IIHS Report Rebuts Study that Questioned Red Light Cameras
A six-page report posted Nov. 6 by the Insurance Institute for Highway Safety calls into question a 2007 study that said using red light cameras at intersections in six northern Virginia communities increased rear-end crashes. The 2007 study, which was commissioned by the Virginia Transportation Research Council and received significant attention when published, did not validly assess whether camera enforcement reduced intersection crashes or not, authors of the report concluded. IIHS supported the report, which is dated May 2008 on its cover page.
The essence of the report is that the VTRC researchers used insufficient analytic methods to isolate the effect of camera enforcement on crashes, and the best method (named the Empirical Bayes Method) was used incorrectly. The selection of comparison sites and use of those sites to calibrate the crash prediction models were not appropriate, and the VTRC study used an unusual, illogical crash prediction model, the IIHS report states.
"All of these problems likely produced unstable and unreliable parameter estimates for many of the models, such that estimates of expected crashes without camera enforcement, a key ingredient of the EB methodology, likely were incorrect," it adds.