0
$\begingroup$

I have a question regarding the use of Multilevel Modeling (aka Hierarchical Linear Modeling) with Sequential Linear Modeling.

I am trying to perform Sequential Linear Modeling (with a binary outcome) but I also want to take into account two levels of the data. Specifically, I have data taken from multiple schools and I want to have a second level for school - which would require me to use Multilevel Modeling.

My question is, is there a way to perform both analyses at the same time?

The alternative I have thought of is to perform Sequential Linear Modeling and separately perform Multilevel Modeling and report there was (not) nesting effects of school.

$\endgroup$
4
  • $\begingroup$ What do you mean with Sequential Linear Modeling? $\endgroup$ Commented Dec 28, 2022 at 17:15
  • $\begingroup$ By Sequential Linear Modeling, I mean the predictors are entered sequentially or in steps/blocks to determine the information gained by adding another predictor or sets of predictors. $\endgroup$
    – Sam16
    Commented Dec 28, 2022 at 18:08
  • $\begingroup$ So that is stepwise methods? Then maybe add the tag stepwise-regression $\endgroup$ Commented Dec 28, 2022 at 18:45
  • $\begingroup$ Yes, and thank you I will add that now. $\endgroup$
    – Sam16
    Commented Dec 28, 2022 at 23:55

1 Answer 1

0
$\begingroup$

Stepwise modelling is a fairly simple approach, so I don't see why it couldn't be done. Simply select your criterion for calling a model "better", and implement the stepwise procedure using a hierarchical model.

The question shouldn't be "can it be done" but rather "should it be done" and the evidence would suggest the answer would be: NO. You're free to search this site for threads on stepwise selection to see why this is so problematic.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.