I strongly concur. Probably none.
I have been using spreadsheets since the late 70's starting on an
interactive main frame and then on PCs. I find spreadsheets very useful
for many purposes. However, doing statistics is not one of them.
The choice of packages to some degree depends on the kinds of analysis
you will be doing, whether you are just doing some exercises or doing
serious research, what discipline you are in, where you can get help,
whether you will do a complete analysis at one sitting or will come back
to complete or refine it, whether others need to be able to follow what
you are doing (QA reviewers, supervisors, other researchers in
disciplines that require sharing data, people from whom you are asking
help, yourself when you have been away from your computer for a while,
etc.).
Also if you are going to do a very wide range of stat methods, or need
more esoteric procedures, you will need to use a more than one package.
In the social and behavioral sciences 80 to 95 percent of an analyst's
time is in getting the data ready for analysis ( Sociology, political
science, psychology, nursing, market research, management, public
policy, evaluation, etc.). After looking at clients needs, we often
recommend starting with SPSS, pasting the syntax generated by the GUI,
for the preparation cleaning and exploration of the data. If there are
procedures not available, then it is easy to transfer the data to other
formats. Some special purpose packages such as SUDAAN can directly read
SPSS system files. SPSS writes SAS files, Excel files, etc.
Art
***@DrKendall.org
Social Research Consultants
Post by John_KanePost by b***@aol.comWhat statistical add-ins (free or commercial) do people recommend for
Microsoft Excel. I have seen criticisms of Excel's built-in statistics
functions. Thanks.
Probably none. Spreadsheets, in general, are not designed for
statistical analysis. In many cases some of the algorithims used are
suspect, the graphics are not very good (although often flashy) and
because of the way spreadsheets work it is pretty well impossible to
document what you are doing.
See http://www.cs.uiowa.edu/~jcryer/JSMTalk2001.pdf for some specific
criticisms of using Excel for statistics but I think the criticisms may
apply both to add-ins and other spreadsheets.
You are almost certainly much better off with a dedicated statistics
package either one of the main comerical packages (SAS, SPSS,
Mintab, ???) or one of the free/open source ones ( R,
http://www.r-project.org, or Dataplot, etc
http://freestatistics.altervista.org/stat.php)
John Kane, Kingston ON Canada