For the t value itself, which depends on the size of the sample and on the
confidence interval, you can enter typical values into a 'table' and make a
lookup once you do the computation. As example, for a 100(1-alpha) percent
confidence interval, not knowing the standard deviation of the two samples,
but known to be the same, the difference of the two populations means,
AVG(x) and AVG(y), is given by:

AVG(x) - AVG(y)       +/-      t (alpha/2, COUNT(x) + COUNT(y) -2 )   *   Sp
* (  1/COUNT(x)  + 1/COUNT(y) )  ^ 0.5

where t is to be read from the table you would have stored, and

Sp=  ((  (COUNT(x)-1) * StDEV(x)  + (COUNT(y)-1) *
tDEV(y)  )/((COU&NT(x)+COUNT(y)-2)) ^0.5

(ref:Statistical Methods for business and economics, by Pfaffenberger and
Patterson, at Richard D.Irwin inc.
http://www.amazon.com/Statistical-Methods-quantitative-analysis-business/dp/0256
036640/ref=sr_1_1/002-7842890-3627202?ie=UTF8&s=books&qid=1183988121&sr=1-1)

Of course, you have to check that the assumptions are right, in your case.
Basically, it is just as if you had to do it by 'hand', with only a t-table
at your disposal. That t-table is what will be entered as a database table,
to supply the t value. Alternatively, you can still retrieve that t-value
from any other package at your disposal, through a VBA function.

Vanderghast, Access MVP