Statisticians can prove almost anything, a new study finds

National Post:

Catchy headlines about the latest counter-intuitive discovery in human psychology have a special place in journalism, offering a quirky distraction from the horrors of war and crime, the tedium of politics and the drudgery of economics.

But even as readers smirk over the latest gee whizzery about human nature, it is generally assumed that behind the headlines, in the peer-reviewed pages of academia, most scientists are engaged in sober analysis of rigorously gathered data, and that this leads them reliably to the truth.

Not so, says a new report in the journal Psychological Science, which claims to show “how unacceptably easy it is to accumulate (and report) statistically significant evidence for a false hypothesis.”

In “False-Positive Psychology: Undisclosed Flexibility in Data Collection and Analysis Allows Presenting Anything as Significant,” two scientists from the Wharton School of Business at the University of Pennsylvania, and a colleague from Berkeley, argue that modern academic psychologists have so much flexibility with numbers that they can literally prove anything.

Read the whole story: National Post