"David A. Heiser" wrote:
> ----- Original Message -----
> From: Eric Turkheimer <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Friday, May 26, 2000 9:09 AM
> Subject: Software for Problem Construction
>
> > Has anyone ever seen software designed to generate data for the
> > construction of statistical problem sets? One might input a correlation
> > matrix among a set of predictors, and amount of variance in Y to be
> > explained, a pattern of means, etc. The program would then generate
> > random data to fit the specified problem.
> >
> > I spend a lot of time in SAS doing this one problem at a time...
> > Eric
> > [EMAIL PROTECTED]
>
> ..........................................
> You position yourself in a very difficult position by teaching students by
> means of artificial data.
>
> Artificial data does not teach the student how to think when he has to work
> with real world data. This is the messy stuff that does not fit any known
> distribution and you get confilcting resuslts when the classic stat tests
> are applied.
I respectfully have to disagree. If the simulation is realistic, then you
won't be able to distinguish simulated data from the real thing. This includes
generating datasets which are "messy" in some sense (with tests disagreeing
and so on).
Moreover, real datasets for illustrating certain problems or concepts are not
always readily available, or do not exist within the particular discipline
that the instructor is teaching. Again, an appropriate simulated dataset can
have the same pedagogical potential as a real dataset.
I've spent a little time creating software (DATASIM) to do this. It makes the
process about as easy and straightforward as one could want for generating
data for experiments and multivariate studies having normal and/or nonnormal
distributions, correlated and/or uncorrelated observations, and with or
without scale limits. One declares a design (onegroup, twogroup, oneway,
twoway, threeway, multivariate, table, etc.) and then specifies the relevant
population parameters (mu, sigma, rho, pop. proportions, etc.) and other
necessary information (sample size, decimal places in data, scale limits,
distribution shape, etc.). The program can then generate single or multiple
datasets and do most of the standard parametric and nonparametric statistical
analyses on each, saving the results in a file for later tabulation. This
allows Monte Carlo experiments to be easily conducted, and also allows
students to explore Type I and II errors and the concept of power empirically
(comparing the results to theory).
For those interested, the references appended to this post provide more
information.
Regards,
Drake R. Bradley
Department of Psychology
Bates College
Lewiston, ME 04240
207-786-6180 or 4113
-------------------------------------------------------------------
Bradley, D. R. (1997). DATASIM: A general purpose data simulator. In L.
Lloyd (Ed.), Technology and teaching: case studies on the use of computers,
networks and multimedia in the classroom (pp. 93-118). Medford, New Jersey:
Information Today.
Lehman, R. (1995). Student statistics programs for the Macintosh. Behavior
Research Methods, Instruments, & Computers, 27, 383-391.
Bradley, D. R., & Fleisher, C. L. (1994). Generating multivariate data from
non-normal distributions: Mihal and Barrett revisited. Behavior Research
Methods, Instruments, & Computers, 26, 156-166.
Bradley, D. R. (1993). Multivariate simulation with DATASIM: The Mihal and
Barrett study. Behavior Research Methods, Instruments, & Computers, 25,
148-163.
Wright, J. (1993). Datasim 1.0. Psychology Software News, 4(1), 15-16.
Bradley, D. R., Hemstreet, R. L., & Ziegenhagen, S. T. (1992). A simulation
laboratory for statistics. Behavior Research Methods, Instruments, &
Computers, 24, 190-204.
Bradley, D. R. (1991). Anatomy of a DATASIM simulation: The Doob and Gross
horn-honking study. Behavior Research Methods, Instruments, & Computers, 23,
190-207.
Bradley, D. R., Senko, M. W., & Stewart, F. A. (1990). Statistical
simulation on microcomputers. Behavior Research Methods, Instruments, &
Computers, 22, 236-246.
Bradley, D. R. (1989). A general purpose simulation program for statistics
and research methods. In G. Garson & S. Nagel (Eds.), Advances in social
science and computers (Vol. 1, pp. 145 186). Greenwich, CT: JAI Press.
Bradley, D. R. (1989). Computer simulation with DATASIM. Behavior Research
Methods, Instruments, & Computers, 21, 99-112.
Bradley, D. R. (1988). DATASIM. Lewiston, Maine: Desktop Press.
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
