One sample t-test and significance

Submitted by David on

Hi,

I have a group of 90 test subjects and I calculated FC for the left Amygdala. I used the DPABI statistical toolbox and calculated a one-sample t-test for the FC zmaps for the ROI with base 0. The results show very high t-values up to 40 in the seed region itself and up to 20 in other regions (see attached picture). I cannot correct for multiple comparison because any voxel with a t-value that high will be significant even when using FDR. I'm not sure if this is correct or if I did something wrong?

Any helpful comments appreciated

YAN Chao-Gan

Sat, 10/03/2015 - 13:08

Hi David,

The high t values of one-sample t-test is normal, especially if you didn't do global signal regression.

You can set a high threshold for one-sample t-test, such as p<0.001. One sample t-test maps are only used for visual inspection.

For two-sample t-test, you can set the normal threshold at p<0.05 corrected.

Here is an example on ALFF results: Wang, Z., Yan, C., Zhao, C., Qi, Z., Zhou, W., Lu, J., He, Y., Li, K., 2011. Spatial patterns of intrinsic brain activity in mild cognitive impairment and alzheimer's disease: A resting-state functional MRI study. Hum Brain Mapp 32, 1720-1740.

Best,

Chao-Gan

Attachment Size
onesamplettest.png 149.61 KB