calls. NEMSIS standardizes and aggregates point of care EMS data at a local, state, and national level, helping policymakers accurately assess EMS needs and performance. The data NEMSIS provides is relevant to traffic safety in rural areas, as the database includes location information (i.e., urban or rural) as well as crash-related injury factors such as “auto vs. bicycle/pedestrian thrown, runover, or >20 MPH impact” and “motorcycle crash > 20 mph.” Despite the possible utility of NEMSIS, the dataset does not represent all incidents in reporting states and is not population-based. More importantly, researchers at the University of Utah, where NEMSIS is hosted, were unable to provide geographic detail on NEMSIS data which limits its usefulness when studying differences in crash patterns or EMS response by county type. Regional Integrated Transportation Information System To assist transportation planning agencies with setting targets, programming projects, and measuring and reporting progress, USDOT established the Regional Integrated Transportation Information System (RITIS) to house the National Performance Management Research Data Set (NPMRDS) and associated performance management tools. NPMRDS contains historical speed and travel time probe data for road segments on the National Highway System and a few additional segments near prominent land border crossings with Canada and Mexico. Unlike other traffic data sources, NPMRDS is available at five-minute intervals and distinguishes passenger traffic from freight. RITIS/NPMRDS is available for free to state DOTs and their governmental partners. RITIS also provides tools so DOTs and local planning agencies can use this information, plan and program projects, and track their progress toward achieving their state’s performance management targets. The fine temporal and geographic scales allow users to conduct analysis ranging from holiday travel impact forecasts to after- action reviews of crashes. The research team sought to gain access to RITIS for its speed, traffic volume, incident, and other measurement data. However, neither the research team nor the project panel was able to find a necessary sponsor. Behavioral Risk Factor Surveillance System The Behavioral Risk Factor Surveillance System (BRFSS) is a system of annual health-related telephone surveys sponsored by the Centers for Disease Control and Prevention (CDC) and other federal agencies. The survey asks individuals about their chronic health conditions and use of preventive health services, in addition to their possible risk behaviors, such as smoking, drinking, or use of a seatbelt. All 50 states collect BRFSS data to monitor trends and track state and local health objectives. Although there are survey questions in the BRFSS relating to risky driving behavior, the data does not provide enough geographic detail for this analysis. The BRFSS data is publicly available at the state level and for certain metropolitan and micropolitan counties. Youth Risk Behavior Surveillance System The Youth Risk Behavior Surveillance System (YRBSS) is a system of telephone surveys similar to the BRFSS. The YRBSS is designed to monitor health behaviors that contribute to leading causes of death and

25 injury among youth. This includes questions on dangerous driving behaviors. As with the BRFSS, the YRBSS does not have comprehensive coverage for geographies below state level. In addition, four states (Minnesota, Oregon, Washington, and Wyoming) do not participate in the YRBSS. Local YRBSS data are only available for a few specifically funded school districts and counties monitored by the CDC. Web-based Injury Statistics Query and Reporting System The CDC’s Web-based Injury Statistics Query and Reporting System (WISQARS) is a national database of fatal and nonfatal injury, violent death, and cost of injury data. WISQARS is intended to help researchers, the media, public health professionals, and the public learn more about the public health and economic burden associated with unintentional and violence-related injury in the U.S., including injuries on the nation’s roadways. While comprehensive, WISQARS injury data is of limited use for analysis for two reasons: first, WISQARS data is available only in summary format for privacy reasons and second, WISQARS data is not available at sub-state geographies. Individual State Data Sources Before gaining access to the state-level HSIS data described above, the research team collected crash data directly from several states to determine their applicability for this analysis. The investigated states include the following: • Alabama: Crash data from 2015-2019, including county, crash severity, date/time, location, and functional class. Source: Alabama DOT Integrated Crash Data. • Alaska: Crash data from 2009-2017, including county/borough, crash severity, date/time, location, and functional class. Source: Alaska DOT and Public Facilities Highway Safety Office. • Iowa: Crash data with several attributes including time/day, environmental conditions, lighting, major cause, and route system, but not functional class. Source: Iowa DOT Iowa Crash Analysis Tool (ICAT). • Ohio: Fatal and injury crashes with many attributes including route system (i.e., U.S. route, state route), but not functional class. Source: Ohio DOT’s Transportation Information Mapping System (TIMS). • Michigan: Query-able crash data including many attributes like time/day, environmental conditions, lighting, major cause, and route system, but not functional class. Source: Michigan Traffic Crash Facts. • Montana: Crash data from 2009-2018 with detail on severity, and some tabulations on behavioral attributes such as driver impairment or seatbelt usage, but not functional class. Source: Montana Department of Transportation (MDT). • Nevada: Crash data from 2015 to 2017 and many attributes, but not functional class. Source: Nevada DOT Traffic Crash Data. These data sources offer a wealth of information, and the identified limitations should not imply that states lack publicly available crash data. Ultimately, the research team decided to use the Alabama and Alaska data described above and HSIS data from Minnesota, North Carolina, and Washington. These data

