- Email: [email protected]
Driving exposure by driver age in Michigan
Journal of Safety Research 42 (2011) 193–197
Contents lists available at ScienceDirect
Journal of Safety Research j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j s r
Driving exposure by driver age in Michigan J.P. Ehsani a, b,⁎, C.R. Bingham a, b, J.T. Shope a, b a b
University of Michigan Transportation Research Institute University of Michigan School of Public Health
a r t i c l e
i n f o
Available online 19 May 2011 Keywords: Driving exposure Motor vehicle injury Young driver High-risk driving conditions Workplace safety
a b s t r a c t Background: This study compared driving exposure between two high-crash-risk groups (16–17 and 18-24-yearolds), with a low-crash-risk group (35-64-year-olds). In addition, patterns of association between driving exposure measures and demographic and driving behavior variables were examined. Methods: Respondent's total miles, minutes, and trips driven were calculated within a 48-hour period, using state-wide survey data collected in 2004 and 2005. Results: The youngest drivers drove fewer miles and minutes, but a comparable number of trips as the two older groups. Employment and high vehicle access were associated with greater driving exposure for 16-17-year-olds and 18-24-year-olds. Employment, high household income, large household size, and low vehicle access were associated with greater driving exposure for 35-64-year-olds. More driving was done alone than with passengers present and during the day than at night across all ages. There was a positive association between two driving exposure measures (miles and minutes driven) and demographic and driving behavior variables, which did not extend to trips driven. Discussion: Driving exposure is directly related to stage of life. The entire sample of 16-17-year-old respondents were in high school, which directly influenced their driving times, destinations, and purpose. Those aged 18–24 years displayed driving behavior patterns that were closer to the older drivers, while retaining some differences. The oldest drivers were likely to be shouldering the greatest household responsibilities, and their greater driving exposure may reflect this reality. Impact on industry: These findings provide new information about driving exposure for two high-risk and one low-risk group of drivers. They also raise concern over potential workplace safety issues related to teens’ higher driving exposure, and concomitant crash risk, related to being employed. Future research should examine this issue more carefully so that evidence based recommendations can be made to enhance the safety of teens who are employed, especially those who are employed as drivers. © 2011 National Safety Council and Elsevier Ltd. All rights reserved.
1. Background Motor-vehicle crashes are the leading cause of death and a leading cause of non-fatal injury among teenagers and young adults in the United States (National Highway Traffic Safety Administration [NHTSA], 2008). For every mile driven, 16-19-year-olds are four times more likely to crash than older drivers. Crash risk is highest at age 16 (Williams, 2003), with the crash rate per mile driven nearly twice as high for 16-year-olds as for 18-19-year-olds (Insurance Institute for Highway Safety [IIHS], 2009). Understanding young driver exposure has been identified as a national priority (Transportation Research Board, 2008). Knowing how much exposure vulnerable driver populations have to high-risk conditions is critical to the formulation of effective intervention and prevention strategies. Travel surveys such as the National Household Travel Survey (NHTS; U.S. Department of Transportation: Bureau of Transportation ⁎ Corresponding author at: Department of Health Behavior and Health Education, University of Michigan School of Public Health, 1415 Washington Heights, MI 48109– 2029, United States of America. Tel.: + 1 734 764 5425; fax: + 1 734 763 5455. E-mail address: [email protected] (J.P. Ehsani).
Statistics, 2006) obtain information about amount, type, and distance of travel of a representative sample of U.S. drivers that have been used to estimate exposure for subgroups of drivers. The number of teens sampled, however, is relatively small and insufficient to allow a detailed analysis by specific geographic region, such as the state-level. Moreover, the NHTS does not obtain demographic information on non-household passengers (U.S. Department of Transportation: Bureau of Transportation Statistics, 2006). Previous studies have informed understanding of young drivers’ exposure (Ehsani, Shope, Bingham, Sunbury & Kweon, 2010) and how it relates to crash risk (Williams, 2003). Recent studies have begun to examine the amount and conditions of travel, as well as driver and passenger behaviors, using in-vehicle recording devices (Farmer, Kirley & McCartt, 2010; Neale, Dingus, Klauer, Sudweeks & Goodman, 2005; Stutts et al., 2005). However, little is known about how the youngest drivers’ exposure compares with other age groups. This study seeks to build on earlier research focusing exclusively on 16-17-year-olds (Ehsani et al., 2010) by quantifying driving exposure across three age groups with varying levels of crash risk (National Highway Traffic Safety Administration [NHTSA], 2010b): 16–17 years (highest risk), 18–24 years (high risk), and 35–64 years (lowest risk)
0022-4375/$ – see front matter © 2011 National Safety Council and Elsevier Ltd. All rights reserved. doi:10.1016/j.jsr.2011.04.002
194
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
using three measures of driving exposure (miles, minutes, and trips). The second objective of this paper is to compare patterns of association between the three driving exposure measures and demographic and driving behavior variables.
Health Sciences and Behavioral Sciences Institutional Review Board approved the research reported in this manuscript. The sample for this study consisted of individuals aged 16–17, 18–24, and 35–64 years who reported driving a ‘motor vehicle, van or truck’ within the 48-hour survey period.
2. Method 2.2. Measures 2.1. Sample Michigan Department of Transportation (MDOT) data from a state-wide self-reported survey of 14,315 households conducted between February 2004 and March 2005 (Michigan Department of Transportation, 2009) were analyzed. The state of Michigan was divided into seven clusters, and within each cluster sampling was stratified according to the household size, the number of vehicles in the household, and the number of workers in the household. Recruitment was conducted in two stages. First, a pre-recruitment letter was sent to a sample of households obtained by matching a sample of random telephone numbers with residential addresses. The pre-recruitment letter informed household residents of the study objectives and notified them that they would be contacted and asked to participate. Second, a household member age 18 or older was contacted by telephone to determine each household's eligibility for participation and to invite them to participate in the survey. For those agreeing to participate, the person contacted was designated as the primary respondent for the household. Phone numbers that could not be matched to an address were still included in the survey sampling frame and used for recruitment calls. Households were requested to report the travel characteristics for every member (including children) in travel diaries during a consecutive 48-hour travel period. The primary respondent for each household provided the basic demographic information (age, working status) for every member of the household during the recruitment call. More detailed information, such as school and work-specific information (name, address, etc.) as well as personal travel information, was included in the individual travel diary completed by each family member. The travel diaries were limited to travel occurring on weekdays between Monday and Thursday during the academic school year, meaning that travel diaries during summer, Fridays and weekends, and school holiday periods were not obtained. Each household member had four options for providing their individual travel diary information: in person on the telephone, by proxy on the telephone (i.e., primary respondent provided the travel information recorded by the individual), by mail, or via a dedicated website. The majority of 16-17-year-olds (60.2%) chose to provide their travel diary information via the primary respondent on the telephone, followed by mail (33.6%), in-person on the telephone (5.1%), and online (1.0%). The majority of 18-24-year-olds (51.0%) also provided their travel diary information via the primary respondent on the telephone, followed by mail (28.7%), in person on the telephone (20.0%), and online (0.4%). In contrast, the majority of 35-74-year-olds (52.7%) provided their travel information in person on the telephone, followed by mail (24.9%), via the primary respondent on the telephone (22.1%), and online (0.3%). No significant difference in driving exposure estimates was found by reporting mode. The response rate for eligible households was 48.6% based on the American Association for Public Opinion Research response rate 3 method (American Association for Public Opinion Research, 2008). For each completed individual response, origin and destination points for reported trips were geo-coded by MDOT and a consultant. Permission to use the data for research purposes was granted by MDOT in a written agreement dated November 12, 2008. The survey conducted by the state did not seek Institutional Review Board approval; however, the University of Michigan
