I want to estimate the difference in association between a single predictor and two different outcomes with the same form. As an example take this dataset. One numeric predictor and two binary no/yes outcomes. exDF.RData (9.6 KB) Here is the model I propose running, a multivariate bernoulli. Apolo…

I’m not sure what you mean by “kosher statistically”, but I’m pretty sure the answer is “yes”. That is to say, if you have two outcomes with the same units, and a single predictor, it’s a perfectly well posed question to ask whether the slope of one regression is bigger than the slope of the other. …

Thank you @jscolar. You actually knew exactly what I meant by statistically kosher. Much appreciated. And yes working out how to allow correlations between outcome residuals is my next port of call. Not possible in brms multivariate at the moment so will involve cracking open the black box and rewir…

One easy solution here is to combine your two binary outcomes into one categorical outcome, them use the categorical distribution. You can then create the marginal parameter estimates for each outcome from the draws of the categorical parameters. You will see very quickly what @jsocolar is saying h…

Thank you @simonbrauer . These are non-mutually exclusive outcomes so creating a single categorical variable is less appropriate I think. Doable with two binary outcomes, you’d get a four-level categorical - [0,0], [0,1], [1,0], [1,1]. Output would be pretty confusing but just manageable. Once you ge…

Estimating difference in effect of a predictor on different (but similar) outcomes in a mutlivariate regression

Modeling

Topic		Replies	Views
Predictions from univariate models differ from those of a single multivariate model General brms	1	73	November 19, 2025
Interpreting multivariate versus univariate models brms	3	1844	June 25, 2020
Multivariate brm models brms	3	2134	July 21, 2021
Between + Within subjects multivariate ANOVA in brms brms	15	3367	August 3, 2019
Setting informative priors for multinomial model and comparisons across levels from parameters brms prior-choice	9	1651	June 18, 2020

Estimating difference in effect of a predictor on different (but similar) outcomes in a mutlivariate regression

Related topics