LLN, le 29/05/07
Dear Anna,
>compare whether the activations from C1 and C4 are significantly
>different, which I presume I should do with a paired-t test, since
>each of the 10 subjects has a pair of .con files that will go into
>the analysis, one fromC1 and one from C4.
Do you mean something like [C1-Baseline]-[C4-Baseline]?
Mathematically speaking, this boils down to [C1-C4], which you should
compute individually at the first level and simply enter in a
second-level 1-sample t-test with a contrast set at [1].
Another option is to enter your 4 individual contrasts in a "Basic
Models"/"One way ANOVA", with the following specifications:
Grand mean scaling [No]
Threshold masking [none]
explicitly mask images [No]
Global calculation [omit]
non-sphericity correction [yes]
replications are over?... [repl(...)]
correlated repeated measures [yes]
The "yes" to the "non-sphericity?" question allows for inhomogeniety
of variance (ie 3 separate hyper-parameters over the "block diagonal"
terms of the error covariance matrix).
The "yes" to the "repeated measures?" question allows for
inhomogeniety of covariances (ie 3 further hyper-parameters over the
three "off-diagonal" block terms of the error covariance matrix).
The "repl(...)" answer to the "replications are over?" question tells
SPM2 that the replications come from the (implicit) factor of
"replications", which in this case corresponds to subjects (and the
other factor, group (=basis function), is the repeated measure).
(I quote Will Penny for this second option - but blame it on me if
this suggestion is wrong! I must confess that I don't see if this is
better/worse than your proposal to use paired t-tests. I would say
"better", but I'm not totally sure...)
Hope this helps,
Mauro.
>Hello everyone,
>
>A really simple question from an SPM newbie about running paired
>t-tests in SPM5 - I would love it if a kind person could help out, I
>have just spent 2 hours searching online and can't find the answer
>anywhere!
>
>Background: I'm analysing data from a study with 10 subjects, and
>four experimental conditions. For each condition I've calculated a
>contrast minus the baseline (C1-4) at the single subject level, and
>now I'm running second-level analyses to allow for population
>inference. I want to compare whether the activations from C1 and C4
>are significantly different, which I presume I should do with a
>paired-t test, since each of the 10 subjects has a pair of .con
>files that will go into the analysis, one fromC1 and one from C4.
>
>I'm stumped by one thing in the paired t-test tree though... the
>Independence option (default "Yes"). Normally one uses a paired
>t-test because the two observations in each pair are NOT
>independent, this is why it is often also called a "dependent
>t-test". So based on this prior knowledge, my instinct is to set
>Independence to "No", because each pair of observations is a
>repeated measure for that subject.
>
>However... is independence referring to something else here? SPM
>help says "By default the measurements are assumed to be independent
>between levels". I wondered if "measurements" and "levels" refers to
>the pairs themselves, as opposed to the items within the pair? In
>this case I should leave independence on "Yes", because each pair
>represents a different subject, and I can assume independence
>between them.
>
>Thanks in advance for any help!!
>
>Cheers,
>Anna Wilson
>
>*****************************************************
>Anna J. Wilson, Ph.D.
>Honorary Research Fellow, Department of Psychology
>University of Auckland
>
--
_____________________________________
Help fighting hunger: http://www.hungersite.com
Just click your mouse and sponsors of The Hunger Site donate a
serving of food to a person in need - at no cost to you.
______________________________________
Mauro PESENTI
Research Associate, National Fund for Scientific Research (Belgium)
Unite de Neurosciences Cognitives
Departement de Psychologie
Universite Catholique de Louvain
Place Cardinal Mercier, 10
B-1348 Louvain-la-Neuve
tel.: +32 (0)10 47 88 22
fax: +32 (0)10 47 37 74
E-mail: [log in to unmask]
http://www.nesc.ucl.ac.be
http://www.nesc.ucl.ac.be/mp/pesentiHomepage.htm
|