0
$\begingroup$

For the data structure, we are fitting the multilevel model to the data. Before fitting the model, we are eager to do bivariate analysis so that we can keep those independent variables in the multilevel model which has an association with the dependent variable. My question is that since in the multilevel model, there is a group variable where the data are nested within each group, do I need to create different bivariate table for each group? Or, will I create single bivariate table where group is an independent variable?

That is, for example, I have $x_1, x_2, x_3, x_4, x_5$ and group variables as independent variables. Here, group has three categories. If before fitting the multilevel model, I want to do bivariate analysis, will I do that analysis separately for thee categories of group variable, or there will be only one unified bivariate analysis where I will check association between group variable and dependent variable too?

Improve this question
$\endgroup$
1
  • 2
    $\begingroup$ Are you planning on using information from a bivariate analysis to decide which variables to include in a multivariable regression? That's a very bad idea. You need to use theory and other relevant research to decide which variables to include in the multivariable model. I'd suggest that you spend some time drawing out a directed acyclic graph of the causal relation between your key predictor and your outcome before moving forward with the analysis. See stats.stackexchange.com/a/445606/87305 $\endgroup$
    – Erik Ruzek
    Commented Feb 3 at 20:48

0

Reset to default

Your Answer

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