Naturalistic Driving Study: Descriptive Comparison of the Study Sample with National Data (2015)

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.

12 The guiding principle of the original SHRP 2 NDS sampling plan was to recruit an equal number of male and female licensed drivers in each of several carefully selected age ranges. These age ranges were designed to sample across the full breadth of the driver age spectrum, oversampling the youngest and oldest seg- ments of the study population. This oversampling was done in light of the elevated crash risk expected for these youngest and eldest of the driving age groups (Stutts et al. 2009) in support of one of the major goals of the study: to observe and learn how best to prevent crash-related events. The original sample design is presented in Table 3.1. A secondary intention of the sample design was to collect data from 350 vehicles featuring any of several advanced vehicle technologies (AVTs), without respect to participant age or gender for those AVT vehicles. Such tech- nologies included crash warning/avoidance technology, inte- grated commu nication and infotainment technology, or brake assist applications. Participant Recruitment and Screening As reflected in Table 3.1, the initial study design called for drivers of vehicles eligible for inclusion in the study (accord- ing to criteria to be elucidated later in this chapter) to enroll for either one or two years. The enrollment period was later made more flexible, allowing participants the opportunity to extend existing enrollment periods beyond their originally planned dates of exit and allowing researchers to enroll new participants for shorter durations of time to maximize use of study equipment and data collection near the end of the study. In addition, as a result of participant attrition for a wide variety of reasons (e.g., moving out of the area, needing to sell the instrumented vehicle), some participants were enrolled for less time than would have been expected based on Table 3.1. Recruitment More than 18,000 individuals were recruited (i.e., were con- tacted using a variety of approaches and agreed to participate in the study), though only 3,100 were fully enrolled. The first recruitment approach, random cold calling to households, was initiated approximately one month in advance of the first anti- cipated installation in October 2010 and continued until early 2011. This approach was based on guidance from statisticians that a subject pool recruited in this quasi-random fashion would provide the least biased and most representative partici- pant sample. Unfortunately, success rates using this approach were disappointingly low (i.e., on the order of less than 2%), as few people contacted in this way were interested in partici- pating, the nature of a household sample made targeting older and younger drivers problematic, and the vast majority of those who were interested did not have a vehicle that met the initial set of eligibility criteria. In the earliest study stages, vehicle eligibility was based on access to data via the onboard vehicle network (details on this topic are expanded below). When it became obvious that a purely random cold calling approach would not be suitably efficient, operational effi- ciency and budgetary constraints dictated implementation of a more focused cold calling approach in which calls were made only to households believed to own an eligible vehicle (based on household data purchased from R. L. Polk & Com- pany and other similar organizations, as well as on customer lists received from some original equipment manufacturers). At the same time, site contractors were authorized to pursue recruitment via convenience methods, including using social media, posting ads in local newspapers or other media, dis- tributing flyers, and making personal appearances at a wide variety of venues. Development of a web-based screening tool allowed interested recruits a convenient and highly efficient means of expressing interest in the study and providing the necessary contact information. C h a P t e R 3 Participant Sample Design and Operations

13 Figure 3.1 illustrates the overwhelming success of con- venience methods as opposed to random or focused calling approaches, particularly among individuals aged 35 and younger. Although at first glance the 16- to 17-year-old sub- group appears to be an exception to this characterization, it is important to note that many of the teen drivers included in the study who identified as coming into the program through the call center did so as additional primary drivers of parents recruited via that method. Goal-Oriented Recruiting As the study progressed, recruitment priorities shifted from fulfillment of the original sample design of the number of participants in each age–gender cell toward a more goal- driven approach of number of vehicle-months in each age– gender cell. Midway through the study, targets for recruitment of drivers aged 26 to 65 were reduced, and targets for younger and older drivers were increased to accentuate even further the oversampling already built into the original study design. During the study’s final months, if no potential participants were available in undersubscribed age–gender cells, partici- pants in oversubscribed cells were enrolled so that available DAS units would collect additional data rather than sitting on a shelf. Figure 3.2 depicts the number of primary participants enrolled in the study for at least four months relative to the original sample design targets (as represented by the horizon- tal bar). Figure 3.2 illuminates the success of the study in reaching recruitment targets for most cells. Drivers aged 16 to 17 represent a notable exception, primarily due to parental consent issues and lower driver populations from which to draw relative to other cells. Concurrent with the shift to a more goal-oriented approach to recruitment was the cessation of targeted recruitment of drivers of AVT-equipped vehicles. Challenges in the recruiting process for drivers of these vehicles were similar to the obstacles confronted in attracting younger and older drivers. Specifically, the random cold calling approach initially implemented, by its nature, made it problematic to target drivers of such vehicles. The focused calling approach pursued later in the study failed Table 3.1. Original Sample Design Gender and Age Range Age Group Description Period of Enrollment No. of Participants DAS Units Vehicle-Years One Year Two Years M 16–17 Minor teen 72 28 172 100 200 M 18–20 Adult teen 72 28 172 100 200 M 21–25 Young adult 72 28 172 100 200 M 26–35 Adult 72 28 172 100 200 M 36–50 Middle adult 72 28 172 100 200 M 51–65 Mature adult 72 28 172 100 200 M 66–75 Younger older adult 72 28 172 100 200 M 76+ Older older adult 72 28 172 100 200 F 16–17 Minor teen 72 28 172 100 200 F 18–20 Adult teen 72 28 172 100 200 F 21–25 Young adult 72 28 172 100 200 F 26–35 Adult 72 28 172 100 200 F 36–50 Middle adult 72 28 172 100 200 F 51–65 Mature adult 72 28 172 100 200 F 66–75 Younger older adult 72 28 172 100 200 F 76+ Older older adult 72 28 172 100 200 Any AVT 0 350 350 350 700 Total 1,152 798 3,102 1,950 3,900 Note: M = male; F = female. Source: Dingus et al. 2014.

