about standard deviation in ttest

[mapi]

Hi,

I have a question about the calculation of standard deviation in ttest. From matlab help, ttest uses the sample
standard deviation which is the total squared error divided by (n-1).

From wiki, it says the grand standard deviation of sample is used (divided by n or n-1?). Are the two the same?

Thanks,

Mapi

 

[IRstuff]

Well, for a large population, it doesn't matter much, does it?

For small populations, the correct answer is n-1, since that's the degrees of freedom in the data.

But, basically, if you have more than 10 samples, the difference is pretty small compared to all the potential error sources, including the standard deviation of the mean.

TTFN

FAQ731-376

 

[mapi]

Thanks.

Just because my sample size is less than 5, I have to be very careful about it.

mapi

 

[IRstuff]

for 5, I would go with n-1, but the variance of the mean will be commensurate, so the standard deviation is going to be pretty rough, no matter what.

TTFN

FAQ731-376