2.2.1. Driving exposure Driving exposure was quantified using three measures: minutes of driving, miles driven, and number of trips taken within the 48-hour survey period. Minutes of driving and number of trips taken were reported by respondents in their travel diaries. Miles driven were calculated by the authors using origin and destination coordinate data points projected onto a road network of Michigan using ArcGIS version 9.3. The shortest on-road route between origin and destination points was used to calculate miles driven using Network Analyst. The survey instrument defined a trip as going from one location to the next. Hence, leaving home and picking up a friend, then a stop at the store, followed by arrival at a destination would be considered three trips. Driving time was quantified by asking respondents: ‘when did you leave location 1’ followed by the question ‘when did you arrive at location 2.’ Individual trips of greater than 120 minutes were excluded from the analysis (n = 3 for 16-17-year-olds, n = 18 for 18-24-year-olds, n = 181 for 35-64-year-olds). The proportion of trips over 120 minutes constituted 3% of all trips overall, and 1% of trips for 16–17 year-olds. As outliers, these cases were likely to be qualitatively distinct from the majority of trips and to not represent typical daily driving. In addition, respondents were asked their mode of transportation. If traveling by car, van, or truck, respondents were asked if they were the driver or a passenger. If they reported driving, the presence and number of passengers were also reported, including if any passenger was from the respondent's household. 2.2.2. Demographic variables Age was structured by aggregated age groupings, rather than individual years. The 16- to 17- and 18- to 24-year-old age groups were retained while the 35- to 44-, 45- to 54-, and 55- to 64-year-old age groups were combined to provide a comparison group for which crash risk is at a lifetime low (National Highway Traffic Safety Administration [NHTSA], 2005). A total of 583 16-17-year-olds, 1,250 18-24-year-olds, and 20,367 35-64-year-olds fit the inclusion criteria. A random sample of 5,000 of the 35-64-year-olds was drawn for inclusion in the analyses, and did not differ significantly from the remainder of the 35-64-year-old drivers for any variables used in the analysis. An individual was considered employed if he or she reported being a full-time or part-time worker. Those who reported being unpaid or volunteer workers, not working, and “not applicable” because they were too young were classified as not employed. Household income was dichotomized as below U.S. $50,000 = 0 and U.S. $50,000 and greater = 1. This dichotomization was based on Michigan's 2004 and 2005 median household income of $47,724 (State of Michigan, 2009). Rural or urban residence was defined for these analyses according to the Rural Urban Commuting Area Codes (RUCAs). The RUCA codes are a 10-tiered classification system that uses population size and commuting relationships at the census tract level as the basic building blocks (Hailu & VanEenwyk, 2009). The 10-level code was designed to be aggregated so that they suited the specific needs of separate studies, depending on relevant aspects of connectivity, rural and urban settlement, and isolation (Hart, Larson & Lishner, 2005). In this study, residences were geocoded and matched to their corresponding census tract to establish the RUCA code for each household. A residence was considered urban if the census tract had a RUCA code between 1 and 3, and rural if the census tract had a RUCA code between 4 and 10. RUCA codes 1 to 3 correspond to commuting flows occurring within urban cores of 50,000 inhabitants or greater, as well as
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
metropolitan counties with high commuting flows to urban cores; while RUCA codes 4 to 10 correspond to commuting patterns in population centers of 49,000 inhabitants or less (WWAMI Rural Health Research Center, 2006). Household size was dichotomized as 3 individuals or less = 0 and 4 individuals or more = 1, based on Michigan's average household size of 2.55 (U.S. Census Bureau, 2009). Respondents’ access to a vehicle was estimated by the ratio of vehicles to licensed drivers within each household. This ratio was then used to create meaningful cut-offs reflecting three levels of vehicle access: 1 = low access (from zero to less than 0.5 vehicles per licensed driver); 2 = medium access (from 0.5 to less than 1 vehicle per licensed driver); and 3 = high access (one or more vehicles per licensed driver). To test whether the dichotomization of household income and household size may have led to a loss of information or concealed any non-linearity in the relation between the variable and outcome (Altman & Royston, 2006), scatterplots of household income and the three driving exposure measures, and household size and the three driving exposure measures were examined for each age group. The association between both household income and household size and driving exposure was symmetrically curvilinear around the median for the youngest drivers, and uniformly distributed for both 18–24 and 35–64 year-old drivers. These results indicate that dichotomization of these variables did not divide the sample into subsets with differing association with the outcome, and provide support for the assumption that using a median split of the household income variable for analysis was not a biasing factor.
2.2.3. Trip characteristic variables Two passenger variables were created for the analysis. The number of passengers in the vehicle for each trip was calculated using the sum of two variables, the number of household passengers, and the number of additional passengers. The total number of passengers was categorized as zero passengers, one passenger, and two or more passengers. If there were one or more household passengers in the vehicle and no additional passengers, a trip was classified as occurring with household passengers. A trip with any non-household passengers (either exclusively or in combination with household passengers) was classified as carrying non-household passengers. Nighttime driving was defined as any trip beginning at 9 p.m. or after and ending before 5 a.m. These times correspond to the period of elevated crash risk for teens (Insurance Institute for Highway Safety [IIHS], 2009). Destination was dichotomized. The three most frequented destinations (home, work, and school) were aggregated and compared with a second group including the remaining 12 destinations measured by the survey (attend childcare, eat out, personal business, everyday shopping, major shopping, religious/community, social, recreation participate, recreation watch, accompany another person, pick-up/drop-off passenger, and turn around). This dichotomization resulted in fewer, more exclusive trip destination categories while
195
maintaining in the analyses data on low frequency trip types by incorporating them with other similar trips. 2.3. Plan of analysis The analysis was conducted in two stages. First, univariate descriptive statistics were calculated for each age group. Second, driving exposure was estimated by demographic and trip characteristics, and age groups. Analyses were conducted using SAS 9.0 PROC SURVEYREG in order to accommodate the complex sampling procedure in which households were stratified within clusters (SAS Institute Inc., 2004). 3. Results 3.1. Descriptive results Approximately half the respondents were male and just over half the sample came from rural residences (Table 1). Among 16-17-year-olds, all participants reported their highest educational level as high school. Among 18-24-year-olds, 44.2% reported having an education level of high school or less, while among 35-64-year-olds, 35.7% reported an educational level of high school or less. The proportion of respondents working for income varied widely by age group: less than half the 1617-year-olds (47%), the majority of 18-24-year-old (81%), and over twothirds of the 35-64-year-olds (70%) reported part-time or full-time employment. Eighty one percent of 16-17-year-old drivers reported an annual household income above $50,000. Approximately three-quarters (74%) of 18-24-year-olds and two-thirds (68%) of 35-64-year-olds reported an annual household income above $50,000. Within the 48-hour survey period, drivers aged 16–17 years drove an average of 29.5 miles (SD: 27.0), 82.7 minutes (SD: 59.2), and 6.5 trips (SD: 3.8). Drivers aged 18–24 years drove an average of 55.8 miles (SD: 60.2), 118.8 minutes (SD: 78.9), and 6.9 trips (SD: 3.8); while 35-64-year-old drivers drove an average of 55.0 miles (SD: 56.8), 125.0 minutes (SD: 82.6), and 7.7 trips (SD: 4.3). The average driving speed for 16-17-year-olds was 25.8 miles per hour (mph). Drivers aged 18-24-year-olds and 35-64-year-olds drove an average of 37.3 mph and 34.8 mph, respectively. Driving exposure patterns were distinct across demographic groups. On average, 16-17-year-olds drove just over half the distance and approximately two-thirds of the duration of 18-24-year-old drivers, but drove a similar number of trips. The driving exposure pattern of 18-24-year-olds more closely resembled 35-64-year-olds than 16-17-year-olds. Men aged 35–64 years drove significantly more minutes than women the same age. Employment and high vehicle access were associated with greater driving exposure for 16-17-yearolds and 18-24-year-olds. Employment, high household income, large household size, and low vehicle access were associated with greater driving exposure for 35-64-year-olds (Table 2).
Table 1 Sample characteristics⁎. Age Group
Male High School Graduate or less Working for income Rural residence Household with income above $50,000
Miles driven Minutes driven Trips driven ⁎ Driving exposure estimates are for the 48-hour survey period.
16–17 (N = 583)
18–24 (N = 1,250)
35–64 (N = 5,000)
%
%
%
49.1 100.0
48.8 44.2
48.7 35.7
47.0 55.2 81.6
81.0 51.1 74.2
69.5 53.1 67.7
Mean (95% CI) 29.5 (24.3 – 34.8) 82.7 (72.6 – 92.8) 6.5 (5.9 – 7.0)
Mean (95% CI) 55.8 (49.4 – 62.2) 118.8 (107.4 – 130.3) 6.9 (6.4 – 7.3)
Mean (95% CI) 55.0 (53.0 - 57.0) 125.0 (121.5 – 128.5) 7.7 (7.5 – 7.9)
196
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
Table 2 Driving exposure estimates and interactions by demographic characteristics⁎. Interaction Groups
Mean Miles (95% CI)
Sex 16-17 x 16-17 x 18-24 x 18-24 x 35-64 x 35-64 x
30.6 28.6 48.8 62.5 52.9 57.0
(23.3 (20.8 (41.4 (52.4 (46.4 (54.1
-
Employment 16-17 x Unemployed 16-17 x Employed 18-24 x Unemployed 18-24 x Employed 35-64 x Unemployed 35-64 x Employed
25.7 33.7 45.3 58.3 54.2 55.4
(19.4 (26.8 (36.5 (50.7 (47.4 (53.0
Household Income 16-17 x b $50000 16-17 x $50000+ 18-24 x b $50000 18-24 x $50000+ 35-64 x b $50000 35-64 x $50000+
28.0 29.9 54.0 56.5 53.1 55.9
Residence 16-17 x Urban 16-17 x Rural 18-24 x Urban 18-24 x Rural 35-64 x Urban 35-64 x Rural
Table 3 Driving exposure estimates and interactions by trip characteristics⁎.
