In the most recent year, about 1/5 articles in AJPS and JOP had a survey experiment. (preliminary result from work in progress with @vincentab.bsky.social and @jonmellon.bsky.social, come see more at SPSA)
Comments
Log in with your Bluesky account to leave a comment
Interesting! Do you have a sense of what they are replacing?
My (uninformed) impression is that articles used to include an unidentified cross-sectional regression analysis of some sort, but we've swept that out.
Extra survey experiments that at least estimate an estimand seems like progress!
Thanks! Not yet, but hopefully soon. This is the first small bit of a much larger project and we are presenting some preliminary results in early January so we need a draft by then
This looks interesting. Do you have more fine-grained data on what types of experiments make up these groups? For example, what share of survey experiments are conjoint experiments, etc.?
We're using LLMs and RAs to code very many features of a very large number of papers, so we have lots of talk about in terms of technical challenges and also interesting substantive results.
Comments
My (uninformed) impression is that articles used to include an unidentified cross-sectional regression analysis of some sort, but we've swept that out.
Extra survey experiments that at least estimate an estimand seems like progress!
I also don’t mourn the bad old days.