The relevance of crash type and severity when estimating crash risk using the SHRP2 naturalistic driving data

November 2015

Crash risk factors may vary by crash severity. Of 1,465 crashes in the second Strategic Highway Research Program (SHRP2) naturalistic driving dataset, 248 were severe (coded “most severe” or “police-reportable”); 597 were “minor;” and 620 were “low-risk tire strikes.” Nearly half (48%) of severe crashes were rear-end, compared with 12% of minor crashes and none of low-risk tire strikes. Most (64%) minor crashes and 99% of low-risk tire strikes were road departures versus 13% of severe crashes. Crash risk associated with five secondary behaviours relative to when no secondary behaviours were present varied by crash severity, especially when including low-risk tire strikes. For example, cellphone conversation was not associated with a significant change in the odds of any crash (OR=1.21; 95%CI[0.89,1.65]) or a low-risk tire strike (OR=0.57; 95%CI[0.32,1.02]), but was associated with a significantly higher odds of a minor crash (OR=2.00; 95%CI[1.31,3.06]) and any crash except low-risk tire strikes (OR=1.97; 95%CI[1.37,2.83]). Researchers should consider crash type and severity when analysing SHRP2 crash data and exclude tire strikes when generalizing to crash risk. Note that the odds ratios assessed risk relative to driving with no secondary behaviours whereas typical driving frequently includes various secondary behaviours.