CODEWITHSUNDEEP
rstatisticsregressionstata
Jnavero
Jnavero

Reputation: 363

Find variance in a Logistical Multilevel Analysis

I am trying to show the variance of different variables in a logistical multilevel analysis. I just find the variance in the second level (contextual).

In a lineal multilevel regression I have the residual variance that I apply the next formula:

variance = [σu(M0) - σu(M4)] / σu(M0)

But in a Logistic multilevel regression, I have not residual variance so,

How can I know what variance explains the individual level and the contextual level?.

Note: I have a variance from level 2 but stata don't give me the level 1 variance.

Upvotes: 0

Views: 143

Answers (1)

Erdne Htábrob
Erdne Htábrob

Reputation: 879

This is a statistical question, not a programming one. But: what you are looking for is the intraclass correlation or icc.

The way to get this measure is to run the empty model in stata, and then type estat icc. For example, if you have a three level model you would do: 

melogit depvar ||third_level_cluster: || second_level_cluster:
estat icc

The output of estat icc tells you what percentage of the variance in the dependent variable is "on the second level" and the "third level". The difference to 100 is the first level variance. Usually if ICC is lower than .1, that is 10%, don't expect independent variables to explain much on the respective level.

Upvotes: 2

Related Questions