You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
I’ve spent the last couple days reading RCT studies from criminology. Almost all restrict analysis to compliers & program-completers, instead of using random assignment as treatment or IV. 
Remember: just because it’s an RCT doesn’t mean results are unbiased. Read the paper.https://twitter.com/amitabhchandra2/status/1075931428322492417 …
Related, I’ve read several papers that compare “random samples” from treatment and comparison groups. The only reason I can think of to do this (rather than using the full samples) is to try to fool readers into thinking you ran an RCT.
Yikes. What is the stated motivation? One (generous) possibility: seems similar to, but not exactly like, coefficient stability across alt. samples (ala sequential exclusion) to rule out leverage points. But a random sample of endogenously treated units does not an RCT make.
No stated reason. I think they just like the word “random”.
Unless data collection on a particular outcome is costly, so one only collects on a random sub sample?
Definitely not the case in these contexts. Usually using admin data w N in the 100s.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
