Identification of Factors Contributing to the Decline of Traffic Fatalities in the United States from 2008 to 2012 (2020)

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.

17 Data series were assembled to cover the vehicle, driver, and environmental factors identified as related to traffic safety. In addition, numerous economic series were collected to reflect the influence of the economy on traffic safety. This chapter provides a list and discussion of the data and sources used. The data series collected were largely at the state level, meaning that the data were collected for each state and year within the period 2001 through 2012. However, data were not available at the state level on vehicle fleets. Accordingly, data on vehicle characteristics at the national level were used in the models. However, unless otherwise identified, all data series are by year and state. The data included all 50 states. The District of Columbia was excluded because it introduced excessive variance and only accounted for 0.1% of traffic fatalities in the period. 5.1 Crash Data The NHTSA FARS is the standard source for data on fatal traffic crashes in the United States. FARS provides a census file of all motor vehicle crashes in the United States that occurred on a trafficway customarily open to the public, and in which one or more persons died of crash injuries within 30 days of the crash. The FARS data set is composed of data compiled by analysts who are housed within each state. Data elements cover crash-level, vehicle-level, and person-level information. The data are collected from police accident reports, death cer- tificates, vehicle registration files, hospital and coroner records, EMS reports, state highway department data, and other state records. There is one record for each crash, vehicle, and person involved in a fatal crash (NCSA 2014). FARS data were used for all analyses of fatal traffic crashes in this report. NHTSA’s General Estimates System data were used where it was informative to examine crashes of all severities. GES is a nationally representative probability sample of police-reported crashes in the United States. Crashes are sampled by a stratified, hierarchical sampling system from about 400 police jurisdictions nationally. About 50,000 crash reports are sampled each year. All data in GES were coded from police reports, without any additional investigation. The variables and code levels are largely consistent with variables and code levels in FARS (NHTSA 2014). GES data are sampled through a national sampling structure, and cannot be used to form state-level estimates. 5.2 Sources of Other Data Used Exposure includes all types of measures that reflect the opportunity or exposure to the pos- sibility of a crash. Table 5-1 lists the primary sources of exposure data. Population data are avail- able from the U.S. Census Bureau (2016) for each state, including counts by state, age cohort, C H A P T E R 5 Data

18 Identification of Factors Contributing to the Decline of Traffic Fatalities in the United States from 2008 to 2012 and year. Years between the census years of formal counts are estimated by interpolation. The bureau also has estimates of state area, which were used to compute population density. Road miles by FHWA function classes were used to normalize state highway expenditures, controlling for differences in the sizes of states. VMT most directly captures exposure to crashes and is probably the most important measure of exposure used. The FHWA Highway Statistics publication provides annual estimates by road- way function class (including urban and rural) and vehicle type. Highway Statistics also includes vehicle registration data by year and state, and many other relevant data series, which will be discussed later (FHWA 1992-2014). Data series on employment, the labor force, and unemployment were obtained from the Bureau of Labor Statistics (2016), based on the Current Population Survey (Table 5-2). The data are available by year and month for each state; annual state-level estimates were obtained by summing across the employment and labor force counts and taking the average. Employ- ment was defined as the total number of persons on establishment payrolls employed full or part time who received pay for any part of the pay period that included the twelfth day of the month. Unemployed persons were defined as all persons 16 or older who had no employment, were available for work, and had made specific efforts to obtain employment. The labor force was defined as all persons either employed or unemployed according to those definitions. The full definitions can be obtained at http://www.bls.gov/sae/790faq2.htm#Ques3. GDP estimates by state and year were obtained from the U.S. Department of Commerce. GDP measures the gross productive output of a state, so it is used as a gross estimate of economic activity. The estimates were divided by population estimates to produce GDP per capita esti- mates. Median household income estimates were obtained from the Bureau of Census Current Population Survey. The estimates available were for 2- to 3-year periods, not for individual years. Estimates for individual years were obtained by averaging over spans of years. For example, to obtain an estimate for 2010, estimates for 2009–2010 and 2010–2011 were averaged. Household income combines all incomes within a household, while GDP per capita is on a per person basis. All monetary estimates were converted to constant 2013 dollars using the Consumer Price Index (CPI) calculator at the Bureau of Labor Statistics (Bureau of Labor Statistics 2013). Data Source Population by state and age Census Bureau (2016) Table 2. Intercensal Estimates of the Resident Population by Sex and Age Square miles by state Census Bureau, Geography, accessed at https://www.census.gov/geo/reference/state-area.html Road miles by roadway function class and year Highway Statistics, FHWA, Table HM-10 for each year, 2001– 2012 VMT by roadway function class, vehicle type, urban/rural, national Highway Statistics, FHWA, Table VM-1 for each year, 2001–2012 VMT by roadway function class, urban/rural, by state Highway Statistics, FHWA, Table VM-202 for each year, 2001– 2012 Vehicle registrations by type and state Highway Statistics, FHWA, Table MV-1 for each year, 2001–2012 Table 5-1. Exposure data series.

