Link: The frequentist Bigfoot

1 minute read

Carl Zimmer connects Bigfoot with an explanation of the history of “null hypothesis” in a Nautilus essay: “Why we can’t rule out Bigfoot”. He discusses a recent paper by Bryan Sykes’ lab, which systematically tested purported Yeti and Bigfoot samples.

Some skeptics have offered up an alternative explanation for Sykes’ finding. It’s possible that the polar bear-like DNA actually comes from a living mammal—perhaps a brown bear—that happened to pick up a few mutations that created a false resemblance to that ancient polar bear DNA.
What these skeptics have done, in effect, is create a null hypothesis. And there’s a straightforward way to set about disproving it. Scientists would need to find more DNA from these mysterious bears. If other regions of the DNA also matched ancient polar bears, then scientists could reject the null hypothesis.

I think this would be a good essay for undergraduate courses introducing science. Zimmer gives a good account of how R. A. Fisher conceived of frequentist statistical testing and the role of the null hypothesis in it.

Applied to Bigfoot, I think we should be very skeptical about the choice of a null hypothesis. That’s in part why I think it would be a good essay for a class discussion. I don’t think that Sykes approached his study with the null hypothesis that the samples would all represent other mammals. I think he tested the hypothesis that each sample could represent an unknown primate, and refuted it in each case by showing some other animal (from raccoons to bears) is a close genetic match to the sample.