26 sources were most compatible with the rural road safety classification system presented in the Task 1 Memorandum. The HSIS data from Minnesota, North Carolina, and Washington was also comprehensive enough for an inferential analysis using regression. While using only three to five states for portions of the analysis limits its generalizability, the research team felt that Alabama, Alaska, Minnesota, North Carolina, and Washington together offer a variety of county types, road types, and other characteristics that will allow researchers and practitioners to draw meaningful conclusions on factors contributing to rural road safety. Although Alaska is different from the lower 48 states in many respects, its uniqueness makes it an important part of rural safety research. Data Analysis Approach The purpose of this research was to support the development of behavioral countermeasures for rural road safety. The team conducted this research using the following analyses: • Crash frequency and incidence analysis: The research team used regression analysis to identify the influence of various road design characteristics and contextual factors on the number of crashes averaged over a three-year period. Although the focus of this study was on behavior, they investigated road design and context because behavioral countermeasures can help address the dangers associated with speeding, driving in high traffic, and driving in mountainous terrain, for instance. This analysis looked at crash incidence through the lens of both road type and county type and a detailed regression methodology is provided in Appendix A. • Crash severity analysis: The research team performed descriptive analysis to assess how crash severity differs based on road type and county type. Certain behavioral countermeasures take crash severity into account, with targets based, in part, on whether crashes involve fatalities, injuries, or no injuries. This analysis supported this targeted approach by highlighting the places where certain severity levels are more common. • Behavioral factors analysis: The research team performed a descriptive analysis to assess several behavioral factors and how they vary by county type, road type, and crash severity level. These factors include speeding, drunk driving, seatbelt and helmet use, and distracted driving. This analysis supported the development of countermeasures by providing nuance on the relationship between behavioral factors and crash likelihood. • Crash reduction analysis: This phase of this research project involved developing case studies of places that experienced reductions in crashes. To support this phase, national safety data was used to identify rural counties where average annual crashes decreased over time. Conceptual Framework Rural areas have various physical, cultural, and socioeconomic characteristics that influence road safety. As discussed in the Task 1 Memorandum, the research team determined that their safety analysis should focus on different road types and county types to capture the wide variation in rural places. Rather than replicate prior research on the factors influencing rural road safety, they identified key parameters from the HSM that are used to predict crashes. The research team then determined which parameters had analogs in their state-level crash data and used descriptive and inferential statistics to characterize safety situations in various road and county type combinations.

27 The results below reflect a road safety analysis that helped the research team identify behavioral countermeasures that are appropriate for local conditions. While their analysis attempted to draw nationally applicable conclusions, available “national” data was really a collation of incomplete state data sources. Therefore, some models that might interest state and local road safety officials were inestimable due to insufficient data coverage, including those using area types as an explanatory variable in regression analysis. State-Level Analysis Results The following sections summarize state-level analyses of crash incidence and crash severity. The research team provided results for Alabama, Alaska, Minnesota, North Carolina, and Washington state for the analysis of crash frequency and severity. They also provided results for the inferential analysis of crash incidence in Minnesota, North Carolina, and Washington. As discussed earlier, these states offer the necessary data and attributes for their analysis, including county identifiers, roadway functional class, and detailed crash information. While not representative of the entire U.S., the research team believed that these states vary enough in climate, topography, and cultural characteristics to offer insights that will apply to other states with similar conditions. Figure 5 shows the county types in each state and Table 8 provides contextual information. Figure 5. Alaska, Minnesota, and Washington State County Types

28 Table 8. Contextual Information for Alabama, Alaska, Minnesota, North Carolina, and Washington Alabama Alaska Minnesota North Carolina Washington Population (2019 ACS) 4,903,190 731,550 5,639,630 10,488,084 7,614,890 Safety Data Source Alabama DOT Alaska DOT HSIS HSIS HSIS Years of Safety Data Used 2015-2019 2015-2017 2013-2015 2016-2018 2016-2018 Total Lane-Miles N/A N/A 293,250 171,808 18,440 Average Annual Crashes 156,435 7,160 89,900 237,529 62,760 Crash Frequency and Incidence The first step in the analysis was to determine crash incidence using the rural road safety classification system. This involved calculating the average annual number of crashes per road mile for each rural county type and rural road type. For Minnesota, North Carolina, and Washington, the research team also calculated crashes per million VMT.9 Table 10 shows the frequency of crashes by rural road type and rural county type in Alabama. Note that in Alabama, Alaska, and Minnesota, there are no roads classified explicitly as Other Freeways and Expressways. These types of roads are likely included under other categories:10 in Alabama, there are no counties classified as agriculture and extraction, destination, or older-age; in Alaska, there are no counties classified as fringe, older-age, or rural towns; in North Carolina, there are no counties classified as agriculture and extraction; and in Washington, there are no counties classified as rural towns. Crash Frequency The analysis of Alabama crash data revealed distinct patterns across rural county types and road types. In all county types except remote, half of crashes occur on Other Principals and Arterials or Minor Arterials (Table 9). In the state’s three Tribal counties, over half of all crashes occur on Other Principals and Arterials alone. Statewide, the smallest share of crashes occurs on minor collectors. 9 Travel volume data was unavailable for Alabama and Alaska. 10 We suspected (but could not confirm) that this road type was grouped with Other Principals and Arterials in Alaska and Minnesota.

