I posted an entry earlier this year in which I said analysts and statisticians – basically, people in my field – are the new prophets. I continue stand by that. We are the new prognosticators, the new seers of the future.
And you should never, ever trust a prophet. It doesn’t matter whether they’re quoting the Bible or Bayesian statistics – don’t trust people who try to predict the future, no matter if they’re interpreting ancient riddles or writing complex algorithms.
We were the statisticians. We were wrong. And now we have to fix it, and the “how did we fuck this up so badly?” conversations have already started. In my classrooms, on the listservs I belong to, across my field this week, we’re asking ourselves…what the hell did we just do? How did this get away from us so badly?
How can we change, as a field, to make this sort of predictive analysis more accurate?
My answer has been simple – it doesn’t matter how many ancient texts you consult, or how many mathematical principles you employ, or how you change the models – when you’re measuring human behavior, there are always variables you cannot control for. Humans are unpredictable.
That’s where we went wrong. It wasn’t just that polls aren’t conducted well, or that people tend to lie in polls, or that the media grossly misrepresents statistical information. Those things all contributed, but at the base of everything, you simply cannot reduce human behavior to a set of algorithms, nor is it possible for any research to lay aside his or her own biases.
Never trust a prophet, no matter what form they appear to you in.