Mean Minutes (95% CI)
Mean Trips (95% CI)
37.9) 36.3) 56.2) 72.6) 59.4) 59.9)
86.4 79.1 116.9 120.7 134.0 116.5
(73.5 - 99.3) (66.7 - 91.5) (103.8 - 130.0) (107.9 - 133.5) (124.8 - 143.2) (112.6 - 120.5)
6.5 (5.7 6.5 (5.5 6.5 (5.8 7.2 (6.5 7.4 (6.8 8.0 (7.7
-
7.3) 7.5) 7.1) 7.9) 7.9) 8.3)
-
32.1) 40.7) 54.1) 65.9) 60.9) 57.9)
70.2 96.6 87.9 126.0 109.5 131.8
(58.2 - 82.3) (83.3 - 109.8) (73.0 - 102.9) (113.0 - 139.0) (99.2 - 119.8) (127.5 - 136.2)
5.6 (4.9 7.5 (6.7 5.7 (5.0 7.1 (6.6 7.4 (6.8 7.8 (7.7
-
6.3) 8.2) 6.5) 7.6) 7.9) 8.0)
(19.6 (23.4 (44.9 (48.3 (46.7 (53.4
-
36.4) 36.4) 63.0) 64.6) 59.5) 58.5)
81.5 82.9 105.9 123.3 113.1 130.7
(64.0 - 98.9) (71.7 - 94.2) (92.7 - 119.1) (110.1 - 136.6) (103.9 - 122.4) (126.5 - 134.8)
5.9 (4.9 6.6 (5.9 6.5 (5.8 7.0 (6.5 7.3 (6.8 7.9 (7.7
-
7.0) 7.2) 7.1) 7.5) 7.8) 8.1)
30.7 28.6 55.9 55.7 55.2 54.8
(24.1 (21.6 (48.2 (46.3 (49.3 (52.3
-
37.4) 35.6) 63.6) 65.2) 61.1) 57.3)
93.4 74.0 127.7 110.4 136.1 115.2
(81.2 - 105.5) (62.8 - 85.3) (116.0 - 139.4) (96.2 - 124.5) (126.8 - 145.5) (111.4 - 119.0)
6.8 (6.1 6.2 (5.4 6.9 (6.3 6.9 (6.3 7.9 (7.3 7.6 (7.3
-
7.5) 7.0) 7.4) 7.4) 8.4) 7.8)
Household Size 16-17 x 3 or less 16-17 x 4 or more 18-24 x 3 or less 18-24 x 4 or more 35-64 x 3 or less 35-64 x 4 or more
26.9 29.7 62.5 54.8 51.6 58.4
(15.8 (22.9 (48.9 (46.6 (45.0 (55.5
-
38.0) 36.4) 76.1) 62.9) 58.2) 61.3)
67.4 83.3 116.6 119.2 120.5 129.5
(41.1 - 93.7) (71.7 - 95.0) (101.3 - 131.9) (105.3 - 133.1) (107.8 - 133.3) (124.7 - 134.2)
6.8 (5.1 6.4 (5.8 7.3 (6.5 6.8 (6.2 7.2 (6.7 8.2 (8.0
-
8.5) 7.1) 8.2) 7.3) 7.7) 8.4)
Vehicle Access 16-17 x Low 16-17 x Medium 16-17 x High 18-24 x Low 18-24 x Medium 18-24 x High 35-64 x Low 35-64 x Medium 35-64 x High
19.2 29.9 30.9 55.3 55.1 56.3 67.8 59.0 51.3
(10.0 (23.1 (24.4 (39.1 (46.6 (48.2 (58.8 (52.5 (49.1
-
28.3) 36.8) 37.4) 71.5) 63.5) 64.3) 76.8) 65.4) 53.6)
60.9 80.0 89.0 88.6 111.9 125.5 127.2 125.8 124.3
(34.5 - 87.3) (65.3 - 94.8) (75.9 - 102.1) (65.9 - 111.3) (96.9 - 126.8) (111.6 - 139.5) (111.6 - 142.9) (113.1 - 138.4) (119.5 - 129.1)
4.3 (2.8 6.4 (5.7 6.8 (6.2 5.8 (4.8 7.0 (6.4 6.9 (6.5 9.1 (8.3 8.2 (7.7 7.3 (7.1
-
5.9) 7.2) 7.4) 6.7) 7.5) 7.4) 9.9) 8.7) 7.4)
Male Female Male Female Male Female
⁎ Driving exposure estimates are for the 48-hour survey period.
Across all ages, significantly more driving was done alone than with passengers present. Drivers aged 35–64 years drove more with non-household passengers. All respondents reported driving significantly more during the day than at night. Those aged 16-17-years drove significantly more to home, work, or school than other destinations while the two older age groups had more destination diversity than the youngest drivers (Table 3). The three measures of driving exposure showed two consistent patterns in their associations with demographic and driving behavior variables. First, there was a positive association between the number of miles driven and the minutes driven, meaning that if the miles driven were higher for a particular demographic group (e.g., those who were employed), then the minutes driven would also be higher for this group relative to those who were not employed. This pattern was replicated for all trip characteristics, with a single exception: although 18-24-year-olds drove significantly more minutes and trips to home, school, and work, they drove significantly more miles to other destinations (the category that includes social and recreational
Interaction Groups
Mean Miles (95% CI)
Mean Minutes (95% CI)
Mean Trips (95% CI)
Number of Passengers 16-17 x 0 16-17 x 1 16-17 x 2 or more 18-24 x 0 18-24 x 1 18-24 x 2 or more 35-64 x 0 35-64 x 1 35-64 x 2 or more
24.9 10.5 12.1 44.0 31.3 32.2 42.4 19.2 31.1
(19.7 - 30.0) 67.0 (58.8 - 75.3) (5.6 - 15.4) 31.9 (23.1 - 40.6) (7.6 - 16.6) 36.9 (28.8 - 45.0) (38.0 - 49.9) 104.4 (92.8 - 116.1) (23.7 - 38.9) 43.9 (34.2 - 53.6) (26.0 - 38.4) 47.6 (38.1 - 57.0) (37.7 - 47.2) 100.4 (91.9 - 108.8) (13.9 - 24.5) 47.5 (39.4 - 55.7) (29.0 - 33.2) 61.4 (58.2 - 64.5)
5.4 (4.8 2.4 (1.9 2.7 (2.3 5.9 (5.4 2.6 (2.1 3.1 (2.6 6.0 (5.5 2.6 (2.1 4.3 (4.1
- 5.9) - 2.9) - 3.2) - 6.4) - 3.0) - 3.6) - 6.5) - 3.0) - 4.5)
Passenger Type 16-17 x Household 16-17 x Non household 18-24 x Household 18-24 x Non household 35-64 x Household 35-64 x Non household
11.4 11.6 35.5 29.9 20.6 30.2
(6.7 - 16.2) (7.2 - 15.9) (28.3 - 42.7) (23.9- 35.8) (16.0 - 25.2) (28.2- 32.2)
2.7 (2.2 2.6 (2.1 2.9 (2.4 3.1 (2.6 2.8 (2.4 4.1 (4.0
- 3.1) - 3.0) - 3.3) - 3.5) - 3.3) - 4.3)
Time of Day 16-17 x Daytime 16-17 x Nighttime 18-24 x Daytime 18-24 x Nighttime 35-64 x Daytime 35-64 x Nighttime
27.5 6.9 50.5 12.0 52.8 10.5
(24.2 - 30.8) 76.3 (69.3 - 83.3) (4.6 - 9.2) 20.9 (16.5 - 25.3) (46.3 - 54.6) 104.3 (96.2 - 112.4) (9.3 - 14.6) 32.1 (27.6 - 36.5) (50.0 - 55.6) 118.8 (114.0 - 123.7) (9.7 - 11.3) 29.2 (27.6 - 30.8)
6.0 (5.6 1.5 (1.3 6.0 (5.8 1.9 (1.7 7.4 (7.1 1.6 (1.5
- 6.4) - 1.6) - 6.2) - 2.0) - 7.6) - 1.6)
Destination Activity 16-17 x Home, Work, School 16-17 x Other 18-24 x Home, Work, School 18-24 x Other 35-64 x Home, Work, School 35-64 x Other
36.6 33.8 49.1 45.3 51.5 57.5
(27.7 (25.9 (39.4 (36.1 (44.2 (54.4
-
45.5) 41.8) 58.8) 54.5) 58.8) 60.6)
21.3 (16.8 - 25.8)
60.6 (53.1 - 68.1)
4.7 (4.2 - 5.1)
12.5 (8.6 - 16.4) 22.7 (18.7 - 26.8)
33.7 (27.5 - 39.9) 84.4 (75.2 - 93.5)
2.7 (2.3 - 3.1) 4.5 (4.1 - 4.9)
44.4 (39.0 - 49.7) 29.7 (25.7 - 33.8)
48.3 (41.8 - 54.7) 76.0 (68.1 - 84.0)
3.3 (2.9 - 3.7) 4.2 (3.8 - 4.6)
30.5 (28.8 - 32.2)
59.5 (57.2 - 61.8)
4.2 (4.1 - 4.4)
⁎ Driving exposure estimates are for the 48-hour survey period.
