Today I posted a new study on MTurk as a Requester. My coauthor Matt and I are working on a paper where we compare MTurk samples to other samples. We have comparisons with snowball samples and students samples, but not with a panel sample.
We had collected some panel data last summer so today I launched a revised (shortened) version of the study. I did an estimate of the time (three minutes, based on other surveys I had taken) and calculated an appropriate payment (.30) based on the 10 cent per minute estimate.
I launched it at 8:39 today, asked for 200 different respondents, and I got those by 9:00 am today. Everyone but one person got the attention check correct. The average completion time was 2 minutes 39 seconds (FIST PUMP) and the estimated hourly compensation rate is $6.71.
What I did differently this time: I wrote a much more descriptive introduction, I noted that there would be one attention check in the study (thanks Rochelle), and I offered a good incentive for a short survey. I think I can call this study a win/win.