When properly used (see posts by Mike Middleton and Jon Peltier) Excel's
polynomial trendline fit is more accurate than lm() in S-Plus (unless
you explicitly do a poly.transform of an orthogonal polynomial fit) and
most other dedicated statistics packages. Excel's statistical fitting
functions are mathematically exact, but implemented in a way that does
not minimize rounding errors with challenging data sets (the OP's data
set is not challenging). For univariate and bivariate statistics, there
are easy workarounds to handle even challenging data
http://groups.google.com/[email protected]
Alternately, Excel 2003 uses much improved algorithms.
Excel does not and probably never will have the statistical power and
flexibility of a dedicated statistics package, but that does not mean
that it is not appropriate for simple calculations such as the OP was
attempting.
Jerry
The statistical tools in Excel are not reliable.
Use a statistical package such as S-Plus or do the
calculations by hand if they are relatively small.
Josie
regression analysis using
trendline is used. Note
....
