Content-Type: text/html
Mark J. Crosten
335 Elan Village Ln.
Apt. 404
San Jose, CA 95134
CompuServe: 70733,14
A CONTENT ANALYSIS OF QUANTITATIVE GRAPHICS
IN THREE U.S. NEWSPAPERS
Presented to the Visual Communication Division of AEJMC
at Kansas City, August 1993
KEYWORDS:
Informational graphics, Newspaper graphics, Edward Tufte
Data density, New York Times, Wall Street Journal
USA Today, Graphics
ABSTRACT:
This content analysis studies an empirical test of the
efficiency of quantitative graphics at displaying information
in three national U.S. newspapers. The results were compared
between and within the newspapers to determine trends in the
content of quantitative graphics over the past decade. It was
found that sampled graphics in the Wall Street Journal
displayed the highest average amount of data while those in
the New York Times and USA Today had significantly lower
average data densities.
CHAPTER ONE
Introduction
"To clarify, add detail" -- Edward R. Tufte
In examining how to display data graphically the above
statement by Edward R. Tufte, a professor of political
science and statistics at Yale University, might seem
contradictory at first glance. He contends, however, that "we
thrive in information-rich worlds because of our immense
human capacities to select, edit, focus, group, harmonize,
condense, reduce, choose, categorize, merge, filter,
abstract, skip, scan, idealize, isolate, check out,
extrapolate, screen, sort, blend, average, lump, smooth,
inspect, approximate, cluster, aggregate, summarize, dip
into, flip through, browse, leaf through, search, skim,
glean, winnow the wheat from the chaff, and separate the
sheep from the goats."<1>
Tufte says that "high information displays are not only an
appropriate and proper match to these human capabilities, but
such designs are often optimal. If the visual task is
contrast, comparison, and choice--as so often it is--then the
more relevant information within eyespan, the better."<2>
This content analysis studies an empirical test of the
efficiency of quantitative newspaper graphics at displaying
information.
Newspaper graphs and graphics, or "infographics," have
increased in number, size, kind, color and importance,
especially over the last decade. Editors and publishers,
eager to position their papers to subscribers they say are
time-starved and spoon-fed by television, have scrambled to
add charts full of bars, hues, pies, cartoon-like characters,
lines, pictures and tables to their pages. Many analysts have
attributed this phenomenon to the arrival of two prominent
products: USA Today in September of 1983 and the Apple
Macintosh in February of 1984. The introduction of concise,
colorful, graphically aware competition in the form of USA
Today in many cities stirred hundreds of newspapers to take
action, and the availability of the Macintosh computer made
taking that action easy.
According to Tufte however, a problem with the rapid rise
in the use of newspaper infographics is one of content. He
alleges that most newspaper infographics--most graphics for
that matter--are superficial to the point of being useless.
Tufte's advice to the creators of such graphics is "to show
readers lots of information. Clutter and confusion are not
defects of information, but of bad design. Don't blame the
information for being cluttered or confused or overrich--
instead, you fix the design."<3> This study will discuss
an experimental test of a theory of the efficiency of
quantitative graphs at displaying information--content over
style. Using this theory, Tufte proposes that data graphics
should be based on large rather than small data matrices and
have a high rather than low data density. In other words,
presenting more information is better than presenting less
information. Specifically, this study will use a sample of
quantitative graphics from three national mass-circulation
newspapers to test the data density ratio, a principle
developed by Tufte. The numbers that go into a graphic can
be organized into a data matrix of observations by variables.
Taking into account the size of the graphic in relation to
the amount of data displayed yields the data density:
By using the data density ratio, Tufte has determined that
a graphic containing .15 numbers per square inch "is thin
indeed," a graphic containing 3.8 numbers per square inch is
"lightweight," a graphic containing 181 numbers per square
inch "does very well," and "the current record" is a map
containing 9,000 numbers per square inch.<4>
Quantitative graphics from three national newspapers, the
Wall Street Journal, the New York Times and USA Today will be
used to test the data-density ratio. The choice of these
newspapers, a renowned, well-respected financial newspaper
that has won many awards for its graphic displays; a
comprehensive, internationally aware newspaper known for its
writing as well as its respect for numbers; and a highly
colorful, easily read and graphic newspaper acknowledged as a
leader with respect to data displays and informational
graphics, will provide an adequate cross-section of graphics
to sample. The results of this study will offer a
detailed analysis of the content of national newspaper
quantitative graphics through the application of an empirical
test. The conclusions will answer (1) the question of how
graphics from the three newspapers compare with respect to
Tufte's data density ratio; and the related questions of (2)
where the graphics are located within each newspaper; (3)
what kinds of statistical graphics are being used; (4) which
graphics are related to a story; (5) which graphics include
pictorial or decorative elements; (6) how the sizes of the
graphics compare; and (7) the overall question of how the use
and content of graphics in the three newspapers have changed
over time.
Hypothesis I
1.) The average data density ratio of all sampled graphics
in the three newspapers will differ widely with (2.) graphics
in the Wall Street Journal exhibiting the highest data
density and those in USA Today exhibiting the lowest data
density. The data density of graphics in the New York Times
will fall just below those of the Wall Street Journal.
Hypothesis II
1.) There will be an increase from the start to the end of
the sampling dates in the A.) overall average data density;
B.) total number; and C.) overall average area (in square
inches) of graphics in all three newspapers.
Hypothesis III
1.) There will be an increase from the start to the end of
the sampling dates in the number of graphics in all three
newspapers that directly relate to stories.
Hypothesis IV
1.) There will be a large increase in the use of pictorial
or decorative elements in the graphics of USA Today during
the sampling period. The use of pictorial or decorative
elements will be negligible in graphics in the Wall Street
Journal and only a minor component of graphics in the New
York Times during the sampling period.
CHAPTER TWO
Literature Review
The results of this study will provide an analysis of the
content of quantitative graphics in three national
newspapers. The conclusions will tell us how the graphics in
the papers compare in their efficiency of displaying data in
the last ten years and recognize any changes in the kind,
placement, story relationship, use of pictorial or decorative
elements and size. It would be a rather simple, though
monumental, task to summarize for the readers of this study a
complete history of statistical graphs and their relevance to
the mass media. However, there exist (thankfully!) very
adequate reviews of the history of statistical graphs in
various books and journals which are noted in the
bibliography. Similarly, numerous articles and studies have
been written on the topic of how infographics infiltrated the
mass media and what it all means to everyone from the
president of a newspaper concern<5> to "120 student observers
in two sophomore level classes taught at the School of
Journalism at Indiana University in the Spring semester of
1987."<6> These have also been duly noted in the
bibliography.
What can and should be discussed here is the relevant
literature concerning what is thought to be the proper status
of statistical graphs or infographics in general, moving on
to survey criticism of their use in the media, specifically
newspapers, and finishing with an analysis of graphical
theory.
In the introduction to The Visual Display of Quantitative
Information, Tufte summarizes the discipline of statistical
graphs:
Data graphics visually display measured quantities by
means of the combined use of points, lines, a coordinate
system, numbers, symbols, words, shading, and color.
The use of abstract, non-representational pictures to show
numbers is a surprisingly recent invention, perhaps because
of the diversity of skills required--the visual-artistic,
empirical-statistical, and mathematical. It was not until
1750-1800 that statistical graphics--
length and area to show quantity, time-series, scatterplots,
and multivariate displays--were invented, long after such
triumphs of mathematical ingenuity as logarithms, Cartesian
coordinates, the calculus, and the basics of probability
theory. The remarkable William Playfair (1759-1823) developed
or improved upon nearly all the fundamental graphical
designs, seeking to replace conventional tables of numbers
with the systematic visual representations of his "linear
arithmetic."
