Poll: What experiment to run on AWS mechanical turk?
11–14 of 14 posts
Re: Poll: What experiment to run on AWS mechanical turk?
#12Re: Poll: What experiment to run on AWS mechanical turk?
#13How about creating an experiment and designing it right on turk? You create a rough sketch and then you auto-correct it using people?
If you're not careful, you could get http://en.wikipedia.org/wiki/Publication_bias on a massive new scale...
It would be interesting to try to find scientifically optimal ways of doing study iteration, or approaches to interpreting the complete (experiment_rev, results)-list from another researcher...
Re: Poll: What experiment to run on AWS mechanical turk?
#14I'd love to see a prisoner's dilemma experiment via MTurk. These tend to be financial in nature, though; can you set variable payment rates per task, depending on external factors?
so you could easily do single-round prisoner's dilemma, and have amazon pair up 2 (different) folks at a time, and when you get the results back, look to see what the folks in each pair did, and award bonuses appropriately.
i'm not sure if there is a way to do an iterated version.