29 Table 9. Frequency of Crashes by Rural County and Rural Road Type in Alabama, 2015-2019 County Type Interstate Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Fringe 15.5% 23.1% 26.6% 18.6% 2.3% 13.9% Micropolitan 7.3% 30.2% 25.2% 18.4% 2.2% 16.6% Remote 8.2% 53.2% 21.7% 4.2% 12.7% Rural Towns 12.4% 28.2% 24.1% 17.7% 3.3% 14.3% Tribal 0.0% 53.6% 20.6% 10.3% 1.1% 14.3% Source: 2015-2019 Alabama Integrated Crash Data from the Alabama DOT averaged over the five-year period. Note: Sixteen percent of crashes have an unknown functional class. This includes 20% of crashes in Metropolitan counties but less than 10% in all other county types. The analysis of Alaska crash data showed noticeable differences among rural road types and rural county types. In Alaska’s sole Agriculture and Extraction county, in the far north region of the state, most crashes occur on either minor collectors or local roads (Table 10), which is unsurprising given the sparse road network. Destination counties in Alaska are quite different; in these counties, most crashes occur on Interstates or Other Principals and Arterials.11 In Alaska’s sole Micropolitan county, most crashes occur on Other Principals and Arterials or on Minor Arterials. In remote counties, most crashes occur on Minor Arterials and Major Collectors. Lastly, in Alaska’s sole tribal county, Interstates account for the largest share of crashes. Table 10. Crash Frequency by Rural County and Rural Road Type in Alaska, 2015-2017 County Type Interstate Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Agriculture & Extraction 10.9% 13.6% 10.9% 30.9% 33.6% Destination 22.9% 39.1% 7.9% 7.3% 13.1% 9.6% Micropolitan 35.4% 33.5% 19.2% 6.5% 5.4% Remote 18.4% 6.5% 25.9% 26.6% 10.3% 12.3% Tribal 35.8% 17.0% 17.7% 17.5% 6.3% 5.6% Source: 2015-2017 Alaska DOT and Public Facilities Highway Safety Office crash data averaged over the three-year period. Note: In Alaska, there are no roads classified as Other Freeways and Expressways. An empty cell indicates that the county type / road type combination does not exist. Crash patterns in Minnesota are more uniform than in Alaska. In all rural county types except agriculture and extraction, the greatest percentage of crashes occur on other principals and arterials (Table 11). Minor Collectors account for the second largest share of crashes in all county types except rural towns and tribal. Crashes on local roads are also considerably more likely in Minnesota than in Alaska. In all rural county types, crashes are least likely to happen on minor collectors. 11 Interstates include Alaska Route 1, Route 2, Route 3 (Parks Highway), and Route 4 (Richardson Highway).

30 Table 11. Crash Frequency by Rural County and Rural Road Type in Minnesota, 2013-2015 County Type Interstate Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Agriculture & Extraction 16.3% 40.2% 13.7% 5.1% 24.7% Destination 5.5% 32.5% 25.7% 17.9% 1.5% 16.9% Fringe 11.5% 25.9% 22.0% 18.1% 4.7% 17.7% Micropolitan 12.5% 28.2% 20.8% 14.4% 3.8% 20.2% Older-age 46.3% 15.6% 15.6% 5.8% 16.8% Remote 5.6% 32.3% 22.4% 17.8% 4.9% 16.9% Rural Towns 12.8% 30.6% 14.9% 16.1% 3.6% 21.9% Tribal 36.8% 17.6% 24.0% 7.2% 14.4% Source: 2013-2015 Minnesota HSIS crash data averaged over the three-year period. Note: In Minnesota, there are no roads classified as Other Freeways and Expressways. An empty cell indicates that the county type / road type combination does not exist. Different patterns emerge when crashes are calculated per million VMT. Using this measure, crashes are most likely to occur on Interstates in Minnesota (Table 12). Interstate crashes are most likely in remote counties, where there are approximately 22 crashes per million VMT annually. In all county types, crashes are least likely to occur on local roads. Table 12. Crashes per Million VMT by Rural County and Rural Road Type in Minnesota, 2013-2015 County Type Interstate Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Agriculture & Extraction 0.51 0.44 0.14 0.06 0.03 Destination 13.27 0.99 2.65 0.90 0.27 0.17 Fringe 11.99 1.31 1.04 0.66 0.33 0.10 Micropolitan 13.96 1.46 1.25 0.72 0.46 0.21 Older-age 0.81 1.04 0.53 0.30 0.07 Remote 21.99 1.11 0.45 0.24 0.08 0.03 Rural Towns 8.37 0.73 0.71 0.42 0.15 0.08 Tribal 0.39 0.41 0.32 0.12 0.06 Source: 2013-2015 Minnesota HSIS crash data averaged over the three-year period. As in Alabama, crashes in North Carolina commonly occur on other principals and arterials and minor arterials (Table 13). In the destination counties located near the Appalachian Mountains and along the Atlantic Ocean, over 30% of crashes happen on other principals and arterials. Crashes in North Carolina are also common on major collectors and local roads. This is especially true in remote, rural town, and tribal counties, where over one-quarter of crashes happen on major collectors. In tribal counties, another quarter of crashes happen on local roads.

31 Table 13. Crash Frequency by Rural County and Rural Road Type in North Carolina, 2016-2018 County Type Interstate Other Freeways & Expressways Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Destination 11.9% 2.9% 31.1% 16.7% 17.0% 4.3% 16.2% Fringe 10.7% 4.2% 27.2% 25.2% 14.9% 5.3% 12.4% Micropolitan 13.7% 1.6% 24.3% 24.1% 18.2% 3.2% 14.8% Older-age 9.2% 1.1% 22.0% 21.7% 20.7% 8.9% 16.4% Remote 8.2% 17.5% 17.9% 27.0% 6.9% 22.4% Rural Towns 19.8% 21.4% 29.0% 10.4% 19.4% Tribal 26.9% 14.0% 14.6% 26.5% 4.3% 26.8% Source: 2016-2018 North Carolina HSIS crash data averaged over the three-year period. Note: Functional class is unknown for 2.3% of crashes. When calculated on a per-VMT basis, crashes in North Carolina appear most common on Interstates and Freeways and Expressways and least common on local roads (Table 14). One exception is in Tribal counties, where crashes are least common on minor collectors. Table 14. Crashes per Million VMT by Rural County and Rural Road Type in North Carolina, 2016- 2018 County Type Interstate Freeways & Expressways Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Destination 8.87 3.11 3.12 1.81 1.09 0.87 0.65 Fringe 8.58 2.37 2.94 1.53 1.02 0.80 0.54 Micropolitan 12.04 1.70 2.08 1.50 0.98 0.89 0.54 Older-age 5.25 3.05 1.49 1.31 1.09 0.80 0.46 Remote 9.03 2.37 1.52 1.21 0.66 0.48 Rural Towns 2.72 0.77 0.86 0.46 0.19 Tribal 15.11 1.07 NA 0.53 0.28 0.29 Source: 2016-2018 North Carolina HSIS crash data averaged over the three-year period. Note: NA indicates that there is not enough travel volume data for analysis. Most crashes in Washington take place on Other Principals and Arterials (Table 15). Otherwise, crashes are common on Minor Arterials and Other Freeways and Expressways, especially in destination counties. Interstate crashes are most common in Micropolitan and Remote counties. In Washington’s sole tribal county, all crashes between 2016-2018 took place on Major Collectors.

