30-11-2012, 10:33 PM
(This post was last modified: 30-11-2012, 10:48 PM by Greg Burnham.)
The Abuse of Statistics: A Brief, But Humorous Study
On 3 occasions back in the 1990's I decided that it would be exhilarating to attempt stand-up comedy. While it was exhilarating, to be sure, it was also terrifying.
My attempt was made on three occasions, but one occasion stands out. I was observing that it is not very wise to quote statistics to teenagers regarding the
dangers of drinking and driving because, unless you are brutally honest with yourself about the statistics, you will not persuade them to refrain from the practice.
Indeed, one must be very careful with how they interpret statistics and must resist the temptation to interpret them in a sloppy manner or with ascertainment bias
run amok.
As you might imagine this did not get any laughs. Then I said, "Each holiday season we hear on the news that there were, for example, 1109 deaths on the highway
in California and that a full 20% of them were alcohol related. Then some spokesperson for Mothers Against Drunk Drivers basically wags her finger in the face of
the camera. But, statistically this means that 20% were alcohol related and 80% were SOBRIETY related! And, if we believe the statistics, it is indicated that those who
are under the influence are so much more relaxed than those who are sober, that they rarely get more than minor injuries, while the sober drivers often get killed.
Moreover, as for collisions with other drivers, there are only statistics relating to sober drivers colliding with drunk drivers, as well as sober drivers colliding with each
other. There are so few drunks colliding with other drunks as to make it statistically insignificant. So, what's the moral of the story using this MADD statistical model:
Never drive unless you've had at least 2 drinks otherwise you are at risk because the odds of you having a collision goes from 1 in 5 when drunk to 4 in 5 when sober
and if you're sober when in an accident you probably will get hurt or killed!"
It got a big laugh, but there is a point to the story and it's not to go ahead and drink and drive. The point is to resist the urge to manipulate and/or misinterpret statistics
even if for all the right reasons.
I went on with my analogy: "I've read statements by prosecutors and journalists who report that they asked heroin addicted inmates being held at the county jail if
heroin was their first illegal drug or if they started first with marijuana. In more than 85% of the cases the answer was answered affirmatively. The authors of these
studies therefore concluded that marijuana usage leads to heroin usage. Moral of the story: Don't smoke pot because you will end up on heroin. Oh, and by the way,
don't breast feed because nearly 94% of the heroin addicted inmates had been breast fed. Don't breath oxygen either because 100% of the heroin addicted inmates
started off on oxygen..."
On 3 occasions back in the 1990's I decided that it would be exhilarating to attempt stand-up comedy. While it was exhilarating, to be sure, it was also terrifying.
My attempt was made on three occasions, but one occasion stands out. I was observing that it is not very wise to quote statistics to teenagers regarding the
dangers of drinking and driving because, unless you are brutally honest with yourself about the statistics, you will not persuade them to refrain from the practice.
Indeed, one must be very careful with how they interpret statistics and must resist the temptation to interpret them in a sloppy manner or with ascertainment bias
run amok.
As you might imagine this did not get any laughs. Then I said, "Each holiday season we hear on the news that there were, for example, 1109 deaths on the highway
in California and that a full 20% of them were alcohol related. Then some spokesperson for Mothers Against Drunk Drivers basically wags her finger in the face of
the camera. But, statistically this means that 20% were alcohol related and 80% were SOBRIETY related! And, if we believe the statistics, it is indicated that those who
are under the influence are so much more relaxed than those who are sober, that they rarely get more than minor injuries, while the sober drivers often get killed.
Moreover, as for collisions with other drivers, there are only statistics relating to sober drivers colliding with drunk drivers, as well as sober drivers colliding with each
other. There are so few drunks colliding with other drunks as to make it statistically insignificant. So, what's the moral of the story using this MADD statistical model:
Never drive unless you've had at least 2 drinks otherwise you are at risk because the odds of you having a collision goes from 1 in 5 when drunk to 4 in 5 when sober
and if you're sober when in an accident you probably will get hurt or killed!"
It got a big laugh, but there is a point to the story and it's not to go ahead and drink and drive. The point is to resist the urge to manipulate and/or misinterpret statistics
even if for all the right reasons.
I went on with my analogy: "I've read statements by prosecutors and journalists who report that they asked heroin addicted inmates being held at the county jail if
heroin was their first illegal drug or if they started first with marijuana. In more than 85% of the cases the answer was answered affirmatively. The authors of these
studies therefore concluded that marijuana usage leads to heroin usage. Moral of the story: Don't smoke pot because you will end up on heroin. Oh, and by the way,
don't breast feed because nearly 94% of the heroin addicted inmates had been breast fed. Don't breath oxygen either because 100% of the heroin addicted inmates
started off on oxygen..."
GO_SECURE
monk
"It is difficult to abolish prejudice in those bereft of ideas. The more hatred is superficial, the more it runs deep."
James Hepburn -- Farewell America (1968)
monk
"It is difficult to abolish prejudice in those bereft of ideas. The more hatred is superficial, the more it runs deep."
James Hepburn -- Farewell America (1968)