Modern data graphics can do much more than simply
substitute for small statistical tables. At their best,
graphics are instruments for reasoning about quantitative
information. Often the most effective way to describe,
explore, and summarize a set of numbers--even a very large
set--is to look at pictures of those numbers. Furthermore, of
all methods for analyzing and communicating statistical
information, well-designed data graphics are usually the
simplest and at the same time the most powerful.<7>
Tufte goes on to list a number of requirements for the
attainment of excellence in statistical graphics. He says
graphical displays should:
- show the data
- induce the viewer to think about the substance rather
than about methodology, graphic design, the technology of
graphic production, or something else
- avoid distorting what the data have to say
- present many numbers in a small space
- make large data sets coherent
- encourage the eye to compare different pieces of data
- reveal the data at several levels of detail, from a
broad overview to the fine structure
- serve a reasonably clear purpose: description,
exploration, tabulation, or decoration
- be closely integrated with the statistical and verbal
descriptions of a data set.<8>
In a 1981 paper by Wainer and Thissen, the usefulness of
any particular graphic display is determined by comparison
with a list of desirable characteristics. These graphical
characteristics were proposed by Gnanadesikan in 1980 and
include their descriptive capacity, versatility, data
orientation, potential for internal comparisons, aid in
focusing attention, degree to which they are self critical of
assumptions, and adaptability to large volumes of data.<9>
In Schmid and Schmid's 1979 book on graphic presentation,
a summary of the qualities and values of charts and graphs
was presented. These values were summarized as follows:
- In comparison with other types of presentation, well-
designed \charts are more effective in creating interest and
in \appealing to the attention of the reader.
- Statistical charts represent a very important form of
visual \communication. They provide a clear, economical, and
precise medium for conveying a message. They are frequently
superior to words or figures. Moreover, visual relationships,
as portrayed by charts and graphs, are more clearly grasped
and more easily remembered.
- The use of charts and graphs saves time since the
essential meaning of large masses of statistical data can be
visualized at a glance.
- Charts and graphs can provide a comprehensive picture of
a problem that makes possible a complete and better-balanced
understanding than could be derived from tabular or textual
forms of presentation.
- Charts and graphs can bring out hidden facts and
relationships and can stimulate, as well as aid, analytical
thinking and investigation.<10>
A 1984 study by Wainer which looked at the ways graphical
data could be displayed badly stated that "the aim of good
data graphics was to display data accurately and clearly." He
categorized methods of data display into three parts: showing
data, showing data accurately, and showing data clearly.
Under the heading of showing data, Wainer refers directly to
Tufte's theory of data density by calling it "surprisingly
informative." He states that a range of .1 to 362 numbers per
square inch was found in a sample of popular and technical
media. Wainer said that "although arguments can be made that
high data density does not imply that a graphic will be good,
nor one with low density bad, it does reflect on the
efficiency of information. Obviously, if we hold clarity and
accuracy constant, more information is better than less."<11>
In Bertin's 1981 book, he states that graphic
communication involves transcribing and telling others what
you have discovered. The aim of graphic communication, he
says, is rapid perception and, potentially, memorization of
the overall information. The imperative is simplicity. He
says this simplicity of forms authorizes the superimposition
of images. Graphic communication poses problems on the level
of simplification and selectivity.<12>
Bertin said in his 1983 book that there are three
functions of graphic representation: recording information,
communicating information and processing information. He
defines the recording information function as the creation of
a storage mechanism--a graphic--which avoids the effort of
memorization. This would most likely be a comprehensive
graphic which could be non-memorable in its totality. In
communicating information, Bertin said the graphic should
create a memorable image which will increase the information
in the viewer's mind. A simple graphic is ideal in this
situation. The processing function of a graphic provides a
simplification and its justification. Bertin said the graphic
should be memorable (for comparisons) and comprehensive (for
choices).<13>
In 1985, Cleveland said that graphs are an exceptionally
powerful tool for data analysis. He went on to say that
graphical methods tend to show data sets as a whole, allowing
summarization of the general behavior and to study detail.
Cleveland said that information can be retained in the data
because a large amount of quantitative information can be
displayed and absorbed.<14>
Ehrenberg said in his 1975 book that data need to be
summarized in analysis. He said the criteria for a good
summary are that it be succinct, complete (i.e., that the
original data could be reconstructed within the stated or
implied limits of approximation), and usable (i.e., that the
results can readily be used when analyzing further data).<15>
A number of studies and articles criticizing the use of
graphics by the mass media and specifically, newspapers, have
appeared parallel to their increased use and perceived
importance. In a 1987 study by Kenney and Lacy examining the
economic forces behind newspapers' increasing use of color
and graphics, it was said that the success of USA Today
attracted reader interest and debate within the journalistic
community concerning the merits, costs and standards of
publishing more color and graphics. Kenney and Lacy said that
quality color and graphics in newspapers is needed and wanted
by advertisers and readers. They state that although many
factors enter into the decision to use color and graphics,
the usefulness of these devices to differentiate a newspaper
and attract readers mean visual communication will continue
to increase in competitive markets.<16>
Shapiro, in a 1982 report on the increasing use of charts
and maps in newspapers, found that this growth can be
attributed to the fact that editors and reporters are coming
to recognize that trends in American life are often
newsworthy. These stories, he said, are sometimes culled from
statistics and sometimes merely enhanced by them and are more
visible than ever. Shapiro attributed the move to brighter
maps and charts in the press to graphics editors who were
willing to explore and the new technology that can help them.
He quoted Roger Fidler, corporate design consultant for
Knight-Ridder newspapers, as saying "the use of statistics in
stories and in graphics all revolves around trying to
communicate effectively. We've become much more aware of
being a visual medium as well as a word medium. We are
initiating efforts to find new ways to hold readers."<17>
In a 1988 study by Smith and Hajash describing the use of
informational graphics in 30 daily U.S. newspapers, it was
found that informational graphics are an important aspect of
how newspapers communicate information. However, the use of
graphics on the level of USA Today was not supported. It was
also found that a large proportion of the sampled graphics
were found in business sections. This was attributed to the
statistical and data based origins of the genre. Maps,
followed by bar charts, were the most-used types of graphic
forms. Little incidence of embellishment was found in the
graphics, leading Smith and Hajash to conclude that they are
being used as basic communication tools and not as creative
devices.<18>
In a 1979 article by Wainer, four reasons were given for
the wide adoption of statistical graphs by the news media:
- they can often display complex data simply
- they can provide an evocative image which is readily
remembered
- they lend seemingly scientific confirmatory evidence to
the contentions of a story; and
- they are often "eye-catching," enlivening the layout of
the page and enticing the viewer to finish the story.