32 Table 15. Crash Frequency by Rural County and Rural Road Type in Washington, 2016-2018 County Type Interstate Other Freeways & Expressways Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Agric. & Extraction 1.9% 98.1% Destination 20.1% 59.9% 16.9% 3.0% Fringe 12.1% 48.9% 19.1% 19.9% Micropolitan 28.1% 15.0% 34.3% 14.1% 8.1% <1% <1% Older-age 9.2% 65.2% 11.7% 13.9% Remote 24.8% 6.1% 30.4% 19.9% 18.8% Tribal 100% Source: 2016-2018 Washington HSIS crash data averaged over the three-year period. Note: In Washington, there are no counties classified as rural towns. An empty cell indicates that the county type / road type combination does not exist in the data. On a per-VMT basis, crashes in Washington are most common on Interstates, which are present in the state’s Micropolitan and Remote counties (Table 16). In counties where Interstates do not exist, crashes are most common on Other Freeways and Expressways (Fringe, Older-age) and Other Principals and Arterials (Agriculture and Extraction, destination). Table 16. Crashes per Million VMT by Rural County and Rural Road Type in Washington, 2016- 2018 County Type Interstate Other Freeways & Expressways Other Principals & Arterials Minor Arterials Major Collector Minor Collector Local Road Agric. & Extraction 0.03 1.19 NA Destination 1.77 1.90 1.77 1.04 Fringe 2.87 1.95 0.73 1.11 Micropolitan 3.75 1.71 1.65 1.42 0.98 Older-age 2.29 1.35 1.75 0.88 Remote 6.22 1.12 0.82 0.49 0.46 Tribal NA Source: 2013-2015 Minnesota HSIS crash data averaged over the three-year period. Note: NA indicates that there is not enough travel volume data for analysis. Crash Incidence by Rural Road Type This section summarizes key findings from the analysis of different types of rural roads in Minnesota, North Carolina, and Washington.12 The research team first analyzed Interstates and Other Freeways and Expressways in rural counties. Their analysis showed that in both Minnesota and Washington, crash incidence on Interstates increases with the number of lanes, lane width, severity of terrain (i.e., 12 Alabama and Alaska data did not have the necessary attributes for regression analysis.

33 mountainous), and whether the Interstate passes through an urban area. In Minnesota and North Carolina, crash incidence on Interstates also increases with greater AADT and posted speed limits. For Other Principals and Arterials, the posted speed limit is negatively correlated with crash incidence in Washington and Minnesota, meaning crashes decrease as speeds increase. One hypothesis for this relationship is that higher-speed arterials have lower intersection density, thereby decreasing crash exposure.13 Crash incidence along Other Principals and Arterials is also negatively correlated with the number of lanes, lane width, and width of the right shoulder. This suggests that crashes are less likely when vehicles have more space, controlling for other factors. Crash incidence is also negatively correlated with terrain on Other Principals and Arterials in Washington and Minnesota. This finding is unique to this road type, indicating that there is something different about it that cannot be explained by other factors. For Other Principals and Arterials and Minor Arterials, crash incidence is higher in urbanized areas or clusters. On Minor Arterials, crash incidence also increases with the combination of segment length and AADT (i.e., VMT), which points to the added crash exposure that comes with more vehicles sharing the same road. In Minnesota and North Carolina, crash incidence on Minor Arterials is positively correlated with posted speed limits and negatively correlated with shoulder width, indicating that having the ability to veer outside the road lines or pull over lowers the likelihood of crashes. For major collectors, minor collectors, and local roads, the research team found no statistically significant predictors of crash incidence in Minnesota and Washington.14 In North Carolina, crashes decline with greater shoulder widths on Major Collectors. Crashes increase when passing through urban clusters on both Major Collectors and Local Roads. For all three states, descriptive analysis shows that lower-volume roads have significantly lower overall crash rates. One complicating factor is that people may be less likely to report crashes on these roads, especially in rural areas. Crash Incidence by Rural County Type This section summarizes the analysis of crash incidence by rural county type in Minnesota, North Carolina, and Washington. As discussed in the Task 1 Memorandum, rural America is a heterogenous place with various cultural and socioeconomic characteristics. The research team’s analysis of three states that comprise a mix of rural county types captured some of this nuance. From this perspective, rural Micropolitan counties behave like urban metropolitan counties. In micropolitan counties, crash incidence is positively correlated with number of lanes, lane width, and the presence of an urban area or cluster. Destination counties in Washington and Minnesota are unique from other rural county types in that crash incidence is negatively correlated with number of lanes, lane width, and location within an urban area or cluster. One possible reason is that destination counties attract visitors who may drive more cautiously due to their unfamiliarity with local roads. Posted speed limits generally have a positive impact on crash incidence. In some Destination and Micropolitan counties, where population densities are relatively high, speed limit appears to act as a proxy for intersection density. As discussed previously in relation to Other Principals and Arterials, this relationship seems to describe the effect of increased crash exposure that comes with more intersecting 13 Intersection density is a measure of the number of intersections in a given area (e.g., one square mile). 14 Our Minnesota and Washington models showed high standard errors, suggesting that a lack of sufficient data on these road types could be an issue.

