Excel's Checkered Statistical Past
|
D O N 'T ! |
ExcelTM is a wonderful tool for many things. Statistics is not among them.
Weibull Topics
- Weibull
- Likelihood
- Loglikelihood Ratio Criterion Animations
- LR is distributed as
- Other models
- lognormal distribution
- normal (Gaussian) distribution
- loglogistic
- logistic
- ExcelTM isn't for Statistics
- Precision and Bias
- Answer shopping
- Software Menu
- Get Weibull Software
I'm actively working on the Weibull pages. Please visit again soon.
Perhaps surprising to engineers, Excel's lackluster performance as a statistics engine has been well-documented by practicing statisticians for more than a decade.
Engineers use Excel for many numerical tasks. They don't use if for finite element computations (although I suppose you could coax Excel to do that too) because it wasn't intended for that purpose. They shouldn't use it for statistics either, and for the same reason - Excel doesn't do a very good job at tasks for which it was not intended.
For statistics
there is
,
which in addition to being the best software for applied statistics
available anywhere, at any price, is FREE. There is
another reason not to use Excel for statistics - it is statistically
unreliable. Here are some documented examples:
- A. Talha Yalta, "The
accuracy of statistical distributions in Microsoft® Excel 2007,"
Computational Statistics & Data Analysis, Volume 52, Issue 10, 15 June 2008,
Pages 4579-4586:
"Abstract - We provide an assessment of the statistical distributions in Microsoft® Excel versions 97 through 2007 ... We find that the accuracy of various statistical functions in Excel 2007 range from unacceptably bad to acceptable but significantly inferior in comparison to alternative implementations. In particular, for the binomial, Poisson, inverse standard normal, inverse beta, inverse student’s t, and inverse F distributions, it is possible to obtain results with zero accurate digits as shown with numerical examples."
- Jeffrey S. Simonoff, " Statistical analysis using Microsoft Excel " (2008) "Each of the nine numbers given above is incorrect! The slope estimate has the wrong sign, the estimated standard errors of the coefficients are zero (making it impossible to construct t–statistics), and the values of R2, F and the regression sum of squares are negative! It’s obvious here that the output is garbage (even Excel seems to know this, as the #NUM!'s seem to imply), but what if the numbers that had come out weren’t absurd - just wrong?"
- Patrick Burns, "Spreadsheet Addiction" often referenced webpage. Excel isn't the only culprit, however, "... Excel is known for its many faults ..."
- Hans Pottel, "
Statistical flaws in Excel" (2003)
"Excel is clearly not an adequate statistics package because many statistical methods are simply not available. This lack of functionality makes it difficult to use it for more than computing summary statistics and simple linear regression and hypothesis testing." - Leo Knüsel, University of Munich, " On the Accuracy of Statistical Distributions in Microsoft Excel 97" (2002) "Microsoft Excel (MS Office 97 for Windows) ... is shown that the computation of some discrete distributions (Binomial, Poisson, Hypergeometric) fails even for probabilities in the central range between 0.01 and 0.99 and even for parameter values that cannot be judged as too extreme."
- Yu-Sung Sua, "It’s easy to produce chartjunk using
Microsoft®Excel 2007 but hard to make good graphs,"
Computational Statistics & Data Analysis,
Volume 52, Issue 10, 15 June 2008, Pages 4594-4601 - Neil Cox, "Use of Excel for Statistical Analysis" (2000) "The conclusion from these tests is that, in many cases, Excel uses naïve algorithms that are vulnerable to rounding and truncation errors and may produce very inaccurate results in extreme cases."
- B.D. McCullough and David A. Heiser, "On the accuracy of
statistical procedures in Microsoft Excel 2007," Computational
Statistics & Data Analysis
Volume 52, Issue 10, 15 June 2008, Pages 4570-4578
Abstract: "Excel 2007, like its predecessors, fails a standard set of intermediate-level accuracy tests in three areas: statistical distributions, random number generation, and estimation. Additional errors in specific Excel procedures are discussed. Microsoft’s continuing inability to correctly fix errors is discussed. No statistical procedure in Excel should be used until Microsoft documents that the procedure is correct; it is not safe to assume that Microsoft Excel’s statistical procedures give the correct answer. Persons who wish to conduct statistical analyses should use some other package." - B.D. McCullough and Berry Wilson, "On the accuracy of
statistical procedures in Microsoft Excel 2003"
Computational Statistics & Data Analysis
Volume 49, Issue 4, 15 June 2005, Pages 1244-1252
Abstract: "Some of the problems that rendered Excel 97, Excel 2000 and Excel 2002 unfit for use as a statistical package have been fixed in Excel 2003, though some have not. Additionally, in fixing some errors, Microsoft introduced other errors. Excel's new and improved random number generator, at default, is supposed to produce uniform numbers on the interval (0,1); but it also produces negative numbers. Excel 2003 is an improvement over previous versions, but not enough has been done that its use for statistical purposes can be recommended." - Jonathan D. Cryer, "Problems With Using Microsoft Excel for Statistics," (2001) talk presented at the Joint Statistical Meetings, Atlanta, GA, August 2001, Summary: "Friends Don't Let Friends Use Excel for Statistics!"
Summary:
Don't use Excel for statistical calculations if you require a credible result. Use R (or JMP, Minitab, SAS, SPSS, S-Plus, Stata, ... all of which have substantial licensing fees - R is free).
|
|
mail to:
Charles.Annis@StatisticalEngineering.com |