A vote for Prism after a comparison of four packages.

by Eric Martz, Ph.D., Professor, Department of Microbiology,
University of Massachusetts, Amherst MA USA,
emartz@microbio.umass.edu.

January, 1996.

For many years, I happily used Fig P for DOS.  It was the
first plotting package which my students used in preference
to graph paper.  However, it did take considerable
hand-holding before a student could use it.  My applications
are general purpose data plotting and curve fitting, with
data entered into the plotting program manually in most
cases.

In late 1995, I decided to upgrade to a Windows plotting
package, and naturally assumed I'd be happy with Fig P for
Windows.  However, I was disappointed with the idiosyncratic
user interface inherited from DOS, so I explored other
packages.

The difficulty I had finding out what packages were on the
market, and the pros and cons of each package (especially
ease of use) led me to create this web page.

I spent some time creating graphs in the following Windows
packages: Fig P, Origin, Prism, and Stanford Graphics. 
Trimetrix kindly provided a full copy of Axum, but I was
unable to evaluate it because neither the tutorial nor the
program itself ran. I have not determined whether this was
due to a defective copy on the diskettes, or bugs in the
programs.

For ease of use, and the detail and quality of its on-line
help, Prism is the winner by a wide margin among the four
programs I test-drove. In fact, Prism has extended my
concept of how easy to use a well-designed user interface
can be.  The program automatically provides explicit
guidance at many points where the novice may not see how to
proceed.  In the case of nonlinear curve fitting, it goes
beyond simple ease of use.  It actually generates a report
interpreting the fitting results, and explains how to do
such fits in some detail.  That is, it teaches some basic
mathematics and statistics as well as doing a smooth job
of fitting.

Prism's ease of use seemed to require some minor sacrifices
in flexibility found in the other programs, at least in the
version I tried (version 1.03).  I haven't yet had a chance
to explore the enhancements in version 2.0 which I've now
ordered.  Also, Prism lacked some capabilities which I
wish it had (and which the other packages had).  It cannot
generate horizontal bar graphs.  It cannot generate text
at angles other than horizontal and vertical.

The other three packages were considerably more powerful
than Prism.  However, I often could not find whole topics in
the on-line help, and generally found the learning curve
steep.  I'm sure that if one uses these packages frequently,
this can be overcome, and the power would then be a
tremendous benefit.  However, I expect that my students will
be able to use Prism without hand-holding time from me, and
that using it occasionly will not result in a significant
penalty of "getting stuck because you can't remember how to
do what you want".  I would not have this expectation about
any of the other three packages.

I chose not to evaluate SigmaPlot because I purchased and
used it for over a year in 1988-89.  I found it
unnecessarily difficult to use then, and my experience is
that such design defects are rarely overcome in later
versions of a program.  Also, that version had a few
annoying bugs, and despite reporting these, the company did
not provide a fixed version for well over a year. Since my
experience is so long ago, I would appreciate contributions
to this web page of critical opinions from users of the
current version of SigmaPlot.

Despite my overall great enthusiasm for, and constant use,
during 1989-94, of Fig P for DOS, there was one dangerous
bug which caused me many frustrating hours.  When a bar
graph is created for multiple datasets, and then additional
datasets are later added, the order of the bars is
scrambled.  Despite reporting this problem to Biosoft more
than once, the problem was never fixed to my knowledge in
several subsequent upgrades.  I would be interested to know
if this problem persists in the Windows version.  I tried to
find out in the demo version but had such difficulty
creating a bar graph at all that I gave up!  Perhaps someone
else can contribute an evaluation of this for this web page.

[THE END]