EJ
About
- Username
- EJ
- Joined
- Visits
- 2,559
- Last Active
- Roles
- Member, Administrator, Moderator
Comments
-
Hi Tom, Let's look at the interaction first. The two main effects model has BF10 = 9.154e10. Adding the interaction makes the model much stronger (compares to the null): BF10 = 2.936e12. So the evidence for including the interaction is 2.936e12/9.…
-
Hi Eniseg the newbie, :-) Bayesian statistics does not require small-sample corrections. With small samples, there is likely to be little information in the data, which means that the prior distributions will not change a lot. So you are likely not…
-
Hi Michif, No, that's a feature. The slight offset prevents the intervals from overlapping with one another and potentially creating a rather messy plot. Cheers, E.J.
-
I think it eliminates the participant as soon as there is missing data for that participant. Easy to check by doing this by hand and seeing whether the results differ. Let me know what you find! Cheers, E.J.
-
Hi Tom, We work through an example in part II of a series available here: https://osf.io/m6bi8/ Cheers, E.J.
-
Here it is: http://bayesfactor.blogspot.co.uk/2015/01/multiple-comparisons-with-bayesfactor-2.html
-
Hi Dion, Right now this is not possible with JASP but it should be possible with the BayesFactor package (although you might have to do some post-processing). I believe Richard once wrote a blog post on this. I will forward this to him and seek his…
-
Hi Martin, A purely subjective Bayesian does not have a multiplicity problem. If the individual tests are "highly hypothesis driven", then I don't have a problem with it, especially for the given case with only three tests. But you can im…
-
If I'm correct then BF21 = 5.521e+234/5.444e+180 = 1.014144e+54. This is a massive number (a 1 followed by 54 zeros). The percentage error on the component BFs is low, so no need to worry about accuracy. E.J.
-
OK. So this is the way I see it. In the Bayesian analysis, it is more about model comparison than about parameter estimation. So we have the model without any predictors: the null model. Then we can compare this to the model with only the covariate …
-
And here "Group" is the factor of interest, and "Variable1" is the covariate?
-
You can check out the "ArthropodHostiliy.jasp" file (https://osf.io/mjft7/). The analysis should run fine if you simply define your factors. It would help if you could send a screenshot of jasp as it gives the error message, or, better sti…
-
Hi 2Elephants, Can you send a .jasp file or a screenshot of the output? [As an aside, you can't use ANCOVAs to control for a covariate that differs between groups. This is a common misunderstanding. You can use ANCOVA to explain away error variance…
-
Part II here (https://osf.io/m6bi8/) explains the principle of marginality. You probably included interactions without also including the main effect? Cheers, EJ
-
Yes, it is probably not perfect, but looks like an acceptable workaround. Perfect would be ordinal regression I guess, where you first enter the nuisance variables and then see whether the addition of the variable of interest matters. E.J.
-
Hi Graham, (1) I am not sure whether those Z-residuals are ordinal, or how they have been obtained. At any rate, regular Pearson correlations on ordinal data become less problematic the more data you have. But JASP now has Bayesian Kendall's tau as…
-
Weird. I know Windows complains and you have to ignore warnings. We are looking into changing that. The trojan warning is new though. I will pass this on the programming team. E.J.
-
Cauchy priors on effect size. The specifics are in a range of papers. You will find the relevant references in this draft paper (partII, the pdf): https://osf.io/m6bi8/ Cheers, E.J.
-
The model that is best supported is the full model. However, the evidence over the two-main effects model is not strong. Also, the model with only one main effect is supported almost as much as the full model. Conclusion: more data are needed. You m…
-
Standard practice is to center on zero. I guess the idea is originally from Jeffreys, who viewed 0 as the "general law" or "invariance", and then allowed H1 to depart from that value. If you don't center on zero you can have a mo…
-
no worries, thanks for asking
-
Yes that is a big inconvenience. This may be a "feature" of the latest release. I'll make a note on GitHub.
-
Hi MJS, For a scale this discrete, a nonparametric density plot (using "density" in R) is not useful. There are no values on the y-axis because the units are not relevant -- the histogram has been rescaled so that the area under the curve …
-
As long as the % is reasonably small (and it almost always is) there's no reason to report it. For ANOVA it can sometimes be a little bigger, but there we added the option to collect more samples to decrease the %. So is a a measure of the quality o…
-
It is this one: http://pcl.missouri.edu/sites/default/files/Rouder.etal_.pbr_.2016.pdf Do you have a reference for the Steyer paper in return? :-) E.J.
-
So Quentin kindly looked into the R code for the BayesFactor package. It seems that the following function is used to calculate the error for the cases where the package does not use the integrate function (obtained from https://github.com/richarddm…
-
Yes, I recall a few papers critiquing Type III sums of squares because of issues with differences due to scale transformations. And there's also a new paper out by Rouder, Morey and colleagues (in PBR) about model-based ANOVA approaches.
-
Great question. Let's ask Richard. Richard, can you explain how this is computed exactly? We will also mention this in our new paper on JASP which will be on my website soon. Cheers, E.J.
-
Hi jjth, I'm not sure why there isn't an error %. JASP calls the BayesFactor package. Of course the error is very small, but that shouldn't be an issue. What do you mean when you say "what is the correct statement for this question?" -- …
-
Also, for the benefit of others in this forum, can you link to those slides from Richard? Cheers, E.J.