EJ
About
- Username
- EJ
- Joined
- Visits
- 2,557
- Last Active
- Roles
- Member, Administrator, Moderator
Comments
-
Hi Nicolas, Not 100% sure, but these high BFs translate into very small probabilities, where rounding can make a big difference. To check, you could take an example where the BFs are modest. E.J.
-
Hi Shuang, Well, statistically this is a tricky issue. I personally do not like the term "controlling for", because it suggests something that is impossible without an experimental setup. But in general, yes, you can do a "hierarchic…
-
No idea! Can you attend us to this problem again please (on GitHub)? It would be good to sort this out E.J.
-
Hi Nicolas, Sure, happy to oblige. I'll ask Tim to send the relevant code, or post it here. Cheers, E.J.
-
I'll Email the student now. We discussed this issue yesterday in the JASP team. It would also help if you made this an issue in GitHub, if you haven't done so already. Cheers, E.J.
-
Hi Ivan, Well, I am not so sure there is a big conflict. The BF for including the interaction is about 2.7, right? And the p value for your last analysis is about .01. That discrepancy is a little larger than what I usually see, but then again, thi…
-
Hi Hannah, I assume you are talking about the t-test. Delta is the population version of the sample Cohen's d. So it is just the mean population difference divided by the population sd. You can find a detailed description in http://www.ejwagenmakers…
-
Hi NT, Yes, mediation, moderation, moderated mediation, mediated moderation (if these exist): they will be implemented. If you want to urge the team to assign more priority to this development, please post a message on our GitHub page. Cheers, E.J.
-
Hi Sarah, This occurs because that particular analysis requires numerical techniques. I am on my cell phone but I believe that under Advanced Optiobs you can increase the number of draws that determine the accuracy of the approximation. You could a…
-
There's an entire statistical field called "objective Bayes", and their goal is to pick priors that fulfill a number of general desiderata and are fit for use across a wide range of situations. In my experience, within the limits of reason…
-
We have these numbers for some analyses (such as the t-test: both in tables and in figures). But for others it seems to be missing; you've already made a feature request on GitHub? Cheers, E.J.
-
I'd report report the entire model comparison table. Yes you can test the two main effects model against the model that also includes the interaction. It seems to me that this is the most intuitive analysis. Maybe the table will look more compelling…
-
Hi CHei & thanks Aram, You are basically on the right track with all your remarks. 1. Yes you can use the analysis with all sample sizes. 2. I would stick to the default prior. 3. If you select "compare to best model", the resulting …
-
Hi Ivan, It's a little hard to say without the data (descriptives). I could assist more easily if the analysis were done in JASP, but Richard is the expert anyway so I'll bring this post to his attention. I'd be pretty impressed if he knows what's …
-
Hi Kevin. The null is always the absence of an effect (i.e., a point null). In one of the next blog posts on BayesianSpectacles.org, I will explain how you can obtain the test for direction from two directed tests against the point null. EJ
-
The new version (hopefully out today!) will change "dependent" to the label of the DV. Right now you cannot change the axis end points. We are working to make that possible. If you want to up it on the priority list, please make it an issu…
-
Hmm I think we have been focusing on other improvements (new version coming out soon). This is actually more of an issue for our GitHub pages -- perhaps you can post the issue there? This way the programmers will see it and can take the appropriate …
-
Hi Christos, Definitely; we probably will roll that out when we do MANOVA too. If you want to keep the pressure on you can add a GitHub feature request so that this moves up our priority list. Cheers, E.J.
-
That's a good BayesFactor question! For a speedy Email, you might want to Email Richard directly (feel free to post his response here, of Richard agrees). Cheers, E.J.
-
Hi Aram, 1. As far as Bayesian analyses are concerned, we are trying to keep BayesFactors and JASP in sync. 2. There is a confusion in the literature about classical NHST and BFs. My perspective is that p-values are not to be recommended (but you mi…
-
If you add group & time as nuisance, then you are comparing the full model to the two-main effects model. This yields BF=6.25 in favor of adding the interaction. So that's a little better than "no support"
-
That level of evidence is almost completely nondiagnostic. If you have strong reasons to include the two main effects then you can focus on the two main effect model compared to the full model that also includes in the interaction (this is the same …
-
The model with the interaction is the best. To see by how much, it is easiest to select "compare to best model". Cheers, E.J.
-
Hi Francesco, Great example. You summarize the results well. So if you already have the main effects in there, there is some evidence for including the interaction. But at the same time, the null model beats all of the other models. As always, I re…
-
There is Bayesian Design Analysis. See https://osf.io/d4dcu/ E.J.
-
Hi Larry, The RM ANOVA requires numerical approximation. When you increase the number of samples (i.e., draws from the posterior) you increase the quality of the numerical approximation. The downside is that the procedure then takes a little longer…
-
Hi Francesco, You are reporting the BF, which, as you indicate, express the relative likelihood of the data under the models at hand. So that means that you can eliminate the text "assuming that H0 and H1 are equally likely" -- the analys…
-
Sound good, thanks for the paper! Cheers, E.J.
-
Hi Nicolas, The "Baws Factors" should be in the next version of JASP; they are computed by neglecting certain models. So the procedure is the same, but what differs is the set of models that is averaged over. E.J.
-
Well the tests are ready, we need to finish up the paper and include them. It is a big of work, but when we have the paper submitted you are welcome to the code so you could run it before it's implemented in JASP