14 to produce the anticipated number of drivers of AVT vehicles, despite the procurement of household data from R. L. Polk & Co. and customer lists from original equipment manufactur- ers. In the final analysis, the operational goal of maximizing efficient use of DAS units to collect as much data as possible superseded the original secondary and ultimately unattainable goal of including a substantive AVT subset in the study cohort. A total of 135 vehicles identified as being equipped with AVT were included in the SHRP 2 NDS vehicle fleet. Figure 3.3 presents a breakdown of these vehicles by technology. The dis- crepancy between the total number of AVT vehicles in the fleet and the numbers presented in Figure 3.3 can be attributed to the presence of multiple AVTs on some vehicles. Participant and Vehicle Eligibility Screening In the early stages of the study, a potential participant was required to satisfy the criteria listed below. The first three were basic eligibility criteria, and the other two criteria were established in the interest of operational efficiency: 1. Is a licensed driver; 2. Is competent to grant informed consent (or for minors, informed assent with consent granted by a parent or guardian); 3. Has an eligible vehicle (the list of eligible vehicles grew as the study progressed); 4. Drives at least three days per week; and 5. Plans to keep the vehicle for the duration of antici- pated study participation (i.e., one or two years for most participants). Participants were initially required to own a vehicle for which the coordination contractor had acquired parameter IDs, which were needed to interpret the data generated by the vehicle’s onboard network (the third criterion above). Later Figure 3.1. Success of call center recruiting versus recruiting via convenience by age group (years). Figure 3.2. Primary participants enrolled in NDS for at least four months relative to original sample design goals. (Blue  male; green  female.) Total – 3,092 Primary Participants Source: Dingus et al. 2014.

15 in the study, vehicle eligibility criteria were expanded to per- mit a greater proportion of recruits to become actual study participants. In particular, the addition of older model years attracted more drivers in the younger age groups. The result- ing expansion of the vehicle fleet to include some newly man- ufactured model years, as well as older vehicle-years, for both of which a less robust set of vehicle network data was avail- able, produced four distinct vehicle classes in terms of the collection of vehicle network data. These vehicle classes are presented in descending order of data richness in Table 3.2. A set of guiding principles was used to shape ongoing recruit- ment strategy and installation priorities. These goals focused on allocating recruitment, inventory, and human resources in a manner aimed at increasing the number and time-in-study of participants in the 16 to 25 and 76+ age groups, with males being prioritized over females in each age group due to their Table 3.2. SHRP 2 NDS Vehicle Classes Class Vehicles Included Vehicle Network Information Collected Vehicle Count Percentage of Fleet Prime Vehicles for which parameter IDs were available and included on the original eligible vehicle list. Speed, plus wiper activation, brake actuation, headlight activation, turn signal activation, and steering data, as available 1,717 51% Subprime Generally, vehicles manufactured after 2009 for which parameter IDs were not available. Speed and accelerator position 488 14% Legacy Vehicles manufactured between 1996 and 2008. Speed and accelerator position 736 22% Basic Vehicles manufactured prior to 1996; no vehicle network. None 421 13% All 3,362 100% Source: Dingus et al. 2014. higher relative risk (NHTSA 2012). Further, site contractors were asked to prioritize, whenever possible, the installation of participants with vehicles in classes that maximized data richness. These principles continued to steer recruitment and installation until the final months of the installation period, when the goal of maximizing use of DAS equipment to collect study data became paramount. The transition from cold calling to a more goal-oriented approach, the cessation of targeted recruiting in the AVT sector of the study cohort, and the adjustment of eligibility criteria to produce a sample more inclusive of the primary populations of interest (i.e., younger and older drivers) are illustrative of the manner in which real-world considerations intersected with study design. Any assessment of the degree to which the resulting sample compares to the target population must take such necessary fine-tuning into account. Figure 3.3. AVT vehicles by technology.