destinations). Second, in contrast to miles and minutes driven, the number of trips provided less variability as a measure of driving exposure. As a result, the number of trips did not consistently correspond to the same patterns observed in the number of minutes or miles driven. 4. Discussion This study compared two young driver groups that have high motor-vehicle crash risk with a low-crash-risk group of adult drivers. The results indicated that driving exposure patterns varied considerably among the three age groups, suggesting that driving exposure is directly related to stage of life. The entire sample of 16-17-year-old respondents were in high school, which directly influenced their driving times, destinations, and purposes. Those aged 18–24 years displayed driving behavior patterns that were closer to those of older drivers, while retaining some differences. The oldest drivers were likely to be shouldering the greatest household responsibilities, and their greater driving exposure relative to the younger age groups might reflect this reality. Michigan's graduated driver licensing law imposes distinct driving restrictions (e.g., supervised driving, night driving restrictions) for each stage of licensure. Therefore, stage of driver licensure might contribute to or account for differences seen in exposure between the youngest and the two older groups. Within the youngest age group there are likely to be learners, restricted license holders, and full license holders and as a result, driving exposure estimates for this age group are likely to be influenced by the proportion of 16-17-year-olds within each stage of
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
licensure and the corresponding driving restrictions imposed on each stage. While the absence of licensure stage data limits the interpretation of 16-17-year-olds’ driving exposure, the mix of licensure status is very likely representative of the age group. Future studies should consider examining driving exposure by licensure stage and age. In addition to the differences in stage of life, the makeup of the drivers within each age group also differed. The youngest drivers came from more affluent households than the older drivers, making the group itself, and not merely their driving exposure, different from the older cohorts. This finding also suggests that less affluent teens are less likely to drive. Like the older age groups, 16-17-year-olds who were employed reported driving more than those who were not employed. This has clear implications for those who are at higher risk of crashing; specifically, employment for 1617-year-olds, while a positive factor in normal adolescent development may also contribute to higher crash and injury risk. There are several limitations to the generalizability of these findings. Eligibility for participation in the Michigan Travel Counts survey was based on household characteristics rather than those of individual drivers. As a result, the driving exposure estimates provided by this sample may not be representative of all drivers in Michigan. In addition, the responses for the Michigan Travel Counts survey were individual self-reports of distance or duration of travel, which have been criticized for underestimating or overestimating the true durations or distances travelled (Energy Information Administration, 2000; Staplin, Gish & Joyce, 2008; Wolf, Oliveira & Thompson, 2003). However, based on the correlations among the three exposure measures, the findings of this study provide a reasonable representation of driving exposure. Miles driven correlated with self-reported minutes driven at r = 0.30, and with self-reported trips at r = 0.55. The lower correlation between miles and minutes is expected, due to the variation in speed that would change the association between miles driven and minutes driven. Overall, these correlations provide support that these driving exposure estimates have validity in terms of their rank-order associations, though the validity of the point estimates may be less. The Michigan Travel Counts survey restricted sampling to travel occurring between Monday and Thursday during the school year (i.e., while classes are in session), thus not including travel during the summer months and on weekends, two high-risk periods for young drivers. Fatality Analysis Reporting System (FARS) data for Michigan over a 20 year period (1990–2009), however, indicated that, unlike the national data, the winter months also correspond to a period of high crash risk for 16-17-year-olds (National Highway Traffic Safety Administration [NHTSA], 2010a). The absence of weekend and summer driving exposure data in this study make it impossible to determine whether the periods included in the survey represent a biased subset of travel (i.e., weekend and summer travel not only differ in the amount, but also in character of exposure). In addition, the nature of driving might differ on weekdays and weekends for different age groups. To determine whether weekday and weekend travel differed similarly for the three age groups, data from the 2001 National Household Travel Survey (NHTS) were analyzed (U.S. Department of Transportation: Bureau of Transportation Statistics, 2010). Results showed that vehicle miles travelled on Monday, Tuesday, Wednesday, and Thursday totaled 60%, 60%, and 59% of all driving for 16–20 year-olds, 21–25 year olds, and 36–65 year-olds, respectively. Assuming the NHTS 2001 data are representative for Michigan, and travel patterns did not significantly change in the time between the two surveys, it can be concluded that the weekly pattern of driving exposure did not differ by age group. References Altman, D. G., & Royston, P. (2006). The cost of dichotomising continuous variables. BMJ, 332(7549), 1080.
197
American Association for Public Opinion Research (2008). Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys (5th edition). Lenexa, Kansas: AAPOR. Ehsani, J. P., Shope, J. T., Bingham, C. R., Sunbury, T. M., & Kweon, B. (2010). Teen driving exposure in Michigan: demographic and behavioral characteristics. Accident; Analysis and Prevention, 42, 1386–1391. Energy Information Administration (2000). Odometer Versus Self-Reported Estimates of Vehicle Miles Traveled. Washington, DC: Author. Farmer, C., Kirley, B. B., & McCartt, A. T. (2010). Effects of in-vehicle monitoring on the driving behavior of teenagers. Journal of Safety Research. Hailu, A., & VanEenwyk, J. (2009). Guidelines for Using Rural-Urban Classification Systems for Public Health Assessment. Retrieved from http://www.doh.wa.gov/ data/guidelines/RuralUrban1.htm Hart, L. G., Larson, E. H., & Lishner, D. M. (2005). Rural Definitions for Health Policy and Research. American Journal of Public Health, 95(7), 1149–1155. Insurance Institute for Highway Safety [IIHS] (2009). Fatality Facts 2008: Teenagers. Retrieved 12 October, 2009, from. http://www.iihs.org/research/fatality_facts_2008/ teenagers.html Michigan Department of Transportation (2009). MI Travel Counts. Retrieved 30th April, 2009, from http://www.michigan.gov/mdot/0,1607,7-151-9615_51690—,00.html National Highway Traffic Safety Administration [NHTSA] (2005). Traffic Safety Facts 2005: A compilation of motor vehicle crash data from the Fatality Analysis Reporting System and the General Estimates System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. National Highway Traffic Safety Administration [NHTSA] (2008). Youth Traffic Safety Statistics. Retrieved 2nd April, 2009, from http://www.nhtsa.gov/portal/site/nhtsa/ template.MAXIMIZE/menuitem.cd18639c9dadbabbbf30811060008a0c/?javax.portlet.tpst=4427b997caacf504a8bdba101891ef9a_ws_MX&javax.portlet. p r p_ 44 2 7b 99 7 ca a cf 50 4 a8 bd b a1 0 18 91 e f9 a_ v i ew ID =d e ta i l _v i e w&i te mID=d91cd60670268110VgnVCM1000002fd17898RCRDviewType= standard#stats National Highway Traffic Safety Administration [NHTSA] (2010a). Fatality Analysis Reporting System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. National Highway Traffic Safety Administration [NHTSA] (2010b). Traffic Safety Facts 2009 Ealry Edition: A compilation of motor vehicle crash data from the Fatality Analysis Reporting System and the General Estimates System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. Neale, V. L., Dingus, T. A., Klauer, S. G., Sudweeks, J., & Goodman, M. (2005). An overview of the 100-car naturalistic study and findings. : Paper presented at the International Technical Conference on Enhanced Safety of Vehicles. SAS Institute Inc. (2004). SAS/STAT® 9.1 User's Guide. Cary, NC: SAS Institute Inc. Staplin, L., Gish, K. W., & Joyce, J. (2008). Low mileage bias' and related policy implications–A cautionary note. Accident; Analysis and Prevention, 40(3), 1249–1252. State of Michigan (2009). Trends in Household Income. Retrieved 15th May, 2009, from http://michigan.gov/hal/0,1607,7-160-17451_18668_41233-175057–,00.html Stutts, J., Feaganes, J., Reinfurt, D., Rodgman, E., Hamlett, C., Gish, K., et al. (2005). Driver's exposure to distractions in their natural driving environment. Accident; Analysis and Prevention, 37(6), 1093–1101. Transportation Research Board (2008). TRBe-Circular Mid-Year Report. Retrieved from http://www.youngdriversafety.org/docs U.S. Census Bureau (2009). Michigan - Selected Social Characteristics in the United States: 2007. http://factfinder.census.gov/servlet/ADPTable?_bm=y&-geo_id= 04000US26&-qr_name=ACS_2007_1YR_G00_DP2&-context=adp&ds_name=&-tree_id=307&-_lang=en&-redoLog=false&-format= Retrieved 30th April, 2009, from U.S. Department of Transportation: Bureau of Transportation Statistics (2006). National Household Travel Survey Washington, D.C. . U.S. Department of Transportation: Bureau of Transportation Statistics (2010). 2001 Online Table Designer. National Household Travel Survey. Retrieved 23rd December, 2010, from. http://nhts.ornl.gov/tables09/ae/TableDesigner.aspx Williams, A. F. (2003). Teenage drivers: patterns of risk. Journal of Safety Research, 34, 5–15. Wolf, J., Oliveira, M., & Thompson, M. (2003). The Impact of Trip Underreporting on VMT and Travel Time Estimates: Preliminary Findings from the California Statewide Household Travel Survey GPS Study. Transportation Research Record, 1854, 189–198. WWAMI Rural Health Research Center (2006). RUCA Data Code Definitions. Retrieved 21st December, 2010, from http://depts.washington.edu/uwruca/ruca-codes.php Johnathon Ehsani is a doctoral candidate at the University of Michigan School of Public Health with a research interest in transportation and injury. Dr. Ray Bingham is a Research Professor at the University of Michigan Transportation Research Institute, in the Department of Psychiatry of the University of Michigan Medical School, and in the Department of Health Behavior and Health Education of the University of Michigan School of Public Health. Dr. Jean T. Shope is a Research Professor at the University of Michigan Transportation Research Institute where she served as head of the Social and Behavioral Analysis Division from 1995 to 2005. She is also a Research Professor in the Department of Health Behavior and Health Education at the University of Michigan School of Public Health.
