Stan Brown -

I think my description of Excel's method is "accurate."

But there are several ways to determine percentiles. I was only trying to
describe Excel's method. Browse to www.google.com, and search for
"percentile," or take a look at Wikipedia's entry.

Personally, I prefer starting the cumulative relative frequencies at 1/(2*n)
for the minimum value and using steps of 1/n up to 1-(1/(2*n)) for the
maximum. (Justification: The minimum value represents the lower 1/n
cumulative relative frequencies, so I associate the minimum value with the
median cumulative relative frequency, i.e., 1/(2*n).)

- Mike Middleton
http://www.DecisionToolworks.com
Decision Analysis Add-ins for Excel



"Stan Brown" wrote in message
...
Mon, 21 Jan 2008 13:23:13 -0800 from Mike Middleton
:
Excel assigns cumulative relative frequency of zero to the minimum value
of
the data set and cumulative relative frequency of one to the maximum
value.

Is that accurate? I should have thought cumulative relative
frequency of the minimum number would be 1/N for a data set of N
items.

--
Stan Brown, Oak Road Systems, Tompkins County, New York, USA
http://OakRoadSystems.com/
"If there's one thing I know, it's men. I ought to: it's
been my life work." -- Marie Dressler, in /Dinner at Eight/