Hi,
Why is the same?
I want to see if my results remain still covariating for age. So, my
hypothesis is that good outcome doesn't depend on the age of my subjects, it
depends on outcome.
So, if, as you say, is the same if covariate for age demeaned or not?
# In your case, it is the same to substract the average of all
subjects or not. However they are different from the case to substract
average for each group.
Why I should have to demean and write this values to GLM for each EVs?
# I think there is no need to demean only if
what you care is the difference of intercept. And It is computational
consuming after demeaning.
If I going to get the same result if I just don't demean and put the age
directly.....Could you explain me this, please?
# It depends on your
data and hypothesis and you can look up in the maillist.
I am just going to by Linux enterprise Red Hat because my university is
close and I can't connect to the remote server because it is off...and I am
going to call an informatic to help me installing this programme and intalling
fsl script (I don't know which is the command for doing so)....if I am going to
get the same results....it is useless and expensive.
By the other way, I separate my sample by age and I didn't get results in
any contrast, and this is very strange for me, althought my sample is very litle
actually ( only 14 subjects).
# I am not sure.
Please, give me an advice.
And many thanks for your helping, you are always helping me....do you know
fMRI well? I ask this because I am searching a group who can explain me all
about this (resting and with a task)....and I would like learning about DTI,
TBSS, VBM, HARDI and Graph Theory but it results difficult for me to find a good
group....
# These just are my understanding and I am learning on the way too.
# Waiting for corrections if there is some
mistakes.
2011/8/1 Xiangzhen Kong
<[log in to unmask]>
Hi, Rosalia.
It depends on your hypothesis.
In your case, I think you just
want to regress out the age, so you should subtract the mean of all subjects
from all group. However, it is the same whether you demean(in this way)
or not.
An understandable case:
Group1(subjects from US.) with (very) high income,
Group2(subjects from North Korea) with (very) low income.
Now you want to regress out the
effect of income(just like age in your case), how do you do?
Hope this helps.
2011-08-01
Xiangzhen Kong
发件人: Rosalia Dacosta Aguayo
发送时间: 2011-07-30
22:57:10
收件人: bnucon
抄送:
主题: Demaning data
Hi,
It was very useful and helping the web you recomended me last day about
ANOVA and the page of GLM and how to do a matrix. But now, I have a problem. I
want to demean my sample for age and, althouhgt I have been reading about this
term in FSL (de-meaning the data) and looking for presentations about how to
do....I don't realise how I have to do it.
If I have two groups with two differents means according to age....how I
have to do for demean the age for every subject in the two groups with these
two means.
Any good formula I can understand, not a criptogenic formula,
physics...engineering...is difficult for me to understaand and I suppose It
must be a very simple process I can do manually because my sample is very
little (only 14 subjects)....please, some example, clear example?
Do I have to sustract the mean of G1 of the real age of every subject in
G1 and write this results in EVs for group1 and the same for group2
?
Thank you !!!
Rosalia.