A specter is haunting statistics: the specter of JASP.
Example diagram made in JASP
Statistics is at the basis of social and behavioural sciences. All social science students have to go through the hardship of learning how to compute the obscure tests in statistics. For a long time, SPSS (Statistical Package for the Social Sciences) ruled undisturbed over most of the statistics courses. SPSS is statistical analysis software owned by IBM, an American multinational technology company. It is a tyrant ruling over students’ grades and finances, charging € 121.00 for a basic subscription (IBM, 2024). To be a social scientist, you had to bow and submit yourself to the almighty software. The poor students were not only required to learn statistics, which is a death sentence in itself, but they were also asked to pay an insane amount for it, without any hope of relief. Furthermore, after they graduated, they were not freed from their shackles: what they had learned in statistics thanks to their hard work could not be used unless they continued paying for a subscription to SPSS.
The issue with SPSS goes beyond the students’ pockets and extends to scientific practice. In fact, when the analyses are done with SPSS, they can only be accessed by someone who owns a subscription. This makes the statistical analyses less accessible and complicates the process towards a more open science, where every step of scientific research is open and transparent.
At our university, as many psychology students may be aware, a revolution has been igniting faith in the lost souls of statistics learners. A free software for statistical analysis has been developed: JASP, Jeffreys’s Amazing Statistics Program. The name was chosen in honour of Bayesian statistics pioneer Sir Harold Jeffreys. It is a free, accessible and intuitive software for both Bayesian and classical frequentist statistical analyses. JASP made a dent in SPSS’ dark empire, rebelling against a private and inaccessible way of doing statistics.
The following transcript reveals an interview with the one and only: Johnny van Doorn. To our greatest honour, Johnny, our beloved statistics professor here at UvA, and a raging fanatic of the JASP sovereignty/ascendancy/insurrection, has devoted his precious time to provide us with a better understanding of what is behind the statistically significant revolution that is JASP.
“It’s important that people have accessible software all over the world. Especially, [to avoid] teaching students statistics using something that they need to pay a lot of money for after they’re done studying, because that’s really conflicting with the ideas of university and science. ”
The Start Of The Revolution
How was the idea of JASP developed in its very first steps? Who were its pioneers?
It was a little bit before I arrived at UvA. The main pioneer is Eric-Jan Wagenmakers (EJ). He started out doing experimental psychology, some more cognitive psychology. But then… he saw the Bayesian light, and he became much more of a methodologist and statistician. He noticed that Bayesian methods were very hard to use: you had to really learn some very obscure programming language or syntax to run your models. Which can be very cool, but it’s not for everybody. By then, of course, SPSS had been around for quite a while. He figured, “Yeah, let’s make a similar thing, but for Bayesian statistics.” He hired a programmer, who became the second pioneer: Jonathon Love.
EJ was more about the design. He had the whole lab group with academics who wrote the analysis code, while Jonathon worked on the actual software, the data loading, and all that. The two of them were the main ones, and then EJ’s lab behind it. I joined the team as a research master’s student. I actually went on Erasmus to Amsterdam. I was studying in Leuven, so technically, even though I’m Dutch, I counted as a Belgian student. It made a lot of sense for me because I really like Bayesian statistics. I loved the JASP thing. And then that turned into a PhD and now my assistant professorship. So I’ve been part of the team for like 11 years or something at this point. I started in the same way as the rest of the lab: I was mainly just doing my PhD, but then ten percent of the time I would also program additional analyses into JASP. And what was also nice was this Bayesian Trojan horse approach where then they figured out, “Oh, let’s also include the frequentist analyses”.
Was the addition of frequentist analyses something that happened afterwards? Was the project intended to be exclusively Bayesian at first?
I wouldn’t dare to say off the top of my head. When I was there, it was already both, at least. I think at some point, the whole idea of JASP morphed a bit more from “oh, people wanna have easy Bayesian analyses” to “open-source statistics software”. I had a similar mental pivoting moment of realization that [while] Bayes is a fun little hobby, it’s much more important that people have accessible software all over the world. Especially, [to avoid] teaching students statistics using something that they need to pay a lot of money for after they’re done studying, because that’s really conflicting with the ideas of university and science. That also changed a bit how we presented JASP. When we would go to conferences, we would have a little booth with a JASP banner and stuff to tell people about it. In the beginning, it was always like, “Oh, but do I need to do Bayesian statistics if I use JASP?” and then slowly we got rid of that stereotype a little bit.
Were there any champions of it? Was there any resistance?
