RT @KhoaVuUmn: Real footage of a synthetic control model:

RT @LabConflict: 📊 Check out the last @Bangladesh's fatalities pattern that mostly led to a decreasing conflict risk in history. Explo…

RT @LabConflict: 📊 Check out the last @Mozambique's fatalities pattern that mostly led to a slightly increasing conflict risk in history.…

📘 New publication in Journal of Forecasting, with T. Schincariol We show that migration flows exhibit recurring temporal patterns. Leveraging these temporal shapes dramatically improves forecasting accuracy, even with limited data. 🔗

Check out our latest article in Political Analysis with Jian Cao! We show how units can react at varying speeds to shocks, which can distort causal estimates in Synthetic Controls. We introduce a more efficient dynamic estimator to correct these biases

RT @LabConflict: 📊 Check out the last @Russia's fatalities pattern that mostly led to a stable conflict risk in history. Explore more:…

RT @LabConflict: 📊 Check out the last @Brazil's fatalities pattern that mostly led to a decreasing conflict risk in history. Explore m…

Social science papers are too long. That's it.

RT @andreamatranga: Modern econometrics

@TLJHuntington We're still teaching realism?

RT @prisonrodeo: Going straight into the syllabus, where it will appear on IV Is A Party Trick, Not A Real Method Day.

@AlecStapp Now just hear me out for a sec. Perhaps, just perhaps, the electoral college is a lousy institution.

@waitbutwhy Old French from 1000 years ago is still pretty readable.

@dylanmatt @jfkirkegaard Slightly more informative:

RT @matt_blackwell: Senior faculty presenting the paper written by their graduate student

@dylanmatt @jfkirkegaard Yes, in a way. But: - *median* wages would much more informative. - this does not account for time worked. - the US lags in median wealth, so these incomes are not really translating into wealth.

RT @tomaashby: For those of you doing PhDs that need to hear it: your dissertation is an exam *not* a published book & totalising perfectio…

@namalhotra I come from conflict prediction and know little about election predictions, but N=59 is surprising. Even with a limited number of elections, we can test models at various geographic levels, even down to individuals. So genuine question: Isn't that the norm in election preds?

RT @JohnHolbein1: "We show that Black families whose ancestors were enslaved until the Civil War have considerably lower education, income,…

RT @LabConflict: 📊 Check out the last @Burundi's fatalities pattern that mostly led to a decreasing conflict risk in history. Explore…