If you're doing unpaired tests, the order shouldn't matter. You should just name the subjects according to your study protocol (i.e. pseudorandomized numbers, or whatever). Renaming subject ID's can get very messy.
________________________________________
From: FSL - FMRIB's Software Library [[log in to unmask]] on behalf of Jason S. Lee [[log in to unmask]]
Sent: Thursday, February 05, 2015 11:57 PM
To: [log in to unmask]
Subject: [FSL] Naming subjects for TBSS
Dear Experts,
As tbss user guide explains, I listed all controls before all patients for TBSS analysis. I performed 2-sample unpaired t-test to compare 16 patients and 16 healthy controls after matching age and gender.
My question is that when I list controls and patients, do I also need to care about the order within one group in terms of matched-age and gender?
For example, let's say I match age in this way:
CON001 (33 yrs) - PAT003 (30 yrs)
CON002 (20 yrs) - PAT001 (19 yrs)
CON003 (24 yrs) - PAT002 (25 yrs)
...
When I list the subjects, can I list them not caring about the order within the group, like this?
CON001 (33 yrs)
CON002 (20 yrs)
CON003 (24 yrs)
...
PAT001 (19 yrs)
PAT002 (25 yrs)
PAT003 (30 yrs)
...
Or should I need to re-name patients for matching age in this way?
CON001 (33 yrs)
CON002 (20 yrs)
CON003 (24 yrs)
...
PAT001 (30 yrs)
PAT002 (19 yrs)
PAT003 (25 yrs)
...
Recently, I am trying to verify my tbss results on my own. Any advice in this would be very appreciated in advance.
All the best,
Jason
|