NEUROPOWER: Incorrect output when requesting statistical power for a certain sample size

[karitaojala]

Dear Neuropowertools developers,

In the "Power calculation" tab of Neuropower, there seems to be an error where requesting the level of statistical power for a specific sample size (and multiple comparisons correction method) gives a clearly completely incorrect text output.

This same error can also be seen even in the Neuropower tutorial image:

Here, the software interface states that with N = 35 subjects, using Random Field Theory correction, the power is 96%. In contrast, the graph in the same figure shows that power should be close to 10%. At first sight, it seems that the software interface computes the power graphs correctly but is deriving the desired power from the wrong curve (Benjamini-Hochberg instead of RFT). However, testing different examples with our own data revealed that it might also be that the power value stated in text is consistent with the graphs when adding to the number of subjects in the enquiry (N = 35), the x-offset of the graph (in this case, around 25 subjects), i.e. the power value would be correct for N = 35 + 25 subjects.

It is of course also possible that the plotted curves themselves are wrong, which would be a more serious issue - although it seems more likely that simply the text output is incorrect.