"[W]hile right angle crash incidents have been reduced, rear-end crashes that resulted in injuries went up 22 percent." Chicago officials recently claimed that the cameras led to a 47% reduction "T-bone" injury crashes, using that statistic as evidence that the program is worthwhile. But the study's authors, who "accounted for declining accident rates in recent years as well as other confounding factors, found cameras reduced right-angle crashes that caused injuries by just 15 percent."
So the article says rear-end went up 22% and T-bone went down 47%. You have to be suspicious whenever you see a news article that says x went down by y%.
per cent of what? What were the base numbers?
Here's some example situations to show why I say that.
suppose before red light camera we had 100 rear-end crashes and 10,000 t-bone crashes at the intersection (all with injuries)
suppose after red light, we have 122 rear-end crashes and 5,300 t-bone crashes. That's 22% rear-end up and 47% t-bone down
But, the total number of injuries dropped 4,678. That's good isn't it? Redlight cameras must be great!
Or, suppose this:
before red-light camera, 10,000 rear-end and 100 t-bone w/injury
after red-light camera: 12,200 rear-end and 53 t-bone w/injury again, 22% increase in rear-end and 46% decrease in t-bone.
so we had an increase of 2,153 injuries total. Oh my, red-light cameras are killers, aren't they?
I used a wide disparity in the numbers to make my point: you cannot make a useful comparison between percent changes in numbers of two different measurements without knowing the base numbers. That is covered in your freshman "Lying with Statistics 101" class.
So, I read the article in the Tribune (it's free if you give them your email address and live out-of-zone)
If you read the Tribune article (and the accompanied "How the Red Light Camera Study was Done" you may come away with a quite different view than the slashdot summary or the ArsTechnica summary. The Tribune article is not as ridiculous as the slashdot summary.
The article does indeed have some raw numbers:
Quoted from the Tribune:
"In raw numbers at the 90 intersections included in the study, the researchers concluded the cameras prevented as many as 76 right-angle crashes and caused about 54 more rear-end injury crashes. The study said that without the red light cameras about 501 angle crashes would have occurred and only 425 were reported. It also said that there were 296 rear-end injury crashes, and there would have been only 242 had the cameras never been installed."
I've been driving for a few decades and have seen many serious injuries and fatalities, but not a single serious injury or corpse in a rear-end crash.
If you give me a choice between trading 76 t-bones crashes for 54 rear-end crashes, I'd take those numbers. As many other posters have observed, t-bone crashes are much more likely to result in serious injuries and deaths than rear-enders.
The two Tribune articles also covers some of the crookedness associated with Chicago's use of the cameras. They are both a good read and covers a lot of why you should be careful about these numbers and problems associated with the data.