National Academies Press: OpenBook
« Previous: Chapter 3 - Participant Sample Design and Operations
Page 16
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 16
Page 17
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 17
Page 18
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 18
Page 19
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 19
Page 20
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 20
Page 21
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 21
Page 22
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 22
Page 23
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 23
Page 24
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 24
Page 25
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 25
Page 26
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 26
Page 27
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 27
Page 28
Suggested Citation:"Chapter 4 - Data Comparisons." National Academies of Sciences, Engineering, and Medicine. 2015. Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data. Washington, DC: The National Academies Press. doi: 10.17226/22196.
×
Page 28

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.

16 C h a p t e R 4 Chapter 3 documents how the SHRP 2 NDS data came from a convenience sample of drivers. Analysts using the data must keep this sampling method firmly in mind. Analysis methods must account for the extent to which the aggregated SHRP 2 data differ substantially from the national distribution of char- acteristics that influence the subject being analyzed. For exam- ple, if driver age is important to the analysis, then the analyst must account for the fact that the SHRP 2 data contain a rela- tively greater proportion of younger and older drivers than the national driver population. As a first step, the SHRP 2 data could be weighted to match the national driver age distribution, as illustrated in this chapter. This chapter compares a variety of SHRP 2 data elements with local and national distributions on key driver, vehicle, and crash variables. These comparisons should help data users understand some of the biases in the aggregated SHRP 2 data. participant Characteristics This section describes the SHRP 2 sample in terms of a vari- ety of participant and socioeconomic factors: age, race, eth- nicity, household income, employment/work status, marital status, and education level. The SHRP 2 sample on these fac- tors is compared with individuals living within the SHRP 2 recruitment counties, the SHRP 2 states, and the nation as a whole. The comparisons include a series labeled “SHRP 2 Sample Weighted According to Age/Gender” that was derived by weighting the SHRP 2 data according to the distribution of licensed drivers across the United States, as presented in Figure 4.1 and more explicitly defined in Appendix A. The series labeled “SHRP 2 Recruitment Site Counties Weighted by Nominal Site Size” is derived according to the method described in Appendix A. The gender distribution in the study compares favorably with that of the U.S. driving population in 2012: 0.48 male in SHRP 2 compared with 0.50 male in the U.S. driving popula- tion (FHWA 2014). The distribution of participants across age groups is com- pared with the U.S. driving population in 2012 (the latest full year of data collection in the SHRP 2 NDS) in Figure 4.1. As noted in the sampling discussion, participants in the youngest and oldest age groups were intentionally oversampled to cap- ture the best information possible for those individuals with an elevated crash risk. Table 4.1 shows the percentage of the SHRP 2 participant pool self-identifying as Hispanic across the data collection sites. (The U.S. Census Bureau recognizes only two ethnicities: Hispanic and non-Hispanic.) Figure 4.2 shows the percentage of Hispanic participants in the SHRP 2 sample compared with the aggregate of the recruit- ment areas, the SHRP 2 states aggregated, and the nation. The figure shows that the SHRP 2 sample was notably less Hispanic than would have been expected within the recruitment areas, the SHRP 2 states, and the nation at large. In fact, nationally the Hispanic percentage of the population is on the order of four times that of the SHRP 2 sample. Figure 4.3 compares the racial distribution of the SHRP 2 sample with that of the SHRP 2 recruitment sites, the SHRP 2 states, and the nation. All are generally very similar, with the SHRP 2 sample including a notably greater proportion of indi- viduals who identify themselves as white and, in turn, a notably smaller proportion who identify themselves as black or African- American. Respondents to the 2010 U.S. Census identified either as single race (97%) or multirace (3%). For the purposes of this representation, the two groups have been combined with the designation indicating the predominant race identified for multiracial respondents. Socioeconomic Factors Figure 4.4 shows participant responses indicating annual household income. The majority of the SHRP 2 sample (56%) had an annual household income lower than the average house- hold income of each of the following comparison groups: the Data Comparisons

