-----Original Message-----
From: G. Marc Turner [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, January 31, 2001 9:47 PM
To: TIPS
Subject: issue with stats course
I need some advice on how to handle an issue in my stats course...
<snip>
So, tomorrow afternoon when I go over this in class, and the students ask
me to explain why the scores are different, what should the answer be? My
immediate reaction is to say that we shouldn't trust the values of SPSS,
and this points out why it is important to know how to do the calculations
by hand. However, I'm not sure how far that will get me in convincing the
students.
Any suggestions are appreciated...
--------------
Marc:
This is true of most stats programs. I actually make it one of my
descriptive stats HW assignments to compare results of calculations using
spreadsheet formulae with the results from the stat program because I think
it helps emphasize the difference in formulae used. It is a fact that SPSS
and many other stat programs only calculate the simple (or exclusive) range
not the true (inclusive) range. If the students want a reason to use the
real limits in the calculation instead of just taking the SPSS result, take
an example from creating intervals for a grouped frequency distribution. Say
you have a distribution with intervals 0-9, 10-19, 20-29, 30-39, etc. What
is the range of one of those intervals as calculated by SPSS? 39-30=9. So is
the interval only 9? Count how many scores are in the interval. 30, 31, 32,
33, 34, 35, 36, 37, 38, 39. There are 10. Why? Because the true range is
from the upper real limit of 39.5 down to the lower real limit of 29.5,
39.5-29.5=10.
In some cases, the simple range is an OK rough estimate of variability but,
for many applications, you will need to calculate the true range. I don't
think it will be helpful to say, "we shouldn't trust the values of SPSS, and
this points out why it is important to know how to do the calculations by
hand" because that leaves the student in the position of having to second
guess all of SPSS' calculations and do all of them by hand (where is the
time savings in that?) Instead, I would focus on pointing out to them those
cases in which they should know that SPSS is printing something different
than they may be looking for. The range is one and many stat programs also
assume sd of a sample used to estimate a population sd (N-1 in the
denominator) when you ask for a standard deviation. You are likely to have
to find another command to actually receive the sd of the population (N in
the denominator). Of course, this is just descriptive stats. With all the
models of multivariate approaches out there, there is a good chance that a
computer program will give you something other than you might calculate by
hand. I think the bottom line is that that doesn't mean that either
calculation is wrong, you are just using different formulae to ask different
questions using different models and you have to know the implications of
those differences to correctly interpret your results. In the final
analysis, maybe stats isn't as easy as everyone thinks ;-}
Rick
Dr. Richard L. Froman
Psychology Department
John Brown University
Siloam Springs, AR 72761
e-mail: [EMAIL PROTECTED]
http://www.jbu.edu/sbs/psych/froman.htm
