Line: 1 to 1  

Problems with Excel
 
Line: 7 to 7  
 
Added:  
> >  
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

Problems with Excel
 
Line: 6 to 6  
 
Added:  
> >  
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

Problems with Excel
 
Changed:  
< < 
 
> >  
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

Problems with Excel
 
Changed:  
< < 
 
> > 
 
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

Problems with Excel
 
Added:  
> > 
 
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

Problems with Excel
 
Changed:  
< <  
> >  

Line: 1 to 1  

 
Changed:  
< < 
See also Patrick Burns' Spreadsheet Addiction page for general and computational problems with spreadsheets.
There is also an interesting article about how the open source gnumeric spreadsheet team corrected computational errors that Microsoft was unable to fix in Excel : http://www.csdassn.org/software_reports/gnumeric.pdf  
> > 
Problems with Excel
 
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK] 
Line: 1 to 1  

 
Changed:  
< < 
See also an interesting article about how the open source gnumeric spreadsheet team corrected computational errors that Microsoft was unable to fix in Excel : http://www.csdassn.org/software_reports/gnumeric.pdf  
> > 
See also Patrick Burns' Spreadsheet Addiction page for general and computational problems with spreadsheets.
There is also an interesting article about how the open source gnumeric spreadsheet team corrected computational errors that Microsoft was unable to fix in Excel : http://www.csdassn.org/software_reports/gnumeric.pdf  
From: P.J.Wells@OPEN.AC.UK [mailto:P.J.Wells@OPEN.AC.UK]  
Line: 294 to 296  
website at http://www.gre.ac.uk/~cd02/eusprig/ containing some interesting material and links.  
Added:  
> > 
On 19Dec04, Ted.Harding@nessie.mcc.ac.uk wrote: There is a huge literature on this topic, some of it published in journals, much of it floating around on the web and in the archives of mailing lists. Tim's reference above is interesting, but only one example. The McCullough and Wilson reference given there, though now somewhat dated, identifies many of the classic problems. Googling on mccullough wilson excel will throw up a host of followups.
Informed statistical comment on the problems of Excel encountered
by serious users can be found by browsing in the mailing list
ASSUME (Association of Statistics Specialists Using Microsoft Excel): http://www.jiscmail.ac.uk/lists/assume.html. The most recent serious issue reported there is the RAND() bug:
see ASSUME archives for Dec 2003 followed up in the March 2004
archives. The latter point to a statement from Microsoft:
SYMPTOMS When you use the RAND function in Microsoft Office Excel 2003, the RAND function may return negative numbers. CAUSE This problem may occur when you try to use many random numbers, and you update the RAND function multiple times. For example, this problem may occur when you update your Excel worksheet by pressing F9 ten times or more. RESOLUTION This problem is fixed in the Microsoft Excel 2003 Hotfix Package that is dated February 29, 2004.http://support.microsoft.com/default.aspx?scid=kb;enus;834520 (and the deeper you probe in this, the worse it gets). While using Excel for statistics has some limited value in the context of initiating to statistics students whose IT experience is limited to exposure to courses on Excel and Word, and the teacher wants to build on such experience, I think that Excel should never be used for serious statistical work, for several reasons.
 
Line: 1 to 1  

gnumeric spreadsheet team corrected computational errors that Microsoft was unable to fix in Excel : http://www.csdassn.org/software_reports/gnumeric.pdf
 
Line: 7 to 7  
Sent: Donnerstag, 8. März 2001 19:34 To: allstat@JISCMAIL.AC.UK Subject: Summary: Excel for statistics  
Changed:  
< <  
> >  
Last month I posted a request for information about the alleged deficiencies  
Line: 33 to 33  
I've tried to give appropriate credit for specific points, but as might be expected a number of references were the subject of multiple independent messages, so I hope that the following generic credit will do justice to  
Changed:  
< <  
> >  
Andrew Bertie Jan Beyersmann Jill Binker  
Line: 67 to 67  
+44 1908 654658 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++  
Changed:  
< <  
> >  
1) Fundamental shortcomings of digital computation
 
Deleted:  
< <  
These naturally apply to all programs, including highend specialist statistical ones; see B.D. McCullough's twopart article in The American Statistician (11/98, Vol 52, parts 4 and 5/99, Vol 53 part 2  available  
Line: 89 to 88  
As a nonspecialist, I found Part 1 especially chastening. For a simple demonstration I devised the following (prompted by a ExcelG post on the autofill feature); try:  
Changed:  
< <  
> >  
Enter "100" in A1 Enter "=A10.1" in A2  
Line: 97 to 96  
Use the autofill facility to copy A2 down through (say) A110 Set Excel to display 13 places of decimals in the relevant cells.  
Changed:  
< <  
> >  
How many of these cells would you want to use as the basis for any further calculations which depended on the assumption that the contents were exact multiples of 0.1?  
Line: 107 to 105  
(a) A similar approach to the above is applied to Excel in:  
Changed:  
< <  
> >  
McCullough, B. D. and Wilson, B. (1999) On the accuracy of statistical procedures in Microsoft Excel 97. Computational Statistics and Data Analysis 31:2737.  
Changed:  
< <  
> >  
For those whose institutions have the necessary subscriptions, this is available online at http://www.elsevier.nl/gejng/10/15/38/37/25/27/article.pdf.  
Line: 141 to 136  
(c) Knut M. Wittkowski is working on a (very impressive) animated PowerPoint presentation on a number of issues, of which the current version is  
Changed:  
< <  available from http://www.rucares.org/Course/Excel.ppv.  
> >  available from http://www.rucares.org/Course/Excel.ppv.  
(d) Nick Cox pointed out the eccentric nature of Excel's approach to  
Line: 201 to 197  
... a tiny dataset that made Excel give a totally meaningless regression output.  
Changed:  
< <  
> >  
x1 x2 x3 y 1 2 3 21 3 2 5 33 5 4 9 45 7 4 11 49 9 6 15 61  
Changed:  
< <  
> >  
You do a model of y = x1 x2 x3 Excel gives you a load of old tosh.  
Line: 298 to 294  
website at http://www.gre.ac.uk/~cd02/eusprig/ containing some interesting material and links.  
Deleted:  
< <  