You can now use LLMs to learn about any chapter of my 𝑀𝑜𝑑𝑒𝑙 𝑡𝑜 𝑀𝑒𝑎𝑛𝑖𝑛𝑔 book, or any function in the `marginaleffects` package for R or Python.

Check out the cool demo video below (sound on).

Install here: github.com/vincentarelb...

#marginaleffects #RStats #pydata

Hot take? Johnson-Neyman plots are useful for basically all cases *except* for bi-linear interactions (for which they are almost exclusively used 🥲).

( #marginaleffects + #ggplot code in alt text)

#stats

The course also had #rstats exercises for all topics developed by my amazing TAs - using #marginaleffects throughout and of course @easystats.github.io!

I hope to have all those materials on GitHub soon.

Using #marginaleffects, we compared model predictions for each outcome variable across three levels of food insecurity: "low", "medium", and "high". Each level represents an "average" participant who either responded "never", "sometimes", or "often" to each food insecurity question.

The Pink Book of #MarginalEffects (aka Model to Meaning) ships next week and I've got a backlog of Zoolander memes.

Hope you're hungry for some spam in your timeline.

#RStats #PyData

Me: shows in class how to understand multilevel binomial models w/ #marginaleffects, supplies detailed and commented R code.

Student: hands in final assignment where all model interpretation was based on results from a badly written custom function that gives incorrect results*.

Why do I bother?

This could make #marginaleffects package even better than it already is!

I'd love feedback on this #marginaleffects WIP.

In GLMs with many parameters, the marginaleffects 📦 can get a 5–20x speedup (and better SEs) by calling JAX for automatic differentiation.

Try the instructions at this link and let me know if you run into trouble.

Thanks!

github.com/vincentarelb...

Even if you're not tackling these super complex questions, {modelbased} is generally just a fantastic tool for really getting your head around your statistical models. Go on, take a peek! You might just fall in love: easystats.github.io/modelbased/

#rstats #easystats #marginaleffects #inference

Okay, so you've crunched your numbers and got some awesome statistical models? Sometimes, just knowing "X predicts Y" isn't enough to really get to the juicy bits. That's where the cool post-hoc stuff comes in – think estimated marginal means, contrasts, pairwise comparisons, or #marginaleffects.

Just clearly define your estimands. #marginaleffects

Model building - both statsmodels and sklearn - pretty smooth.

But that's where the smoothness ends it seems.
Trying to doing stuff with these models other than just making predictions is oh so very clunky.

(I don't know what people did overthere before #marginaleffects...)

Making sense of logistic regression using #marginaleffects? ✅
Using Bayesian #stats with non-flat priors? ✅
Fitting the model with #brms? ✅

Give Shachar's post a read!

Dive in for code, visuals, and a clearer path through the log-odds fog → cogpsychreserve.netlify.app/posts/logist...

#NLP #Kaggle #marginaleffects #BayesianStatistics #DataScience #SignificantTesting

2/3

• NLP + PCA to capture toxicity/incoherence
• Cohen’s d ➡️ log-odds priors in one line using #brms
• #marginaleffects → 0–100 % probability shifts you can explain
• Inference with HDI-ROPE. It flags which effects are big enough to matter. Great for researchers and anyone shipping spam filters!

Note to self (and in case anyone else has got tripped up not realising), #marginaleffects, #emmeans etc. functions tend to keep the dataset in the object attributes when you assign them to an object. That's caught me recently for big targets simulations with large population datasets...

The #rstats #package #marginaleffects do more that just #equivalence #testing have a look marginaleffects.com

How to interpret and report nonlinear effects from Generalized Additive Models | GAMbler Generalized additive models (GAMs) are incredibly flexible tools that fit penalized regression splines to data. But interpreting nonlinear effects from GAMs is not as easy as interpreting linear model...

#statstab #340 How to interpret and report nonlinear effects from Generalized Additive Models

Thoughts: Maybe you heard about GAMs and spline, but not sure how to use them. Here's a #R tutorial.

#GAM #spline #nonlinear #tutorial #guide #marginaleffects

ecogambler.netlify.app/blog/interpr...

Took the opportunity to flex my #marginaleffects muscle and write a little guide for computing ranks and rank-based contrasts from ordinal regression models.

Check it out!

#rstats

Code of a function that convers objects from the marginaleffects package to an emmGrid object from the emmeans package.

Don't mind me... just working on an abomination over here...

#rstats #marginaleffects

#statstab #223 Conformal predictions w/ {marginaleffects}

Thoughts: Sometimes you need a range of likely future values. To get an assumption-free Prediction Interval, use conformal methods.

#r #stats #marginaleffects #prediction #conformalprediction

marginaleffects.com/bonus/confor...

#statstab #213 Marginal means and Average predictions

Thoughts: Do you prefer Estimated Marginal Means or Average Predictions? Care to report an Average Counterfactual Adjusted Prediction or an AME? Here are some options:

#r #emmeans #marginaleffects

marginaleffects.com/bonus/altern...

I loooove this post! So much so that I plan to include a section with similar diagrams in the #marginaleffects book (With proper attribution; thanks for the CC license!). Hope you don't mind! Here's a concept of a plan: marginaleffects.com/chapters/fra...

📚🚨 I posted 11 new chapters of my upcoming book!

Model to Meaning: How to Interpret Statistical Results with #marginaleffects for #RStats and Python.

These are early drafts and I really need your feedback! Errors, content requests, improvements, etc.

marginaleffects.com

Equivalence Tests with {marginaleffects} Reproducing the Clark and Golder (2006) example from Rainey (2014)

#statstab #205 Equivalence Tests Using {marginaleffects}

Thoughts: Easily testing for "no effect" is important. Here's a tutorial by @carlislerainey using @VincentAB r pkg.

#equivalencetests #equivalence #TOST #marginaleffects #tutorial #nullresults #r

www.carlislerainey.com/blog/2023-08...

#statstab #202 Distribution regression in #R @VincentAB

Thoughts: "distribution regression...allows us to measure the association b/w the predictor of interest and the outcome at different quantiles of the outcome"

#regression #marginaleffects #quantile

arelbundock.com/posts/distri...

Making apples from oranges: Comparing noncollapsible effect estimators and their standard errors after adjustment for different covariate sets We revisit the well-known but often misunderstood issue of (non)collapsibility of effect measures in regression models for binary and time-to-event outcomes. We describe an existing simple but largel....

Not everyone agrees that ORs/HRs, and indeed non-collapsibility, are inherantly problematic. I don't want to get into those arguments.

If you want to estimate effect estimates that are not model/covariate dependent, then you can do this using #marginaleffects
onlinelibrary.wiley.com/doi/full/10....

But there is another 'best of both worlds' solution.

Run a logistic regression model, but instead of calculating ORs from the coefficients you calculate RRs (or risk differences) using @vincentab.bsky.social's #marginaleffects package!

NEW to {bayestestR} dev version - support for #marginaleffects!

Take it for a test drive: remotes::install_github("easystats/bayestestR")

#easystats #rstats

It's official. I'm writing the #marginaleffects book!
The provisional title is: “Model to Meaning: How to Interpret Statistical Results in R and Python”
Please reply with better titles, artwork ideas, content suggestions, etc. What do you want to read?