<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#333300" bgcolor="#FFFFFF">
Dear Eric,<br>
Thank you for responding.<br>
<blockquote type="cite">
<pre wrap="">Performing a second level t-test (in your case, between different groups of
subjects) on first level t-tests is very unusual. In fact, this would imply
that you second level t-test is about an hypothesis that pertains to first
level t-tests.
</pre>
</blockquote>
<br>
The intention was that all t-values from the stim vs. baseline will
be entered into the 2nd level t-test, not only the ones that are
significant.<br>
But either way I guess the inference regarding the implication of
the second level t-test on a first-level t-test hypothesis still
holds..? <br>
<br>
<blockquote type="cite">
<pre wrap="">Instead, one always performs a second level t-test on
first-level averages (typically baseline-normalized to deal with individual
differences in distance to the helmet or baseline power). Differences in
number of trials per conditions is never a problem as long as your
first-level measure (the baseline-normalized mean) is unbiased.</pre>
</blockquote>
Good to know.<br>
<br>
Thanks again,<br>
May<br>
<br>
<br>
<div class="moz-cite-prefix">On 16/10/2013 10:19,
<a class="moz-txt-link-abbreviated" href="mailto:fieldtrip-request@science.ru.nl">fieldtrip-request@science.ru.nl</a> wrote:<br>
</div>
<blockquote
cite="mid:mailman.363.1381915193.1015.fieldtrip@science.ru.nl"
type="cite">
<pre wrap="">
----------------------------------------------------------------------
Message: 1
Date: Wed, 16 Oct 2013 06:58:00 +0200 (CEST)
From: "Eric Maris" <a class="moz-txt-link-rfc2396E" href="mailto:e.maris@psych.ru.nl"><e.maris@psych.ru.nl></a>
To: "'FieldTrip discussion list'" <a class="moz-txt-link-rfc2396E" href="mailto:fieldtrip@science.ru.nl"><fieldtrip@science.ru.nl></a>
Subject: Re: [FieldTrip] Questions relating to MEG nonparametric
testing and uneven trial numbers
Message-ID: <a class="moz-txt-link-rfc2396E" href="mailto:005d01ceca2c$4add4530$e097cf90$@maris@psych.ru.nl"><005d01ceca2c$4add4530$e097cf90$@maris@psych.ru.nl></a>
Content-Type: text/plain; charset="utf-8"
Dear May,
2 subject groups: Controls (C) and Patients (P).
For a particular experimental condition (A), on average, subjects within
each group have the following number of trials:
condA_Ntrials_C ~= 100
condA_Ntrials_P ~= 80
I plan to perform a 2-step statistics-test to test for differences between C
and P groups on a particular experimental condition.
Using a 1st level within subject t-test (or active-vs-baseline test) on
baseline vs active period for each group, deriving subject-specific t-stats
which are to be subsequently used in a 2nd level test between subject group
contrast, using nonparametric test.
Question 1) Does it matter if the average trials per subject in each group
is different? Should I try to equalize the trial numbers by e.g. random
removal of some trials before computing any statistics? Is this advisable?
Performing a second level t-test (in your case, between different groups of
subjects) on first level t-tests is very unusual. In fact, this would imply
that you second level t-test is about an hypothesis that pertains to first
level t-tests. Instead, one always performs a second level t-test on
first-level averages (typically baseline-normalized to deal with individual
differences in distance to the helmet or baseline power). Differences in
number of trials per conditions is never a problem as long as your
first-level measure (the baseline-normalized mean) is unbiased.
Question 2) Minimum trials required for source-level significance?
The supplementary info. accompanying the 2007 paper illustrated some minimum
trial numbers required for obtaining a significant effect at the
sensor-level analysis, given some threshold (e.g. cluster>250 sensor-time
pairs).
Has anyone systematically shown what is the minimum number of trials
required to obtain significant clusters/fdr stats at the source-level
analysis?
If not, is there a sensible way to find out? (presumably involving some form
of bootstrapping and simulation?)
There is no sensible way to find out this minimum number of trials, because
the true effect size is unknown.
This relates to another issue I have with regards to 'within-subjects'
comparison for 2 experimental conditions A and B where there are average
trial number differences both within and between subject groups:
condA_Ntrials_C ~= 100 condB_Ntrials_C ~= 60 (min. =35)
condA_Ntrials_P ~= 80 condB_Ntrials_P ~= 50 (min. =35)
A similar concern of uneven trials arises if say I wish to perform a within
subjects comparison between the experimental conditions.
I think I?ve dealt with this question.
In general, Q: Would the intended 2-step statistical test (as described
above) be appropriate, OR would it best to control for 'equal' number of
trials for all subjects and conditions of interest?
I think I?ve answered this.
Best,
Eric Maris
I would really appreciate it if someone could kindly comment or offer advise
where appropriate.
Thank you very much in advance for your time.
Yours sincerely,
May
</pre>
</blockquote>
<br>
</body>
</html>