You don't *have* to include TIV. It depends on what you are looking at, your hypotheses, etc.
See O'Brien et al. 2006.
But yes, if you want TIV as a covariate you have to include it yourself. SPM doesn't automatically do it.
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Ben Becker [[log in to unmask]]
Sent: Friday, February 04, 2011 3:08 PM
To: [log in to unmask]
Subject: [SPM] DARTEL-VBM: total intracranial volume in second level analysis
Iíve computed a VBM analysis using the DARTEL-toolbox to investigate GM density differences in SPM8. The aim of the study was to investigate differences in GM density in two groups of patients. I set up the analysis according to the SPM8 manual (chapter 26 Dartel tools) using unmodulated GM images of the patients.
Do I have to incorporate the individual total intracranial volume in the second-level between-group comparison as a covariate or does the analysis already account for this nuisance variable? So is it possible to analyze the date using an unpaired t-test or do I have to use the total intracranial volume as a covariate in the second-level analysis?
Thanks and kind regards