[R] Rounding problem R vs Excel

[Marc Schwartz]

On Wed, 2003-06-04 at 08:09, Paul, David A wrote:
> I don't have the reference, but a biologist friend of mine
> once showed me a refereed journal article that purported
> to demonstrate numerical errors made by MSExcel.  This 
> would have been Excel97 or Excel2000... In any case, the
> journal's scope was biological in nature and the article 
> was of interest since Excel is heavily used in that community.
> 
> -david paul


There is a series of articles here:

http://www.stat.uni-muenchen.de/~knuesel/elv/accuracy.html


In addition, there are additional references on Excel specifically:


On the accuracy of statistical procedures in Microsoft Excel 2000 and
Excel XP
B.D. McCullough and B. Wilson, (2002), Computational Statistics & Data
Analysis, 40, pp 713 - 721
http://www.elsevier.com/gej-ng/10/15/38/85/51/28/abstract.html


On the accuracy of statistical procedures in Microsoft Excel ‘97
B.D. McCullough and B. Wilson, (1999), Computational Statistics & Data
Analysis, 31, pp 27-37
http://www.elsevier.com/gej-ng/10/15/38/37/25/27/article.pdf


Problems with using Microsoft Excel for statistics
J.D. Cryer, (2001), presented at the Joint Statistical Meetings,
American Statistical Association, 2001, Atlanta Georgia
at http://www.stat.uiowa.edu/~jcryer/JSMTalk2001.pdf


Use of Excel for statistical analysis
Neil Cox, (2000), AgResearch Ruakura
at http://www.agresearch.cri.nz/Science/Statistics/exceluse1.htm


Using Excel for statistical data analysis
Eva Goldwater, (1999), Univ. of Massachusetts Office of Information
Technology
http://www.umass.edu/acco/statistics/handout/excel.html


Statistical analysis using Microsoft Excel 
Jeffrey Simonoff, (2002)
at http://www.stern.nyu.edu/~jsimonof/classes/1305/pdf/excelreg.pdf


Testing the Intrinsic Functions of Excel
National Physical Laboratory, UK
http://www.npl.co.uk/ssfm/ssfm1/validate/testing/excel.html



There are also some general articles on several stats applications by
McCullough.

http://www.amstat.org/publications/tas/mccull-1.pdf
http://www.amstat.org/publications/tas/mccull.pdf


It has been some time since I looked at many of these papers, but if my
memory is correct, in general, not much has changed in Excel since "97".
However, from McCullough's most recent article:

"The problems that rendered Excel 97 unfit for use as a statistical
package have not been fixed in either Excel 2000 or Excel 2002 (also
called "Excel XP"). Microsoft attempted to fix errors in the standard
normal random number generator and the inverse normal function, and in
the former case actually made the problem worse."


Many of the above articles have an overlap on references, some
published, some are online resources or lecture notes.


HTH,

Marc Schwartz