Wainer says, however, that the last reason seems to be the
primary motivator for the New York Times, but that this
purpose is often at odds with that of effective and accurate
communication in that by making a statistical graph "flashy,"
the artist often obscures or, worse, distorts the data which
generated the graph. He goes on to say that the use of
statistical graphs can be a very powerful tool for
journalistic communication, they can misinform either by
design or by poor design. Wainer notes sadly that graphs in
current usage are at least as often misused as used correctly
in the popular media.<19>
Tankard, in a 1987 study, identified 10 pitfalls to avoid
in the use of informational graphics by newspapers. Among the
problems Tankard found were graphics that were misleading,
unclear, inaccurate, or unnecessary. He also addressed some
recommendations by graphics experts such as Tufte, Bertin and
Wainer by saying that their rather stringent procedures would
eliminate or drastically alter about 90% of the graphs now
appearing in newspapers. Tankard spoke to Tufte's criticism
that graphics should never be adorned with "chartjunk" or
decorative elements by saying that his recommendations may be
too severe for newspaper graphics. Tankard said that the use
of quantitative graphics in newspapers requires striking the
proper balance--a balance between being accurate and clear,
and being attention-getting and entertaining.<20>
In a study by Pasternack and Utt in 1990, an attempt was
made to understand when newspaper readers read graphics in
relation to headlines and text, and whether readers read
graphic devices for appearance or content reasons. The
results of the study found that readers turned to graphs and
charts for both content and appearance-
based reasons. A call for more emphasis on the "info" in
infographics is given by Pasternack and Utt. They said that
based on their data readers expect charts and graphs to
fulfill an information gathering need, perhaps assuming color
and attractiveness as a given. The first burden for editors
and graphic artists, they said, was to make graphs
understandable since for many readers, infographics are
either intimidating or fall into the category of something
that should "be seen but not heard."<21>
The effects of three-dimensional and cartoon-like
informational graphics--called "chartoons" by Tankard in a
1989 study--were discussed with regard to reader interest and
information gain. Tankard found general support for the
hypotheses that chartoons and three-dimensional graphs are
perceived as more appealing than plain graphs and little
support for the contention that chartoons and three-
dimensional graphs will lead to less retention of information
than plain graphs. Tankard said that Tufte's concept that
chartoons and three-dimensional graphs contain unnecessary
decorative elements--"chartjunk"--may not make any real
difference to the average newspaper reader.<22>
In 1989, Kelly tested Tufte's concept that graphs should
draw the viewer's attention to the sense and substance of the
data, not to something else. Tufte's data-ink ratio, which
holds that a large share of the ink on a graphic should
present data-information, was used to answer the question of
whether graphics which contain a large amount of "non-data
ink" affected a reader's accuracy of recall of quantitative
display. Kelly's results were inconclusive as to whether
Tufte's theory holds true partly because of the small number
(ten) of graphics used as stimuli and the observers were
forewarned of the stimuli's subject which increased their
attention to detail.<23>
An article by Moen in 1989 reprimanded editors of
informational graphics for not giving the same scrutiny to
informational graphics that they give to stories or
headlines. Moen said that to many editors, any information
graphic is a good graphic but too many are "misinformation"
graphics. He urged editors to exercise quality control by
spot checking graphics against a number of standards. Moen
said graphics should be created like good writing, simple and
direct. He said improper format as well as artwork can
obscure the message a graphic is trying to communicate.<24>
In an 1991 interview, Tufte described his general
experience and opinions with regard to newspapers and their
use of quantitative graphics. He said that he is quite
impressed by what is being done at the New York Times in
answer to a question asking him to identify publications that
consistently produce good information graphics. He said
"their business graphics are terrific, their temperature
chart at the end of the year is very good--they show
inflation-adjusted dollars, instead of just skyrocketing
inflation." Tufte also said the Wall Street Journal is "good-
-their tables are well designed." When asked to identify what
unsuccessful publications are doing wrong when producing
information graphics, Tufte had this to say, "The visual
problem is that data are a lot more interesting than the
designers think. Their readers have come to this paper, they
have selected themselves to this story, to this topic. And in
most cases, the readers--believe it or not--know more about
the topic, and care more about the topic, than the graphics
illustrator who has just come to this particular topic for
half an hour. The graphics illustrator should design upwards,
rather than dummy things down. My advice is to show readers
lots of information."<25>
A 1992 article in the New York Times said that "some
practicing graphic designers see Tufte's standards as overly
academic and rigorous. Time's Nigel Holmes, a creator of
graphics Tufte has criticized in his books, was
understandably irked. Holmes admits his work has sometimes
been exaggerated, but feels Tufte, in his insistence on
absolute mathematical fidelity, remains trapped in 'the world
of academia' and insensitive to 'the world of commerce,' with
its need to grab an audience. But even Holmes admires and
admits to having been inspired by Tufte's books."
Opinion about how graphics should be used and presented is
abundant and easily found; however, reasoned theories
concerning the same topics are more difficult to locate. An
article by Cleveland and McGill in 1984 stated the problem
clearly:
Today graphs are a vital part of statistical data analysis
and a vital part of communication in science and technology,
business, education, and the mass media.
Still, graph design for data analysis and presentation is
largely unscientific. This is why Cox argued "there is a
major need for a theory of graphical methods," and why
Kruskal stated "in choosing, constructing, and comparing
graphical methods we have little to go on but intuition, rule
of thumb, and a kind of master-to-apprentice passing along of
information....there is neither theory nor systematic body of
experiment as a guide."
Cleveland and McGill attempted to apply some theory to
graphical perception with their study which examined the
relative accuracy with which various graphical forms conveyed
quantitative information. Their results showed that when
graphics are "read," position judgments about certain graphic
elements such as bars in a bar chart and lines on a line
chart were more accurate than length and angle judgments such
as those made when viewing pieces of a pie in a pie chart.
They also attempted to order ten elementary perception tasks
on the basis of accuracy with which people can extract
quantitative information. These tasks included those of
position on a common scale, position on non-aligned scales,
length, direction, angle, area, volume, curvature, and
shading. Their hypothesis stated that graphical forms which
involved elementary perceptual tasks that lead to more
accurate judgments than another graphical form which shows
the same quantitative information will result in better
organization and increase the chances of a correct perception
of pattern and behavior. The conclusions from this hypothesis
called for a dismissal of some of the more popular chart
forms--bar charts, divided bar charts, pie charts, and
statistical maps with shading--in favor of dot charts, dot
charts with grouping, and framed-rectangle charts.<26>
Macdonald-Ross in a comprehensive 1977 review of research
on the presentation of quantitative data in texts urges
restraint in applying too much importance to rigor in
examining graphical methods. He says that for practical
communication to be effective it is necessary for empirical
research to be studied, but that is not sufficient by itself.
He said many subtle tricks of the expert designer have never
been studied, especially the art of adjusting the various
parts of a design so that one coherent whole is produced. He
said a great deal of experience, intuition, critical skill,
and trial of alternative solutions go toward the design of
good communication: this artful process will never be
replaced by formal laboratory investigations. Macdonald-Ross
stated that many issues that interest practical communicators
have never been investigated and the ones that have been
investigated were not done in a working environment.
Formative evaluation is rare, he said; most tests were
conducted on material produced by the researcher himself. He
concluded that in general, the advice given by the best
practical communicators was substantially vindicated. One or
two workers, he stated, quite openly suggested that such
advice was just "a matter of opinion" and hence as likely to
be wrong as right. Macdonald-Ross said that on the contrary:
there is no known case where a well articulated and deeply
held practical viewpoint has been overturned, and many cases
where the practitioner has been upheld.<27>
In Bertin's 1983 book he proposes the theory that
efficiency in representing quantitative data graphically can
be defined by the following proposition: If, in order to
obtain a correct and complete answer to a given question, all
other things being equal, one construction requires a shorter
period of perception than another construction, we can say
that it is more efficient for this question. But how does
Bertin define efficiency and how does that definition relate
to the simplification of data? He answers the questions by
saying that efficiency comes through simplification of data
into the presentation of only those essential data which will
accomplish the communication of the desired message in the
shortest possible time. Surprisingly, Bertin introduces two
contrary methods of simplification of data into its essential
parts. The first is obvious; take out all the data which is
not necessary to the communication. Bertin proposes that this
be done through the calculation of moving averages and other
mathematical operations and, with maps, the structural
generalization of geographic forms. The second method is a
foreshadowing to Tufte's theories of graphical display.
Bertin proposes that data be simplified by adding more data.
Specifically, he says that the information being produced can
be introduced into a higher set. In other words, the
representation of the initial data can sometimes be made more
clear--simplified--by comparing or introducing a collection
of related data which can be used to derive general
tendencies, which can then be applied to the original data.
Bertin admits that this second method of simplification of
data relies heavily on the expert skill and knowledge of the
graph creator to decide which related data to introduce.<28>
Conveniently, this leads directly to Tufte's theories of
data graphics. The first is that a large share of ink on a
graphic should present data-information, the ink changing as
the data change. Data-
ink is the non-erasable core of a graphic, the non-redundant
ink arranged in response to variation in the numbers
represented. Tufte says that the larger the share of a
graphic's ink devoted to data, the better (other relevant
matters being equal). He qualifies this theory by saying that
the principle of the data-ink ratio makes good sense and
generates reasonable graphical advice--for perhaps two-thirds
of all statistical graphics. For the others, the ratio is
ill-defined or is just not appropriate.<29>
Kelly tested Tufte's theory of data-ink maximization in a
1989 study. He found that his results did not permit him to
conclude that the ratio of data-ink to non-data-ink had no
effect on accuracy judgments based on graphically displayed
data. Kelly did call into question Tufte's theory somewhat by
saying that based on his test, high ratios are not sufficient
by themselves to consistently decrease accuracy of data
recall.<30>
Tufte's second principle for graphical displays involves
the idea of multifunctioning graphical elements. He says the
same ink should often serve more than one graphical purpose.
