Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Cross-classification question

From   Cameron McIntosh <>
Subject   RE: st: Cross-classification question
Date   Mon, 9 Jan 2012 23:10:20 -0500

Hi Sarah,
You might get more feedback on this type of question on the Multilevel discussion list: ;

It's true that the data you describe don't fit neatly into the usual cross-classified modeling scenarios. The closest example I have sometimes seen used is a study of intellectual growth where the students change classrooms over time... within your model the cross-classification by students and classrooms is contemporaneous but perhaps similar methods might be useful. I think you are generally on the right track w.r.t. the random effect modeling you suggest, but the following papers may give you some additional guidance:

Luo, W., & Kwok, O.-M. (2011). The Consequences of Ignoring Individuals' Mobility in Multilevel Growth Models: A Monte Carlo Study. Journal of Educational and Behaviorial Statistics, Online First.

Fielding, A., & Goldstein, H. (2006). Cross-classified and Multiple Membership Structures in Multilevel Models: An Introduction and Review. Research Report RR791. Birmingham, UK: Department  for Education and Skills, University of Birmingham.

Goldstein, H., Burgess, S., & McConnell, B. (2007). Modelling the effect of pupil mobility on school differences in educational achievement. Journal of the Royal Statistical Society, Series A, 170, 941-954.

Leckie, G. (2009). The complexity of school and neighbourhood effects and movements of pupils on school differences in models of educational achievement. Journal of the Royal Statistical Society, Series A, 172(3), 537-554.

Chung, H. & Beretvas, S.N. (2011). The impact of ignoring multiple membership data structures in multilevel models. British Journal of Mathematical and Statistical Psychology, Early View.

Hope this helps,


> Date: Mon, 9 Jan 2012 15:02:51 -0500
> Subject: st: Cross-classification question
> From:
> To:
> Hello All,
> I am working on a project that I think requires a cross-classified approach.
> My data is from 1,000 students, surveyed in approximately 200
> classrooms in 1 school.  Students took the survey in at least one and
> often three different classrooms-- i.e. in Math, English, and Science.
>  The data set is organized such that each row is one students' survey
> responses in one classroom.  So, if a student took the survey three
> times, their responses would be in three different rows.  My dependent
> variable is engagement, which is continuous and specific to whatever
> classroom a student happened to to be in when they took the survey.
> My independent variables include several student characteristics as
> well as student perceptions of their teachers' classroom practices.
> I believe that my data is cross-classified but I am not sure.  Most
> students belong to more than one classroom in the entire data set,
> which makes me think that it is cross-classified data.  But within
> each observation every student belongs to just one classroom and my
> dependent variable is classroom-specific, which is contrary to every
> example of cross-classification that I have found.  Can any of you
> shed some light on this dilemma for me?
> Also, if my data is indeed cross-classified, it seems like I can deal
> with this issue by adding a random-effect for class subject to my
> model, since each student is only in one classroom for Math, one for
> English, etc.  If so, is this the correct stata syntax for the null
> model?
> xtmixed engagement || _all.R.subject || classid:, mle
> Or, is it:
> xtmixed engagement || _all.R.subject || _all.R.classid: , mle
> Thanks so much for your help!
> Best,
> Sarah
> *
> *   For searches and help try:
> *
> *
> *
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index