34 roads. In remote counties, where population densities are very low, speed limits are a more reliable predictor of crash incidence. The research team found no statistically significant indicators in Older-age and Rural Town counties. In North Carolina, there were also no statistically significant findings in Tribal counties. There are too few Agriculture and Extraction and Tribal counties in Minnesota and Washington to draw significant conclusions. In North Carolina, there are no Agriculture and Extraction counties. Crash Severity In addition to crash frequency and incidence, crash severity was evaluated using three categories: crashes involving no injuries, crashes involving one or more injuries, and crashes involving one or more fatalities.15 The research team performed this analysis for Alabama, Alaska, Minnesota, North Carolina, and Washington. In all states analyzed, crashes involving no injuries are most common and crashes involving fatalities are least common. Crash Severity by Rural Road Type Table 17 shows the percentage of crashes in Alabama in each severity category by rural road type. Note that in Alabama, there are no roads classified as Other Freeways and Expressways. Fatal crashes are most common on minor collectors and least common on Other Principals and Arterials and local roads. Crashes involving an injury, but no fatality, are most common on minor collectors and least common on Interstates. On Interstates, nearly 80% of crashes involve no injuries or fatalities, which is the highest share among all rural road types. Table 17. Percent of Crashes by Rural Road Type and Crash Severity in Alabama, 2015-2019 Road Type Fatality Injury No Injury Interstates 1.2% 19.1% 79.7% Other Principals & Arterials 0.8% 24.7% 74.5% Minor Arterials 1.0% 26.4% 72.6% Major Collectors 1.5% 29.1% 69.3% Minor Collectors 2.1% 33.4% 64.5% Local Roads 0.8% 23.4% 75.8% Source: 2015-2019 Alabama Integrated Crash Data from the Alabama DOT averaged over the five-year period. Note: In Alabama, there are no roads classified as Other Freeways and Expressways. Sixteen percent of crashes have an unknown functional class. This includes 20% of crashes in Metropolitan counties but less than 10% in all other county types. Table 18 lists percentage of crashes in Alaska by severity category and road type. As in Alabama, there are no roads classified as Other Freeways and Expressways in Alaska. Crashes involving a fatality are most common on Interstates. They are least common on Other Principals and Arterials (the same is true in Alabama) and on Minor Arterials. Crashes involving an injury, but no fatality, are most common on Interstates. These types of crashes are least common on local roads. 15 Includes crashes in rural counties only.

35 Table 18. Percent of Crashes by Rural Road Type and Crash Severity in Alaska, 2015-2017 Road Type Fatality Injury No Injury Interstates 1.5% 33.5% 64.9% Other Principal & Arterials 0.4% 33.4% 66.1% Minor Arterials 0.6% 31.4% 67.9% Major Collectors 1.1% 31.9% 67.0% Minor Collectors 1.1% 31.9% 67.0% Local Roads 1.3% 27.1% 71.5% Source: 2015-2017 Alaska DOT and Public Facilities Highway Safety Office crash data averaged over the three-year period. Note: In Alaska, there are no roads classified as Other Freeways and Expressways. In Minnesota, Interstate crashes are less likely to be fatal than in Alabama and Alaska, which could be due to Minnesota’s relatively flat terrain (Table 19). In fact, over 75% of Interstate crashes involve no fatality or injury. Minor Collectors have the greatest share of both fatal crashes and crashes with injuries but no fatalities. Minor Arterials and Major Collectors also have relatively high shares of the crashes that involve injuries only. Local Roads are like Interstates, where fatal crashes are rare and nearly three-quarters of crashes involve no fatalities or injuries. Table 19. Percent of Crashes by Rural Road Type and Crash Severity in Minnesota, 2013-2015 Road Type Fatality Injury No Injury Interstates 0.2% 24.2% 75.6% Other Principal & Arterials 0.6% 29.7% 69.7% Minor Arterials 0.4% 30.5% 69.0% Major Collectors 0.7% 30.5% 68.7% Minor Collectors 2.0% 38.5% 59.4% Local Roads 0.3% 24.7% 74.9% Source: 2013-2015 Minnesota HSIS crash data averaged over the three-year period. Note: In Minnesota, there are no roads classified as Other Freeways and Expressways. In North Carolina, as in Alabama and Minnesota, fatal crashes and crashes involving injuries, but no fatalities are most common on minor collectors (Table 20). Fatal crashes are least likely to occur on Interstates and Other Principals and Arterials in North Carolina. Interstates are also where crashes involving injuries but no fatalities are least likely to occur, and where crashes involving no injuries are most likely to occur. Similar patterns exist in Alabama and Minnesota.

