Our post yesterday mentioned in passing that the Virginia Department of Transportation’s 2007 study of red light cameras found that the total number of accidents dipped slightly in Fairfax City after red light cameras were installed. Every other city in Virginia reported a large increase in accidents and injuries at the intersections with cameras. A closer examination of the VDOT report’s fine print shows that this conclusion is far from clear and that Fairfax City’s results do not support the need for red light cameras.
It turns out that at two of the seven intersections where red light cameras were used in Fairfax City, the “before installation” data were incomplete. To compensate, VDOT used statistical methods to, in effect, guess what that those missing data might be. It used two different sets of assumptions on how to fill in the missing numbers. Fair enough. But what happens when you look at the raw before & after data for just the intersections with complete datasets? (page 74)
Fairfax City Camera Crashes Per Intersection Year per Million Entering Vehicles
|Route 29/50 and Route 123||398.7||466.0||+16.9%|
|Route 123 and North Street||632.2||399.3||-36.8%|
|Route 50 and Jermantown||196.1||277.0||+41.3%|
|Route 236 and Pickett||60.8||346.9||+470.6%|
Doesn’t look so good, does it? Well, it turns out the control intersections in Fairfax City — the ones free from Big Brother cameras — saw their own increases in accidents. To sort out what this means, VDOT used three different equations to figure out whether the cameras provided a benefit or not. These are all commonly accepted statistical methods. Page 33 shows a summary of the effects based on which of these methods you select:
Total Crashes in Fairfax City
- Paired t-test: nonsignificant increase
- General Linear Models: nonsignificant increase
- Empirical Bayes: nonsignificant decrease
Two methods tell you accidents went up a tiny bit, one popular method says they went down a tiny bit. In other words, what we’re looking at is essentially a statistical draw.
So tell me again why Fairfax City wants to install these cameras?