Method used for parameter constraints

[OliverTS]

I have been experimenting with creating models using sets of constraints for both the fixed and random effects parameters, but have encountered some difficulty in working out how exactly these are incorporated into the procedures used for parameter estimation in MLwiN (using IGLS). Would anyone be able to point me in the direction of any useful resources?

Do the methods used in MLwiN match those described in Goldstein's Multilevel Statistical Models - second edition (section 3.3)? as available at http://www.bristol.ac.uk/cmm/team/hg/msm-2nd-ed/

Many thanks

Oliver

[ChrisCharlton]

I have now checked this, and yes MLwiN does do the equivalent of the formulae that you referenced.

[OliverTS]

Ok great, many thanks for the confirmation.