The guy trained at LSE and everyone acts stunned that he's a closet Marxist. In many ways, his work is a correction of Marxist thinking that never made much sense (collapse of capitalism, historical determinism) *and* naive takes from American liberals about markets & government.

This is such an important part of modelling. The best model is not the one that fits every data point nor the one that fits none of them. It’s the one that does the best job without assuming too much—or only as much model as you need to answer the Q.

One of the big problems in my mind is how conversations about grading and assessment only deal with what’s fair, not what’s effective. This was always true for students but seems to the MO for admin increasingly.

I don't quit academia because I am academia.

Academia is not owned by those who have tenure or positional power. As long as we are doing honest research and not a crackpot, we are as academic as anyone.

'Publishing negative results is good for science' | Elisabeth Bik | Microbiology Society | buff.ly/3vO9RBX

Prospective PhD students have until Monday to decide where they want to study!

My new piece in Research & Politics with Dominik Stecuła may be informative. It shows that programs that pay their students better place more of them into tenure-track jobs. A brief 🧵 1/

journals.sagepub.com/doi/10.1177/20…

We (me and Łukasz Chrostowski) are organising a workshop on the {nonprobsvy} package: an R package for inference based on non-probability samples (e.g. #bigdata , #websurveys )! If you are interested please register via the google form and spread the word forms.gle/n1BZCsCGzJD9h2…

.Vincent Arel-Bundock now has a 1) model summary 📦, 2) table-making 📦 and 3) marginal effects 📦.
Can Vincent Arel-Bundock use his market power to stifle innovation by buying out competitors?
Or lobby the R foundation to get lucrative earmarks in the code base?

The EU should investigate... 👀

Hopping on this bandwagon a bit late 😅, but why not use an entropy measure for inequality rather than Gini? Isn't entropy a bit more holistic? (Plus various types of entropy to use).

This is why you write blog posts folks: no way any self respecting journal would let me get away with this overly dramatic title.

Stephen Wild Demetri My take is that using OLS on a binary outcome is a bit like doing free verse poetry. Once you understand grammatical rules, it's OK to break them. But if you only use OLS because logit never made sense... that could be bad.

Demetri I totally get it.

So long as your predictions are contained in [0, 1], LPMs are usually fine. But it's still 110% okay to make fun of them

📢New discussion paper on what is wrong with #economics #peerreview and what can be done about it.

Link: ssrn.com/abstract=44029…

Ever annoyed by contradictory or excessive referee asks? There are fixes!

A short thread 1/n
#EconTwitter #AcademicTwitter

People have needed rituals to turn data into truth for many years. Why would we be surprised if many people now need procedural reforms to work?
statmodeling.stat.columbia.edu/2024/04/08/peo…