DADRL remarks on NHTSA's preliminary study

Here are some points that we make against the NHTSA preliminary report.

• NHTSA's preliminary assessment shows no improvement for DRL equipped vehicles in preventing fatalities in two vehicle crashes. Repeat, no improvement. This is from the fairly comprehensive FARS database. While not statistically significant, the Odds Ratio technique shows a net decrease in effectiveness for two vehicle fatal collisions.
• NHTSA's assessment for non-fatal two vehicle collisions uses the SDS database for 4 out of 17 states. This does show an improvement for DRL equipped vehicles, but only when analyzed by using the Simple Odds analysis technique. No statistically significant benefit was found when using the Odds Ratio technique.
• NHTSA did not account for vehicle wear and technology advances (ABS, traction control, etc.) when doing 'before and after' comparisons of the SDS database. Perhaps this is why newer vehicles were less likely to be in a collision than an older one, whereas when comparing DRL and non-DRL vehicles of the same age (e.g., GM vs. Ford), a benefit was found in only one case using Simple Odds.
• Why didn't NHTSA report data from the other 13 states providing data to the SDS database? Was the data not favorable to DRLs?
• NHTSA's assessment admits to difficulty in defining the collision types for the SDS data.
• Both techniques have been sharply criticized by Stephen Prower, Research Officer for the British Motorcycle Federation. Further analysis of European studies finds less than favorable results for DRL proponents. Instead of relying on total vehicle miles for the comparison, why didn't NHTSA simply compare the ratios of collisions at daytime vs. those at night?
• NHTSA did not analyze data from collisions involving light trucks. Why? Was the data not favorable to DRLs?
• NHTSA did not analyze multi-vehicle collisions. Why?
• NHTSA's pedestrian fatality analysis appears to be flawed. In each individual case, no statistically significant benefit was found. Using a statistician's slight of hand, they combine the results and come to the conclusion that there is a benefit.
• The NHTSA assessment failed to examine DRL effectiveness in low-light conditions (dawn, dusk, poor weather). We agree that having lights on during these times will reduce collisions.
• To our knowledge, NHTSA's assessment has not been externally reviewed. Given the bias of the assessment exhibited in its introductory section, one must question the true motive of the study.

THE BOTTOM LINE: Granted, NHTSA's report, as its title implies, is preliminary. However, there's a clear indication of bias toward a desired result, as evidenced by the attack on the HLDI data. We have filed FOIA requests to NHTSA to find the true motivation for this study. The FARS data, perhaps the more reliable of between the two sets, failed to show a benefit of DRLs. The SDS data analysis was far from complete, failing to clearly define a two vehicle collision and analyzing only 4 out of 17 states. In both cases, the analysis techniques seem antiquated. This report as a whole appears to be very incomplete and has not been reviewed by academics and experts in the field of statistical crash analysis. Take it with a grain of road salt.

[ About the Association ] [ The DRL Story ] [ News ] [ Studies ] [ FAQ's ] [ Disable ] [ Join DADRL ] [ Related  Links ]