On Sensational Statistical Results

Opening Post by the Rabbi

On Sensational Statistical Results

Posted on 6/3/2014

| |

On Sensational Statistical Results

From time to time, the press publishes a sensational statistical result. Many of them have two characteristics: 1. they are surprising (counterintuitive). 2. they support a fashionable principle (politically correct, that is, convenient and useful for advancing some agenda). Here are a few examples (with fictitious numbers, since I have not checked at the moment, though such findings and similar ones have indeed been published), starting with a result like: one in three people is homosexual, one in four beats his wife. Every woman has been raped at least one and a half times in her lifetime. Usually, when these two characteristics are present, one should be suspicious of the result, and of the limited statistical ability of the journalist who publishes it (the problem is not always with the researchers themselves, but sometimes that certainly happens too). Yesterday I heard Yael Dan on the midday news program reporting with absolute confidence (the kind reserved for “scientific” results grounded in statistics) that if you want to reach your destination safely, let a woman drive, because she drives more safely. A quick search brought me to the Globes website:

23% of all drivers involved in road accidents between 2003 and 2010 were women. Among the drivers involved in fatal accidents, women constitute only 10%, while the percentage of women among all licensed drivers stands at about 42%. This is reported by the Or Yarok association, based on an analysis of data from the Central Bureau of Statistics. Shmuel Abuav, CEO of Or Yarok: “Women are characterized by more cautious driving than men in all age groups. This is reflected in their significantly lower involvement in severe and fatal road accidents.” According to Abuav, “Women are capable of exerting a positive influence on male drivers and are willing to voice opposition to unsafe driving. This is reflected in the fact that they drive less aggressively than men do; they view dangerous driving as unacceptable and undesirable behavior, they drive more slowly, overtake less often, and drive drunk less often.” There is nothing like data from the Central Bureau of Statistics. But I assume (I have not checked) that the Central Bureau of Statistics did not determine that it is indeed safer to travel with women. I suspect that Abuav and Yael Dan determined that. And I wondered whether anyone had weighted the question of how much mileage is driven per woman per unit of time. How many women drive demanding vehicles that involve many driving hours and long distances (trucks, large and heavy vehicles). How much of the intercity driving is done by women. And so on and so forth. I believe (without checking) that the results would look completely different with that weighting. When I said this yesterday to my female students, the question arose: what do the insurance companies do? Are premiums for women higher or lower, since they certainly examine this in the most reliable way. Opinions were divided, and again I did not check the actual facts. But I suspect this would not say anything, since insurance companies do not care what the figure is per kilometer per unit of time. From their perspective, a car driven by a woman means fewer accidents, and that is what matters. But regarding the question of how to reach the destination more safely, what matters is precisely the weighted result. And we have not yet spoken about the ability to reverse (I am only waiting for the next study that will refute that fact). Or about courtesy versus stubbornness on the road (to my impression, men are much more courteous, perhaps because they have more confidence and feel less as though they are in a war. And no, this has no necessary connection to aggressive driving, in which too, in my impression, men excel as well).

I am merely pointing to this amusing phenomenon, which rests on a combination of fashion and sensation. Readers are invited to offer additional examples. They are not hard to find. I hardly know any cases reported in the media that do not suffer from biases of this kind. Above I brought a few examples from memory (with made-up numbers, because I did not check).

Source (forum ‘Stop Here, Think’): http://www.bhol.co.il/forums/topic.asp?topic_id=3042854&forum_id=1364