STUDENT T TEST

[petergod]

Anyone with a LabVIEW implementation of the Excel "test.student" statistical function ?

See attached Excel example.

I need to use the "unilateral" argument for the distribution as well as the type 2 (sample with equal variances), unlike in the Excel original example.

The computed value I'm after is also known as the "p value".

I guess this will seem pretty basic to many, but I have very little statistics background, so help is very welcome.

LabView 8.2, 8.6 or 10

[vt92]

Have a look at the Hypothesis VIs under the Mathematics pallete.

[vt92]

[Darin.K · Accepted solution]

The T test VI in the Hypothesis testing palette is for a single sample, but located inside that VI is the Student distribution CDF subVI which is very useful to construct the two sample test.  Here is an example for the one-tailed test for equal variances as well as pairwise test, you'll have to look up the formulas in any standard text.  It matches the Excel result, I always worry about using N or N-1 in various locations.

For the two-tailed result, just double the p-value.

[petergod]

Thanks for looking. I found that already but was lloking for a more complete implementation. Darin.K did a great job on this one - see rest of the thread. CU

[petergod]

Great, this is fast & efficient :-) Just revised it a little bit to completely mimic (I hope) the Excel 2003 implementation I'm after. Great help CU

[petergod]

Well in fact there is an issue, depending on the datasets. Either the pvalue is exactly the one returned by Excel, or it is completely different. I see no obvious reason. Any idea what's going on ? See attached XL file with examples & the same in LV 8.20

[Darin.K]

Seem to be missing attachments.

[petergod]

for some reason, attachements do not work today!? New trial

[Darin.K]

If you preview the message, the attachments will be stripped.  Otherwise, just post the data as text in a message along with the Excel result.