stever@tektronix.UUCP (Steven D. Rogers) (03/11/86)
Here is something that might be of interest to this
news group:
*************************************************************
ENHANCED COMPUTER GRAPHICS FOR DECISION MAKERS
a dissertation defense and oral examination
feb 28, l986
by Floyd J. Brock
Words need a context to be understood. Visual
patterns also need a context to convey their meaning.
When patterns represent quantities in business
graphics, decision makers (DMs) depend on contrasting
visual contexts to discern patterns and discover
relationships. Depending on the context in which DMs
see trends, differences between two trends may point
to a problem, continuity, or an opportunity.
Can enhancing the context in computer graphics
help DMs visualize problems? To answer this research
question, three experiments were done in the field on
computer graphics. One hundred five DMs tried 17
different contexts for time-series trends displayed on
a microcomputer monitor. The research objective was
to find out whether or changing the context in graphics
affected the decision efficiency (accuracy/response
time) of DMs in determining relationships among
trends. Essential for measuring the effect were
interactive computer programs that displayed random
trends in graphics of differing contexts, collected
the DMs' answers to questions about the trends, and
graded 1133 graphics based on the answers, response
times, and trend data.
The experimental results supported the hypothesis
that computers can enhance the visual context
surrounding time-series trends so that DMs can better
visualize problems. Results were based on comparisons
of DMs' decision efficienies between trial graphics
with differing contextual enhancements and on answers
to questions about the trial graphics. The results
were tested with nonparametric statistics at the 0.05
significance level. Specific findings were:
--Computer-suppolied forecasts, as an enhancement,
significantly helped DMs find differences among
trends.
--Although not statistically significant, stratified
presentation of trends and fading chartjunk tended
to increase DMs' efficiencies.
--Adding two colors, as an enhancement, made no
difference in efficiency over black and white.
--Paired trends in windows did not affect efficiency
significantly.
--Sequentially traced trends and composites of
enhancements did not affect efficiency signifi-
cantly.
--DMs preferred stratified trends most and had the
most confidence in graphics with fading context.
They liked least and had the least confidence in
black-and-white graphics.
***********************************************************
Floyd Brock can be reached through the Systems Science Ph.D.
Program, P.O. Box 750, Portland State University, Portland, OR.,
97207. 503-229-4960.