> TFA mentions the low sample size of the original study, and augments it with Twitter polls. But TFA author's Twitter follows aren't exactly a representative sample of the population...

I agree, and even if their followers were, all it takes is a single retweet by a popular user, or post in a high-traffic subreddit or Facebook share to wildly skew the results.

Offering the Twitter poll as any sort of evidence at all without major disclaimers makes me quite skeptic of the author's qualifications, to be honest...