Oct
Continuing with our “Please Comment” campaign, this month’s topic is: “A Passionate Plea to Read the Data Fairly.”
Nate Cohn of the New York Times has produced a very insightful article entitled “After a Tough 2016, Many Pollsters Haven’t Changed Anything.”
Indeed. They haven’t.
The article points out that most pollsters are still inventing excuses about why their predictions on Brexit and the 2016 US presidential election were so far off, and advances the idea (from pollsters themselves) that “weighting for education” might have resolved the problem.
“Weighting for education?” That’s akin to saying “let’s throw a number into the statistics that accounts for the village idiots in the sample.”
Please.
A novel idea: read and report on the data as it actually comes in
What’s wrong with our industry? Why do we insist on retaining our biases as we read and report on data, then “weight” for the biases we have?
How about the idea of leaving our biases at the curb and reporting on the data simply and naturally, as it has come in? If that had been done already, we wouldn’t need articles from Nate Cohn. He could write about something else.
Apparently the human instinct to bend reality (as represented by the statistics) to our preexisting view of the world is too strong. Thus, there aren’t any top executives at YouGov or Ipsos saying “News Flash! We’ll start reading the data fairly, as it really came in.”
What’s wrong with us? How shocking would it be if a major executive of a major polling firm actually said that? Would anybody have the will and the courage to do that?
Re-education required
This would require a re-education of staffers and decision makers at polling and survey research companies on a scale equivalent to our current social efforts to combat discrimination in hiring and housing. We’d have to commit ourselves to a program whose goal would be: persuade decision makers to think of consumers and voters not as village idiots but as people.
At present we don’t have the fortitude to do that. Stasis is too comfortable.
Your view?
What your opinion about this? Please tell us by adding a comment below. Feel free to use all the bias you want.