JASP — “a fresh way to do statistics — has been around for a while now, but this is really a project I am watching. Even though it explicitly does not refer to just another statistic programme, that’s certainly a useful mnemonic until Google ranks the page higher. JASP comes with a clean interface that will feel familiar to SPSS users, but actually improves on SPSS on many fronts to make it easier to use. That’s a nice touch.
JASP uses a journal system like we know it from IPyhton and Jupyter with live preview. The live preview is great, as users can immediately see what consequences their choices have. Unfortunately, this only really works for relatively small datasets, experiments or a simple population survey with a thousand respondents or so. Better than other similar solutions, the code is not visible to the user, which leads to nice outputs. At the same time, we can go back to the analysis and modify the output at any time. That’s slightly easier than finding the corresponding code in say Rmarkdown and recompiling.
As a nice touch, there is integration with OSF and SocArxiv. This means if I wrote a paper based on analysis carried out in JASP, I could upload this alongside the paper, and anyone can see the output file and modify it — online.
JASP uses R to do the calculations, which gives it a bright future in the kind of things it can offer. Unfortunately, at the moment, what’s on offer remains limited. This means for my purposes, JASP is not (yet) a replacement for SPSS, just like PSPP. The way JASP implements what it offers, however, is the reason why we should watch this project: it is easy to use, and it lets users choose Bayesian analysis for everything.
While JASP uses R underneath, there does not appear to be a plugin system (other than getting involved in developing for the project). I think that user-provided extensions might be just what is needed to make this project take off, because generally speaking, this is the kind of program I can see myself use in teaching (alongside Rstudio) and for simple analyses. For more advanced analyses, R remains the go-to application.