control for clustering?

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

control for clustering?

Postby GerineL » Tue Oct 01, 2013 9:17 am

I collected data in aproximately 60 different classrooms.
i want to control for this with my analyses.

In mplus this is easily done by adding:
variables: cluster = ClassNo; [CLASSNO is a variable that indicates classroom in my dataset].
analysis: TYPE=complex;

Can this be achieved in spss also?
Posts: 2760
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: control for clustering?

Postby statman » Tue Oct 01, 2013 12:13 pm

Covariance Analysis as well as Complex Sampling allows for this if I understand your question but exactly what are you trying to accomplish and also, see the Note below
See the note below

NOTE: Please read the Posting Guidelines and always tell us your OS, the SPSS version and information about your study and data!

Statistical Services
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: control for clustering?

Postby GerineL » Wed Oct 02, 2013 8:41 am

I use spss 20.

I have data of approximately 1200 students, collected in about 60 different classrooms.
I want to do different kinds of analyses, but to keep things simple, let's say I have only scale variables and I want to do some correlations and some multiple regressions.
I want to predict variabel Y (which is loneliness).
However, students from the same classrooms may share some common attitudes or behaviours (i.e., clustering) that may account for some of the variance in loneliness. This clustering should therefore be taken into account when doing analyses.

So basically, the question is, how can I do correlations / regressions while taking the fact into account that I don't have truly individual data but rather individual data nested within clusters.

In Mplus this is easily done, by just adding some lines to the syntax. It would be great if this could be done in spss also, by some sort of pre-setting that can just be added to the syntax, rather than a complete new set of analyses.
But I am open to any suggestion.

Who is online

Users browsing this forum: No registered users and 1 guest