Contents lists available at ScienceDirect
Journal of Safety Research j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j s r
Driving exposure by driver age in Michigan J.P. Ehsani a, b,⁎, C.R. Bingham a, b, J.T. Shope a, b a b
University of Michigan Transportation Research Institute University of Michigan School of Public Health
a r t i c l e
i n f o
Available online 19 May 2011 Keywords: Driving exposure Motor vehicle injury Young driver High-risk driving conditions Workplace safety
a b s t r a c t Background: This study compared driving exposure between two high-crash-risk groups (16–17 and 18-24-yearolds), with a low-crash-risk group (35-64-year-olds). In addition, patterns of association between driving exposure measures and demographic and driving behavior variables were examined. Methods: Respondent's total miles, minutes, and trips driven were calculated within a 48-hour period, using state-wide survey data collected in 2004 and 2005. Results: The youngest drivers drove fewer miles and minutes, but a comparable number of trips as the two older groups. Employment and high vehicle access were associated with greater driving exposure for 16-17-year-olds and 18-24-year-olds. Employment, high household income, large household size, and low vehicle access were associated with greater driving exposure for 35-64-year-olds. More driving was done alone than with passengers present and during the day than at night across all ages. There was a positive association between two driving exposure measures (miles and minutes driven) and demographic and driving behavior variables, which did not extend to trips driven. Discussion: Driving exposure is directly related to stage of life. The entire sample of 16-17-year-old respondents were in high school, which directly influenced their driving times, destinations, and purpose. Those aged 18–24 years displayed driving behavior patterns that were closer to the older drivers, while retaining some differences. The oldest drivers were likely to be shouldering the greatest household responsibilities, and their greater driving exposure may reflect this reality. Impact on industry: These findings provide new information about driving exposure for two high-risk and one low-risk group of drivers. They also raise concern over potential workplace safety issues related to teens’ higher driving exposure, and concomitant crash risk, related to being employed. Future research should examine this issue more carefully so that evidence based recommendations can be made to enhance the safety of teens who are employed, especially those who are employed as drivers. © 2011 National Safety Council and Elsevier Ltd. All rights reserved.
1. Background Motor-vehicle crashes are the leading cause of death and a leading cause of non-fatal injury among teenagers and young adults in the United States (National Highway Traffic Safety Administration [NHTSA], 2008). For every mile driven, 16-19-year-olds are four times more likely to crash than older drivers. Crash risk is highest at age 16 (Williams, 2003), with the crash rate per mile driven nearly twice as high for 16-year-olds as for 18-19-year-olds (Insurance Institute for Highway Safety [IIHS], 2009). Understanding young driver exposure has been identified as a national priority (Transportation Research Board, 2008). Knowing how much exposure vulnerable driver populations have to high-risk conditions is critical to the formulation of effective intervention and prevention strategies. Travel surveys such as the National Household Travel Survey (NHTS; U.S. Department of Transportation: Bureau of Transportation ⁎ Corresponding author at: Department of Health Behavior and Health Education, University of Michigan School of Public Health, 1415 Washington Heights, MI 48109– 2029, United States of America. Tel.: + 1 734 764 5425; fax: + 1 734 763 5455. E-mail address: [email protected] (J.P. Ehsani).
Statistics, 2006) obtain information about amount, type, and distance of travel of a representative sample of U.S. drivers that have been used to estimate exposure for subgroups of drivers. The number of teens sampled, however, is relatively small and insufficient to allow a detailed analysis by specific geographic region, such as the state-level. Moreover, the NHTS does not obtain demographic information on non-household passengers (U.S. Department of Transportation: Bureau of Transportation Statistics, 2006). Previous studies have informed understanding of young drivers’ exposure (Ehsani, Shope, Bingham, Sunbury & Kweon, 2010) and how it relates to crash risk (Williams, 2003). Recent studies have begun to examine the amount and conditions of travel, as well as driver and passenger behaviors, using in-vehicle recording devices (Farmer, Kirley & McCartt, 2010; Neale, Dingus, Klauer, Sudweeks & Goodman, 2005; Stutts et al., 2005). However, little is known about how the youngest drivers’ exposure compares with other age groups. This study seeks to build on earlier research focusing exclusively on 16-17-year-olds (Ehsani et al., 2010) by quantifying driving exposure across three age groups with varying levels of crash risk (National Highway Traffic Safety Administration [NHTSA], 2010b): 16–17 years (highest risk), 18–24 years (high risk), and 35–64 years (lowest risk)
0022-4375/$ – see front matter © 2011 National Safety Council and Elsevier Ltd. All rights reserved. doi:10.1016/j.jsr.2011.04.002
194
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
using three measures of driving exposure (miles, minutes, and trips). The second objective of this paper is to compare patterns of association between the three driving exposure measures and demographic and driving behavior variables.
Health Sciences and Behavioral Sciences Institutional Review Board approved the research reported in this manuscript. The sample for this study consisted of individuals aged 16–17, 18–24, and 35–64 years who reported driving a ‘motor vehicle, van or truck’ within the 48-hour survey period.
2. Method 2.2. Measures 2.1. Sample Michigan Department of Transportation (MDOT) data from a state-wide self-reported survey of 14,315 households conducted between February 2004 and March 2005 (Michigan Department of Transportation, 2009) were analyzed. The state of Michigan was divided into seven clusters, and within each cluster sampling was stratified according to the household size, the number of vehicles in the household, and the number of workers in the household. Recruitment was conducted in two stages. First, a pre-recruitment letter was sent to a sample of households obtained by matching a sample of random telephone numbers with residential addresses. The pre-recruitment letter informed household residents of the study objectives and notified them that they would be contacted and asked to participate. Second, a household member age 18 or older was contacted by telephone to determine each household's eligibility for participation and to invite them to participate in the survey. For those agreeing to participate, the person contacted was designated as the primary respondent for the household. Phone numbers that could not be matched to an address were still included in the survey sampling frame and used for recruitment calls. Households were requested to report the travel characteristics for every member (including children) in travel diaries during a consecutive 48-hour travel period. The primary respondent for each household provided the basic demographic information (age, working status) for every member of the household during the recruitment call. More detailed information, such as school and work-specific information (name, address, etc.) as well as personal travel information, was included in the individual travel diary completed by each family member. The travel diaries were limited to travel occurring on weekdays between Monday and Thursday during the academic school year, meaning that travel diaries during summer, Fridays and weekends, and school holiday periods were not obtained. Each household member had four options for providing their individual travel diary information: in person on the telephone, by proxy on the telephone (i.e., primary respondent provided the travel information recorded by the individual), by mail, or via a dedicated website. The majority of 16-17-year-olds (60.2%) chose to provide their travel diary information via the primary respondent on the telephone, followed by mail (33.6%), in-person on the telephone (5.1%), and online (1.0%). The majority of 18-24-year-olds (51.0%) also provided their travel diary information via the primary respondent on the telephone, followed by mail (28.7%), in person on the telephone (20.0%), and online (0.4%). In contrast, the majority of 35-74-year-olds (52.7%) provided their travel information in person on the telephone, followed by mail (24.9%), via the primary respondent on the telephone (22.1%), and online (0.3%). No significant difference in driving exposure estimates was found by reporting mode. The response rate for eligible households was 48.6% based on the American Association for Public Opinion Research response rate 3 method (American Association for Public Opinion Research, 2008). For each completed individual response, origin and destination points for reported trips were geo-coded by MDOT and a consultant. Permission to use the data for research purposes was granted by MDOT in a written agreement dated November 12, 2008. The survey conducted by the state did not seek Institutional Review Board approval; however, the University of Michigan
