In Excel 2003, LINEST (and some other stat functions) was improved. For
details, including why R2 differs in XL97 and XL2003, see
http://support.microsoft.com/kb/828533
You may wish to see how the chart trendline formulas compare.
best wishes
--
Bernard V Liengme
www.stfx.ca/people/bliengme
remove caps from email

"Philip J Smith" wrote in message
...
I use LINEST to derive a multiple regression formula in Excel 2003.

I have to share this with a client who uses Excel 97.

While the results for the gradients (m) are the same, the coefficient of
determination (r2) in Excel 97 is lower than that given in Excel 2003.
The
results are 0.87 and 0.97 respectively.

Does anyone know why the two versions give different answers? Which if
any
is the more reliable calculation?

Regards