Public Health Researcher Invents New Graphing
Method
Above are examples of the new diamond graph, developed by
epidemiology professor
Alvaro Muñoz, that represents
all the
variables equally in a form that is easy to read. He
believes the
method could replace the 3D bar graph in software commonly
used
in business and science.

'Diamond graph' corrects the longstanding errors of
traditional 3D bar graph
By Tim Parsons School of Public Health
Looks can be deceiving, and that's one of the problems
with today's threedimensional bar graph. While these
graphs may look correct, researchers from the Johns Hopkins
Bloomberg School of Public
Health believe they are, in fact, inaccurate and
misleading.
Currently, the 3D bar graph is used in countless
computer programs, scientific journals and newspapers to
display financial, medical and other information in which
two variables lead to an outcome. Alvaro Muñoz, a
professor of
epidemiology at the School of Public Health, has
developed the new "diamond graph," which corrects these
errors and represents all the variables equally in a form
that is easy to read. He believes the new graphing method
could replace the traditional 3D bar graph in software
commonly used in business and science. Muñoz and his
colleagues described the Diamond Graph method in an article
published in the August edition of the peerreviewed
journal The American Statistician.
So what was wrong with the old method? According to
Muñoz, the 3D bar graph has three main flaws.
First, the variables, which equally contribute to an
outcome, are not equally represented in the diagram; this
gives the impression that one variable is more important
than another. Second, it is sometimes difficult, if not
impossible, to distinguish the true value of the bars
because of the problems of representing a threedimensional
image on a twodimensional page; because of perspective,
some bars appear to be of greater or lesser value when they
are actually of equal value. The third drawback of the 3D
graph is that it cannot be used to present overlapping
data; in some cases, parallel bars with higher values may
obscure those with lower values, making the graph
useless.
"The inaccuracies of the traditional 3D bar graph may
seem trivial, but they can be significant when you're
dealing with important information like predicting your
risk for a heart attack or plotting the performance of
your company's investments," Muñoz said.
The 3D bar graph now commonly used.

The new diamond graph method corrects the inaccuracies
and limitations of the 3D bar graph by representing all
variables equally on a 2D graph. The diamond graph is
essentially the view of the bar graph from above rather
than from the side. Instead of using rising parallel bars,
the diamond graph uses expanding polygons within a
diamondshaped grid to represent values. The researcher
experimented with other shapes but found that the sixsided
polygon was the only shape to represent the outcomes
equally within the grid as it expanded.
Over the years, other researchers have attempted to
develop a better graphing method, but the diamond graph is
the first to represent equally the relationships between a
continuous outcome and each of the two categorical
predictors in a single plot.
"We call our new method of display the diamond graph
[because] it has the shape and, hopefully, the value of a
diamond. Perhaps more importantly, it is reminiscent of the
baseball diamond that The American Statistician
equiponderantly loves. Who would have thought we would
still be inventing new methods of graphing in the 21st
century?"
Authors of the paper were Xiuhong Li, Jennifer M.
Buechner, Patrick M. Tarwater and Alvaro Muñoz. The
research was sponsored by grants from the National
Institute of Allergy and Infectious Diseases.
GO TO AUGUST 18, 2003
TABLE OF CONTENTS.
GO TO THE GAZETTE
FRONT PAGE.
