I was wondering what the t-statistic means for if I only have one beta value for each channel. For example, if I just have one trial data for one subject included 10 repeatations, I could get the beta by GLM. I see many papers, they said the t-statistic is calculated by t = c'*beta/(c'*cov(beta)*c), but if the beta's dimension is 2*1 (2 is the number of regressors and 1 is a channel), what is the cov(beta) stands for?
![[2022.05.13] fNIRS Homer3 Analysis](http://i.ytimg.com/vi/bHhn2vBXF0Y/mqdefault.jpg)
![[2022.05.13] fNIRS Homer3 Analysis](/img/tr.png)
Very helpful. Thank you.
I was wondering what the t-statistic means for if I only have one beta value for each channel.
For example, if I just have one trial data for one subject included 10 repeatations, I could get the beta by GLM.
I see many papers, they said the t-statistic is calculated by t = c'*beta/(c'*cov(beta)*c), but if the beta's dimension is 2*1 (2 is the number of regressors and 1 is a channel), what is the cov(beta) stands for?
hats off to you :)