Cyrus Samii
@cdsamii
Followers
5K
Following
6K
Statuses
8K
NYU prof. I also post at the other place.
Joined February 2009
RT @michael_wiebe: How do you get an event study that looks like this? By having almost no treatment variation. 🧵 1/ #econtwitter https:/…
0
39
0
RT @RobinsonAmandaL: Scholars of political economy in Africa: please remember to apply for the 2025 #WGAPE meeting at OSU on April 25-27.…
0
82
0
RT @EGAPTweets: The Mercury Project’s Health Ambassadors researchers are gathered this week in Nairobi to collaborate on data analysis & re…
0
3
0
@namalhotra If it’s not rush hour into the city, Lyft/Uber gets you there in 20 min. If it is, then you can take NJ Transit and takes about 45 min. Bit of a pain honestly to get to the airport station, but may beat sitting in tunnel or bridge traffic.
0
0
1
Maximal covariate leverage strikes again (cf. Alwyn Young's work on randomization inference and degrees of freedom corrections), balance test edition.
In today's blog, a new paper by @jt_kerwin, Nada Rostom & @OlivierSterck shows that joint orthogonality tests of balance between treatment and control can over-reject (so you conclude not balanced) and that RI can solve this
0
0
4
Effect heterogeneity restrictions in studying mechanisms and in mediation analysis
1
5
29
Junior faculty recruitment in IR and American Politics at NYU:
NYU's Politics Department is seeking outstanding and innovative scholars for two positions at the assistant professor level: International Relations and US Politics. For full descriptions and to apply visit (IR) and (US).
0
0
0
RT @NYUpolitics: NYU's Politics Department is seeking outstanding and innovative scholars for two positions at the assistant professor leve…
0
19
0
Modeling out the situation and checking cross partial comparative statics can help. Maybe you also need to transform the outcome to be less vulnerable to ceiling effects, or analyze conditional on pre-treatment levels.
1
0
1
Interesting analysis: “sample sizes typical of many real-world clinical trials datasets often constrain regression adjustment methodologies from attaining their theoretical potential in practice.” Proposes methods to “borrow observations from large-scale observational datasets.”
The economic value of this finite-sample var. red. could be huge: it could be used to decrease N, while holding statistical power fixed. At scale, this could be worth a billion dollars of clinical trial cost savings/year. Details are in the paper: 22/22
0
2
4