2.2.1. Driving exposure Driving exposure was quantified using three measures: minutes of driving, miles driven, and number of trips taken within the 48-hour survey period. Minutes of driving and number of trips taken were reported by respondents in their travel diaries. Miles driven were calculated by the authors using origin and destination coordinate data points projected onto a road network of Michigan using ArcGIS version 9.3. The shortest on-road route between origin and destination points was used to calculate miles driven using Network Analyst. The survey instrument defined a trip as going from one location to the next. Hence, leaving home and picking up a friend, then a stop at the store, followed by arrival at a destination would be considered three trips. Driving time was quantified by asking respondents: ‘when did you leave location 1’ followed by the question ‘when did you arrive at location 2.’ Individual trips of greater than 120 minutes were excluded from the analysis (n = 3 for 16-17-year-olds, n = 18 for 18-24-year-olds, n = 181 for 35-64-year-olds). The proportion of trips over 120 minutes constituted 3% of all trips overall, and 1% of trips for 16–17 year-olds. As outliers, these cases were likely to be qualitatively distinct from the majority of trips and to not represent typical daily driving. In addition, respondents were asked their mode of transportation. If traveling by car, van, or truck, respondents were asked if they were the driver or a passenger. If they reported driving, the presence and number of passengers were also reported, including if any passenger was from the respondent's household. 2.2.2. Demographic variables Age was structured by aggregated age groupings, rather than individual years. The 16- to 17- and 18- to 24-year-old age groups were retained while the 35- to 44-, 45- to 54-, and 55- to 64-year-old age groups were combined to provide a comparison group for which crash risk is at a lifetime low (National Highway Traffic Safety Administration [NHTSA], 2005). A total of 583 16-17-year-olds, 1,250 18-24-year-olds, and 20,367 35-64-year-olds fit the inclusion criteria. A random sample of 5,000 of the 35-64-year-olds was drawn for inclusion in the analyses, and did not differ significantly from the remainder of the 35-64-year-old drivers for any variables used in the analysis. An individual was considered employed if he or she reported being a full-time or part-time worker. Those who reported being unpaid or volunteer workers, not working, and “not applicable” because they were too young were classified as not employed. Household income was dichotomized as below U.S. $50,000 = 0 and U.S. $50,000 and greater = 1. This dichotomization was based on Michigan's 2004 and 2005 median household income of $47,724 (State of Michigan, 2009). Rural or urban residence was defined for these analyses according to the Rural Urban Commuting Area Codes (RUCAs). The RUCA codes are a 10-tiered classification system that uses population size and commuting relationships at the census tract level as the basic building blocks (Hailu & VanEenwyk, 2009). The 10-level code was designed to be aggregated so that they suited the specific needs of separate studies, depending on relevant aspects of connectivity, rural and urban settlement, and isolation (Hart, Larson & Lishner, 2005). In this study, residences were geocoded and matched to their corresponding census tract to establish the RUCA code for each household. A residence was considered urban if the census tract had a RUCA code between 1 and 3, and rural if the census tract had a RUCA code between 4 and 10. RUCA codes 1 to 3 correspond to commuting flows occurring within urban cores of 50,000 inhabitants or greater, as well as
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
metropolitan counties with high commuting flows to urban cores; while RUCA codes 4 to 10 correspond to commuting patterns in population centers of 49,000 inhabitants or less (WWAMI Rural Health Research Center, 2006). Household size was dichotomized as 3 individuals or less = 0 and 4 individuals or more = 1, based on Michigan's average household size of 2.55 (U.S. Census Bureau, 2009). Respondents’ access to a vehicle was estimated by the ratio of vehicles to licensed drivers within each household. This ratio was then used to create meaningful cut-offs reflecting three levels of vehicle access: 1 = low access (from zero to less than 0.5 vehicles per licensed driver); 2 = medium access (from 0.5 to less than 1 vehicle per licensed driver); and 3 = high access (one or more vehicles per licensed driver). To test whether the dichotomization of household income and household size may have led to a loss of information or concealed any non-linearity in the relation between the variable and outcome (Altman & Royston, 2006), scatterplots of household income and the three driving exposure measures, and household size and the three driving exposure measures were examined for each age group. The association between both household income and household size and driving exposure was symmetrically curvilinear around the median for the youngest drivers, and uniformly distributed for both 18–24 and 35–64 year-old drivers. These results indicate that dichotomization of these variables did not divide the sample into subsets with differing association with the outcome, and provide support for the assumption that using a median split of the household income variable for analysis was not a biasing factor.
2.2.3. Trip characteristic variables Two passenger variables were created for the analysis. The number of passengers in the vehicle for each trip was calculated using the sum of two variables, the number of household passengers, and the number of additional passengers. The total number of passengers was categorized as zero passengers, one passenger, and two or more passengers. If there were one or more household passengers in the vehicle and no additional passengers, a trip was classified as occurring with household passengers. A trip with any non-household passengers (either exclusively or in combination with household passengers) was classified as carrying non-household passengers. Nighttime driving was defined as any trip beginning at 9 p.m. or after and ending before 5 a.m. These times correspond to the period of elevated crash risk for teens (Insurance Institute for Highway Safety [IIHS], 2009). Destination was dichotomized. The three most frequented destinations (home, work, and school) were aggregated and compared with a second group including the remaining 12 destinations measured by the survey (attend childcare, eat out, personal business, everyday shopping, major shopping, religious/community, social, recreation participate, recreation watch, accompany another person, pick-up/drop-off passenger, and turn around). This dichotomization resulted in fewer, more exclusive trip destination categories while
195
maintaining in the analyses data on low frequency trip types by incorporating them with other similar trips. 2.3. Plan of analysis The analysis was conducted in two stages. First, univariate descriptive statistics were calculated for each age group. Second, driving exposure was estimated by demographic and trip characteristics, and age groups. Analyses were conducted using SAS 9.0 PROC SURVEYREG in order to accommodate the complex sampling procedure in which households were stratified within clusters (SAS Institute Inc., 2004). 3. Results 3.1. Descriptive results Approximately half the respondents were male and just over half the sample came from rural residences (Table 1). Among 16-17-year-olds, all participants reported their highest educational level as high school. Among 18-24-year-olds, 44.2% reported having an education level of high school or less, while among 35-64-year-olds, 35.7% reported an educational level of high school or less. The proportion of respondents working for income varied widely by age group: less than half the 1617-year-olds (47%), the majority of 18-24-year-old (81%), and over twothirds of the 35-64-year-olds (70%) reported part-time or full-time employment. Eighty one percent of 16-17-year-old drivers reported an annual household income above $50,000. Approximately three-quarters (74%) of 18-24-year-olds and two-thirds (68%) of 35-64-year-olds reported an annual household income above $50,000. Within the 48-hour survey period, drivers aged 16–17 years drove an average of 29.5 miles (SD: 27.0), 82.7 minutes (SD: 59.2), and 6.5 trips (SD: 3.8). Drivers aged 18–24 years drove an average of 55.8 miles (SD: 60.2), 118.8 minutes (SD: 78.9), and 6.9 trips (SD: 3.8); while 35-64-year-old drivers drove an average of 55.0 miles (SD: 56.8), 125.0 minutes (SD: 82.6), and 7.7 trips (SD: 4.3). The average driving speed for 16-17-year-olds was 25.8 miles per hour (mph). Drivers aged 18-24-year-olds and 35-64-year-olds drove an average of 37.3 mph and 34.8 mph, respectively. Driving exposure patterns were distinct across demographic groups. On average, 16-17-year-olds drove just over half the distance and approximately two-thirds of the duration of 18-24-year-old drivers, but drove a similar number of trips. The driving exposure pattern of 18-24-year-olds more closely resembled 35-64-year-olds than 16-17-year-olds. Men aged 35–64 years drove significantly more minutes than women the same age. Employment and high vehicle access were associated with greater driving exposure for 16-17-yearolds and 18-24-year-olds. Employment, high household income, large household size, and low vehicle access were associated with greater driving exposure for 35-64-year-olds (Table 2).
Table 1 Sample characteristics⁎. Age Group
Male High School Graduate or less Working for income Rural residence Household with income above $50,000
Miles driven Minutes driven Trips driven ⁎ Driving exposure estimates are for the 48-hour survey period.
16–17 (N = 583)
18–24 (N = 1,250)
35–64 (N = 5,000)
%
%
%
49.1 100.0
48.8 44.2
48.7 35.7
47.0 55.2 81.6
81.0 51.1 74.2
69.5 53.1 67.7
Mean (95% CI) 29.5 (24.3 – 34.8) 82.7 (72.6 – 92.8) 6.5 (5.9 – 7.0)
Mean (95% CI) 55.8 (49.4 – 62.2) 118.8 (107.4 – 130.3) 6.9 (6.4 – 7.3)
Mean (95% CI) 55.0 (53.0 - 57.0) 125.0 (121.5 – 128.5) 7.7 (7.5 – 7.9)
196
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
Table 2 Driving exposure estimates and interactions by demographic characteristics⁎. Interaction Groups
Mean Miles (95% CI)
Sex 16-17 x 16-17 x 18-24 x 18-24 x 35-64 x 35-64 x
30.6 28.6 48.8 62.5 52.9 57.0
(23.3 (20.8 (41.4 (52.4 (46.4 (54.1
-
Employment 16-17 x Unemployed 16-17 x Employed 18-24 x Unemployed 18-24 x Employed 35-64 x Unemployed 35-64 x Employed
25.7 33.7 45.3 58.3 54.2 55.4
(19.4 (26.8 (36.5 (50.7 (47.4 (53.0
Household Income 16-17 x b $50000 16-17 x $50000+ 18-24 x b $50000 18-24 x $50000+ 35-64 x b $50000 35-64 x $50000+
28.0 29.9 54.0 56.5 53.1 55.9
Residence 16-17 x Urban 16-17 x Rural 18-24 x Urban 18-24 x Rural 35-64 x Urban 35-64 x Rural
Table 3 Driving exposure estimates and interactions by trip characteristics⁎.
