I agree with much of what has been said about the shortcomings of Excel.
After all this strand of discussion was arose from problems with its ranking
procedure. However, I don't think that users should rely on Excel's
statistical functions or analysis tools. Rather, I think that users should
treat a spreadsheet like "automated pencil-and-paper". To perform a repeated
measures t test, for example rather one would get the spreadsheet to
generate difference scores, would calculate their standard error by "long
hand" (again not using the Excel formula). The most complex functions
required are +, -, *, /, ^2, logs etc. In the process one gets a good look
at the data and a feel for it.
Dennis said;
>, if they
>become professionals in the field .. then they should know that sometimes
>you need to purchase "tools" for your work
But you are teaching statistics undergraduates whereas I am teaching
psychology, business studies and computer science students. After graduating
they are highly unlikely to use a statistical package frequently enough to
justify the cost (to their employer).
Bob said;
>Anybody can do statistic with Minitab, but you need a Ph.D. in statistics
to (safely) do statistics with Excel. The LAST >people in the world who
should beusing Excel for statistics are beginners.
I disagree. My students can do the simple stuff (descriptives, t tests,
correlations, linear regression, non-parametric tests etc) using tried and
tested routines.
My experience is that non-statisticians tend to treat statistical software
as a black-box. They do not seek to understand its inner workings. This
often leads to a poor grasp of statistics, at best a "cookbook"
understanding. At least with the Excel approach they are confronted with
contents of the black box.
Dr Graham D. Smith
Psychology Division
School of Behavioural Studies
University College Northampton
Boughton Green Road
Northampton
NN2 7AL
Tel (01604) 735500 Ext 2393
Email [EMAIL PROTECTED]
