SHARE `AHA’ STATISTICS ARE FREQUENTLY MISLEADING

A statistics teacher used to play a little game with his class. He would give them some simple experiments to do while he was out of the room. They were supposed to write down the numbers that came out of these experiments. They were also told to write down another set of numbers, made up out of their heads.

When the instructor returned to the classroom, he was supposed to guess which set of numbers was real and which set was phony. Although the students tried their best to fool the teacher, he would usually just glance at the two sets of numbers and immediately tell them which set was real.The students were baffled as to how he could tell. After a while, he finally let them in on the secret: The numbers they made up were too even, too orderly. Real statistics from the real world are seldom like that.

Real statistics are not even or orderly. They are irregular. They jump around or are lopsided.

It is a shame that this statistician didn't teach this lesson to the deep thinkers and the morally anointed in the media, in academia, in Congress and in the courts. Whenever they see statistics that are uneven or don't match their expectations, they conclude immediately that there is something wrong with the world - and they are off on another crusade.

Whenever people have a strong suspicion that something is so, any uneven numbers that are consistent with that belief are likely to be seized on and treated as heavy evidence. This is the "Aha!" school of statistics.

Back in the early days of World War II, for example, someone discovered that Japanese Americans were living concentrated near all sorts of military installations, railroads, power stations and other strategically important installations vulnerable to sabotage. Aha!

The conclusion was that Japanese Americans were positioning themselves to sabotage the American war effort against Japan. This belief played a role in the decision to round up Japanese Americans and ship them off to prison camps.

Only much later did cooler heads prevail and the facts come out. Japanese Americans, mostly farmers then, were living where they were before the strategic installations were built in these rural areas. Cheap land attracted these installations, as it had attracted the Japanese Americans.

For decades, American antitrust laws have been attacking "monopoly" on the basis of statistics. Whenever different human beings do the same thing, some do it better than others and lopsided statistics result.

The same formula works in many employment discrimination cases, where you can't find a single woman or minority member who has personally been discriminated against by the particular employer - but where the statistics don't please the judge.

In recent years, courts have started backing away from heavy reliance on statistics by themselves as showing employment discrimination. This has set off hysteria in Congress, leading to a bill sponsored by Sen. Ted Kennedy and Rep. Gus Hawkins, which will make "Aha!" statistics the law of the land.