36 Table 20. Percent of Crashes by Rural Road Type and Crash Severity in North Carolina, 2016-2018 Road Type Fatality Injury No Injury Interstates 0.4% 22.7% 76.9% Freeways & Expressways 0.5% 26.6% 72.9% Other Principals & Arterials 0.4% 29.6% 70.0% Minor Arterials 0.5% 30.3% 69.2% Major Collectors 0.8% 30.5% 68.7% Minor Collectors 1.0% 31.1% 67.9% Local Roads 0.8% 28.7% 70.5% Source: 2016-2018 North Carolina HSIS crash data averaged over the three-year period. In Washington, local roads and Interstates have the smallest share of fatal crashes among rural road types (Table 21). Crashes are most likely to be fatal on Major Collectors, although the share is less than 1% of all crashes on that road type. Crashes are most likely to involve an injury but no fatality on Major Collectors, Minor Arterials, and Other Principals and Arterials. As in Alaska, Washington crashes involving no injuries and no fatalities are most likely on local roads. They are also relatively unlikely on Interstates, which is similar to Minnesota. Table 21. Percent of Crashes by Rural Road Type and Crash Severity in Washington, 2016-2018 Road Type Fatality Injury No Injury Interstates 0.4% 25.3% 74.3% Other Freeways & Expressways 0.5% 28.9% 70.5% Other Principals & Arterials 0.4% 31.2% 68.4% Minor Arterials 0.7% 31.4% 67.9% Major Collectors 0.9% 32.9% 66.1% Minor Collectors 0% 26.3% 73.7% Local Roads 0% 24.0% 75.9% Source: 2016-2018 Washington HSIS crash data averaged over the three-year period. Crash Severity by County Type Table 22 shows crash severity by rural county type in Alabama. Note that there are no agriculture and extraction, destination, or older-age counties in Alabama. Among rural county types, fatal crashes are most common in the state’s remote counties and least common in its tribal counties (metropolitan counties are included for comparison). Crashes involving injuries but no fatalities are most common in remote counties. Crashes involving no injuries are most common in micropolitan counties (again among rural county types).

37 Table 22. Percent of Crashes by Rural County Type and Crash Severity in Alabama, 2015-2019 Type Fatality Injury No Injury Fringe 1.1% 24.5% 74.4% Metropolitan 0.4% 19.3% 80.3% Micropolitan 1.0% 24.0% 75.0% Remote 2.2% 37.4% 60.4% Rural Towns 1.5% 31.2% 67.3% Tribal 0.6% 26.4% 73.0% Source: 2015-2019 Alabama Integrated Crash Data from the Alabama DOT averaged over the five-year period. Note: In Alabama, there are no counties classified as Agriculture and Extraction, Destination, or Older-age. Table 23 shows the percentage of crashes by county type and crash severity in Alaska. Crashes are more likely to be fatal in all rural county types than they are in metropolitan counties. They are most likely to involve fatalities in remote counties, which may be because these counties are the most mountainous. Crashes are most likely to involve injuries but no fatalities in Alaska’s sole agriculture and extraction county. In destination counties, nearly 68% of crashes involve no injuries or fatalities, which is the highest share among all county types. Table 23. Percent of Crashes by Rural County Type and Crash Severity in Alaska, 2015-2017 County Type Fatality Injury No Injury Agriculture & Extraction 1.9% 52.9% 45.2% Destination 1.3% 30.7% 67.9% Metropolitan 0.4% 31.9% 67.6% Micropolitan 0.8% 40.8% 58.4% Remote 3.2% 32.4% 64.3% Tribal 1.7% 33.7% 64.6% Source: 2015-2017 Alaska DOT and Public Facilities Highway Safety Office crash data averaged over the 3-year period. Note: In Alaska, there are no counties classified as fringe, older-age, or rural towns. As in Alaska, fatal crashes in Minnesota are least common in metropolitan counties (Table 24). Minnesota’s sole tribal county has the highest share of fatal crashes followed by the state’s agriculture and extraction counties and its sole older-age county. The older-age county also has the greatest share of crashes involving injuries but no fatalities. In destination counties, nearly 73% of crashes involve no injuries or fatalities, which is the highest share among all county types.

38 Table 24. Percent of Crashes by County Type and Crash Severity in Minnesota, 2013-2015 County Type Fatality Injury No Injury Agriculture & Extraction 2.1% 32.8% 65.0% Destination 0.7% 26.7% 72.5% Fringe 1.4% 33.4% 65.1% Metropolitan 0.3% 27.8% 71.8% Micropolitan 0.8% 29.7% 69.5% Older-age 1.8% 38.5% 59.6% Remote 1.7% 35.9% 62.4% Rural Towns 1.4% 30.3% 68.3% Tribal 2.4% 37.6% 60.0% Source: 2013-2015 Minnesota HSIS crash data averaged over the three-year period. In North Carolina, fatal crashes are most common in Rural Town counties, of which there are two (Table 25). Crashes with injuries but no fatalities are most common in Tribal counties and crashes with no injuries are most common in destination counties. Destination counties are concentrated near the Great Smoky Mountains and along the Atlantic coast in North Carolina’s Outer Banks region. Table 25. Percent of Crashes by County Type and Crash Severity in North Carolina, 2016-2018 County Type Fatality Injury No Injury Destination 0.6% 26.8% 72.6% Fringe 0.6% 32.8% 66.6% Metropolitan 0.6% 28.5% 70.9% Micropolitan 0.6% 29.3% 70.1% Older-age 0.9% 28.6% 70.5% Remote 1.1% 30.3% 68.5% Rural Towns 2.1% 25.9% 71.9% Tribal 1.2% 33.4% 65.4% Source: 2016-2018 North Carolina HSIS crash data averaged over the three-year period. Note: In North Carolina, there are no counties classified as Agriculture and Extraction. Fatal crashes in Washington are most likely in remote counties and least likely in metropolitan counties, which is the same as in Alaska (Table 26). Like in Minnesota, Washington crashes with injuries but no fatalities are most likely to happen in older-age counties. The state’s sole agriculture and extraction county has the greatest share of crashes that involve no injuries or fatalities.