Data 19 Fuel prices were obtained from the U.S. Energy Information Administration, State Energy Data System (2016). Prices for regular-grade gasoline were selected for fuel prices because they represent the most common grade of fuel used. The prices were converted from prices per million BTUs to gallons, and then converted to constant 2013 dollars. Fuel taxes, in terms of cents per gallon, were available in the FHWA’s Highway Statistics Series, Table MF-205, which tabulates fuel taxes for each state. Again, tax values were converted to constant 2013 dollars and summed with the fuel cost to produce an estimate of the price at the pump. The Insurance Institute for Highway Safety (IIHS) maintains a valuable set of digests of state laws concerning critical aspects of traffic safety. These data were used to develop the indexes on the strength of state belt laws and motorcycle helmet requirements. Belt-use rates are available from the continuing National Occupant Protection Use Survey (NOPUS), published annually by NHTSA (reported in Chen 2014; Chen and Ye 2009). The National Institute of Alcohol Abuse and Alcoholism publishes estimates of per capita consumption of beer, wine, and alcoholic spirits. These are available by state and year. Kathleen Klinich of UMTRI (2016) has been com- piling state laws related to drunk driving and shared data that were used to develop an index of state penalties and regulations. ESC penetration rates were estimated from a Highway Loss Data Institute report (2014). And finally, the penetration of post-1991 model year vehicles into the fleet was estimated using quasi-induced exposure methods. The rate of penetration was used as a surrogate for the spread of more crashworthy vehicles, in response to NHTSA’s New Car Assess- ment Program (2007) and the strengthening of the Federal Motor Vehicle Safety Standards (see Table 5-3). Data series on state highway expenditures are available in the FHWA’s Highway Statis- tics series (1992–2014) (see Table 5-4). This series is an exceedingly valuable resource for highway safety. Each year, states report highway spending disaggregated by several types of activities, using a set of common forms, definitions, and instructions. Funding under the Highway Safety Improvement Program (HSIP) was compiled from FHWA funding tables under SAFETEA-LU (2005) and MAP-21 (2012), which are available on the FHWA website (see FHWA 2016a). Data Source Employment, total counts of employed by state, month, and year Bureau of Labor Statistics, Current Population Survey, Local Area Unemployment Statistics Labor force, by state, month, and year Bureau of Labor Statistics, Current Population Survey Unemployment rate, by state, month, and year Bureau of Labor Statistics, Current Population Survey State GDP by year U.S. Department of Commerce, Bureau of Economic Analysis, Regional Economic Accounts: Download State median household income by year U.S. Census Bureau, Current Population Survey, Annual Social and Economic Supplements Fuel tax by state by year Highway Statistics, FHWA. Table MF-205 Fuel costs U.S. Energy Information Administration (2016), State Energy Data System, prices for regular gasoline. Data are converted from prices per million BTUs. Table 5-2. Economic data series.

20 Identification of Factors Contributing to the Decline of Traffic Fatalities in the United States from 2008 to 2012 Data Source Seat belt, primary vs. secondary, by state and year Compiled from Insurance Institute for Highway Safety, digest of state laws, available at http://www.iihs.org/iihs/topics/laws/safetybeltuse Belt-use rates Compiled from NHTSA’s NOPUS program (Chen 2014; Chen and Ye 2009) Blood alcohol concentration limit, per se, other alcohol-related laws and penalties, by state by year Compiled from state laws, index developed from Klinich (2016) Motorcycle helmet by state by year Digest of motorcycle helmet laws from IIHS website, accessed at http://www.iihs.org/iihs/topics/laws/helmetuse/helmethistory? topicName=Motorcycles#tableData Alcohol consumption Compiled from National Institute of Alcohol Abuse and Alcoholism (Haughwout, LaVallee et al. 2015) ESC penetration Compiled from Highway Loss Data Institute report on the penetration of collision avoidance technologies (Highway Loss Data Institute 2014) Post-1991 model year Estimated from GES using a quasi-induced exposure technique Table 5-3. Driver- and vehicle-related framework. Data Source Capital expenditures Compiled from Highway Statistics, FHWA, Table SF-2, includes construction, relocation, resurfacing, restoration, rehabilitation and reconstruction, widening, capacity improvements, restoration of failed components, additions and betterments of roads and bridges. See Federal Highway Administration (N.D.) Maintenance Compiled from Highway Statistics, FHWA, Table SF-2, includes preserving the entire highway, including surface, shoulders, roadsides, structures, and traffic control devices, as close as possible to the original condition as designed and constructed Administration, research, planning Compiled from Highway Statistics, FHWA, Table SF-2, including all general and miscellaneous expenditures not related to a specific project, expenditures for administration, research, and planning Law enforcement and safety Compiled from Highway Statistics, FHWA, Table SF-2, including all relevant federal safety programs, sections 402, 403, 405, 406, 407, 408, 410, and 411 of Title 23 of the United States Code, as well as the Motor Carrier Safety Assistance Program (MCSAP). Also includes capital expenditures designated by states as safety related Highway Safety Improvement Program Compiled from FHWA funding tables under SAFETEA-LU and MAP-21, available from https://www.fhwa.dot.gov/safetealu/fundtables.htm and https://www.fhwa.dot.gov/map21/funding.cfm Table 5-4. Highway expenditures.

Data 21 Highway spending was used in the statistical models to capture the effects of infrastructure and state highway programs on safety. Clearly, highway spending is an imperfect surrogate because the cost-benefit ratios of projects differ. However, it is believed that this surrogate is the best currently available. There are evaluations of specific projects, and crash-modification factors (CMFs) have been developed for different types of projects (AASHTO 2010a, 2010b). But there are no comprehensive data to translate CMFs into variables that capture the effects of modifications in a system-wide fashion. For example, there is ample literature evaluating the safety effect of installing rumble strips on shoulders and centerlines, but no comprehensive data on the penetration of rumble strips into the roadway system. Moreover, a safety-related spending variable was constructed that aggregates all spending that states themselves identi- fied as safety related: law enforcement, state educational safety programs, and the portion of capital spending that the states declared to be safety related. Finally, it is assumed that state departments of transportation attempted to deploy their resources effectively. There are, no doubt, variations in effectiveness, but in light of currently available data, highway spending should be a reasonable approximation.