A lesson indeed Peter........

Every time I come to these groups for an answer, I usually go away with more
than I was smart enough to ask for in the first place......and even extra
stuff that I can use on the next project. Obviously I know nothing about
statistics, in this instance I was only trying to help a Chemist friend with
his Excel charting problems.....(hence my insistance that I only wanted to
access the label formula) but I certainly learned a lot in the process.

Again, I want to thank you for all your time and efforts in my behalf.

Vaya con Dios,
Chuck, CABGx3


"Peter T" wrote:

Tom - thanks for the link. Not exactly bed time reading <g but I've tried
to get to grips with it.

Ron - thanks also for your followup. As you say there is some detail of
LINEST problems in this ng. Presumably all statisticians are fully aware I
but don't suppose it does any harm to have the issue raised from time to
time.

Chuck - did a quick test in XL97 and get same results as in XL2K. Don't
think you should have any problems with LINEST with the particular data set
you have.

Clearly though there may be some sets that error with pre XL2003 and, from
what I interpret, how the function is used. In some of these cases applying
common sense might indicate a problem, such as comparing calculated results
with what one might expect.

On which point neither you nor I did with the early results, eg test the
formula on known x's & y's. (which I did by prior to my second followup to
you).

Your client's original request to go with the simplified formula, at first
glance, seemed very reasonable. It gave coefficients down to very small
albeit rounded decimals. He probably only wanted a result to the nearest
whole number, even that as a rough projection. Might well have expected the
simplified formula to be good enough. But it gave a very wrong result - a
lesson!

Regards,
Peter T