39 Table 26. Percent of Crashes by County Type and Crash Severity in Washington, 2016-2018 County Type Fatality Injury No Injury Agriculture & Extraction 0.9% 21.1% 77.9% Destination 0.7% 30.0% 69.2% Fringe 1.0% 30.5% 68.4% Metropolitan 0.4% 29.1% 70.5% Micropolitan 0.9% 27.3% 71.7% Older-age 0.7% 31.5% 67.7% Remote 1.6% 30.6% 67.8% Tribal 0% 0% 100% Source: 2016-2018 Washington HSIS crash data averaged over the three-year period. Note: In Washington, there are no counties classified as rural towns. National Analysis Results In the following sections, the research team summarized their national analysis of behavioral factors that influence rural road safety and counties that experienced crash reductions. The results of their behavioral analysis pointed to opportunities for targeted behavioral countermeasures that reflect local conditions. The results of the crash reduction analysis supported case study identification in Task 4 of this research. Behavioral Factors Using FARS data, the research team conducted a national analysis to determine how certain behavioral safety factors vary by rural county type. Their work drew on an ongoing FHWA project related to safety and mobility needs in rural areas, to which members of the research team were contributing. As discussed in the Task 1 Memorandum, exploring behavioral factors using the rural road safety classification system provides insight into the potential role of cultural and socioeconomic characteristics. Because of data limitations, the research team was unable to analyze all behavioral factors in the same way. The research team began their analysis by looking at fatal crashes involving drunk driving. When compared with the national average, fatal crashes are most likely to involve a drunk driver in remote, agriculture and extraction, destination, tribal, and older-age counties (Figure 6). In remote counties, 28.9% of fatal crashes involve a drunk driver—the highest among rural county types. This rate is higher than both the national average of 19.6% and the rate for metropolitan counties (18.7%). This finding is especially concerning in places like Alaska and Washington, where crashes of all types are more likely to be fatal in remote counties than in other types.

40 Figure 6. Percent of Fatal Crashes Involving a Drunk Driver The next behavioral factor the research team investigated was speeding. They found that in remote, agriculture and extraction, and destination counties, fatal crashes are more likely to be speeding-related than in all other county types, including metropolitan (Figure 7). Remote counties are the most extreme in this regard; in these counties, 25.5% of fatal crashes are speeding-related. Rural Town counties have the lowest share of speeding-related fatal crashes, at 16.4%. This is lower than the national average of 19.6% and metropolitan average of 17.7%. The research team’s prior analysis showed that in destination counties, there is a negative correlation between crash incidence and posted speed limits. This is particularly true on arterials where intersection density is low. The analysis showed that while speeding may not be a predictor of overall crash likelihood in destination counties, fatal crashes caused by speeding are still more likely than in other rural county types.

41 Figure 7. Percent of Fatal Crashes Involving Speeding Next, the research team looked at differences in restraint use by county type. This includes seatbelt use for vehicle occupants and helmet use for motorcyclists. In all rural county types, vehicle occupants and motorcyclists killed in fatal crashes are less likely to be restrained relative to metropolitan counties (Figure 8). This pattern is most extreme in remote, rural town, and tribal counties, where more than 50% of occupants killed in fatal crashes are unrestrained. golden and destination counties have fatality rates closest to the national and metropolitan county averages. Figure 8. Percent of Occupants Killed in Fatal Crashes Who Were Unrestrained or Unhelmeted

42 Lastly, the research team analyzed the percentage of fatal crashes involving a distracted driver. In all but destination counties, fatal crashes are more likely to involve a distracted driver than the national average (Figure 9). This trend is most extreme in remote counties, where 14.7% of fatal crashes involve a distracted driver. As mentioned previously, crashes of all types are more common in remote counties in two states. Figure 9: Percent of Fatal Crashes Involving a Distracted Driver Figure 6 through Figure 8 show the performance rankings of each county type when considering individual behavioral factors (i.e., drunk driving, speeding, restraint use, and distracted driving). To provide a more comprehensive view, the research team created the following combined ranking: 1. Remote 2. Agriculture & Extraction 3. Tribal 4. Destination 5. Rural Towns 6. Fringe 7. Micropolitan 8. Older-age As the list above shows, fatal crashes involving risky behaviors are most common in remote counties and least common in older-age counties. Across all measures, remote counties and agriculture and extraction counties exceed the national average for percent of fatal crashes involving a risky behavior. Crash Reductions Task 4 of this research included case studies of jurisdictions that have experienced rural road safety improvements. To support this effort, the research team performed a national analysis to identify rural counties where crash rates declined between two recent time periods.

43 To identify rural counties that have experienced safety improvements, the project team relied on a methodology derived from the FHWA High-Risk Rural Road (HRRR) program’s Special Rule for High- Risk Rural Road Safety. This Special Rule requires states to dedicate funds to the program if the fatality rate on rural roads increases. Under MAP-21 and the FAST Act, the HRRR Special Rule applies when the fatality rate on a state’s rural roads increases over the most recent two-year period. The Special Rule also instructs states to calculate the fatality rate for rural major and minor collectors and rural local roads using data from the HPMS and FARS. The research team recreated the Special Rule analysis by calculating the change in average annual crash fatalities on a county basis between the periods 2012-2016 and 2014-2018 using all rural U.S. counties. They first used FARS to gather fatal crash counts on rural roads by county for the two time periods. They then used county-level VMT estimates from the 2018 HPMS to generate annual rural fatal crashes per 100 million VMT. To ensure broad geographic representation, the research team categorized rural counties by NHTSA region. Figure 10Error! Reference source not found. presents the location of the 50 candidate counties identified according to the methodology described above. This figure also shows the five rural counties within each region that experienced the largest decrease in their rural road fatal crash rate between the two time periods. Table C- 1 in Appendix C provides the full list of these counties and additional information on county type and fatal crash rate. More information on how this data was used to create the case studies is shared in the Task 4 Chapter.

