Hacker News Reader

Also worth pointing out that this was a sample of only Mechanical Turk users in America, and members of the Danish Military in Denmark.

View on HN

Draftees in Denmark, which includes all men, plus women who volunteer. So it's skewed, but not quite as skewed as if you used only self selected people.

I am told by people who do this kind of research that using Mechanical Turk isn't as pointless as it would sound. I have to take them at their word in that.

What do they say about Mechanical Turk? Is it that it's more representative of non-American point of view?

In this case they filtered for Americans. They put some effort into validating identity, and into ensuring that you are taking the survey seriously (trap questions, for example).

It doesn't strike me as sufficient but it's not my field.

I can say (but not visibly prove, unfortunately, NDAs) that large mechanical Turk surveys tend to get distinctly different results to mail and phone call surveys. Enough so that the commissioners of the study were not interested in replacing existing surveys with mechanical Turk.

Reply on news.ycombinator.com