Using a psiturk experiment outside of Mturk (Prolific academic)

[Luke Strickland]

Hi,

I'm looking at some existing code for an experiment run through psiturk, which I'm hoping to modify for a future experiment with the option of recruiting on prolific academic.

I wanted to check whether this is still a sensible thing to do - are there any major drawbacks to using psiturk outside of Mturk? Would love if somebody could point me to any resources that would help. 

I found an old thread below on recruiting externally for psiturk experiments, which seems somewhat promising

https://groups.google.com/g/psiturk/c/WW-PKoL6z6A/m/oA98NMS7CQAJ

[Dave Eargle]

I've successfully used psiturk for non-aws, there's no problem with that approach. You just have to change the /complete.html template to your liking, and you have to set up a landing page that will generate arbitrary workerid, hitid, assignmentid for your given participant. Then you pass those to exp.html, and as long as they're the right format / length, experiment.py won't complain.

Prolific specifically has come up a few times. I've never used it, but I have heard that it's totally do-able. See, for instance, https://github.com/NYUCCL/psiTurk/issues/312#issuecomment-378578815 and this fork https://github.com/NYUCCL/psiTurk/compare/master...kahana-sysadmin:master

[Luke Strickland]

Thanks Dave, that's really encouraging, I'll get started and look into those threads!