Hello,
I am doing Multilevel modelling where my dependent variable is income and my dependent variables are age, type of job (categorical) (fixed effect) and sex (categorical) (fixed effect).
I have been modelling in MLwiN and everything has gone pretty good. I fitted a quadratic function in level 2 and everything was perfect.
My confusion occurred when I modelled level 1 variance. I modelled the variance and there is a clear heterocedasticity, both coefficients are significant. But, in that process my level 2 variance becomes none significant and my standard error increases a lot.
My only explanation is that the whole majority of the variance is within districts (level 1) instead of between districts (level 2). Therefore there is a lot of confounding variance across levels but the majority is in level 1.
Any suggestions for this?
Thank you very much!
I think your argument is entirely plausible. Indeed it is probably best to model level 1 complex heterogeneity first before examining level 2 - the results would not then be a paradox.
Given income has a sort of compound interest element to it - people get paid 5% more - not the absolute $5k more- it is often a good idea to model the loge of income and then that often simplifies the variance at level 1. You can then expo the results to get back to the original scale.
Dear Professor Jones,
Thank you very much for your answer. Then I will go the other way around and start from level 1 complex heterogeneity.
If that does not solve the problem I will try to model the log of income.
Thank you very much for your help!
Another plausible explanation may be the fact that you increase the estimates of the total variance, resulting in higher Standard Errors. I would recommend to evaluate both models (with and without level 2 variance) on fit as well, by making use of AIC or BIC.
- Badges
- Science topic
- Similar topics
- Mathematics