"Jerry W. Lewis" wrote:
The example on p.481 of the TI-83 Plus/TI-83 Plus Silver
Edition Graphing Calculator Guidebook
http://education.ti.com/guidebooks/graphing/83p/83m$book-eng.pdf
is equivalent to Excel's
=TTEST({12.207,16.869,25.05,22.429,8.456,10.589},
{11.074,9.686,12.064,9.351,8.182,6.642},2,3)

If I had the raw data, I wouldn't be asking the question, now
would I?! The TI83 also has the option of entering just two
sample means and std devs. That is the __only__ case I want
to discusss.

Excel does not have a pre-programmed equivalent of TI's
Input=Stats mode

Okay, I'll take your word for it. I really am very surprised.

There does not seem to be an MS KB equivalent of p.603 of
the old Excel 5 User's Guide

Is that available online today? I don't know how to get it.

but formulas are widely available, for example
http://www.statsdirect.com/help/para...ethods/utt.htm

That seems to be the same formulas that are in the Excel
"about statistical tools" help page. And I realize now that they
would work for my case of a "u1 = u2" null hypothesis. I am
just (still) surprised that I must enter all the formulas. I would
think this is a not-uncommon requirement, albeit perhaps also
not necessarily the most common one either.

But according to my (remedial) stat book, the complete numerator
for the t-score is (m1 - m2) - (u1 - u2), where m1 and m2
are the sample means, and u1 and u2 are the population
means. That is the genesis of "delta" on the Excel help page.
Of course, we do not know u1 and u2. So it is unclear how I
would use that formula for a "u1 <= u2" null hypothesis.

Oh well, that question is better discussed in a stat newsgroup.
All I was asking here was what Excel function did the job. I
am hearing a resounding "none!" :-(.