17 aggregated geographic areas from which the sample was drawn ($63,419), data collection sites weighted according to nominal DAS allocation ($65,384), and the national average ($70,883). Figure 4.5 shows the work status of the SHRP 2 sample. To achieve consistency with Census data, students are included in the employed population. A majority of SHRP 2 participants were employed in some capacity. Further, many participants who self-identified as not working outside the home were found on deeper analysis to be engaged in gainful labor on a full-time or part-time basis. Figure 4.6 shows the percentage of the SHRP 2 sample employed compared with the aggregate of the recruitment areas, the SHRP 2 states aggregated, and the nation. Consistent with Census data, the percentage of the SHRP 2 sample identi- fied as employed comprises those identifying as employed full time or part time and as students. Due to lack of availability of Census data on individuals aged 65 and older and in the interest of providing a meaningful comparison between the study sample and the Census data, only data from study par - ticipants between the ages of 16 and 64 are presented in Fig- ure 4.6. Although the percentages are all very similar, the employed percentage of the SHRP 2 sample was slightly greater than in the comparison areas. Figure 4.7 shows the percentage of the SHRP 2 sample employed full time compared with the aggregate of the recruit- ment areas, the SHRP 2 states aggregated, and the nation. Due to lack of availability of Census data on individuals aged 65 and older and in the interest of providing a meaningful com- parison between study sample and Census data, only data from study participants between the ages of 16 and 64 are pre- sented in Figure 4.7. Approximately 60% of SHRP 2 partici- pants aged 16 to 64 were employed full time, but the percentage was lower (40%) when the entire sample was considered, as is the case in Figure 4.5. The percentages associated with the comparison data are all nearly identical. Figure 4.8 shows the marital status of the SHRP 2 sample. Figure 4.9 compares the two largest groups from Figure 4.8, single and married, with the aggregate of the recruitment areas, the SHRP 2 states aggregated, and the nation. Although the percentages are nearly identical for the comparison areas, the SHRP 2 sample has a notably higher percentage of single and a notably lower percentage of married individuals relative to the comparison data. Figure 4.10 shows the level of education attained by the SHRP 2 sample compared with that of the aggregate of the recruitment areas, the SHRP 2 states aggregated, and the nation. In the interest of accurate comparison with the 2010 Census, which only includes respondents aged 25 and older in its representation of educational attainment, only participants Figure 4.1. SHRP 2 participant versus U.S. driving population percentages by age group (years). Source: Dingus et al. 2014. Table 4.1. Characterization of SHRP 2 NDS Sample by Ethnicity SHRP 2 Sample by Site Percentage Hispanic New York 2% Florida 12% Washington 4% North Carolina 2% Indiana 2% Pennsylvania 1% (text continues on page 22)

18 Source: U.S. Census Bureau 2010d. Figure 4.2. Percentage of Hispanic participants in SHRP 2 compared with the U.S. population. Source: U.S. Census Bureau 2010e. Figure 4.3. Racial distribution of SHRP 2 participants compared with the U.S. population.

19 Source: U.S. Census Bureau 2010f. Figure 4.4. Annual household income of SHRP 2 sample and mean annual household income of relevant populations. Figure 4.5. SHRP 2 sample work status.

20 Source: U.S. Census Bureau 2010h. Figure 4.6. Percentage of employed SHRP 2 participants compared with the U.S. population. Source: U.S. Census Bureau 2010h. Figure 4.7. Percentage of SHRP 2 participants employed full time compared with the U.S. population.

21 Figure 4.8. SHRP 2 sample marital status. Source: U.S. Census Bureau 2010g. Figure 4.9. Percentage of single and married SHRP 2 participants compared with the U.S. population.