A graphical element may carry data information and perform a
design function usually left to non-data-ink or it might show
several pieces of data. The theory, then is to mobilize every
graphical element, perhaps several times over to show the
data.<31>
Wainer and Thissen describe a method of constructing
multifunctioning graphical elements in their 1981 paper.
Relying on human ability to perceive and remember even small
variations in the structure of human faces, Chernoff
developed and tested his idea in 1973 and 1975. The scheme
involved letting the size, shape, or orientation of each
feature of a cartoon face represent a particular variable.
Thus one might let the size of the eyes represent on
variable, the width of the mouth another, the length of the
nose a third, and so on. The performance of Chernoff's scheme
has not been fully explored<32> but it conforms to Tufte's
idea by using facial element to represent specific variables
and the locations of the faces within a grid to represent
comparison as a whole.
A third principle explored by Tufte attempted to enhance
the idea of graphical integrity. He stated that the
representation of numbers, as physically measured on the
surface of the graphic itself, should be directly
proportional to the numerical quantities represented. Tufte
measures violations of this principle by calling into effect
a measurement of the Lie Factor. The Lie Factor is determined
by dividing the size of the effect shown in a graphic by the
size of the effect in the data. Tufte said that if the Lie
Factor is equal to one, then the graphic might be doing a
reasonable job of accurately representing the underlying
numbers. Lie Factors greater than 1.05 or less than .95
indicate substantial distortion, far beyond minor
inaccuracies in plotting.<33>
Tufte's Lie Factor was studied by Jarvenpaa in a 1989
paper. Jarvenpaa examined Tufte's principle as it related to
decision making in a financial forecasting context. He found
that the differences in overall performance or learning rates
were not significantly different. The results suggest,
Jarvenpaa said, people may be able to adapt format to task
and that violations of Tufte's law may be tolerable when
graphs are used as decision support tools.<34>
Through the theories listed previously, it is seen that
Tufte shows obvious concern for statistical graphs in
general, however the subject of this study will only
specifically address his contentions that relate directly to
his theory of data density, namely that graphics should
"present many numbers in a small space and make large data
sets coherent."<35>
Tufte says that our eyes can make a remarkable number of
distinctions within a small area. With the use of very light
grid lines, it is easy to locate 625 points in one square
inch, he says. Tufte asks the questions: How many statistical
graphics take advantage of the ability of the eye to detect
large amounts of information in small spaces? And how much
information should graphics show? Tufte describes a measure
called the data density to determine the number of numbers
represented in the space a graphic uses. It has been
mentioned previously that Tufte called a graphic with a data
density of .15 numbers per square inch "thin indeed," one
with 3.8 numbers per square inch "lightweight," one with 181
numbers per square inch as "doing very well," and a map with
9,000 numbers per square inch as holding "the current
record."<36>
Tufte measured the data densities of graphics sampled from
scientific and news publications with results published in
his 1983 book. He describes the results by saying that "the
table records an enormous diversity of graphical performances
both within and between publications. A few data-rich designs
appear in nearly every publication. The opportunity is there
but it is rarely exploited: the average published graphic is
rather thin, based on about fifty numbers shown at the rate
of ten per square inch."<37> Tufte said that "among the
world's newspapers, the Wall Street Journal, The Times
(London), and Asahi publish data-rich graphics, with data
densities equal to those of the Journal of the American
Statistical Association. Most of the American papers and
magazines, along with Pravda, publish less data per graphic
than major papers of other industrialized countries."<38>
CHAPTER THREE
Methodology
A comparison was made of three U.S. national daily
newspapers: the New York Times, the Wall Street Journal, and
USA Today. Six years from the last two decades were
studied for the New York Times and the Wall Street Journal:
1981, 1983, 1985, 1987, 1989, 1991. Five years from the last
two decades were studied for USA Today: 1983, 1985, 1987,
1989, 1991. A sample from 1981 was not included for USA Today
since the paper began publication in 1983. The study started
with the year 1981 for the New York Times and the Wall Street
Journal and in 1983 for USA Today since it has been
acknowledged that the increase in the use of informational
graphics in newspapers has for the most part occurred over
the last dozen or so years.<39>
Four weeks from each year for each newspaper were sampled.
The third, 16th, 34th and 48th complete weeks of each year
were randomly chosen with only weekday editions used in the
sample. This decision arose from the fact that the Wall
Street Journal only publishes weekday editions and USA Today
publishes a combined Friday, Saturday, Sunday edition while
the New York Times publishes seven days a week. To keep the
sample fair and equal, the weekend edition of USA Today and
the weekend editions of the New York Times were not included.
DEFINITION OF CATEGORIES:
(1.) GRAPHICS CATEGORIES:
For the purposes of this study, a "graphic" is defined as
a deliberately designed or rendered visual presentation of
quantitative data. The categories of graphics included seven
categories as defined by Smith and Hajash:
- pie chart-the division of a whole into components,
usually in percentages;
- bar chart-comparative quantities represented by bars or
other images of varying length;
- line or fever chart-quantities plotted over time by a
trend line;
- table-visual display of numbers in columns;
- scatterplot-multiple dots or other symbols showing the
relationship of one variable with another;
- spatial-a visual representation of a region of the earth
or heavens which is used to show relational numerical data;
and
- other-a miscellaneous form of displaying quantitative
data not included within the other categories. Television
schedules, weather maps, stock listings, architectural and
engineering drawings, recipe lists, box scores and
statistics, and lists of numbers not in relational graphic
form were excluded as were advertising communications.<40>
Only staff-produced graphics were considered for this
study since graphics available through a syndicate or wire
service--because they are made available to a wide range of
publications at once--do not reflect directly on the efforts
of the individual newspapers' staff. (2.) DATA ENTRIES
(3.) WIDTH/HEIGHT
(4.) DATA DENSITY
Since the overriding purpose of this study was to examine
the data density of graphics, the width and height in inches
as well as the number of numbers presented was counted for
each graphic. All graphics conforming to the requirements of
the category list were counted. If a graphic was encountered
which included more than one specific type of charting method
in a defined area or box--such as the use of two bar charts
and four pie charts together to display the results of a
survey--each occurrence of the individual method was counted.
However for the purpose of determining such a graphic's data
density, the total area and total number of numbers given to
the entire graphic was counted--not each individual chart's
area or numbers separately. Using the example previously
mentioned, if a graphic showing the results of a survey
contained two separate bar charts and four separate pie
charts, the data density of that graphic was determined by
dividing the total area given to the graphic's elements by
the sum of the number of numbers presented by all the charts
within the graphic's area.
(5.) STORY RELATIONSHIP
This category was included for the purposes of determining
the number of graphics which were supplements to textual
information or stories versus those that were meant to stand-
alone with no textual or story information provided. If a
graphic was included with the presentation of a story or
related to a story, this was noted.
(6.) PICTORIAL ELEMENTS
This category was included for the purposes of determining
the number of graphics which used pictorial or decorative
elements along with the quantitative data within their
boundaries versus graphics which displayed nothing but
quantitative data. A pictorial or decorative element was
defined as any drawing, sketch, rendering, illustration,
photograph, caricature, model, trademark or logo which was
not quantitative in nature in its relationship to the data
presented. For example, this category would include a
photograph over which a graphic was superimposed, a company
logo such as the Apple Computer, Inc. symbol, an artist's
drawing of baseball bats used as bars in a bar chart, or a
woodcut of an executive's face placed within a graphic's
frame.
CHAPTER FOUR
Findings
A total of 3,510 graphics was examined for this study. To
avoid coder reliability problems, all data were collected by
one coder. To test the categories and repeatability of this
study, two other coders compared their results to the
original coding. One week for each year of each newspaper
studied (seventeen total issues) was used for this
comparison. A 97 percent rate of duplication was recorded.
Table 1
NEWSPAPER QUANTITATIVE GRAPHICS
OVERALL AVERAGE DATA DENSITY
NEWSPAPER NUMBER AVERAGE DATA DENSITY
New York Times 859 12.74
USA Today 1,064 3.63
Wall Street Journal 1,587 22.67
Data densities represented as numbers per square inch.
The results indicated by Table 1 support the first part of
Hypothesis I which states: The average data density ratio of
all sampled graphics in the three newspapers will differ
widely.
