Dr Jennifer Marino
drjenmarino.bsky.social
Dr Jennifer Marino
@drjenmarino.bsky.social
Adolescent and women's health epidemiologist, feminist. They/them. Black lives matter. Trans lives matter. Sovereignty was never ceded.
Opinions expressed are solely my own and do not express the views or opinions of my employer.
(I do understand that your primary question is about having too many variables in your model relative to the number of observations, and the particular vulnerability of intx terms, neither of which is the topic of that paper - but picking through the paper's arguments and references might help)
December 11, 2025 at 11:01 PM
this might be helpful? it is about linear regression models specifically, but I think the overall argument about the assumed interaction structure holds. www.cambridge.org/core/journal...
How Much Should We Trust Estimates from Multiplicative Interaction Models? Simple Tools to Improve Empirical Practice | Political Analysis | Cambridge Core
How Much Should We Trust Estimates from Multiplicative Interaction Models? Simple Tools to Improve Empirical Practice - Volume 27 Issue 2
www.cambridge.org
December 11, 2025 at 10:54 PM
(I reckon they just (manually) copied the previous datapoint’s label and failed to update the content.)
November 29, 2025 at 1:52 AM
Yeah, I totally know where you’re coming from, and you know I know, so we are just laying it all out for drive-by-ers
November 28, 2025 at 2:25 AM
And of course, as with racism and economic anxiety - the genuineness of the underlying suffering doesn’t excuse its expression as hatred and epistemic injustice and all types of violence.
November 27, 2025 at 11:31 PM