1. Yes, you should always correct for multiple comparisons.
2. See the attached document by Rik Henson. (I don't remember where I
found it).
Ben Becker wrote:
> Dear SPMers,
>
> we are planning to examine functional (potential compensatory) neural activity in one patient with a structure-specific lesion. As a control group we want to examine 10 matched controls.
> Because it’s the first time I’m conducting a single-case fMRI study, I would like to call for advice. My questions:
>
> 1. When interpreting the activity pattern of specific active conditions vs. baseline (first level analysis) from the lesion patient: is it necessary to FWE-correct for multiple comparisons?
>
> 2. How to statistically compare the one lesion patient with the controls on the second level:
>
> a) Compare the first level results from the lesion patient with the mean activity levels of the controls (summarizing the activity of the controls by means of a 1-sample t-test)?
> This would only allow interpreting differences on a descriptive level – right?
>
> b) To compare the patient and the controls on an inference-level statistic: would it be appropriate to compare the patient and the controls by means of the second-level random fx 2sample t-test?
>
> c) However, I think the second-level t-test might not be valid: what other analysis might lead to valid results? (Maybe you can recommend some studies / papers which used appropriate methods for comparing a single subject with a control group)
>
> Thanks in advance & kind regards,
>
> Ben
>
|