RESULTS: The average data density of graphics in USA Today
was 3.63 numbers per square inch while the Wall Street
Journal exhibited 22.67 numbers per square inch. Graphics in
the New York Times had an average data density of 12.74
numbers per square inch.
The results indicated by Table 1 only partially support
the second part of Hypothesis I which states: Graphics in the
Wall Street Journal will exhibit the highest data density and
those in USA Today will exhibit the lowest data density. The
data density of graphics in the New York Times will fall just
below those of the Wall Street Journal.
RESULTS: The average data density of graphics in the Wall
Street Journal was 22.67 numbers per square inch. Those in
USA Today exhibited 3.63 numbers per square inch--a
difference of 19.04 numbers per square inch. However, the
average data density of graphics in the New York Times did
not "fall just below those in the Wall Street Journal." The
average data density of graphics in the New York Times was
12.74 numbers per square inch--9.93 number per square inch
less.
Table 2
NEWSPAPER QUANTITATIVE GRAPHICS
AVERAGE DATA DENSITY AND AVERAGE AREA
NEWSPAPER/ NUMBER AVG. AVG.
YEAR DATA DENSITY AREA
New York Times 1981 111 7.25 10.27
1983 124 9.13 9.76
1985 129 11.57 9.01
1987 141 7.11 12.55
1989 172 21.84 12.71
1991 182 19.53 11.64
USA Today
1983 301 3.28 9.49
1985 280 2.86 9.18
1987 189 2.85 11.50
1989 144 4.40 11.23
1991 150 4.78 11.35
Wall Street Journal
1981 142 27.45 9.51
1983 146 19.94 9.75
1985 266 25.72 7.52
1987 281 20.21 9.20
1989 377 21.44 8.63
1991 375 21.24 8.97
Data densities represented in numbers per square inch.
Areas represented in square inches.
The results indicated by Table 2 show mixed support for
part A of Hypothesis II which states: There will be an
increase from the start to the end of the sampling dates in
the overall average data density of graphics in all three
newspapers. RESULTS: The overall average data density of
graphics in the New York Times and USA Today did indeed
increase from the start of the sampling period to the end.
However, overall average data densities in the Wall Street
Journal decreased during the sampling period.
For the New York Times in 1981 the average data density of
sampled graphics was 7.25 numbers per square inch and in 1991
the average data density was 19.53 numbers per square inch.
However, this increase was far from steady. Sampled graphics
in the New York Times for the years 1981, 1983 and 1985
showed an increase from 7.25 numbers per square inch in 1981
to 11.57 numbers per square inch in 1985. But in 1987 the
average data density showed a decrease of 4.46 numbers per
square inch from the average in 1985. In 1989, the average
data density increased again to 21.84 numbers per square inch
from 7.11 numbers per square inch in 1987.
For USA Today in 1983 the average data density of sampled
graphics was 3.28 numbers per square inch and in 1991 the
average data density was 4.78 numbers per square inch. This
increase was also not steady. Sampled graphics in USA Today
showed a small decrease in 1985 and 1987 of approximately .42
numbers per square inch. These numbers increased in 1989 and
1991 to 4.40 and 4.78 numbers per square inch respectively.
The results for the Wall Street Journal did not agree with
part A of Hypothesis II. The average data density of sampled
graphics decreased from 27.45 numbers per square inch in 1981
to 21.24 in 1991. This decrease, like increases for graphics
in the New York Times and USA Today, was unsteady. The
average data density for graphics in 1983 decreased to 19.94
numbers per square inch and increased to 25.72 in 1985. This
was followed by a decrease to 20.21 numbers per square inch
in 1987 and an increase to 21.44 in 1989.
The results indicated by Table 2 show mixed support for
part B of Hypothesis II which states: There will be an
increase from the start to the end of the sampling dates in
the total number of graphics in all three newspapers.
RESULTS: The total number of sampled graphics in the New York
Times and the Wall Street Journal did increase from the start
of the sampling period to the end. However, the total number
of graphics in USA Today decreased during the sampling
period.
The number of sampled graphics in the New York Times
increased steadily from 111 in 1981 to 182 in 1991. Likewise,
the number of graphics in the Wall Street Journal increased
steadily from 142 in 1981 to 377 in 1989 with a tiny decrease
to 375 in 1991. The number of sampled graphics in USA Today
decreased from 301 in 1983 to 144 in 1991 with a small
increase to 150 in 1991.
The results indicated by Table 2 show mixed support for
part C of Hypothesis II which states: There will be an
increase from the start to the end of the sampling dates in
the overall average area (in square inches) of graphics in
all three newspapers.
RESULTS: The average area of sampled graphics in both the
New York Times and USA Today showed an increase from the
start of the sampling period to the end while the average
area of graphics decreased in the Wall Street Journal during
the sampling period.
The average area of a graphic in the New York Times was
10.27 square inches in 1981 and increased to 11.64 square
inches in 1991. However, the average fluctuated during the
sampling period. The average area of a graphic was 9.76 in
1983, 9.01 in 1985, 12.55 in 1987, and 12.71 in 1989. The
average area of a graphic in USA Today was 9.49 square inches
in 1983 and 11.35 in 1991. This average also fluctuated. The
average area was 9.18 in 1985, 11.50 in 1987 and 11.23 in
1989.
The average area of a graphic in the Wall Street Journal
decreased from 9.51 square inches in 1981 to 8.97 in 1991.
This average also showed unsteadiness. The average was 9.75
in 1983, 7.52 in 1985, 9.20 in 1987 and 8.63 in 1989.
Table 3
NEWSPAPER QUANTITATIVE GRAPHICS
GRAPHICS RELATED TO STORIES
NEWSPAPER/ NUMBER PERCENT
YEAR
New York Times
1981 81 73
1983 91 73 1985 99 77 1987
109 77
1989 109 63 1991 121 66
USA Today
1983 72 24 1985 58 21 1987
46 24 1989 55 38 1991 57 38
Wall Street Journal
1981 31 22 1983 31 21 1985
71 27 1987 91 32 1989 155
41 1991 166 44
The results indicated by Table 3 partially support
Hypothesis III which states: There will be an increase from
the start to the end of the sampling dates in the number of
graphics in all three newspapers that directly relate to
stories. RESULTS: The number of sampled graphics which
related to stories in both the New York Times and the Wall
Street Journal did show an increase from the start of the
sampling period to the end while the number of graphics which
related to stories decreased in USA Today during the sampling
period. However, when the corresponding percentages are
compared, a different picture emerges.
In 1981, the New York Times had 81 graphics, or 73
percent, which related to stories and in 1991, the number
increased to 121 but the percentage decreased to 66 percent.
The numbers for the New York Times were also unsteady during
the sampling period. In 1983, there were 91 graphics, or 73
percent, which related to stories; in 1985, there were 99
graphics, or 77 percent, which related to stories; in 1987,
there were 109 graphics, or 77 percent, which related to
stories; and in 1989, there were 109 graphics, or 63 percent,
which related to stories.
The Wall Street Journal had 31 graphics, or 22 percent,
which related to stories in 1981 and in 1991 the number
increased to 166, or 44 percent. Furthermore, the numbers
during the sampling period were relatively steady in their
increase. In 1983, there were also 31 graphics, or 21
percent, which related to stories; in 1985, there were 71
graphics, or 27 percent, which related to stories; in 1987,
there were 91 graphics, or 32 percent, which related to
stories; and in 1989, there were 155 graphics, or 41 percent
which related to stories.
While USA Today did exhibit a decrease in the number of
stories which related to graphics, 72 in 1983 to 57 in 1991,
the percentage actually increased, from 24 in 1983 to 38 in
1991. The numbers for USA Today during the sampling period
showed unsteadiness with 58 graphics, or 21 percent, related
to stories in 1985; 46 graphics, or 24 percent, related to
stories in 1987; and 55 graphics, or 38 percent, related to
stories in 1989.
