The logical dependence in the outcome would seem to rule out many standard methods for panel causal inference. In other contexts one would fit a survival model, but that sort of thing isn't too common in panel CI. Thoughts? Maybe PanelMatch would work because it matches on lagged Y?

Re methods research, Stephen Jessee and @jerzakconnor.bsky.social have a working paper on dealing with measurement error in latent variables. It too deprecates MOC and proposes an IV-based approach instead.

Put differently: even the best scholars produce bad work, and even more often produce work that is not a good fit for a given outlet. A paper is not a person.

Yeah I hear you on me(). MO is faster. The advantage of both over MOC is allowing other variables in the analysis model to inform estimates of the missing values rather than assuming them to be independent as MOC does (see the appendix to Treier and Jackman).

It’s true! Though I have come to the conclusion that MOC is more assumption-dependent than is commonly realized, and it is often better to use @mattblackwell.bsky.social et al.’s “multiple overimputation” or to use Bayesian measurement-error model, which brms makes pretty easy.

I am Gen X (barely) and have said “on accident” as long as I can remember. Didn’t realize it was nonstandard until I was made fun of for it it as an adult.

Theoretically either is sufficient to drive the approximate bias to zero, but given that this is impossible in practice, it’s definitely better to do both. Also, doing it for the outcome has the additional advantage of reducing variance.

For a theoretical treatment see Sarndal and Lundstrom’s 2006 textbook, which shows that the approximate bias under calibration is zero only if the outcome or inverse response probabilities are a linear function of the auxiliary vector. In raking the aux vec is the marginal dist; in PS the joint.

I mean short “I”

On my first listen I was pretty sure that it was just an aborted mispronunciation of “migrants” with a short a (as in “immigration”), but after a second listen it sure does sound a lot like something else.

If you’re looking for more inspiration, you can check out my “political science scope and methods” syllabus, which has a few sessions on philosophy of (social) science: devincaughey.github.io/files/syllab....

For general overviews, I like Godfrey-Smith’s textbook and the Cartwright and Montuschi Philosophy of Social Science edited volume.

Looks great! And just in time to make my syllabus for the fall.

I’m honestly a little surprised bc my impression was that speed improvements from within-chain parallelization were nonmonotonic in the number of cores due to the fixed costs of each additional core, and 64 is way beyond the
point where I thought performance declined. But maybe that’s outdated?

64!