Home » Status Report » 2002 » Article
Status Report, Vol. 37, No. 5 | SPECIAL ISSUE: AUTOMATED ENFORCEMENT | May 4, 2002 Subscribe

Results vary, but studies find red light cameras reduce injuries

Diahann Hill

Red light running results in an estimated 260,000 crashes each year in the United States, and 750 of the crashes are fatal. Cameras that detect red light running deter violations. The findings of numerous studies suggest that photo enforcement also reduces injury crashes at intersections. However, the results of these studies vary widely in terms of the extent of the crash reductions. An Australian study found an injury crash reduction of 7 percent, another Australian study found a reduction of 46 percent, and an Institute study found a 29 percent reduction.

The Institute has critically reviewed crash-based studies of red light camera programs, virtually all of which have reported reductions — some large, some relatively small — in injury crashes. The variability of the findings reflects methodological problems evident in many of the studies. Such problems may have led researchers to either over- or underestimate the crash effects of the cameras.

Most studies also reported increases in rear-end crashes. This isn't surprising. The more people stop on red, the more rear-end collisions there will be if motorists behind them are following too closely or not paying attention.

Despite this effect, the body of evidence indicates that red light cameras are beneficial. "Although rear-end crashes tend to go up, when you look at all crash types — in particular those involving injury — photo enforcement leads to significant overall reductions in crashes," says Susan Ferguson, the Institute's senior vice president for research.

Studies conducted in Australia, the United States, and Singapore met the criteria for the Institute's review. To be included, a study had to present crash data for intersections before and after red light cameras were installed. The studies had to include data from comparison sites without cameras.

During the review, Institute researchers found that many of the studies had methodological weaknesses, including failure to control for the statistical tendency known as regression to the mean. This happens when red light cameras are applied to the worst intersections with the highest crash frequencies, and then the intersections are compared with ones that haven't been treated. "If crashes are already unusually high at a given location, over time they're likely to decline or normalize with or without intervention," Ferguson explains. "Studies that didn't compare sites with comparably high crash rates and didn't adequately control for regression to the mean probably overestimated the crash reductions."

Another common bias was not adequately controlling for the spillover effects of cameras on other intersections. Photo enforcement tends to produce generalized changes in driver behavior, so violations and crashes decline throughout the area where cameras are used, not just at the specific intersections equipped with cameras. In fact, many jurisdictions structure their programs to promote a perception of community-wide camera enforcement. The studies that only looked at crashes within the same area and, therefore, didn't control for spillover probably understated crash reductions.

To account for these methodological weaknesses, Institute researchers segregated the studies into groups and compared findings among groups:

  • The first group of studies, all from Australia, didn't control for either regression to the mean or spillover effects. Combined results from this group indicate an overall 39 percent drop in injury crashes at camera sites. A study from Queensland, Australia, reported the largest injury crash reduction (46 percent) of all the studies reviewed. Two studies from Adelaide, South Australia, also reported injury crash reductions.
  • The second group of studies partially controlled for regression to the mean by using comparison sites similar to the camera sites. But these studies didn't control for spillover, which means they probably underestimated the crash reductions. Combined findings of this group indicate a statistically nonsignificant 10 percent reduction in injury crashes. Included among these are two studies from Australia and one from Singapore, site of one of the few red light camera evaluations conducted outside Australia.
  • The sole study conducted in the United States, in Oxnard, California, controlled for both regression to the mean and spillover effects. Researchers looked not only at crashes at Oxnard's camera-equipped sites but also at overall changes in crashes at intersections with signals throughout Oxnard compared with nonsignalized intersections. Comparisons then were made with three other California cities without cameras. The finding of this Institute study was a 29 percent reduction in injury crashes for all 125 signalized intersections in Oxnard (see "Red light cameras yield big reductions in crashes and injuries," April 28, 2001).

Studies in all three groups found increases in rear-end crashes at camera-equipped sites. "To the extent they encourage more stopping, red light cameras may exacerbate rear-end crashes," Ferguson says. She adds that "when you put the available research together, the evidence indicates that red light cameras reduce injury crashes by about 25 to 30 percent, and that's after accounting for some small increases in rear-end crashes." Ferguson concludes that "automated enforcement cameras work, and the international evidence is consistent when you consider the different methodological weaknesses of the various studies."

Injury crash reductions attributed to red light cameras
in Australia, Singapore, and the United States

Graph image

Note: The first 3 studies (Queensland and 2 in Adelaide, Australia) did not control for regression to the mean or spillover. The next 3 studies (Victoria and New South Wales, Australia, plus Singapore) controlled for regression to the mean but not spillover. The 7th study, conducted in the United States (Oxnard, California), controlled for both effects.

Reanalysis confirms Oxnard findings

Institute researchers took another look at data from an earlier red light camera study at the suggestion of a California lawmaker. The reanalysis confirmed the crash reductions.

©1996-2016, Insurance Institute for Highway Safety, Highway Loss Data Institute | www.iihs.org