Statistical questions

Submitted by nabila11 on

Hi experts,

I use REST toolbox to analyze  resting state data. My study is a placebo-controlled, 2-period crossover study with 16 subjects who participated in two visits (V1 and V2) in random order, where they underwent fMRI protocol. So there is a treatment group and a placebo group. I did the statistical analysis (paired t test) on the mREHO and mALFF to compare these maps between the two groups; the results do not reach significance (p corrected <0.05); So I wonder if I can use uncorrected p-value for my results, or is it too difficult to interpret results with uncorrected p-value ?

Thank you for your answer,

Best regards,

Nabila

YAN Chao-Gan

Thu, 01/07/2016 - 02:10

Hi Nabila,

It's very difficult to publish uncorrected results nowadays given its high probablity of false positive.

Best,

Chao-Gan

Hi Chao-Gan,

Thank you for your reply.

When i open my T-contrast with spm and use a p-FWE<0.05, I have almost nothing but when i open it with the slice viewer of REST, i have some results even when i enter 0.05 for p-value. I understood that it isn't the same way to calculate between the 2, but i don't know exactly why ? How is it possible to obtain 2 differents results from with same p-value ?

Thank you again,

Best,

Nabila

Hi,

If you click 0.05 on REST slice viewer panel, that means uncorrected p values.

You can perform AlphaSim or Gaussian Random Theory Correction, through "Cluster", either at REST Slice Viewer or DPABI_VIEW.

Best,

Chao-Gan