Or, "How I couldn't go to sleep and stayed up to 2am doing fake research and rejected all my theories about something no one else but me probably cares about"
This was one of the discussion we were having in the department as well. My personal take on this rise (might not align with the time you show upwards but might as well if you check) of CS estimator is the Paper "What’s trending in difference-in-differences?". Mainly because it draws the comparison about CS and other estimators. However the CS estimator gets really detailed and intuitive explanation throughout the process. And small things (not really but on grand scheme of things) like CS being able to deal with serially correlated outcomes and parallel trends being relative easy to justify over long time periods. Also, the paper discussed about the no carryover assumption of of dCDH and a stronger parallel trends, which also kind of tips the choice of estimator towards CS. This is coming from my PhD experience and why I choose CS (which is mainly due to context of my research).
That’s an interesting theory. I guess I would’ve thought that ‘what’s trending’ might have been a more general boon (is it boon or boom?) but as Pedro is one of the coauthors on it, I could imagine it getting more of a bump from him. SA shares many of those same attributes, though SA doesn’t handle covariates the same way that CS does, so that might matter, but nonetheless, that’s a good point you make.
Although, with that said, What's Trending is published in 2023. It does appear in 2021, and it does get 170 cites in 2022. I find almost 400 additional cites under a parallel trends assumption to some unknown 2022 treatment for CS, or at minimum from a pivoting of trend relative to the others trend. Although that's not impossible -- you could imagine a multiplier effect from a survey article such that each 1 cite of the survey article supports, say, 3 of another person's paper.
That does make sense. Because I remember at the time, at least the people I knew around me were always a bit unclear what to do with all the estimators. And you could see in the earlier papers and to this day, authors try to chart all the estimators in the same graph to showcase robustness. However, this survey article at least gave people clear comparison and choices, assumptions needed when making those choices, and does that so in a very digestible way.
This was one of the discussion we were having in the department as well. My personal take on this rise (might not align with the time you show upwards but might as well if you check) of CS estimator is the Paper "What’s trending in difference-in-differences?". Mainly because it draws the comparison about CS and other estimators. However the CS estimator gets really detailed and intuitive explanation throughout the process. And small things (not really but on grand scheme of things) like CS being able to deal with serially correlated outcomes and parallel trends being relative easy to justify over long time periods. Also, the paper discussed about the no carryover assumption of of dCDH and a stronger parallel trends, which also kind of tips the choice of estimator towards CS. This is coming from my PhD experience and why I choose CS (which is mainly due to context of my research).
That’s an interesting theory. I guess I would’ve thought that ‘what’s trending’ might have been a more general boon (is it boon or boom?) but as Pedro is one of the coauthors on it, I could imagine it getting more of a bump from him. SA shares many of those same attributes, though SA doesn’t handle covariates the same way that CS does, so that might matter, but nonetheless, that’s a good point you make.
Although, with that said, What's Trending is published in 2023. It does appear in 2021, and it does get 170 cites in 2022. I find almost 400 additional cites under a parallel trends assumption to some unknown 2022 treatment for CS, or at minimum from a pivoting of trend relative to the others trend. Although that's not impossible -- you could imagine a multiplier effect from a survey article such that each 1 cite of the survey article supports, say, 3 of another person's paper.
https://scholar.google.com/citations?view_op=view_citation&hl=en&user=BVWmSY4AAAAJ&citation_for_view=BVWmSY4AAAAJ:ZeXyd9-uunAC
That does make sense. Because I remember at the time, at least the people I knew around me were always a bit unclear what to do with all the estimators. And you could see in the earlier papers and to this day, authors try to chart all the estimators in the same graph to showcase robustness. However, this survey article at least gave people clear comparison and choices, assumptions needed when making those choices, and does that so in a very digestible way.