Table 4
NEWSPAPER QUANTITATIVE GRAPHICS
GRAPHICS CONTAINING PICTORIAL ELEMENTS
NEWSPAPER/ NUMBER PERCENT
YEAR
New York Times
1981 10 9.1 1983 9 7.3
1985 3 2.3 1987 10 7.1
1989 3 1.7 1991 19 10.4
USA Today
1983 28 9.3
1985 102 36.4 1987 82 43.3 1989
93 64.6 1991 84 56.0
Wall Street Journal
1981 2 1.4 1983 4 2.7 1985
1 0.3 1987 1 0.4 1989 0
0.0
1991 2 0.5
The results indicated by Table 4 support Hypothesis IV
which states: There will be a large increase in the use of
pictorial or decorative elements in the graphics of USA Today
during the sampling period. The use of pictorial or
decorative elements will be negligible in graphics in the
Wall Street Journal and only a minor component of graphics in
the New York Times during the sampling period. RESULTS: The
number of sampled graphics which contained pictorial elements
in USA Today increased sharply from the start of the sampling
period to the end. The number of graphics containing
pictorial elements was indeed negligible in the Wall Street
Journal and a minor component in graphics of the New York
Times. The number of pictorial elements in graphics of USA
Today increased greatly from 28, or 9.3 percent of the total,
in 1983 to 84, or 56 percent in 1991. A large increase was
made from 1983 to 1985 where 102 graphics with pictorial
elements, or 36.4 percent of the total, were counted. In
1987, 82 graphics containing pictorial elements, or 43.3
percent, were counted and in 1989, 93 graphics, or 64.6
percent, were counted.
Graphics with pictorial elements were almost nonexistent
in the Wall Street Journal sample. In 1981, 2 graphics
containing pictorial elements, or 1.4 percent of the total,
were counted. The other years showed similar numbers: 1983, 4
graphics, or 2.7 percent; 1985, 1 graphic, or 0.3 percent;
1987, 1 graphic, or 0.4 percent; 1989, zero graphics; and
1991, 2 graphics, or 0.5 percent.
The New York Times sample contained 10 graphics with
pictorial elements, or 9.1 percent, in 1981; 9 graphics, or
7.3 percent, in 1983; 3 graphics, or 2.3 percent, in 1985; 10
graphics, or 7.1 percent, in 1987; 3 graphics, or 1.7
percent, in 1989; and 19 graphics, or 10.4 percent, in 1991.
CHAPTER FIVE
Discussion
Graphics editors from the three newspapers used for this
research project were contacted and asked to react to the
data and conclusions drawn. The following discussions are
graphics editors' commentary on the content of their
informational graphics.
New York Times
Richard Meislin, a graphics editor at the New York Times
since 1988, said that graphics in the Times must reach high
standards every day in order to be considered for
publication. "The way the Times news judgment works is that a
graphic has to justify the space it takes on its news value.
We have to convince a national editor, for example, that a
two by five-inch graphic tells the story as well as, or
better than, ten extra inches of text. When you come in with
that theory you end up with a different kind of graphic than
a paper like USA Today."
Meislin said the Times tends to be more closely aligned
with the content-driven "Tufte school" than the visual impact
"Nigel Holmes school" when it comes to the substance of their
graphics. "However, we have been trying over the past four
years to make our graphics a bit less gray, a bit more
interesting. We're trying now to strike a balance between
high content and visual appeal in our graphics."<41>
USA Today
Richard Curtis, graphics director at USA Today had "no
comment" on the results of this study.<42>
Wall Street Journal
Jim Condon, graphics director at the Wall Street Journal,
could not be reached for a response to the results of this
study.
SUMMARY:
The overall average data density of a quantitative graphic
in the three newspapers differed widely. Sampled graphics in
the Wall Street Journal had the highest average data density
while those in the New York Times and USA Today had
significantly lower average data densities.
The overall average data density and the overall average
area of quantitative graphics in the New York Times and USA
Today increased over the sampling period while those in the
Wall Street Journal decreased. The total number of sampled
graphics in the New York Times and the Wall Street Journal
increased over the sampling period while the total number of
sampled graphics in USA Today decreased.
The number of sampled quantitative graphics which had a
story relationship increased during the sampling period in
both the New York Times and the Wall Street Journal and
decreased in USA Today. However, when percentages of graphics
with a story relationship are compared, USA Today and the
Wall Street Journal showed increases while the New York Times
showed a decrease.
The number of sampled quantitative graphics which
contained pictorial elements in USA Today increased sharply
during the sampling period. The number was negligible for the
Wall Street Journal and only a minor component of graphics in
the New York Times.
CONCLUSIONS:
Overall, the results of this study were not earthshaking,
nor were they expected to be. The lack of steadiness in some
of the increases and decreases provided some concern, but can
be generally blamed on differences in news coverage during
the years. Some results might raise an eyebrow but it is
hoped that they raise questions instead. The intention of
this work was not to provide the speculation but to start a
foundation which it can be based on.
It is easier to look back at this study and analyze what
could have been done than look to the future and predict what
will come. However, there are some concerns which might have
been addressed more clearly if this type of study was to be
conducted again. Three newspapers were used for this study:
the New York Times, the Wall Street Journal and USA Today.
One concern is that the circulations of the newspapers
selected were too large for a truly representative analysis
of the content of quantitative graphics in the country today.
What would the results have been if the sample of newspapers
used had a much broader range of circulations? An obvious
second concern is that only three newspapers were sampled.
Furthermore, two are located in New York City while the other
is produced in Arlington, Virginia. What would the results
have looked like if a greater number of newspapers which
represented more locations in the United States--or for that
matter, the world--were used? Was there an "East Coast" slant
on the results? What about the use of some (or many!) local
newspapers instead of three which were more national in their
scope?
Another concern is that the sampling period only allowed
the counting of graphics for certain weeks from odd-numbered
years starting in 1981. Was relevant data missed by sampling
only from odd-
numbered years and only from certain weeks? Informational
graphics have appeared in newspapers for decades. What would
the results show if the sampling period started in 1900 or
1920 and counted each year instead of every other year?
The nature of the data density theory tested in this study
only allowed for quantitative graphics to be sampled. The
number of numbers and areas were easily counted. However,
this limitation made necessary the total exclusion of
informational graphics which were explanatory or spatial by
nature since qualities associated with explanations and maps
are difficult, if not impossible, to quantify. How does one
accurately measure the information content of a weather map
or a graphic showing a surgical procedure? It seems that
until such a method is developed much of the analysis of
informational graphics will necessarily remain subjective.
While every attempt was made to conduct and present this
study in a fair and equitable manner, an analysis of the
results does provide some small opportunities for
subjectivity. It should come as no surprise to those familiar
with the three newspapers examined that the results generally
confirmed Hypothesis I. However, the statement that the data
density of graphics in the New York Times will fall just
below those of the Wall Street Journal was, in fact,
confirmed when looking at the most recent results. Graphics
in the New York Times exhibited a large jump in average data
density for the years 1989 and 1991, equaling average data
densities of graphics in the Wall Street Journal for the
period. This can be attributed to a renewed commitment to
increased graphic content on the part of the staff of the New
York Times in recent years.
A look at the results corresponding to Hypothesis II
provided some surprising findings. While the number of
graphics rose modestly and at a steady pace in the New York
Times, the increase was dramatic in the Wall Street Journal.
An overall design change in the Wall Street Journal--from two
to three daily sections in 1984--probably accounted for the
rise. In the case of USA Today, a dramatic fall in the number
of graphics--accompanied by an overall increase in their
average data densities and average areas--could imply that an
effort has been made to make the paper, or maybe just the
graphics, more content-driven.
That contention becomes more likely when the results
corresponding to Hypothesis III are analyzed. The percentage
of graphics related to stories--an indication that they are
being used as adjuncts to text and not simply as context-less
numbers--jumped markedly for USA Today during the sampling
period. Curiously, the same rise occurred for graphics in the
Wall Street Journal, signaling their efforts to accompany
stand-alone graphics with text.
Finally, the results corresponding to Hypothesis IV were
rather surprising. While the percentages of graphics
containing pictorial elements were pretty much as expected
for the New York Times and the Wall Street Journal, those for
USA Today were unexpected. A jump in excess of 45 percent was
experienced during the sampling period. The ease with which
drawings and illustrations can be produced using personal
computers such as the Macintosh--and USA Today's propensity
to employ them--has no doubt contributed to the rise.
