I don't care for change scores for non-Gaussian data. But in addition to the ordinal change score models in Senn's textbook, some folks in sports analytics use the Skellam distribution to model change in Poisson-distributed goals.

Actually, now I think about it I do recall Senn entertained change score analyses with ordinal data in his crossover trial book. He's the only person I've ever see discuss that. I thought it was remarkably bold.

... even if the adults do sometimes make mistakes... [ducks and runs]

It's not the job of the children to change the opinions of the adults

Here are my thoughts on change scores. In short, I don't personally like them, but they can be perfectly valid in some contexts. solomonkurz.netlify.app/blog/2023-06...

👍

I'm not trying to be obtuse, here, but that still doesn't help me understand what your research goals are. Without that, I don't think there's a clear answer

The "only sensible default model" for what?

it's awful

'[There's a good chance] a strongly informative prior would pop out like a bright flashing light in peer review.'

There's also the president Bayesians have largely used very weak priors for the past several decades. A strongly informative prior would pop out like a bright flashing light in peer review

It also goes the other way if one wanted to qualify that this would only hold for "reasonable priors." Even among scientists, one person's "reasonable" is another person's "abomination."

Yeah, I wouldn't be so sure about that. At least not without first being very clear about what you mean by "reasonable amount of data."

I can empathize. Broadly speaking, if common sense were good enough, we wouldn't need science. Extend that to statistics. Without extensive training and practice, common sense will be a poor guide for understanding statistics. I wish it weren't, but that's just how it is for most of us.

Ah. Yeah, I can see it showing up in that context

No. I don't teach at the moment. Search committees are free to slide into my DMs, though

Or collaborate with others who have already worked through Bayesian textbooks.

Thanks for the shout out. If you want to start using Bayes in your professional projects, I recommend you work though a textbook.