Calculating CI variables in multi level multiple regression models

[HHawleyHague]

Hi everyone
I have used MLwIN as part of a study to do linear and logistic multiple MLM and we calculated the CI for the variables manually using 1.96 calculation. However, I have since been advised by the statistician (who I worked with)- that he has identified that these are wrong as it does not consider the different levels (we hand calculated the p values as well and I did notice that some of the CI overlapped when the p-values were significant). I really want to amend this now I know, but we cannot establish how we calculate them.

STATA provides CI but calculates the models in a slightly different way.

are you able to give us any advice to do this from the MLWiN output (manually Is fine)?

I am not a statistician by any means but I can share your response with the statistician I have worked with if I do not understand!

Many Thanks
Helen

[GeorgeLeckie]

Hi Helen,

In the latest version of MLwiN, the Model Comparison window now gives you p-values and 95% confidence intervals for each regression coefficient if you tick the relevant check boxes.

Best wishes

George

[HHawleyHague]

Hi George
Thank you, that is excellent news! So all I have to do now is download the new software and enter my final models again.

Brilliant

Helen