I tried that, but got confused by the output I was getting, so did it by hand on the basis that I understand what is going on bsky.app/profile/chan...
Has anyone tried calculating average effects for conditional logit models using @vincentab.bsky.social 's marginaleffects package? I tried two ways (using `comparisons` and averaging predictions by hand), and I get different results. MWE at gist.github.com/chrishanrett...
Two different ways of calculating "average effects" in conditional logit models - clogit_mes.R
I think "crunch time" as an expression predates its use in software development
Relevant for people who are modelling contests or anything where you need a conditional logit model
New blog post: "How much do gender and ethnicity affect wannabe MPs’ chances of being selected?" chrishanretty.co.uk/posts/candid...
All of you need to read this article now, because I have just read the words "Hummingbirds have to do this, or they would piss themselves to death" and must share the experience with you arstechnica.com/science/2024...
Nightly suspended animation, sugary diet, backwards flight, and long migrations help.
Congratulations!
Is the full text available yet?
Good example of things social scientists study only b/c it illustrates a mechanism: a lottery win makes you more right-wing (papers.ssrn.com/sol3/papers....), but no one's using "lottery winnings" to explain election trends
There are lots of people like that, but I would still have expected asset ownership to increase the *probability* of voting Conservative, even if it's not a huge effect (say, two to five percentage points).