home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!asuvax!asuacad!atg2g
- Organization: Arizona State University
- Date: Saturday, 9 Jan 1993 13:56:00 MST
- From: <ATG2G@ASUACAD.BITNET>
- Message-ID: <93009.135601ATG2G@ASUACAD.BITNET>
- Newsgroups: bit.listserv.stat-l
- Subject: Re: Roundoff of various formulae in spreadsheets
- References: <9301081401.AA01015@cbis.ECE.Drexel.EDU>
- Lines: 20
-
- As a matter of curiosity (since I don't deal in ppm scales) and to
- extend the "platform" question, I tried to replicate the STDEV results
- on an SE running Excel 4.0, no FPU. I found discrepancies beginning
- roughly at the point that the interquartile range falls below
- one 10-millionth of the mean. Apparently the precision has tightened
- up a bit (no pun intended, but there it is) vis-a-vis 3.0, but the
- formula is as before.
-
- There is a LINEST function in 4.0 that returns multiple regression
- statistics very fast and very accurately for most data, but I was
- able to make it return an error value using data that other Excel 4.0
- functions (determinant, matrix mult, etc.) are able to handle.
-
- Unrepentant, I continue to trust Excel, for statistical purposes,
- neither more nor less than anything else that comes shrink-wrapped.
-
- (Flame deflectors on.)
-
- Conrad Carlberg
-
-