pantelispa
commented
3 years ago This all sounds reasonable!
Before running an experiment with 2 sessions I would run another pilot with a single session. The analysis will be much easier now, without the back and forth between tasks, and it will be provide us a platform where we can get all our programming routines ready for the 2-session experiment.
What additional hypothesis can we test with the 2 session experiment? It would be great if you can write a few things about it!
I'm currently thinking of 2 sessions, 60 games each, and the bonus pay would be 10cents, with 0.1 or 0.2 cent as safe pay. This is analogous to our previous setup, just proportionally smaller. This would make the it possible to win 6euros (or whatever currency we will be using) if all goals are reached (not the maximum possible, but still an overly optimistic estimation, so I'd assume winnings per session would be closer to 4 - 4.5).
Not sure how long a single session will take, but certainly less than an hour. We could probably get away with setting the fixed pay to be smaller ( 2 / 2.5).
If we aim to treat this as Experiment 1 of a paper, I would suggest getting at least 30 participants (so 60 sessions). If you prefer running another pilot, we could get away with less.
Let me know your thoughts!