Extending with R code

JonPeirce

Hi there,

Loving the super-slick JASP interface. Nice work! It will be so much easier to teach our undergrads this than teaching them how to avoid the various foibles of SPSS.

The thing that concerns me is whether we'll always be tied to the particular tests that have been implemented. I don't want to find my analysis went almost to the end and then I (or a manuscript reviewer, or colleague teaching stats) needed some further (obscure) test that would never reasonably be included in the main set of analyses that JASP delivers. The issue is then finding I have to go back and redo everything in another package.

So I just wonder if it's possible to allow extensions of the gui in code, or even export the R syntax for the analyses that are being done, so the user can tweak/extend and rerun with additional R code. PsychoPy, for instance, allows you to export the Python script that the GUI represents so you can read/hack/learn Python. Although PsychoPy's graphical interface won't handle anything you can get at the raw code it compiles so if you want something non-standard you aren't stuck. Similarly, SPSS had the option to do fancier stuff using the syntax and you could "paste" the analysis to help you generate those scripts. Is it too late for such a thing to be engineered into JASP? Or did I miss it and it's there?

thanks again,
Jon

JonPeirce

Ah, I just saw on the FAQ that this is essentially a "no", or was when the FAQ was written anyway. Shame.

sebastiaan

In case the JASP developers are reading this: It would be interesting to know a bit more about how JASP interacts with R.

Are analyses executed in a single-shot way, so that an R instance is launched to run the analysis and immediately destroyed afterwards? That would make interacting with R from within JASP difficult.

Or is an R instance running in the background the whole time? In that case, it might be possible to allow the user to interact with R from within JASP, for example by embedding an R terminal in the same way that OpenSesame and the PsychoPy coder embed a Python terminal. I think a lot of users would like that.

EJ

The R code issue has been on our agenda from the start. Unfortunately, our programming team at the time resisted my frequent urging to implement this, but it is still on the agenda. In terms of the percentage of users that actually want this, I think it is actually rather low. In general, R programmers (and I'm one of them) tend to overestimate R's appeal to students and pragmatic researchers. But, like I said, it is definitely on the agenda, and with the recent 4 million euro in acquired grant support we can do this and a lot of other things for sure. I will run this past the programmers too.
Cheers,
E.J.

sebastiaan

with the recent 4 million euro in acquired grant support we can do this and a lot of other things for sure.

Yes, wow, a VICI and an ERC—congrats!