I think JASP was one of the first open-source counterparts to SPSS. There were some other ones, for example, a very old-school one called PSPP, which is exactly SPSS but then reversed and free. One of the challenges of open source software is that it’s very hard to monetize it. I don’t know a lot about that PSPP project, but it did seem that they just didn’t have enough resources to really keep going and make it a very attractive project. With JASP, we have the privilege that all of our team members are embedded in the university system, and they do it sort of on the side. Also, EJ is very good at getting research grants; he knows what the demand is and also how to write a good, clear plan. The concept of JASP is very powerful in grant applications because you’re like, “Oh, we’re gonna figure out this one analysis method, or we are taking a frequentist method and translating it to Bayesian, and then we put it in JASP so everybody can use it”. That’s just very nice: any grant proposal is strengthened by that. In general, over the years there have been more alternatives popping up that are open source. Original JASP team member Jonathon went to make his own sort of JASP called jamovi, which is extremely similar to JASP. At first I was like, “Yeah, but we’re supposed to work together to combat SPSS.” This episode also has a silver lining because it is inspiring to see how they are approaching their software. We are learning from each other. It is only a win because it means the market is just getting bigger. And a little bit of competition is also healthy. They are a bit more focused on the R community, and JASP is more aimed at just the general research and teaching community. We also develop a lot of educational materials, such as tutorial-style articles or blog posts. We try to make it really accessible in terms of being open source, but also accessible in terms of how scary it is to use it. All of these developments have caused SPSS to have a reduced market share, which is the end goal.
“ JASP is powered by friendship. ”
The Hardship of a Revolution
How do you manage to keep JASP free?
Although the research grants are always good, they have a limit, or they run out after a couple of years. We have been putting quite some work into making JASP a bit more sustainable. We’ve started an initiative called the JASP community. It is a way for universities to join the community and get some benefits, for example, higher-priority feature requests. A teacher who is using JASP in their course but is missing an option can be like, “Oh, hey, can you add this little checkbox to add this little thing?” And then we can give it higher priority while still keeping JASP free. And that way we are also still improving it. There are also different workshops. We put some effort into this project, but it turns out to be really, really hard to get the universities to join. Many teachers or other people at the university are super excited about the idea, but the moment you say something like, “Okay, so now you’re, yeah, like you have to pay us money.”, they back down. It could be like 1,000 or 10,000 a year, while an SPSS license is like 80,000, just to put it in perspective. It has been getting a little better in past years simply because we have been getting more users. Quite some universities use it, as do students and graduates. There are already quite some people who have graduated with their bachelor’s just knowing JASP. Then they carry that to their follow-up studies or where they work. That really helps; people take it a bit more seriously, and it is making these conversations easier. For example, two weeks ago I was over at another university, and they are very interested, as soon as we have version 1.0. We’ve always been operating with version number 0. something, because one of the things we really want to have before 1.0 is syntax mode. This 1.0 would also signal that now we can be trusted. Even though we can already be trusted and the core analyses have been checked many times. Hopefully, this is another small factor that will help convince universities to trust it. Then we would have whole universities switching to JASP. The process of switching to JASP really differs per university. At the UvA, for instance, all these policies are per department. I’ve approached various other departments in the Social sciences, for example, business administration. They still use SPSS. The trickiest thing is that everybody thinks, “Oh, that’s a great idea. We should use it.” But then, of course, there’s always a cost with switching. Teachers do need to update their slides, their screenshots, their homework assignments. I’ve been trying to make more and more things to make the switching as painless as possible, with a lot of tools like the Field’s book, for instance, and the data examples. I think that’s a very important additional thing: you can ask to switch software, as a university, but you also need to make sure that the teachers are then properly supported in the switch with additional hours.
What’s the main struggle of developing an open source software, and what keeps you motivated?
The main struggle is the fact that any change in a curriculum will have some cost tied to it. But that’s short-term thinking, because it’s maybe a higher cost this year, but then you’re saving a lot of money by not having to pay for your SPSS license. You don’t wanna be too paternalistic, being like “Oh, everybody needs to use it now” because that’s also not how it works. I mean, we’ve tried. What actually works a little better is having, for instance, students who have used it go up to their teacher. For example, it was really funny with the bachelor’s thesis last year, I think, or the year before that, because I had already used JASP a bit in the Scientific and Statistical Reasoning course. It was a year where I did both softwares. And some people ended up in the bachelor thesis basically begging the supervisor to use JASP and not SPSS. That’s the most priceless reward. So this also answers the question of what is the most rewarding thing to me: it is this kind of stuff, to actually see that it makes the whole statistics experience less painful for people. In my own experience as a teacher, JASP makes it a little easier to show certain statistical behaviors. But, you know, I’m a JASP team member, so it’s always nice to have a bit less biased opinions coming from students.
