Infinite probability? about bayestestr HOT 2 OPEN

Dear Easystats authors, Thank you of your message! Now that I have the correct email to send comments I will use it. The comments quoted below are complete. The first one is resolved. I will think about suggestions to resolve the second. Regards, Emilio A. Laca, Professor Emeritus One Shields Avenue, 2306 PES Bldg. ***@***.*** Plant Sciences Mail Stop 1 voice: (530) 754-4083 University of California fax: (530) 752-4361 Davis, California 95616 mobile: (530) 220-5315 ANTI-RACISM resources: https://health.ucdavis.edu/diversity-inclusion/racial-justice/anti-racism-resources.html On Oct 15, 2023, at 03:05, Dominique Makowski ***@***.***> wrote: Hi Emilio, thanks for your comment and for sponsoring. By all means, do add here all comments that you might have. I'm going to list your points (I'm taking the liberty of also adding bits from your email) here and then we can aim at improving them. * Replace "(the probability of it being different from a single point being infinite)" by "(the probability of it being different from a single point nearing 100%)" * About model-based estimates "However, these raw means might be biased, as the number of observations in each group might be different. Moreover, there might be some hidden covariance or mediation with other variables in the dataset, creating a “spurious” influence on the means. …” "Note that the means computed here are not that different than the raw means we created above. From which we can surmise that there are not many spurious influences that we need to worry about in the iris dataset. But this might not be the case for your dataset.” * "I think that with a single factor in the model, raw and “marginal” (model-based) means are always identical (if bayesian, with uninformative priors and within random variation due to the MCMC sampling)." * "My understanding is that the raw average is un unbiased estimator of the mean. I think I know what you (and coauthors) want to convey with that, but a beginner might take it to signify that averages are biased. The case for model-based vs raw marginal means can be made nicely when you have more than one factor and different cell sizes (n per cell). I see that you added a covariate to obtain a similar effect effect." — Reply to this email directly, view it on GitHub<#630 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACZXHYJHBXHA2HJZ4PHHY5LX7OYM5AVCNFSM6AAAAAA6AMJBCKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONRTGM2DCNZXHE>. You are receiving this because you authored the thread.Message ID: ***@***.***>