Dear Hernando,
At 10:02 AM 10/14/98 +0100, you wrote:
| From: Hernando Ombao <[log in to unmask]>
| To: John Ashburner <[log in to unmask]>
| Subject: voxel-level Z vs. uncorrected Z
|
| Dear John,
|
| Suppose the SPM output gives:
|
| cluster-level {k,Z} voxel-level {Z} uncorrected k and Z
| ------------------ -------------- -------------------
| 0.572 (5, 3.81) 0.487 (3.81) 0.280 0.0000
|
| I interpret these values to be:
|
| 1. The probability of getting, by chance, a cluster of size 5 or more and
| a maximum Z value of 3.81 or more in this cluster is 0.572.
Yes, accounting for the multiple comparisons problem of searching over the
entire common brain volume.
| 2. The probability of getting, by chance, a maximum value of 3.81 or more
| in this cluster is 0.487 (regardless of the size of the cluster).
Not quite: It's the probability of getting a maximum value of 3.81 or more
in the entire image (...and is therefore corrected for multiple comparisons).
| 3. The probability of getting, by chance, a cluster of size 5 or more is
| 0.281.
Yes, though it's conditional on there being a suprathreshold cluster in the
first place, and is thus not corrected for multiple comparisons.
| 4. The probability of getting, by chance, a Z value of 3.81 or more
| is 0.000.
Yes. Again, this is not corrected for multiple comparisons, and is obtained
by looking the Z-value up in statistical tables for the standard Normal
distribution.
| Am I interpreting these p-values correctly?
Pretty much...
| Thanks for the help!
No problem.
Hope this helps,
-andrew
+- Dr Andrew Holmes [log in to unmask]
| ___ __ __ Wellcome Department of Cognitive Neurology |
| ( _)( )( ) Functional Imaging Laboratory, Stats & |
| ) _) )( )(__ 12 Queen Square, Systems |
| (_) (__)(____) London. WC1N 3BG. England, UK |
+------------------------------------- http://www.fil.ion.ucl.ac.uk/ -+
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|