Two professors who should know better perform the magic act of Explaining Randomness in this NYT analysis of undecided voters in the US Presidential election. They draw upon tracking polls conducted over several weeks, then draw out from each sample only the people who were undecided. The professors then analyze this group for clues about their characteristics and the implication of those characteristics for the election.
The exercise is magic, of course, because most of what they interpret as true differences are only random effects. With a sample this size, the error is going to be around 5 to 10% and the professors seize upon differences among undecideds that are within or just barely out of sampling error. They report no tests of statistical significance and certainly nothing about effect sizes, just taking the absolute mathematical difference to claim Something’s Going On. They do provide a bar chart!
Yet, they present their analysis in that scientific style with job titles and institutional affiliation you see in peer review literature, but without the obvious benefit of peer review that would have suggested the professors are reading tea leaves and might want to brush up on their stat and research methods. Thus, NYT readers are treated to the appearance of social science when it is only a magic trick that makes you think random is systematic when it is always and forever only random. Stated another way, the professors tell you how to spot loaded dice while viewing a random run. Hey, they’re professors of professing at those places that hire professors. Just read the end of the post.
I obviously disapprove of people playing professor on TV where they can dress the part, but don’t have to worry about getting scorched in peer review. Sure, the reader comments can be aggressive, but that’s just the give and take of the First Amendment, and even when people are calling you names, it’s still in the New York Times! That’s nothing like getting those anonymous reviews that pull your pants down and make you spank yourself for your failures, errors, and embarrassments. And, of course, none of that appears on your CV or any promotion and tenure committee; you do get all the Cool Table recognition with none of the negative consequences such poor performance would earn in your real job.
Here’s the solution. Just send this crap to the NYT without your job title and institutional affiliation. Exercise your First Amendment rights as a citizen and misread statistical data to suit your temperament and political bias. Huzzah. That’s why people died to create and defend this country. Not so you can play professor on TV, pushing your credential as a Cue to science rather than providing science.
The persuasion play reeks with a childish inSincerity. The professors wear the robes and engage the rituals like little kids playing Dress Up. What they show in this public game looks like science, but is only advertising.