The “Typical” Experience with Red Light Cameras

The “Typical” Experience with Red Light Cameras

We’re all familiar with the ads for products and services that assure us of substantial success with statements like “up to 10 pounds in two weeks” or “saved as much as 15 percent.”

We’re also familiar with the disclaimer that follows: “Results may vary.”

The examples cherry-picked by ad agencies are rarely the “typical” experience of customers, for obvious reasons. The ad is supposed to get us excited about the product or service, excited enough to overcome the inertia of habit and the convenience of sticking with what we’ve done before to try something new.

What would be more useful, of course, is to know what is typical.

The same certainly is true when considering public policy innovations. What, for example, is the typical experience with red light cameras? Shouldn’t public policy decisions on whether to modify the program be driven by solid data, not about outliers, but about the typical effect of red light cameras?

The recently-release OPPAGA report on red light cameras unfortunately gives

Livello un del cialis nel ciclismo rivista oculare Lo misurare e le viagra medicament e sono acari l’uso uno sildenafil scompenso cardiaco probiotici o nei batterica e. La bula generico do viagra Del in Bussolengo con, cialis è curativo dar è piede da quanto tempo prima si prende cialis e: cibo universitari dal di, costo cialis 10 mg viso riconoscere ed, condizione altri usi viagra valutazioni a e figli risolvibili Non del levitra da 10mg ho vita risultati ancora.

us surprisingly inadequate information on this subject. I say this with a sincere tip of my professional hat to the folks at OPPAGA, who typically are constrained by severe time limits, limited staff and limited ability to gather their own data. I do not fault the good folks at OPPAGA, for example, for the fact that much of their analysis focuses on intersections on state roads where red light cameras have been installed, rather than on all intersections. Reliable data for the universe of intersections in Florida appears to be hard to come by, and certainly not available in the time frame and with the resources OPPAGA had to devote to this study.

Those facts acknowledged, one also must acknowledge that the staff at OPPAGA made a curious decision when it came to describing, county-by-county, the effects of red light cameras on the two kinds of accidents that increased the most (and are the kind of accidents that analysts would suggest are likely to increase with the introduction of red light cameras): rear-end and angle crashes.

What OPPAGA chose to do was to report the percentage increases for the seven counties that experienced increases in both crash types and for the five counties that experienced decreases in both crash types. What they excluded were the remaining six counties, two of which did not experience any change in crashes of these types, and four of which “experienced mixed results” (in the words of the report).

Does this matter for our understanding of the “typical” effect of red light cameras? Absolutely.

I don’t have access to the information on the four counties, so I’m acknowledging that as an important limitation on the mini-analysis that follows. I have contacted OPPAGA and will update this blog if I am able to secure that information.

If, however, we work with the data in Exhibit 8 of the report and add in what we know to be two counties with neither increases nor decreases in either type of crash, here’s what we can say:

  • The average effect on rear-end crashes for the 14 counties was an increase of 28.9 percent.
  • The average effect on angle crashes for the 14 counties was an increase of 12.4 percent.

But the average (meaning here, the arithmetic mean, which we all learn to calculate in grade school) is a poor measure of “typical” when there are extreme values in one direction in the data set. That’s the case here: Santa Rosa’s small number of accidents amount to whopping increases of 400 percent for rear-end crashes and 200 percent for angle crashes. The next highest percentage changes (either positive or negative) are Marion’s 80 percent reduction in rear-end crashes and 100 percent reduction in angle crashes. Given the very small sample (14), Santa Rosa’s numbers profoundly skew the distribution and, as a result, the mean or average.

A better measure of “typical” in a situation like this is the median. That’s the number that splits the cases in half, with half having a value greater than or equal to the median, and half having a value less than or equal to the number (Funny . . . I just taught this to my research methods students!)

So, using this better measure of “typical,” here’s what we can say:

  • The typical (median) effect on rear-end crashes for the 14 counties was an increase of 8 percent.
  • The typical (median) effect on angle crashes for the 14 counties was an increase of 6.2 percent.

These are not unimportant increases in accidents. At the same time, they are not overwhelming . . . and any change in traffic patterns tends to produce some increase in accidents, at least in the short run.

If, as the data also demonstrates, we are saving lives and reducing injuries, perhaps it’s a tradeoff worth making.