The state of informational graphics in newspapers has
received a flurry of attention in recent years. Nearly every
aspect of their production, makeup and use has been
scrutinized and argued over in the professional press. While
much of the writing has been helpful, it is too often full of
opinion and bias and not full enough of careful, reasoned
examination. While some analysts see informational graphics
as a savior for sagging circulations with numbers cloaked in
tiny multi-
hued columns, lines and circles to appeal to those with
little time for quantitative analysis over their morning
cereal bowls, others lament the "dumbing down" of the day's
news into so many nuggets of "eye candy" to be gobbled up at
the pace of a music video. This study is a small attempt to
apply some thoughtful theory into the fray with the hope that
it will begin to form the foundation for further discussion
on a far more objective level.
BIBLIOGRAPHY
Aiges, Scott, "The New Storytellers," Columbia Journalism
Review, March/April, 1991, pp. 36-37.
Anonymous, "Graphics Advice Column," DeadlineMac, October,
1989, p. 1.
, "The Evolving Front Page," Editor & Publisher, April
23, 1988, p. 93.
Anscombe, F.J., "Graphs in Statistical Analysis," American
Statistician, 27:17-21, 1 (1973).
Arkin, Herbert and Colton, Raymond R., Graphs: How to Make
and Use Them (New York: Harper & Brothers, 1940).
Astor, David, "Popularity Rise for Syndicated Graphics,"
Editor & Publisher, June 23, 1984, pp. 34-35.
Baird, Russell N., Turnbull, Arthur T. and McDonald, Duncan,
The Graphics of Communication (New York: Holt, Rinehart and
Winston, 1987).
Baldwin, William, "Publish and Prosper," Forbes, May 18,
1987, pp. 121-123.
Barnhurst, Kevin G., "News as Art," Journalism Monographs,
130:1-
39, December (1991).
and Nerone, John C., "Design Trends in U.S. Front
Pages, 1885-1985," Journalism Quarterly, 68:796-804, 4
(1991).
Beniger, James R. and Robyn, Dorothy L., "Quantitative
Graphics in Statistics: A Brief History," American
Statistician, 32:1-11, 1 (1978).
Beniger, James R. and Westney, D.Eleanor, "Japanese and U.S.
Media: Graphics as a Reflection of Newspapers' Social Role,"
Journal of Communication, 31:14-27, Spring (1981).
Berryman, Gregg, Notes on Graphic Design and Visual
Communication (Los Altos, CA: Crisp Publications, Inc.,
1990).
Bertin, Jacques, Semiology of Graphics (Madison, WI: The
University of Wisconsin Press, 1983).
, Graphics and Graphic Information Processing (Berlin:
Walter de Gruyter, 1981).
Biderman, Albert D., "The Graph as a Victim of Adverse
Discrimination and Segregation: Comment Occasioned by the
First Issue of the Information Design Journal," Information
Design Journal, 1:232-241, 4 (1980).
Brinton, William C., Graphic Methods for Presenting Facts
(New York: McGraw Hill Book Company, Inc., 1914).
Chambers, John M., Cleveland, William S., Kleiner, Beat and
Tukey, Paul A., Graphical Methods for Data Analysis (Belmont,
CA: Wadsworth International Group, 1983).
Chartrand, Robert Lee, Critical Issues in the Information
Age, (Metuchen, NJ: The Scarecrow Press, Inc., 1991).
Cleveland, William S., "Graphics Methods for Data
Presentation: Full Scale Breaks, Dot Charts, and Multibased
Logging," The American Statistician, 38:270-280, 4 (1984).
, The Elements of Graphing Data (Monterey, CA:
Wadsworth Advanced Books and Software, 1985).
and Terpenning, Irma J., "Graphical Methods for
Seasonal Adjustment," Journal of the American Statistical
Association, 77:52-62, 377 (1982).
and McGill, Robert, "Graphical Perception: Theory,
Experimentation, and Application to the Development of
Graphical Methods," Journal of the American Statistical
Association, 79:531-554, 387 (1984).
and McGill, Robert, "Graphical Perception and
Graphical Methods for Analyzing Scientific Data," Science,
August 30, 1985, pp. 828-833.
Covert, Douglas C., "Color Preference Conflicts in Visual
Compositions," Newspaper Research Journal, 9:49-59, 1 (1987).
Curtis, Richard, graphics director for USA Today. Telephone
interview, 20 May 1992.
Cvengros, Stephen, "A New Force in Newsrooms," Design, pp.
34-35, 29 (1988).
DeSanctis, Gerardine, "Computer Graphics as Decision Aids:
Directions for Research," Decision Sciences, 15:463-487, 4
(1984).
Easterby, Ronald and Zwaga, Harm, Information Design,
(Chichester, England: John Wiley and Sons Ltd., 1984).
Ehrenberg, A.S.C., Data Reduction (London: John Wiley & Sons,
1975).
Evangelauf, Jean, "An 'Information Designer' Strives for the
'Elegant and Beautiful'," The Chronicle of Higher Education,
37:A3, May 1 (1991).
Finberg, Howard I., Visual Editing (Belmont, California:
Wadsworth, Inc., 1990).
, "A Rich Tradition," Design, April/May/June, 1991, pp.
17-24.
Fitzgerald, Mark, "Freelance Graphics," Editor & Publisher,
November 9, 1985, p. 18.
, "Front Page Still Off-Limits to Art Departments,"
Editor & Publisher, November 9, 1985, p. 19.
, "Computers Rapidly Replacing Graphic Artists'
Mechanical Tools," Editor & Publisher, November 9, 1985, pp.
19, 36.
, "Graphics Software," Editor & Publisher, September 3,
1988, pp. 12pc, 43pc.
, "Will Artists Become Computer Nerds?," Editor &
Publisher, September 3, 1988, pp. 10pc-11pc, 42pc.
, "Graphics on the Gulf," Editor & Publisher, March 9,
1991, pp. 12-13.
Flury, Bernhard and Riedwyl, Hans, "Graphical Representation
of Multivariate Data by Means of Asymmetrical Faces," Journal
of the American Statistical Association, 76:757-765, 376
(1981).
Forester, Tom, Computers in the Human Context, (Cambridge,
MA: MIT Press, 1989).
Friendly, Jonathan, "USA Today's Fast Gains Raise Both Hopes
and Doubts," The New York Times, April 11, 1983, A20.
Fruitrich, Chris, "Many Hands Make Graphics Work,"
DeadlineMac, December, 1989, p. 1.
Garcia, Mario R., Contemporary Newspaper Design (Englewood
Cliffs, NJ: Prentice Hall, 1987).
and Stark, Pegie, Eyes on the News (St. Petersburg,
FL: The Poynter Institute, 1991).
Garneau, George, "Improving Wire Service Graphics," Editor &
Publisher, March 21, 1987, p. 44.
, "Graphics," Editor & Publisher, March 21, 1987, pp.
46, 48.
, "Infographics," Editor & Publisher, October 22, 1988,
p. 44.
, "Modernizing Through Design," Editor & Publisher,
December 2, 1989, pp. 14-15.
, "Combating Visual Illiteracy," Editor & Publisher,
December 2, 1989, pp. 14-15.
Geraci, Philip C., "Comparison of Graphic Design and
Illustration Use in Three Washington D.C., Newspapers,"
Newspaper Research Journal, 5:29-39, 1 (1983).
Gilbert, E.W., "Pioneer Maps of Health and Disease in
England," Geographical Journal, 124:172-183, (1958).
Goltz, Gene, "Weather Pages," Presstime, March, 1987, pp. 16-
19.
Hartman, John K., "USA Today and Young-Adult Readers: Can a
New-Style Newspaper Win Them Back?," Newspaper Research
Journal, 8:1-14, 2 (1987).
Hilliard, Robert D., "The Graphics Explosion: Questions
Remain About Roles," Journalism Quarterly, 66:192-194, 1
(1989).
Holmes, Nigel, Designer's Guide to Creating Charts and
Diagrams (New York: Watson-Guptill, 1984).
