Excel Statistical Accuracy?

We have been told that Excel statistics calculations can be inaccurate.
While that may be true for precise leading edge scientific work is it also
true for college engineering students learning undergraduate statistical
analysis?

Thanks for any enlightenment,

WDA
ballensr@adelphia.net

end



----------------------------------------
I am using the free version of SPAMfighter for private users.
It has removed 496 spam emails to date.
Paying users do not have this message in their emails.
Try www.SPAMfighter.com for free now!

W. D. Allen Sr. wrote...
>We have been told that Excel statistics calculations can be inaccurate.
>While that may be true for precise leading edge scientific work is it also
>true for college engineering students learning undergraduate statistical
>analysis?

Excel can't really handle any leading edge statistical analysis (or
have I missed the generalized matrix inversion and
eigenvalue/eigenvector functions, stepwise regression and ARIMA time
series facilities?). Certain versions choke on simple statistical
calculations.

Excel 10 (2002) and prior choke on

=VAR(1E8+{0,1})

returning 0 rather than 0.5. Related to this, the regression functions
(LINEST, LOGEST, GROWTH, TREND) don't handle collinearity well. There
are issues with most of the inverse continuous distribution functions
prior to Excel 11 (2003), except for the normal distribution functions
which were fixed in Excel 10 (2002). There are still issues with the
discrete distributions returning #NUM! for values they should be able
to calculate. RAND wasn't all that random prior to *patched* Excel 11.

That said, if all you're doing is calculating univariate means and
standard deviations of relatively low magnitude values with no more
than 4 significant digits, Excel 95 and on are adequate. What stats
does your syllabus cover?