www.cmm.bristol.ac.uk/forum

Hello, I am having an issue which I am not sure is methodological or technical...anyway: I have estimated a linear 2 levels random slope model, having individuals at level 1 and countries at level 2.

I want to see how an economic indicator (X) affects a demographic indicator (Y) and I want to see how this relationship varies between Countries (level 2). So I will have random intercept and random slope on the economic indicator (X).

I am able to get the estimates for fixed and random part. And my question is: how can I know the estimates for each Country (which is my level 2) ? My final aim will be to have a measure of the relationship between Y and X for each country and to draw a graph having country level lines in different colours. Thanks!

I suspect that what you want is the level 2 residuals, see chapters 3 and 4 of the MLwiN User Guide (http://www.bris.ac.uk/cmm/software/mlwi ... al-web.pdf) for information on how to calculate and use these.

Thank you for your kind reply and suggestion. Yes, I was able to get level 2 residuals. but just wondering if there is a way to get slope and intercept at level 1 grouped for each country. My final aim would be to say something about the relationship between X and Y at Level 2 and to show their relationship (and possibly to have a measure of correlation) at level 2. Does it make sense? thanks again!

Chapter 4 of the MLwiN user guide goes through an example of calculating and graphing predictions for a similarly structured model. Page 57 shows the graph of a random intercepts model and page 64 shows the graph of a model where the slope is allowed to vary for each school. The model -> predict window allows you to easily include/exclude different parts of the model within your prediction. You might also want to read through Chapters 5 and 15, which examine various ways of exploring multilevel models graphically.

Thank you!!!!