EJ
About
- Username
- EJ
- Joined
- Visits
- 2,530
- Last Active
- Roles
- Member, Administrator, Moderator
Comments
-
A paper on how to go from AIC or BIC to model weights: Wagenmakers, E.-J., & Farrell, S. (2004). AIC model selection using Akaike weights. Psychonomic Bulletin & Review, 11, 192-196. https://www.ejwagenmakers.com/2004/aic.pdf
-
I do not immediately see how the problems with WAIC and LOO would imply problems with BIC. The BIC might be "unreliable" if the MLE is borderline identified. EJ
-
Yes, but there is another possibility. It may be that the data provide evidence for a "pure" interaction -- so there is only the interaction, and no main effects. This violates the "principle of marginality" which states that if …
-
@Michael_Jasper : yes you can multiply but it needs to be clear what hypotheses you are comparing. @andersony3k: Consider four factories, each of whom creates four products: X, Y, M, Z. The factories create these products in different ways. In Facto…
-
Hi Marissa, Sorry about the tardy reply. I will make some inquiries Cheers, E.J.
-
Sorry for the tardy response. It is important to check that your covariate does not overlap (substatially) with PTSD severity, or you will be removing the very thing you are interested in. As far as the results go, the null model predicts the data b…
-
Could you provide the data? (or fake data with the same structure?) EJ
-
This seems a reasonable explanation. Difficult to say more without the data and screenshots. EJ
-
I don't think there is, at the moment. There may be, but it will probably involve some effort to get it to work. EJ
-
Yes you can conduct those two test, but when you want to combine them then it would be prudent to add those other hypotheses. Of course you can add the negative correlations as well, expanding the hypothesis space. I had hoped you have some substant…
-
I suspect that some cell entries are empty. If they are not, this is a bug, and you can post the issue on our GitHub page, which will bring you in direct contact with the programming team (for details see https://jasp-stats.org/2018/03/29/request-fe…
-
Well, I would still like to argue in favor of the 4 hypotheses I outlined above. And these hypotheses might just be associated to particular people/forecasters: John's believes that XY>0 and MZ >0 Mary believes that XY>0 and MZ=0 Amy belie…
-
Good point! It may be argued that there are at least four hypotheses: XY>0 and MZ >0 XY>0 and MZ=0 XY=0 and MZ>0 XY=0 and MZ=0
-
More precisely: if the size of the correlation between x and z under H1 affects the knowledge about the size of the correlation between z and m under H1, then independence is breached EJ
-
You can multiply the BFs if knowledge of the result for x and y would not alter your knowledge for the correlation between z and m. Now you know that x and z are highly correlated, but I don't think this is relevant for the correlation between z and…
-
This is not the same issue I don't think. Did you post yours?
-
Hi Dan, I am not 100% sure what you did, but the BF is based on relative predictive performance for observed data y. It so happens that the t-value and sample size is a sufficient summary of the data, so that no information is lost when you enter t …
-
I'll ask the team member who added it. Did you check the help file for background information and references? I am doing so now, and I see that it shows "the residual correlations (standardised residual covariance matrix) of the indicators"…
-
Hmm. But maximum likelihood is not Bayesian. I will ask the relevant team member about missing data in the Bayesian process module. EJ
-
The Bayesian Process module does not yield Bayes factors directly, but it does provide BIC and BIC weights (which are a transformation of Bayes factors under a unit-information prior). See the help file for background information and references. EJ
-
I think we have seen a handful of these messages. I recommend that you search for this issue in our GitHub page -- I think these issues were solved. Regardless, the JASP team is always willing to assist, and your best bet if something does not work…
-
This is not possible in the current data editor. Your best bet is perhaps to save the plot in a different format (Powerpoint, pdf) and post-process it in a graphics program. EJ
-
So set everything else to fixed by default? But this seems to go against the popular mantra to "keep it maximal"
-
Yes but we have to do something.
-
The best way of course is to attend our annual workshop in Amsterdam :-) Otherwise, working through some of our papers and course books also helps. (e.g., https://jasp-stats.org/jasp-materials/#papersJASP and https://www.bayesianspectacles.org/free-…
-
More samples yield more reliable results, but at a computational cost. This balancing act is yours to make as it will depend on a number of person-specific considerations. EJ
-
But is this bad behavior? What alternative procedure would you recommend? Not give any output until these effects are specified as fixed or random?
-
Dmartin427: maybe a good feature request? EJ
-
This is also best handled through our GitHub page! EJ
-
This is already included in the Survival module, isn't it? Regardless, the upcoming new version will have expended functionality for survival analysis. Best to take this issue to our GitHub page (under feature requests, see https://jasp-stats.org/20…