Sharing study assets folder across experiments?
I'm going to have a lot of variants of experiments that will end up using the same study assets. I've got images, movies, etc. that take up space and I've got HTML / JS files that are going to be used across experiments. So, both to save space and to ensure that bug fixes propagate across experiments, I'd really like a single folder inside of study_assets that gets shared across the experiments.
But, when I try to create a new experiment and have it point to an existing study_assets folder, it pops up an error saying (rightfully) that it can't create said folder (as it already exists) and won't let me continue. Cloning the study makes an ORIG_clone asset folder.
Is this baked into JATOS or is there a way to share assets across studies?
Thanks,
Craig
Comments
Hi Craig,
This is a feature, not a bug. It’s there partly to keep the JATOS user accounts properly separated.
In your case, it sounds like you should create many batches of the same study. You can then use each of the batch properties to specify values that can be read from your script to create the different variants.
let me know if you’d like more details
If it is about huge asset files like audio, video or images that you want to share between your slightly different studies: JATOS doesn't force you to use it's study assets for file delivery. You can use any other file hosting service in the internet (even Google Drive or Dropbox might work for you). Of course the HTML file has to come from JATOS but all other files can come from anywhere in the internet. On the downside is that it won't work off-line anymore and you have an additional point of failure (what if the file hoster does not work?).
Yea, I was wondering if that was more a feature than a bug... It also has the benefit of letting you lock an experiment's code in place and ensure that each run is consistent.
But, I'm not sure that batches will really work as it'll get needlessly complex. The idea is that we're running a lot of experiments and have certain tasks that'll be in common across experiments (each experiment has a battery of tasks). The batch mode works nicely for having a fixed set of tasks and varying parameters in them, order of them, etc. But, it gets awkward and we'll lose out on some of the nice JATOS features if I try to make one giant batch experiment that, in truth, runs different experiments.
Anyway, I'll have to balance the trade-off between having copies of thousands of files across experiments and having these resources in a common location outside of JATOS. But, I know at least know to stop trying to find a way to make it work within the JATOS zeitgeist. :)
Thanks!