creating lists
Hi All,
I was wondering if there was a way to make lists in JATOS, and then assign participants randomly to one of the lists.
To clarify: I have an experiment with three components and I am counterbalancing the items across participants, so each participant sees only half of the items (because they are similar and because the task would be very long if not).
Should I put different versions of the same task in the same experiment and then have JATOS assign the participants (based on the number of completed experiments), or should I simply create the lists myself, as two separate experiments on JATOS and send the links accordingly?
Thank you!
Comments
Hi,
It depends a bit on what your other plans are.
Having two independent studies is the simplest. But if means that if you will change parameters of the task down the line (timing, number of trials, whatever) you'll have to do it twice.
Have you seen the examples study for how to randomize workers? https://www.jatos.org/Example-Studies. If that doesn't sound too complicated, it's the more sustainable solution. BUT sometimes you don't need sustainability, so there's no point making your life more complicated.
Hope this helps, let us know if you need any more help
Elisa
Thank you.
I will have 3 lists- same version of the tasks containing different items. The full experiment has 3 tasks. If it would be able to put all three versions in a JATOS study and have JATOS assign the participant to a list depending on the number of finished trials in each list, prioritising the list with the lowest number.
If not, i will have to create 3 JATOS studies and send the links accordingly. I can control how many links to each list I send, but I cannot control how many will be finished. So I will have to constantly reassess the list that has to be sent to a new participant. So if there would be a way to automatise that, it would be great.
Marta
Hi, sorry for the delay.
The thing is, in the same way that you can't control how many people will actually finish the task, JATOS also has (so far) no way to automatically assign people according to how many studies have been finished. Part of the reason for this is that a study that is 'finished' might still be data that is not up to your standards (performance too low, responses to fast, etc). So we thought that it would be too difficult/impossible to have an automatic solution for this.
I think Kristian might nevertheless implement some kind of basic check in the future, at least for finished studies. But for now there is none. What you can do is automatically assign links based on how many people started each version of the study. If that's good enough, that's what those examples I sent you would accomplish.
Best
Elisa
Hi Elisa! Thank you, I will look into the examples.
having an automatised way for assigning people to start the task, and then adjusting according tot he obtained results might be a good solution.
Best
Marta