I’m very interested in studies that examine the viability of MTurk studies for research– is it a convenience sample or is it representative of the US? Or both (if that is possible)?
I found this recent study that examined attitudes towards BPA (a chemical that acts as a hormone in the body). Many studies show that BPA can have harmful effects. But that’s another (and important) topic.
In a pair of studies, both a panel from Survey Sample International and a panel from MTurk were asked questions about BPA. The MTurk sample demographically was younger and slightly more educated. And then there’s this:
While our MTurk participants were more familiar with issues surrounding BPA than our Survey Sampling International participants, neither indicated a moderate to high familiarity with BP and the issues surrounding the substance. The higher level of familiarity seen in the second study isn’t surprising given the bias of MTurk toward engaged, active internet users. In this respect, the first study is more likely to be closer to representing US consumers.
This doesn’t have anything to do with samples, but there is a concern with these findings:
But it does worry me that, where perceived pubic opinion dictates risk-based decisions, there is little accountability if those decisions turn out to be ill-informed.