If you have a three level model of students classes and countries - how are you calculating the ICC at the student level ? Or is this a case where it is really four level and below students there is a measurement level? ( Do you have multiple plausible values?)

Lots of multilevel resources at

http://www.bristol.ac.uk/cmm/software/mlwin/mlwin-resources.html

and here is some spss syntax fro a three-level model

http://www.ats.ucla.edu/stat/spss/code/three_level_model.htm

and a 'course'

http://www.ats.ucla.edu/stat/spss/library/spssmixed/mixed/Mixed_table.htm

and a little bit of digging found the Pisa SPSS volume 470 pages free to download 

http://www.oecd-ilibrary.org/docserver/download/9809031e.pdf?expires=1454781820&id=id&accname=guest&checksum=463F8BA42C3AE8053530E6A9BAA96E8A

Was this of any use>?

I find it extremely disconcerting when some one asks a question - someone replies (me in this case) and then the trail disappears................ Anyway I am half way through the PISA SPSS book and I am very impressed with it.  It discusses the analysis of the 5 plausible values in detail and gives a lot of advice and code for using weights and replicates.

Hi Kelvyn,

Thank you very much for your answer, it was indeed very useful. 

I ended up running the models in MLWin, and I was able to do it thanks to the excellent materials provided on the websites you suggested.

The PISA manual is also very good, but it does not cover several important details like the need to create unique school identifiers or changing the country IDs from alphabetic to numeric, etc.  -but this depends on what software you use... and as they say, the devil is in the details!

I'm attaching my SPSS syntax to prepare PISA 2012 data for three-level models in case it is useful for someone. This should work for MLWin and MPlus. 

Also, please receive my sincere apologies for not answering immediately. On the one hand, this is not the only project I'm currently working on; and on the other, I don't like to reply saying only "thanks" because replies to questions are also used to calculate your RG score and I don't think it is fair to get "points" just for saying 'thank you'. Anyway, I could, of course, sent you a private message!

That's useful - thanks

Kelvyn