Reader John Wund left this. Moment about the power of data to mislead those of blind faith:

“Once, after I had an MS in physics, I was hired by a medical school to work on building a ‘physical biochemistry lab’. I also have a degree in Astronomy, so I think I know something (though, perhaps not everything) about statistics.

“Any useful statistic needs to be accompanied by an ‘error estimate’ and needs to be measuring something clearly agreed upon by everyone (brightness, position, etc.). When psychology pretends to measure ‘intelligence’ every rule is broken, and it sends Gauss whirling in his grave.

“These new tests that pretend to measure ‘teaching’ or ‘skill’ or ‘learning’ are simply an extension of the attempt to (or pretense to) measure ‘intelligence’. There is no clear (non-circular) definition of the trait to be measured, or of it’s importance to a larger society. And, no indication of uncertainty (statistical, rather easy to calculate for those who ‘like numbers’) is ever attached.

“It is not statistics, but their bogus use that is the problem. Most people think that numbers and computers are ‘scientific’, and so they must be honored. I have a funny story about that.

“When I went to the Biochemistry Dept., I found that there was only only one other person who had even a passing acquaintance with computers (this was in the late 60’s). Even that person had little understanding of statistical analysis. At first, I took information, analyzed using a calculator it and wrote out the result for the various profs and post-docs. I found they often questioned the result, so I wrote simple programs (Basic or Fortran) and ran them on a computer hooked up to a teletype machine. Suddenly, the questions stopped. After all, if a machine typed it out, it must be factual! Never mind that I was the one programming the machine.

“People are attracted to the certainty that numbers seem to provide. Unfortunately, those numbers can be used to control people, even if they have no value. The ancient Greeks understood that inductive logic always had more value than deductive. We have forgotten that lesson.

“But, it goes even deeper. Because people venerate numbers (and certainty), they are easily manipulated by those who spout them in a blatantly (to me) inaccurate way. Sadly, numbers have become a way to buffalo people. Remember ‘Ivory Soap’, 99 and 44 one hundredths percent pure? Pure what? Pure bullshit.

“Statistics teaches that there is no certainty (well before Heisenberg). There is only a ‘best guess’ at any particular time, subject to change. Statistics revolving around a poorly defined concept are worthless. As a saying in my past explains, GIGO (garbage in, garbage out). Computers and statistics aren’t the problem, our blind belief in the stuff that comes out of machines is.”