22 Source: U.S. Census Bureau 2010g. Figure 4.10. Education level of SHRP 2 participants compared with the U.S. population. aged 25 and older in the SHRP 2 sample are included in Fig- ure 4.10. SHRP 2 participants who did not respond to the question are excluded. The SHRP 2 sample seems relatively well-educated, with a greater percentage holding a college degree than in the SHRP 2 states or the nation. Vehicle Fleet Composition The SHRP 2 fleet sample was compared with that of the U.S. fleet based on a snapshot of the latter taken as of January 1, 2012. The two fleets were compared in terms of type (Figure 4.11), model year (Figure 4.12), and make (Figure 4.13). The SHRP 2 fleet included cars, trucks, SUVs/crossovers, and vans, with cars being the dominant vehicle type. This distribution contrasts with the national light vehicle fleet, which, as of January 1, 2012, featured a nearly even split between cars and the other three vehicle types combined. The model years of the vehicles included in the SHRP 2 fleet ranged from 1987 to 2013. These model years made up over 95% of the total U.S. fleet on Janu- ary 1, 2012. Figure 4.12 illustrates a somewhat similar pattern to the U.S. fleet over the model years from 1987 to 2005, but the SHRP 2 fleet starts to spike in model year 2006 due to the study’s strong recruitment preference for these and other Source: R. L. Polk & Co., personal communication, 2014. Figure 4.11. Comparison of SHRP 2 fleet with U.S. light vehicle fleet by vehicle type. (continued from page 17)

23 Source: R. L. Polk & Co., personal communication, 2014. Figure 4.12. SHRP 2 vehicle fleet versus U.S. fleet by vehicle model year. Source: R. L. Polk & Co., personal communication, 2014. Figure 4.13. SHRP 2 vehicle fleet versus U.S. fleet by vehicle make.

24 Figure 4.14. SHRP 2 crashes by severity level (confirmed crash evaluations as of June 30, 2014). later-model vehicles due to the generally greater accessibility to network data in these model years. Interestingly, both graphs show a dip for model year 2009, possibly due to the effects of the globally depressed auto industry during that model year’s sales period. The 19 vehicle makes included in the SHRP 2 NDS accounted for almost 98.5% of the U.S. fleet as of January 1, 2012. Further, the top 17 makes in the SHRP 2 study were also the top 17 makes in the national fleet as of that same date. Note that in the data represented in Figure 4.13, certain makes have been logically grouped based on close affiliation, so Ford, Lincoln, and Mercury are all grouped together simply as Ford. Simi- larly, all GM brands with fundamentally similar vehicles (e.g., Chevrolet, Buick, and Cadillac) have been grouped together under the GM umbrella. However, when one manufacturer owns another, but each continues to produce fundamentally different sets of vehicles, no attempt was made to group them together. For instance, even though Ford Motor Co. owned Volvo’s Car Division from 1999 to 2010, these are still counted and listed as two distinct original equipment man- ufacturers in this report. Figure 4.13 shows that although the same seven makes make up the top percentages in both the SHRP 2 and national fleets, the relative proportions of these seven differ. There are 275 unique make–model combinations repre- sented in the SHRP 2 data and 1,505 in the national fleet as of January 1, 2012. There is a moderate and highly significant correlation between the ranks of the top 100 SHRP 2 make– model combinations and their corresponding national ranking (r = .4854, p < .0001). Notably, each of the top seven make– model combinations in the SHRP 2 data are within the top 20 nationally with one exception: the Toyota Prius. It ranked sec- ond in SHRP 2, comprising 5.62% of the sample fleet, com- pared with a national ranking of 63rd, comprising only 0.42% of the national fleet. Due to the overrepresentation of the Toyota Prius among the early participants, a moratorium was established as of June 1, 2012, on further recruitment or instru- mentation of this model vehicle. Crash Rates Preceding sections of this report describe and compare the characteristics of study participants, their vehicles, and the geographical areas in which the study was primarily conducted with appropriate reference data sets. In contrast, this section compares actual SHRP 2 NDS data to external and comparable sources of information on crashes. Crashes in the SHRP 2 NDS were classified based on level of severity using the following schema: • Level I: airbag deployment, injury, rollover, high delta-V crashes (virtually all Level I crashes would be police-reported [PR] crashes); • Level II: police-reportable crashes (including PR crashes, as well as other crashes of similar severity that were not reported); • Level III: crashes involving physical contact with another object; and • Level IV: tire strike; low-risk crashes. The nature of the NDS data collection methods meant that crashes were not always immediately identified by the research team. In fact, a few crashes may continue to be identified dur- ing subsequent data reduction and analysis activities months, and sometimes even years, after data collection. Available counts as of June 30, 2014, for SHRP 2 crashes are used in the tables and figures that follow. Figure 4.14 illustrates the num- ber of crashes observed in the SHRP 2 NDS by severity level.

