EJ
About
- Username
- EJ
- Joined
- Visits
- 2,532
- Last Active
- Roles
- Member, Administrator, Moderator
Comments
-
Definitely. Not the next version (this is in a few weeks), but perhaps the one after that.
-
Hi Biva, Yes, please send us your data set. You can just Email it to EJ.Wagenmakers@gmail.com and I'll forward it to Johnny. E.J.
-
Suppose you compute a Spearman's rho and conduct the standard analysis on the ranks. Suppose you have three observations, and the ranks are the same in the two variables. This yields: X Y 1.00000... 1.00000... (rank 1 for X and Y) 2.…
-
Hi Guillaume, Ah, but nonparametrics is a little more tricky with Bayes, as it inverts a generative model. The rank transform will produce OK result but only with large N (from a Bayesian perspective). Anyway, perhaps this is of interest: http://ge…
-
I'll ask Johnny what's up. I think the issue came up before. Do you have a concrete example? Cheers, E.J.
-
I think we might just add your ANOVA inclusion idea to JASP. Let me discuss this with the team. It is just another re-organization of the massive model table, with probability zero assigned to a specific subset of models. E.J.
-
Hi Sebastiaan, Yes, your understanding is correct. I think that the key aspect where your Baws Factor deviates from the inclusion BF is in the selection of the relevant models. For judging a main effect A, you disregard the models of higher complexi…
-
I just implemented the inclusion BF as one would for regression. For the case of ANOVA it is perhaps worthwhile to consider more carefully the kind of comparisons one would like to make. Different inclusion options could be made available in JASP. W…
-
OK, I have briefly considered the Baws factor and I think it makes sense! However, I'd compute the result slightly differently. The problem is that you have to be careful about averaging BFs (e.g., the average of 1/3 and 3 is not 1). So suppose you …
-
OK I will address your comments one at a time. First off, yes, the inclusion BF need not give the same result as the direct model comparison. The direct comparison focuses on two models, and the inclusion BF averages across all of them. You state: …
-
About the strong evidence for the main effects. A more informed assessment requires a look at the Bayes factors for the individual models. However, the Bayesian approach here is different from the classical approach, as it considers all of the possi…
-
A few quick responses: 1. Consider your very first example. The evidence for including the interaction (over the two main effects model) is .448/.066 = 6.79. I am not sure whether this will be exactly the same as the value from your recoded "va…
-
Well, this is interesting. Let me start by stressing that this is the result that probability theory gives us. So given the prior and the likelihood, these are the unique results. Our job here is to understand why it all makes sense (because probabi…
-
Hi Jeff, Right now, JASP does not offer mediation and moderation analysis. It is on the agenda of course. Cheers, E.J.
-
Ah, yes, well spotted. We'll take it into consideration! As an aside, the examples will be expanded and organized better in one of the next versions. E.J.
-
Yes, that's correct.
-
Hi peterose1, JASP does not allow "pure" interaction effects (meaning without the constituent main effects). This is the principle of marginality, discussed in part II available at https://osf.io/ahhdr/ Cheers, E.J.
-
Well I am of two minds here. On the one hand I want those Bayesian assumption tests. On the other hand, the purpose of an assumption test is not to test the presence of a violation -- rather, you want to estimate its impact. So an estimation approac…
-
Hi Niklas, Re. (1): Yes. Re. (2): Good points. No the best model is not convincing. And I am not sure that it is correct to assess the three-way interaction by comparing the most complex model to the model that lacks the three-way interaction (beca…
-
Hi Anna, Thanks for sending the file. What has happened is that you conducted a one-sided analysis for the July test, probably by mistake. The first one made sense (Hornsund > Edinburgh), but the second one (Edinburgh > Hornsund) is so strong…
-
Hi Anna, I think you have to make this public, or else invite me (EJ.Wagenmakers@gmail.com) Cheers, E.J.
-
Ah, I see. Good to know. Well we are almost there, but it looks like version 1 is probably going to take until Christmas. If it makes your institution happy you can point them to me, and I'll explain that the "beta" nature of JASP pertains…
-
Hi Bryan, Yes, that must be frustrating. Our programmers are looking into it. In general, if you have any questions or concerns of a more technical nature, you are better off posting your message on our GitHub page -- this way you can correspond wi…
-
This is difficult to tell without the data. Maybe a screenshot helps. E.J.
-
Hi MP, In the BANOVA, no contrast are implemented (yet). But yeah, entering a file containing only those two groups' results is the way to go. Actually, we are working on Bayesian post hoc pairwise comparisons too...but if your a priori interest was…
-
Hi Jeremy, Currently, you can only edit the data file and execute the JASP analyses on the edited file. So the Select Cases trick won't work, and instead you'd really have to remove the cases you don't want to include in the analysis. We do realize…
-
Hi Anna, No, pizza plots aren't available for ANOVA. I guess the reason is that there are so many models, and so many Bayes factors. There is such a thing as too much pizza. With respect to the priors, you raise an important point. If you have a s…
-
The binomial test allows a subjective prior (through the parameters of the Beta distribution). In the next versions we add a subjective prior for the t-tests, and possibly the correlation. The ANOVA requires more work though. Honestly, I would simpl…
-
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 thin…
-
Hi Inge, My advice is to use Bayesian statistics -- if you have few observations you will probably be told that the evidence is inconclusive (but that does not need to be the case). Also, the posterior distributions will just be wide. E.J.