Centering variables in multilevel model

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

Zwenner
Posts: 2
Joined: Tue Jun 05, 2012 6:16 pm

Centering variables in multilevel model

Postby Zwenner » Wed Apr 03, 2013 3:48 pm

Hello everyone,

I am currently working on the centering of my predictor variables to conduct a multilevel analysis. I found the respective syntax to center variables at the grand mean, and to center them at the group mean (in my case, organisational ID or "orgid"). Here are some syntax examples, one for CGM and one for CWC:

Centering at the grand mean:
-------------------------------
compute allmean=1.
execute.

AGGREGATE
/OUTFILE=* MODE=ADDVARIABLES
/BREAK=allmean
/rolecla_mean=MEAN(rolecla).

compute rolecla_cgm=rolecla-rolecla_mean.
execute.
-------------------------------

Centering at the group mean:
-------------------------------
AGGREGATE
/OUTFILE=* MODE=ADDVARIABLES
/BREAK=orgid
/rolecla_mean=MEAN(rolecla).

compute rolecla_cwc=rolecla-rolecla_mean.
execute.
-------------------------------

Okay, so far, so good. While most of the predictor variables I use are on the individual level, some predictors are organisational-level variables that have been measured at the individual level, for example "climate for quality". I have already calculated ICC(1) and ICC(2) to justify the aggregation to the organisational level - but what about centering? I know that it only makes sense to center at the grand mean at the organisational level in this case, so will I have to calculate the mean of the means per organisation? If this is the case, would my syntax look like this (for quality)?
-------------------------------
compute allmean=1.
execute.

AGGREGATE
/OUTFILE=* MODE=ADDVARIABLES
/BREAK=allmean
/quality_mean2=MEAN(quality_mean).

compute quality_mean_cgm=quality_mean-quality_mean2.
execute.
------------------------------

I would appreciate any feedback and advise on this!
Kind regards,
Sven

Who is online

Users browsing this forum: No registered users and 1 guest

cron