25 The number of crashes increased as the crash severity level decreased. The data represented in Figure 4.14 are shown by data col- lection site in Table 4.2. Any publicly available source of data on vehicle crashes deals strictly with PR crashes—obviously, if a crash was never reported to the police, it would be virtually impossible for it to have been counted or otherwise recorded in any formal data collection. This lack of reporting makes data comparisons between public data sources and the SHRP 2 crash database problematic, because it is not always completely clear in the SHRP 2 data set whether a particular crash was reported to the police. If there is a police accident report, then that is an obvious and definitive indication of a PR crash. But if the site contractor indicated that a police report was not filed, then that crash would be counted definitively as not having been reported to the police. However, there are crashes for which it is simply not clear whether the crash was reported to the police. A crash was considered possibly police reported (PPR) if it was known to have been reported or if any of the follow- ing took place: • Notable injury; • Air bag deployment; • Vehicle rollover; • Significant property damage; • Vehicle towed; • Delta-V of greater than 20 mph or an acceleration on any axis greater than 1.3 g (excluding curb strikes); • Large animal strike; or • Sign or roadway furniture strike. As of July 2014, there were 74 PR crashes (i.e., crashes that had been positively confirmed as having been reported to the police) and a possible total of 224 PPR crashes (i.e., crashes that may or may not have been reported to the police) in the SHRP 2 NDS. To be truly meaningful and comparable across data sets, raw crash numbers must be expressed as rates (e.g., per person, data-year, or miles driven). The comparisons drawn in this instance are PR and PPR crashes per data-year identified through July 2014 in the SHRP 2 data set com- pared with crashes per licensed driver for a single year from existing data sources. That is, each licensed driver ostensi- bly contributed one data-year per year, making this a valid comparison. Of course, not everyone listed as a licensed driver in 2012 actually contributed a full year of driving data, as newly licensed drivers could have received their licenses at any point during the year. Similarly, other drivers may have ceased driving for any reason during the year. However, it is also assumed these partial data-year contributors are a small minority. For instance, even if all drivers younger than 20 years of age (approximately 4% of the total) were excluded, pre- sumably the vast majority of even this age group would have still contributed full data-years. Therefore, the overall point estimate for crash rate is considered to be sufficiently accu- rate for comparison purposes. Table 4.3 and Table 4.4 pre- sent 2012 national crash data and data from the SHRP 2 NDS, respectively. The overall comparison between the crash rates observed in the SHRP 2 data set and the national rate (in 2012) is illus- trated in Figure 4.15. Figure 4.15 shows that the crash rates observed in the SHRP 2 NDS are clearly in the same order of magnitude as the national rate in 2012, with the U.S. rate situated between the low (known PR) and high (PPR) SHRP 2 point esti- mates. The low SHRP 2 point estimate of the PR crash rate is around 70% of the U.S. rate in 2012; the high SHRP 2 Table 4.2. Basic Crash Severity Level Data Across Sites (as of June 30, 2014) Crash Severity Level New York Florida Washington North Carolina Indiana Pennsylvania Total I 15 18 19 16 5 1 74 II 27 25 24 10 8 7 101 III 36 56 36 33 23 11 195 IV 57 108 82 47 20 17 331 Total 135 207 161 106 56 36 701 Table 4.3. U.S. Crash Data, 2012 PR crashes 5,615,000a No. of licensed drivers 211,814,830b Crash rate per 1,000 licensed drivers 26.5 a NHTSA 2014b. b Federal Highway Administration 2013.

