How to find the significant clusters when there is only one condition?

💬 Support & Discussions
1
  • MNE-Python version: 0.24.dev0
  • operating system: linux

Hi! guys,
Here is another question:
I want to find the significant cluster of the 2-D or 1-D decoding performance of one condition, one group of subjects. I used the function permutation_cluster_1samp_test(), but I’m not sure whether this is the right move because I see this in the statements of this function which I quoted here:

Because a 1-sample t-test on the difference in observations is mathematically equivalent to a paired t-test, internally this function computes a 1-sample t-test (by default) and uses sign flipping (always) to perform permutations. This might not be suitable for the case where there is truly a single observation under test; see Statistical inference.

In my case, I only have one condition instead of two, so the input data is not the difference of the two conditions, but the signal of one condition. I don’t know what method or function should I use if permutation_cluster_1samp_test() cannot be applied in my case? Any suggestions?

Thank you a lot! :heart: :heart: :heart:
Best

2

Hi @YuZhou,
I think this is fine as long as you make sure that random decoding performance is close to 0 (otherwise the sign flipping may give you incorrect results). You should just be aware that sign-flipping permutations have additional assumptions (compared to the standard between subjects permutation test) - notably that the null distribution is symmetric. Studies show that its control of type I errors is only approximate and may be too conservative or too liberal in some cases (see https://www.pnas.org/content/113/28/7900 or https://www.biorxiv.org/content/biorxiv/early/2020/01/09/685560.full.pdf).

3 Likes
3

Thank you! Mikolaj, I’m aware.
Best