Emil O W Kirkegaard

Yeah, one thing I liked about this study is that it didn't use the typical big5 scoring but instead maximized variance in that particular dataset (which is much larger than the data used to define big5).

More generally I think a reconstruction loss over questionnaire data is a bit silly. Should be a prediction loss over interesting outcomes. Preferably data is observational instead of survey as well.

Well, psychologists are big fans of theory, and also big fans of whatever data one can get from slave labor, ie by forcing college students to take surveys while promising them it teaches them Something Important about psychology (that you need course credit to finish a degree).

What’s wrong with theory bruh?

Theory is for people who can't code.

That is insolence!

You can fit a covariance matrix to any data bro. Remember the dictum “All models are wrong but some are more useful than others.” Lols