psychopy.org | Reference | Downloads | Github

Control experiment

How do people normally run a control experiment versus the actual experiement? Do you create two experiments and then run either one or the other or do you combine them into one?

My specific situation is that I have used the export functionality on the builder to export into a psychojs experiment which, barring a few bugs in the JS, is online now. So I wondering whether I should have two cloud instances - one for the control experiment and one for the actual experiment or whether I can combine them somehow - and use 1 url. Perhaps it’s cleaner just to have two machines and two urls.

Any thoughts welcome.

It’s simply a matter of how the experiment is needed. The options are:

  • control conditions interleaved trials with the others (avoiding any order effects)
  • in some cases you want to create a “block” design where you have a block of trials of one type and then a block of another and you want the order of the blocks to be randomised or counterbalanced
  • if the control experiment and the “main” experiment are going to take too long to run in one go then, yes, I’d naturally create two different experiment files for it

If you do need to run two different experiments though they could sit right next to each other on the server - you don’t need different server instances. Either have them in different folders to keep the data separate or even in the same folder (make sure that the experiment name in the Experiment Settings differs so that you can distinguish which data files came from which version).