26 PPR crash rate point estimate is around twice that of the U.S. rate in 2012. Figure 4.16 compares the PR and PPR crash rates observed in the SHRP 2 data set with national crash data across age groups. In one NDS series, only those SHRP 2 crashes con- firmed as being PR are shown; in the second NDS series, all PPR crashes in the SHRP 2 data set are shown. The national crash data are represented by vehicle crash involvement rates (i.e., vehicles involved in crashes per vehicle-year of exposure). The SHRP 2 crash rates may underestimate national rates as unlicensed drivers were excluded from participation. NHTSA (2014a) noted that 19% of fatal crashes in 2012 involved at least one unlicensed driver. The national data illustrate a clear and monotonically decreasing crash rate with increasing age. There are so few data in each of the age bins that it may be difficult to accu- rately interpret the SHRP 2 confirmed crash rate data series. The SHRP 2 possible crash rate data series somewhat more closely matches that of the national data, but the pattern is neither clear nor monotonic. Table 4.5 compares crash data for the SHRP 2 sample states to national data. Figure 4.17 compares the PR and PPR crash rates observed in the SHRP 2 data set with SHRP 2 state and national crash data. The first series presents only those SHRP 2 crashes con- firmed as being PR; state crash rate data are depicted in the second series. In the third series, all PPR crashes in the SHRP 2 Table 4.4. SHRP 2 NDS Confirmed (PR) and Possible (PPR) Crashes New York Florida Washington North Carolina Indiana Pennsylvania Total NDS data-years 878 905 851 660 345 318 3,957 Minimum NDS PR crashes 8 12 25 15 8 6 74 Maximum NDS PPR crashes 48 55 60 31 15 15 224 Source for U.S. rate: NHTSA 2014b. Figure 4.15. SHRP 2 confirmed (PR) and possible (PPR) crash rates compared with the national rate.

27 Table 4.5. 2010 Crash Data by State State No. of Crashes Licensed Drivers (1,000s) Crash Rate (per 1,000 Licensed Drivers) New York 315,377 11,286 27.94 Florida 236,528 13,950 16.96 Washington not available 5,106 not available North Carolina not available 6,537 not available Indiana 193,323 5,550 34.83 Pennsylvania 121,101 8,737 13.86 All 50 states; Washington, D.C.; and Puerto Rico 5,419,000 210,115 25.79 Source: Federal Highway Administration 2014; Volpe National Transportation Systems Center, U.S. DOT, personal communication 2014. Source: NHTSA 2012. Figure 4.16. SHRP 2 confirmed (PR) and possible (PPR) crash rates versus 2012 national crash involvement rate by age group. Figure 4.17. Crash rate comparison by state.

28 data set are displayed. The national crash rate (i.e., from all 50 states; Washington, D.C.; and Puerto Rico) is represented by the horizontal black line. State crash rates were unavail- able for Washington and North Carolina, so they are excluded from Figure 4.17. A primary goal of the SHRP 2 NDS was to amass a suffi- ciently large data set to study crashes from a naturalistic per- spective. By design, this approach facilitates a glimpse into the vehicle to allow for a consideration of the roles that driver dis- traction, drowsiness, and other nondriving behaviors play in the moments leading up to a crash event. With data analysis still an ongoing effort, the preceding graphs bear out the suc- cess of the study in collecting a wealth of such information. Despite the inherent challenges associated with establishing a definitive crash rate for a data set acquired using the natural- istic method to be used as a point of comparison with existing data, overall PR and PPR crash rates observed in the SHRP 2 NDS are clearly in the same order of magnitude as the national crash rate. Specifically, the lower estimate of the SHRP 2 PR crash rate, representing confirmed PR crashes in the data set, is 70% of the national rate, and the upper estimate, corresponding to all possible PPR crashes in the SHRP 2 data set, is approximately double the national rate. Although neither the confirmed nor the possible crash rates observed in the SHRP 2 data set repli- cate the trend seen in the national data of a monotonically decreasing crash rate per licensed driver with increasing age, the possible crash rate more closely matches it.

Next: Chapter 5 - Summary and Conclusion »
Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s second Strategic Highway Research Program (SHRP 2) has released Report S2-S31-RW-1: Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data that provides technical support to users of the SHRP 2 Naturalistic Driving Study (NDS) data. Specifically, the report provides guidance for analysts with weighting SHRP 2 NDS data so they may make comparisons with the U.S. population.

  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!