Mean Minutes (95% CI)
Mean Trips (95% CI)
37.9) 36.3) 56.2) 72.6) 59.4) 59.9)
86.4 79.1 116.9 120.7 134.0 116.5
(73.5 - 99.3) (66.7 - 91.5) (103.8 - 130.0) (107.9 - 133.5) (124.8 - 143.2) (112.6 - 120.5)
6.5 (5.7 6.5 (5.5 6.5 (5.8 7.2 (6.5 7.4 (6.8 8.0 (7.7
-
7.3) 7.5) 7.1) 7.9) 7.9) 8.3)
-
32.1) 40.7) 54.1) 65.9) 60.9) 57.9)
70.2 96.6 87.9 126.0 109.5 131.8
(58.2 - 82.3) (83.3 - 109.8) (73.0 - 102.9) (113.0 - 139.0) (99.2 - 119.8) (127.5 - 136.2)
5.6 (4.9 7.5 (6.7 5.7 (5.0 7.1 (6.6 7.4 (6.8 7.8 (7.7
-
6.3) 8.2) 6.5) 7.6) 7.9) 8.0)
(19.6 (23.4 (44.9 (48.3 (46.7 (53.4
-
36.4) 36.4) 63.0) 64.6) 59.5) 58.5)
81.5 82.9 105.9 123.3 113.1 130.7
(64.0 - 98.9) (71.7 - 94.2) (92.7 - 119.1) (110.1 - 136.6) (103.9 - 122.4) (126.5 - 134.8)
5.9 (4.9 6.6 (5.9 6.5 (5.8 7.0 (6.5 7.3 (6.8 7.9 (7.7
-
7.0) 7.2) 7.1) 7.5) 7.8) 8.1)
30.7 28.6 55.9 55.7 55.2 54.8
(24.1 (21.6 (48.2 (46.3 (49.3 (52.3
-
37.4) 35.6) 63.6) 65.2) 61.1) 57.3)
93.4 74.0 127.7 110.4 136.1 115.2
(81.2 - 105.5) (62.8 - 85.3) (116.0 - 139.4) (96.2 - 124.5) (126.8 - 145.5) (111.4 - 119.0)
6.8 (6.1 6.2 (5.4 6.9 (6.3 6.9 (6.3 7.9 (7.3 7.6 (7.3
-
7.5) 7.0) 7.4) 7.4) 8.4) 7.8)
Household Size 16-17 x 3 or less 16-17 x 4 or more 18-24 x 3 or less 18-24 x 4 or more 35-64 x 3 or less 35-64 x 4 or more
26.9 29.7 62.5 54.8 51.6 58.4
(15.8 (22.9 (48.9 (46.6 (45.0 (55.5
-
38.0) 36.4) 76.1) 62.9) 58.2) 61.3)
67.4 83.3 116.6 119.2 120.5 129.5
(41.1 - 93.7) (71.7 - 95.0) (101.3 - 131.9) (105.3 - 133.1) (107.8 - 133.3) (124.7 - 134.2)
6.8 (5.1 6.4 (5.8 7.3 (6.5 6.8 (6.2 7.2 (6.7 8.2 (8.0
-
8.5) 7.1) 8.2) 7.3) 7.7) 8.4)
Vehicle Access 16-17 x Low 16-17 x Medium 16-17 x High 18-24 x Low 18-24 x Medium 18-24 x High 35-64 x Low 35-64 x Medium 35-64 x High
19.2 29.9 30.9 55.3 55.1 56.3 67.8 59.0 51.3
(10.0 (23.1 (24.4 (39.1 (46.6 (48.2 (58.8 (52.5 (49.1
-
28.3) 36.8) 37.4) 71.5) 63.5) 64.3) 76.8) 65.4) 53.6)
60.9 80.0 89.0 88.6 111.9 125.5 127.2 125.8 124.3
(34.5 - 87.3) (65.3 - 94.8) (75.9 - 102.1) (65.9 - 111.3) (96.9 - 126.8) (111.6 - 139.5) (111.6 - 142.9) (113.1 - 138.4) (119.5 - 129.1)
4.3 (2.8 6.4 (5.7 6.8 (6.2 5.8 (4.8 7.0 (6.4 6.9 (6.5 9.1 (8.3 8.2 (7.7 7.3 (7.1
-
5.9) 7.2) 7.4) 6.7) 7.5) 7.4) 9.9) 8.7) 7.4)
Male Female Male Female Male Female
⁎ Driving exposure estimates are for the 48-hour survey period.
Across all ages, significantly more driving was done alone than with passengers present. Drivers aged 35–64 years drove more with non-household passengers. All respondents reported driving significantly more during the day than at night. Those aged 16-17-years drove significantly more to home, work, or school than other destinations while the two older age groups had more destination diversity than the youngest drivers (Table 3). The three measures of driving exposure showed two consistent patterns in their associations with demographic and driving behavior variables. First, there was a positive association between the number of miles driven and the minutes driven, meaning that if the miles driven were higher for a particular demographic group (e.g., those who were employed), then the minutes driven would also be higher for this group relative to those who were not employed. This pattern was replicated for all trip characteristics, with a single exception: although 18-24-year-olds drove significantly more minutes and trips to home, school, and work, they drove significantly more miles to other destinations (the category that includes social and recreational
Interaction Groups
Mean Miles (95% CI)
Mean Minutes (95% CI)
Mean Trips (95% CI)
Number of Passengers 16-17 x 0 16-17 x 1 16-17 x 2 or more 18-24 x 0 18-24 x 1 18-24 x 2 or more 35-64 x 0 35-64 x 1 35-64 x 2 or more
24.9 10.5 12.1 44.0 31.3 32.2 42.4 19.2 31.1
(19.7 - 30.0) 67.0 (58.8 - 75.3) (5.6 - 15.4) 31.9 (23.1 - 40.6) (7.6 - 16.6) 36.9 (28.8 - 45.0) (38.0 - 49.9) 104.4 (92.8 - 116.1) (23.7 - 38.9) 43.9 (34.2 - 53.6) (26.0 - 38.4) 47.6 (38.1 - 57.0) (37.7 - 47.2) 100.4 (91.9 - 108.8) (13.9 - 24.5) 47.5 (39.4 - 55.7) (29.0 - 33.2) 61.4 (58.2 - 64.5)
5.4 (4.8 2.4 (1.9 2.7 (2.3 5.9 (5.4 2.6 (2.1 3.1 (2.6 6.0 (5.5 2.6 (2.1 4.3 (4.1
- 5.9) - 2.9) - 3.2) - 6.4) - 3.0) - 3.6) - 6.5) - 3.0) - 4.5)
Passenger Type 16-17 x Household 16-17 x Non household 18-24 x Household 18-24 x Non household 35-64 x Household 35-64 x Non household
11.4 11.6 35.5 29.9 20.6 30.2
(6.7 - 16.2) (7.2 - 15.9) (28.3 - 42.7) (23.9- 35.8) (16.0 - 25.2) (28.2- 32.2)
2.7 (2.2 2.6 (2.1 2.9 (2.4 3.1 (2.6 2.8 (2.4 4.1 (4.0
- 3.1) - 3.0) - 3.3) - 3.5) - 3.3) - 4.3)
Time of Day 16-17 x Daytime 16-17 x Nighttime 18-24 x Daytime 18-24 x Nighttime 35-64 x Daytime 35-64 x Nighttime
27.5 6.9 50.5 12.0 52.8 10.5
(24.2 - 30.8) 76.3 (69.3 - 83.3) (4.6 - 9.2) 20.9 (16.5 - 25.3) (46.3 - 54.6) 104.3 (96.2 - 112.4) (9.3 - 14.6) 32.1 (27.6 - 36.5) (50.0 - 55.6) 118.8 (114.0 - 123.7) (9.7 - 11.3) 29.2 (27.6 - 30.8)
6.0 (5.6 1.5 (1.3 6.0 (5.8 1.9 (1.7 7.4 (7.1 1.6 (1.5
- 6.4) - 1.6) - 6.2) - 2.0) - 7.6) - 1.6)
Destination Activity 16-17 x Home, Work, School 16-17 x Other 18-24 x Home, Work, School 18-24 x Other 35-64 x Home, Work, School 35-64 x Other
36.6 33.8 49.1 45.3 51.5 57.5
(27.7 (25.9 (39.4 (36.1 (44.2 (54.4
-
45.5) 41.8) 58.8) 54.5) 58.8) 60.6)
21.3 (16.8 - 25.8)
60.6 (53.1 - 68.1)
4.7 (4.2 - 5.1)
12.5 (8.6 - 16.4) 22.7 (18.7 - 26.8)
33.7 (27.5 - 39.9) 84.4 (75.2 - 93.5)
2.7 (2.3 - 3.1) 4.5 (4.1 - 4.9)
44.4 (39.0 - 49.7) 29.7 (25.7 - 33.8)
48.3 (41.8 - 54.7) 76.0 (68.1 - 84.0)
3.3 (2.9 - 3.7) 4.2 (3.8 - 4.6)
30.5 (28.8 - 32.2)
59.5 (57.2 - 61.8)
4.2 (4.1 - 4.4)
⁎ Driving exposure estimates are for the 48-hour survey period.