Proliferation of the Revolutionary Spirit
As a statistics and research professor, how do you transmit the mission of JASP (i.e., open source science) to your students?
It’s always better to show it more than just tell it. For instance, demonstrating how you can easily export your results, how you can share your results and having people figure it out a bit more for themselves, rather than using top-down understanding. It underscores a lot of these points about how science is a lot better if it’s collaborative, because you can have different perspectives.
What other faculties at our university are teaching JASP? Why and why not?
So in Psychology, we are now completely switching, and now the transition is starting in the master’s programs as well. I actually just wrapped up my SKO, which is like a senior teaching qualification, a year-long project about teaching. [The project] was about [navigating] the large degree of heterogeneity in the master’s students. Around 50% of students come from outside of the UvA; everybody starts at a different statistics level. Then you have 16 master’s tracks that all differ in what statistics they offer. I translated SSR into a website that any master’s student can start out with to refresh statistics, but also to get refreshed on JASP, because maybe not everybody uses JASP. I hope that with more and more of these resources, it’s also less painful for other departments to switch. I’ve been in touch with communication science, I’ve been in touch with sociology and business administration, and there it goes understandably slowly. We’re having a lot more success outside of UvA, ironically enough. The JASP website has this little page, uh, it’s in one of the drop-downs, ‘JASP Around the World’ where we have a general JASP icon for every university that has a course that uses it. In the Netherlands, it’s being used in psychology in Utrecht, in Groningen, and Tilburg. Those are the bigger ones. It is also used at university colleges. It’s a bit tricky also to define it because it’s not happening that a full university is switching. It happens more at the department or at the curriculum level. The studies where it’s most relevant would definitely be business administration or economics, medical sciences. All use similar types of tests like just T-tests and regression and so on. There is a lot of territory to be gained, conquered.
Revolutionary Visions and Future: The Path Towards Utopia
What do you envision for JASP for the future? What are your hopes?
I hope that our community gets bigger. At the moment, we have 14 institutions there. It’s already a little bit better than none. There are also some contributing in-kind, so now we also have JASP team members at different universities. To me, this is really beautiful because it’s not just our little team at the UvA Psychology Department, but it’s like you collaborate, and that’s the beauty of open source software. In theory, anybody could contribute.
And more technically speaking, do you have any features?
It is kinda cool that we are kind of running out of things we should add. There are a lot of things that we could add, but we are kinda running out of things that should be added. The two main things to add are data wrangling, so data transformations and all sorts of things. I’ve been having chats with a lot of the different specializations here, and I was like, “Oh, yeah, you are using a lot of data wrangling when you’re doing real research.” And nobody in our team knew, because we’re a bit in our ivory tower of methodology, where sometimes we’re not entirely aware of how important certain features are. That’s also what I’ve been liking: working together with my colleagues who teach other statistics courses; they can have input, give their perspective. I like that it’s a bit more of a grassroots movement, because there’s no way you can send an email to IBM and be like, “Okay, add this checkbox”- and, like, two months later, the checkbox is there. Then there is syntax mode; those are the two big ones. And then the documentation around JASP: having more resources, especially for fellow teachers who are switching. For instance, at the moment, I’m also working on having all the Discover JASP exercises in an ANS format. This way, other teachers can just import a zip file then they have all the exercises.
That’s my personal passion: combining JASP and teaching, and also how you can use one to improve the other. That is more for my own personal enjoyment.
Is there anything else you would like to add or talk about that you feel like you haven’t?
One funny detail is that now we have three developers. And one of the developers, Joris, has been one of my best friends since high school. At some point we had a change in the development team, and then I was having a beer with him, and he was complaining about how his current programming job was boring. And I was like, “Oh, do you wanna, like, join JASP?” And he’s been a part of the team for eight years already. That’s really cool, working together. He’s more of a programmer, and I’m a bit more in the academic part. JASP is powered by friendship.
What’s your favorite JASP feature?
It’s maybe very basic, but just the idea that the results are updated in real time – which makes p-hacking incredibly easy because you can tick all the options to get the best p-value (laughs). But for teaching, it’s so nice because you can change that option and then see how the results immediately change. So I think for teaching purposes, it is just a lot handier than SPSS.
