LOST-STATS/lost-stats.github.io

Multilevel Models

pmeiners opened this issue · 4 comments

pmeiners commented

I couldn't find any mention of general multilevel models on the Desired Nonexistent Pages page.
Did I miss the right page, or is there a specific reasons against providing a basic explanation of multilevel models?

NickCH-K commented

There's one mention there I think, for hierarchical Bayes for multinomial logit, but in general I think just nobody thought to add it! Definitely feel free to make multilevel model pages, or add it to the Desired Nonexistent Pages page.

grantmcdermott commented

There's also this page.

TBH though, I've been wondering whether we should restructure the page/index nesting for this part of the site to make the sub categories clearer. (Like we did for Nonstandard errors.) @NickCH-K what do you think about nesting like:

  • Linear models
    • OLS
    • Fixed effects
    • etc
  • Generalised models
    • logit
    • etc.
  • Multilevel models
    • Random effects
    • Bayesian hierarchical models
    • etc
NickCH-K commented

Breaking it up makes sense to me! There was naturally going to have to be a point where we'd have to break it up as we got more pages.

Figuring out hte categories is a bit tougher, as whatever we pick there's likely to be some overlap. For example, many multilevel models are linear, so I'm not sure "Linear" is the right word for OLS/FE. But maybe Ordinary Least Squares (FE would count I think) / Generalized Linear Models / Random-Coefficient Models / Multinomial Models /(Other?)?

grantmcdermott commented

Yeah, overlap cases are unfortunate, but I think the pros of nesting outweigh the negatives. Let me open up a new issue where we can brainstorm categories and then I'll try to implement soon.