three-level model with no level2 variance

Welcome to the forum for MLwiN users. Feel free to post your question about MLwiN software here. The Centre for Multilevel Modelling take no responsibility for the accuracy of these posts, we are unable to monitor them closely. Do go ahead and post your question and thank you in advance if you find the time to post any answers!

Remember to check out our extensive software FAQs which may answer your question: http://www.bristol.ac.uk/cmm/software/s ... port-faqs/
Post Reply
blois
Posts: 1
Joined: Mon Mar 14, 2011 9:39 pm

three-level model with no level2 variance

Post by blois »

Hello.

I am estimating a random intercept model using 3-level data. Level 1 is a mother, Level 2 is the measurement year and Level 3 is children. My primary interest in using multilevel modeling is to adjust standard errors for clustering (children within years, within mothers).

I have a large number (over 2,000) of mothers and children, but only four measurement years. In preliminary analyses, I estimated a simple variance components model. I found that Level 2 variance was close to 0 probably reflecting the fact that I have only four measurement occurrences at this level.

Does someone have any suggestions as to how to approach such a situation where there is virtually no variance at Level 2, yet I need to account for the three-level clustering?

Would thinking of the data as 2-level data (L1: mother and L2:children), while controlling for the year of observation be a good alternative? This seems to violate the fact that the data are fundamentally organized as 3-level data.

Would sticking with the three-level model and reporting the 0 Level 2 variance be acceptable?better?

Any references to theoretical or empirical examples dealing with similar situations would be greatly appreciated.

Thanks,
CNM
Post Reply