destinations). Second, in contrast to miles and minutes driven, the number of trips provided less variability as a measure of driving exposure. As a result, the number of trips did not consistently correspond to the same patterns observed in the number of minutes or miles driven. 4. Discussion This study compared two young driver groups that have high motor-vehicle crash risk with a low-crash-risk group of adult drivers. The results indicated that driving exposure patterns varied considerably among the three age groups, suggesting that driving exposure is directly related to stage of life. The entire sample of 16-17-year-old respondents were in high school, which directly influenced their driving times, destinations, and purposes. Those aged 18–24 years displayed driving behavior patterns that were closer to those of older drivers, while retaining some differences. The oldest drivers were likely to be shouldering the greatest household responsibilities, and their greater driving exposure relative to the younger age groups might reflect this reality. Michigan's graduated driver licensing law imposes distinct driving restrictions (e.g., supervised driving, night driving restrictions) for each stage of licensure. Therefore, stage of driver licensure might contribute to or account for differences seen in exposure between the youngest and the two older groups. Within the youngest age group there are likely to be learners, restricted license holders, and full license holders and as a result, driving exposure estimates for this age group are likely to be influenced by the proportion of 16-17-year-olds within each stage of
J.P. Ehsani et al. / Journal of Safety Research 42 (2011) 193–197
licensure and the corresponding driving restrictions imposed on each stage. While the absence of licensure stage data limits the interpretation of 16-17-year-olds’ driving exposure, the mix of licensure status is very likely representative of the age group. Future studies should consider examining driving exposure by licensure stage and age. In addition to the differences in stage of life, the makeup of the drivers within each age group also differed. The youngest drivers came from more affluent households than the older drivers, making the group itself, and not merely their driving exposure, different from the older cohorts. This finding also suggests that less affluent teens are less likely to drive. Like the older age groups, 16-17-year-olds who were employed reported driving more than those who were not employed. This has clear implications for those who are at higher risk of crashing; specifically, employment for 1617-year-olds, while a positive factor in normal adolescent development may also contribute to higher crash and injury risk. There are several limitations to the generalizability of these findings. Eligibility for participation in the Michigan Travel Counts survey was based on household characteristics rather than those of individual drivers. As a result, the driving exposure estimates provided by this sample may not be representative of all drivers in Michigan. In addition, the responses for the Michigan Travel Counts survey were individual self-reports of distance or duration of travel, which have been criticized for underestimating or overestimating the true durations or distances travelled (Energy Information Administration, 2000; Staplin, Gish & Joyce, 2008; Wolf, Oliveira & Thompson, 2003). However, based on the correlations among the three exposure measures, the findings of this study provide a reasonable representation of driving exposure. Miles driven correlated with self-reported minutes driven at r = 0.30, and with self-reported trips at r = 0.55. The lower correlation between miles and minutes is expected, due to the variation in speed that would change the association between miles driven and minutes driven. Overall, these correlations provide support that these driving exposure estimates have validity in terms of their rank-order associations, though the validity of the point estimates may be less. The Michigan Travel Counts survey restricted sampling to travel occurring between Monday and Thursday during the school year (i.e., while classes are in session), thus not including travel during the summer months and on weekends, two high-risk periods for young drivers. Fatality Analysis Reporting System (FARS) data for Michigan over a 20 year period (1990–2009), however, indicated that, unlike the national data, the winter months also correspond to a period of high crash risk for 16-17-year-olds (National Highway Traffic Safety Administration [NHTSA], 2010a). The absence of weekend and summer driving exposure data in this study make it impossible to determine whether the periods included in the survey represent a biased subset of travel (i.e., weekend and summer travel not only differ in the amount, but also in character of exposure). In addition, the nature of driving might differ on weekdays and weekends for different age groups. To determine whether weekday and weekend travel differed similarly for the three age groups, data from the 2001 National Household Travel Survey (NHTS) were analyzed (U.S. Department of Transportation: Bureau of Transportation Statistics, 2010). Results showed that vehicle miles travelled on Monday, Tuesday, Wednesday, and Thursday totaled 60%, 60%, and 59% of all driving for 16–20 year-olds, 21–25 year olds, and 36–65 year-olds, respectively. Assuming the NHTS 2001 data are representative for Michigan, and travel patterns did not significantly change in the time between the two surveys, it can be concluded that the weekly pattern of driving exposure did not differ by age group. References Altman, D. G., & Royston, P. (2006). The cost of dichotomising continuous variables. BMJ, 332(7549), 1080.
197
American Association for Public Opinion Research (2008). Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys (5th edition). Lenexa, Kansas: AAPOR. Ehsani, J. P., Shope, J. T., Bingham, C. R., Sunbury, T. M., & Kweon, B. (2010). Teen driving exposure in Michigan: demographic and behavioral characteristics. Accident; Analysis and Prevention, 42, 1386–1391. Energy Information Administration (2000). Odometer Versus Self-Reported Estimates of Vehicle Miles Traveled. Washington, DC: Author. Farmer, C., Kirley, B. B., & McCartt, A. T. (2010). Effects of in-vehicle monitoring on the driving behavior of teenagers. Journal of Safety Research. Hailu, A., & VanEenwyk, J. (2009). Guidelines for Using Rural-Urban Classification Systems for Public Health Assessment. Retrieved from http://www.doh.wa.gov/ data/guidelines/RuralUrban1.htm Hart, L. G., Larson, E. H., & Lishner, D. M. (2005). Rural Definitions for Health Policy and Research. American Journal of Public Health, 95(7), 1149–1155. Insurance Institute for Highway Safety [IIHS] (2009). Fatality Facts 2008: Teenagers. Retrieved 12 October, 2009, from. http://www.iihs.org/research/fatality_facts_2008/ teenagers.html Michigan Department of Transportation (2009). MI Travel Counts. Retrieved 30th April, 2009, from http://www.michigan.gov/mdot/0,1607,7-151-9615_51690—,00.html National Highway Traffic Safety Administration [NHTSA] (2005). Traffic Safety Facts 2005: A compilation of motor vehicle crash data from the Fatality Analysis Reporting System and the General Estimates System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. National Highway Traffic Safety Administration [NHTSA] (2008). Youth Traffic Safety Statistics. Retrieved 2nd April, 2009, from http://www.nhtsa.gov/portal/site/nhtsa/ template.MAXIMIZE/menuitem.cd18639c9dadbabbbf30811060008a0c/?javax.portlet.tpst=4427b997caacf504a8bdba101891ef9a_ws_MX&javax.portlet. p r p_ 44 2 7b 99 7 ca a cf 50 4 a8 bd b a1 0 18 91 e f9 a_ v i ew ID =d e ta i l _v i e w&i te mID=d91cd60670268110VgnVCM1000002fd17898RCRDviewType= standard#stats National Highway Traffic Safety Administration [NHTSA] (2010a). Fatality Analysis Reporting System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. National Highway Traffic Safety Administration [NHTSA] (2010b). Traffic Safety Facts 2009 Ealry Edition: A compilation of motor vehicle crash data from the Fatality Analysis Reporting System and the General Estimates System (No. DOT HS 809 631). Washington, D.C: U.S. Department of Transportation. Neale, V. L., Dingus, T. A., Klauer, S. G., Sudweeks, J., & Goodman, M. (2005). An overview of the 100-car naturalistic study and findings. : Paper presented at the International Technical Conference on Enhanced Safety of Vehicles. SAS Institute Inc. (2004). SAS/STAT® 9.1 User's Guide. Cary, NC: SAS Institute Inc. Staplin, L., Gish, K. W., & Joyce, J. (2008). Low mileage bias' and related policy implications–A cautionary note. Accident; Analysis and Prevention, 40(3), 1249–1252. State of Michigan (2009). Trends in Household Income. Retrieved 15th May, 2009, from http://michigan.gov/hal/0,1607,7-160-17451_18668_41233-175057–,00.html Stutts, J., Feaganes, J., Reinfurt, D., Rodgman, E., Hamlett, C., Gish, K., et al. (2005). Driver's exposure to distractions in their natural driving environment. Accident; Analysis and Prevention, 37(6), 1093–1101. Transportation Research Board (2008). TRBe-Circular Mid-Year Report. Retrieved from http://www.youngdriversafety.org/docs U.S. Census Bureau (2009). Michigan - Selected Social Characteristics in the United States: 2007. http://factfinder.census.gov/servlet/ADPTable?_bm=y&-geo_id= 04000US26&-qr_name=ACS_2007_1YR_G00_DP2&-context=adp&ds_name=&-tree_id=307&-_lang=en&-redoLog=false&-format= Retrieved 30th April, 2009, from U.S. Department of Transportation: Bureau of Transportation Statistics (2006). National Household Travel Survey Washington, D.C. . U.S. Department of Transportation: Bureau of Transportation Statistics (2010). 2001 Online Table Designer. National Household Travel Survey. Retrieved 23rd December, 2010, from. http://nhts.ornl.gov/tables09/ae/TableDesigner.aspx Williams, A. F. (2003). Teenage drivers: patterns of risk. Journal of Safety Research, 34, 5–15. Wolf, J., Oliveira, M., & Thompson, M. (2003). The Impact of Trip Underreporting on VMT and Travel Time Estimates: Preliminary Findings from the California Statewide Household Travel Survey GPS Study. Transportation Research Record, 1854, 189–198. WWAMI Rural Health Research Center (2006). RUCA Data Code Definitions. Retrieved 21st December, 2010, from http://depts.washington.edu/uwruca/ruca-codes.php Johnathon Ehsani is a doctoral candidate at the University of Michigan School of Public Health with a research interest in transportation and injury. Dr. Ray Bingham is a Research Professor at the University of Michigan Transportation Research Institute, in the Department of Psychiatry of the University of Michigan Medical School, and in the Department of Health Behavior and Health Education of the University of Michigan School of Public Health. Dr. Jean T. Shope is a Research Professor at the University of Michigan Transportation Research Institute where she served as head of the Social and Behavioral Analysis Division from 1995 to 2005. She is also a Research Professor in the Department of Health Behavior and Health Education at the University of Michigan School of Public Health.