Assistant Professor of Sociology, University of Michigan
Ann Arbor, MI
Joined July 2011
- Tweets91
- Following115
- Followers205
- Likes19
SOCIOLOGISTS: we are writing an annual review article about word embeddings in sociology. If you have a published article or draft manuscript you've written using word embeddings (including LLMs like BERT/GPT3/...), please drop me a line with a URL so we can include it.
Andrei Boutyline retweeted
Female schooling has soared,
But men are still more likely to be called ‘intelligent’.
‘Genius’ is still reserved for men.
New analysis of a 200 million word corpus of US print media (1930-2009) by @_Andrei_B_ @arsenievK & Cornell
Trying to decide on a paper title: (1) "What Exactly are Cultural Schemas: A Conceptual Reorientation & User's Guide" vs (2) "Cultural Schemas: What They Are, How to Find Them, and What to Do Once You've Caught One" [or (2b): "... Once You Have One"]. Thoughts?
1
4
This review is literally the stuff of nightmares. The argument is roughly: 'the scientific community is the only rightful evaluator of quality of published work; thus, evaluations of the quality of published work by members of this community should not be published'
Here's a thin slice of one reviewer's comments (not editor comments) to provide a flavor from behind the curtain. I don't think the whole review (and others) are identifying, but I'm reluctant to post more. As you might expect, reviews were polarized.
1
2
3.6 million instances of (very, very small-scale) pro-social behavior. #theinternet xitter.gq/carterjwm/status/84981…
Andrei Boutyline retweeted
Belief Network Analysis: A Relational Approach to Understanding the Structure of Attitudes bit.ly/2pnXJ3p #sna #PoliSci
Huh, R takes ~0.02 sec to run an endline..? Compare "a <- Sys.time(); b <- Sys.time(); a-b" to same script w endlines instead of semicolons.
A physical UI of sliders/knobs to control any software: dpreview.com/news/2660638409… Any way to make these useful with statistical software?
538 now gives Hillary 70.8%, her best odds since Sep 5. Guess that 1 debate was enough to undo whole month of Trump pretending to be Romney.
Appealing take on new Penn Sta: nytimes.com/interactive/2016…… May be better than orig, which was as oppressively monumental as Farley post is now
Srsly, sugar-lobby oppo to oakland soda tax may be highest per-voter funded campaign I've seen. Inundated by ads. Lobby clearly scared.
Seeing online ads and getting mailed glossy fliers& branded grocery bag frm sugar-lobby oppo to #OakSodaTax. Well-funded & wildly misleading
Guess the correlation from scatter plot: guessthecorrelation.com. Fantastic for building intuitions for what, e.g., r=0.5 means.
Andrei Boutyline retweeted
The most successful "public sociology" project the discipline is doing, BY FAR, is the General Social Survey.