SHRP2 Naturalistic Engagement of Secondary Tasks (NEST) Distracted Driving Dataset: Baseline Sampling Method and Implications for Crash Risk Approximations

• Activity Number: 655 - Statistical Analysis of Naturalistic Driving Study Data: How You Slice and Dice Matters
• Type: Topic Contributed
• Date/Time: Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Transportation Statistics Interest Group
• Author(s): Birsen Donmez*
• Companies: University of Toronto
• Keywords: naturalistic driving study ; SHRP2 ; case-control design ; NEST dataset

Abstract

The NEST dataset was created from SHRP2 raw data for the purpose of investigating secondary task engagement while driving; it was also originally planned to help researchers to evaluate in detail how distraction affects crash/near crash risks. NEST provides useful and detailed information for distraction-related safety critical events (SCEs: crashes, near-crashes) and associated baseline epochs. While NEST creation concluded in October 2014, SHRP2 data reduction was completed in early 2015. Thus, NEST neither consists of all SHRP2 distraction-related SCEs, nor was drawn from the complete set of SHRP2 events and baselines. Further, there was no known order (e.g., by time, through a statistically random order) to the data reduction process of SHRP2. Given that NEST data was extracted before SHRP2 data was ingested entirely, there is no way of knowing what type of a sub-sample NEST represents. Both SHRP2 and NEST utilized a case-control design. While SHRP2 controls were sampled strategically to approximate crash risk through exposure odds ratios, biased estimates should be expected from NEST. Future users of NEST should be made aware of this limitation.