44 Figure 10: Top Five Rural Counties Within Each NHTSA Region by Decline in Annual Rural Fatal Crashes per 100 Million VMT

45 Task 2 Key Takeaways Below are high-level and statistically significant takeaways from the crash incidence, crash severity, and behavioral factors analyses. The findings provided insights the research team explored further in the literature review and case studies. Findings by county type: • Agriculture & Extraction – In Alaska and Minnesota, Agriculture and Extraction counties are unique because local roads account for the greatest share of crashes. These findings suggest a need for countermeasures targeted toward high-volume roads in primarily non-agricultural counties and low-volume roads in Agriculture and Extraction counties. – Crashes are more likely to be fatal in rural counties than they are in metropolitan counties. In Minnesota, crashes in Agriculture and Extraction counties are most likely to be fatal (along with Tribal counties). This finding suggests a need to design countermeasures for counties with exceptionally low population densities. – Agriculture and Extraction counties have the highest shares of speeding-related fatal crashes and fatal crashes involving a drunk driver (along with destination and remote counties) and fatal crashes involving a distracted driver (along with remote counties). This suggests that there is something unique about this county type as it relates to culture and norms surrounding speeding, drunk driving, and distracted driving. • Destination – On arterials where intersection density is likely low, crash incidence decreases even as speed increases. (It is possible that the crashes that occur are more likely to be fatal in areas with low intersection density. However, this was not a statistically significant finding in the analysis.) The positive relationship between intersection density and crash incidence is particularly apparent in destination and micropolitan counties, and not apparent in remote counties where intersection densities are low. In remote counties, higher speeds do lead to higher crash incidence. – Destination counties are unique among other rural county types, suggesting that visitors play some role in safety outcomes. This calls for a better understanding of the types of drivers in destination counties. – Destination counties have the highest shares of both speeding-related fatal crashes and fatal crashes involving a drunk driver (along with Agriculture and extraction and remote counties). This suggests that there is something unique about this county type as it relates to culture and norms surrounding speeding and drunk driving.

46 • Fringe – Our analysis did not reveal anything particularly unique about Fringe counties with regard to crash frequency, incidence, severity, and behavioral factors. This suggests an opportunity for future research that focuses on rural counties located near metropolitan areas. • Micropolitan – Micropolitan counties behave similarly to metropolitan counties regarding the factors that influence crash incidence—i.e., number of lanes, lane widths, and presence of urban areas or clusters. This suggests that behavioral countermeasures designed for rural areas may not be as effective in Micropolitan counties as they are in other rural county types. • Older-age – In Minnesota and Washington, crashes with injuries but no fatalities are most likely to take place in Older-age counties. – When compared with the national average, fatal crashes are more likely to involve a drunk driver in Older-age counties. This is based on an analysis of all U.S. counties. • Remote – In Alaska and Washington, crashes in remote counties are the most likely to be fatal. This finding suggests a need to design countermeasures for counties with exceptionally low population densities. – Remote counties have the highest shares of both speeding-related fatal crashes and fatal crashes involving a drunk driver (along with Agriculture and Extraction and destination counties). This suggests that there is something unique about this county type as it relates to culture and norms surrounding speeding and drunk driving. – Fatal crashes involving unrestrained or unhelmeted occupants are especially common in remote counties (along with rural town and tribal counties). This points toward something unique about seatbelt and helmet use in this county type. – Remote counties have the highest shares of fatal crashes involving a distracted driver (along with Agriculture and Extraction counties). This finding indicates a need for place-specific behavioral countermeasures that discourage distracted driving. • Rural Towns – Fatal crashes involving unrestrained or unhelmeted occupants are especially common in Rural Town counties. This points toward something unique about seatbelt and helmet use in this county type. • Tribal – In Minnesota, crashes in Tribal counties are most likely to be fatal (along with Agriculture and Extraction counties). This finding suggests a need to design countermeasures for counties with exceptionally low population densities. – Fatal crashes involving unrestrained or unhelmeted occupants are most common in remote, rural town, and tribal counties. This points toward something unique about seatbelt and helmet use in this county type.

47 Findings by select road types: • As measured by crashes per road mile, most rural crashes take place on just three road types: Other Principals and Arterials, Minor Arterials, or Major Collectors. As measured by crashes per million VMT, most rural crashes take place on Interstates. • Minor Arterials show greater crash incidence as VMT increase, suggesting a need to focus on safe driving practices in heavy traffic. (It is possible that this relationship holds true for all road types, but there may be exceptions when traffic is greater, but road conditions are safer, leading to lower crash incidence. Only Minor Arterials showed a statistically significant relationship.) • In three of the states analyzed—Alabama, Minnesota, and North Carolina—Minor Collectors stand out as road type where fatal crashes and crashes involving injuries but no fatalities are most common. In Alabama, Alaska, and Washington, fatalities are least common on Other Principals and Arterials. • Local Roads are among the least injurious road types in each state analyzed (i.e., least likely to involve an injury or fatality). • Crashes are more likely on rural road types that pass through urban clusters. This suggests that behavioral countermeasures should target safe driving practices in high traffic areas.

Next: Literature Review (Task 3) »
Highway Safety Behavioral Strategies for Rural Areas Get This Book
×
 Highway Safety Behavioral Strategies for Rural Areas
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Rural roads have a higher risk of fatality or serious injury than urban roads due to factors such as varying terrain, wildlife, and long distances between services.

BTSCRP Web-Only Document 4: Highway Safety Behavioral Strategies for Rural Areas, from TRB's Behavioral Transportation Safety Cooperative Research Program, documents the overall research effort that produced BTSCRP Research Report 8: Highway Safety Behavioral Strategies for Rural and Tribal Areas: A Guide. Supplemental to the document is a PowerPoint presentation that outlines the project.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!