, Designing Pictorial Symbols (New York: Watson-
Guptill, 1985).
, Pictorial Maps: History, Design, Ideas, Sources (New
York: Watson-Guptill, 1991).
Huff, Darrell, How to Lie With Statistics (New York: W.W.
Norton & Company, 1954).
Hundertmark, Jane, "Picturing the News," Publish, April,
1991, pp. 67-70.
Jarvenpaa, S.L., "Empirical Investigation of Tufte's "Lie
Factor" with Computer Generated Graphics," presented to the
Third IFAC/IFIP/IEA/IFORS Conference, Oulu, Finland, 1988.
Johnson, Linda Lee, "Learning Across the Curriculum with
Creative Graphing," Journal of Reading, 32:509-519, 2 (1989).
Kelly, James D., "The Data-Ink Ratio and Accuracy of
Newspaper Graphs," Journalism Quarterly, 66:632-639, 4
(1989).
, "The Effect of Visual Attractiveness on the Accuracy
of Information Recalled from Newspaper Graphs," Dissertation,
Indiana University, 1990.
Kenney, Keith and Lacy, Stephen, "Economic Forces Behind
Newspapers' Increasing Use of Color and Graphics," Newspaper
Research Journal, 8:33-41, 3 (1987).
Kolata, Gina, "The Proper Display of Data," Science, October
12, 1984, pp. 156-157.
Kolers, Paul A., Wrolstad, Merald E. and Bouma, Herman,
Processing of Visible Language 2, (New York: Plenum Press,
1980).
Kornegay, Van, "How Teaching Informs Research in Graphics
Field Experience," Journalism Educator, 46:38-45, 1 (1991).
Kosslyn, Stephen M., "Graphics and Human Information
Processing," Journal of the American Statistical Association,
80:499-512, 391 (1985).
Lester, Paul, "Use of Visual Elements on Newspaper Front
Pages," Journalism Quarterly, 65:760-763, 3 (1988).
Lordan, Edward James, "Informational Graphics as a Component
of Comprehension in Print Media News Articles," Dissertation,
Syracuse University, 1990.
Macdonald-Ross, Michael, "How Numbers are Shown," Audio-
Visual Communication Review, 25:359-409, 4 (1977).
Malcolm, Andrew H., "Data-Loving Japanese Rejoice on
Statistics Day," The New York Times, October 28, 1977, pp.
A1, A7.
McGill, Robert, Tukey, John W. and Larsen, Wayne A.,
"Variations of Box Plots," The American Statistician, 32:12-
16, 1 (1978).
Meggs, Philip B., "Post War, Post Mortem," Print,
September/October, 1991, pp. 54-63, 132.
Meislin, Richard, graphics editor for the New York Times.
Telephone interview, 22 May 1992.
Menichini, Mike, "Readers Flock to Get Unpublished Persian
Gulf Graphics Package," Editor & Publisher, September 29,
1990, p. 10.
Mintz, Louis, "Graphics Networks," Presstime, September,
1986, pp. 12-14.
Moen, Daryl, "Graphics Require Careful Editing," Presstime,
August, 1989, pp. 15-17.
, "When TV and Newspapers Go to War," Design,
April/May/June, 1991, pp. 3-5.
, "Anatomy of a Trauma," Design, April/May/June, 1991,
pp. 6-16.
Monmonier, Mark and Pipps, Val, "Weather Maps and Newspaper
Design: Response to USA Today?," Newspaper Research Journal,
8:31-42, 4 (1987).
Needles, Loren R., "Torture a Statistic and It Will Agree to
Anything," Hard Copy, Fall, 1989, p. 6.
Pasternack, Steve and Utt, Sandra H., "Reader Use &
Understanding of Newspaper Infographics," Newspaper Research
Journal, 11:28-41, 2 (1990).
Patton, Phil, "Up From Flatland," The New York Times, 1992,
pp. 29-31.
Pedersen, B. Martin, Graphis Diagram, (Zurich: Graphis Press
Corp., 1988).
Perri, Lynne, "Ignorance Keeps Him Searching," Design, March,
1988, pp. 4-7.
Robertson, Bruce, How to Draw Charts and Diagrams
(Cincinnati, OH: North Light Books, 1988).
Schmid, Calvin F. and Schmid, Stanton E., Handbook of Graphic
Presentation (New York: Ronald Press, 1979).
Seelye, Katharine, "Al Neuharth's Technicolor Baby," Columbia
Journalism Review, March/April, 1983, pp. 27-35.
Shapiro, Howard S., "Giving a Graphic Example: The Increasing
Use of Charts and Maps," Nieman Reports, 36:4-7, 1 (1982).
Shiffrin, Richard M., "Capacity Limitations in Information
Processing, Attention, and Memory," in Estes, W.K., eds.,
Handbook of Learning and Cognitive Processes, (Hillsdale, NJ:
Lawrence Erlbaum Associates, 1976).
Silverstein, Louis, Newspaper Design for the Times (New York:
Van Nostrand Reinhold, 1990).
Silverstone, Stuart, "Saying It with Images," Aldus Magazine,
May/June, 1991, pp. 27-29, 56.
, "Visualizing Asthma," Aldus Magazine, July/August,
1991, pp. 40-41, 43.
, "The Picturing of Information," Aldus Magazine,
September/October, 1991, pp. 18-19, 21, 23-24, 55.
, "Signposts of our Time," Publish, November, 1991, p.
120.
and Webb, Craig L., "Many Newspapers, Small and Large,
Turn to 'The Mac'," Presstime, April, 1986, pp. 23-26.
Sissors, Jack Z., "Some New Concepts of Newspaper Design,"
Journalism Quarterly, 42:236-242, (1965).
Smith, Edward J. and Hajash, Donna J., "Informational
Graphics in 30 Daily Newspapers," Journalism Quarterly,
65:714-718, 1 (1988).
Smith, Ron F., "How Design & Color Affect Reader Judgment of
Newspapers," Newspaper Research Journal, 10:75-85, 2 (1989).
Spence, Ian, "Visual Psychophysics of Simple Graphical
Elements," Journal of Experimental Psychology: Human
Perception and Performance, 16:683-692, 4 (1990).
Spoehr, Kathryn T. and Lehmkuhle, Stephen W., Visual
Information Processing (San Francisco: W.H. Freeman and Co.,
1982).
Strugatch, Warren, "Disposing of Graphics Garbage," Editor &
Publisher, May 27, 1989, p. 24.
Tankard, James W., Jr., "Effects of Chartoons & Three-
Dimensional Graphs on Interest & Information Gain," Newspaper
Research Journal, 10:91-103, 3 (1989).
, "Quantitative Graphics in Newspapers," Journalism
Quarterly, 64:406-415, Summer/Fall (1987).
Terrell, Pamela M., "Art," Presstime, February, 1989, pp. 20-
27.
Thrower, Norman J.W., "Edmond Halley as a Thematic Geo-
Cartographer," Annals of the Association of American
Geographers, 59:652-676, December (1969).
Tufte, Edward R., The Visual Display of Quantitative
Information (Cheshire, CT: Graphics Press, 1983).
, "The Power of Graphics," PC Computing, August 1988,
p. 88.
, "Attention to Detail, or Less is a Bore," PC
Computing, Nov., 1988, p. 110.
, Envisioning Information (Cheshire, CT: Graphics
Press, 1990).
Tukey, John W., "Some Graphic and Semigraphic Displays," in
Bancroft, T.A., ed., Statistical Papers in Honor of George W.
Snedecor, 1st. (Ames, IA: Iowa State University Press, 1972).
Utt, Sandra H. and Pasternack, Steve, "Front Pages of U.S.
Daily Newspapers," Journalism Quarterly, 61:879-884, (1984).
and Pasternack, Steve, "How They Look: An Updated
Study of American Newspaper Front Pages," Journalism
Quarterly, 66:621-627, 3 (1989).
Verity, John W., "Graphically Speaking with Dr. Edward R.
Tufte," Datamation, 1985, pp. 88-92.
and Nathans, Jessie, "The War on Information Clutter,"
Business Week, April 29, 1991, p. 66.
Wainer, Howard, "Making Newspaper Graphs Fit to Print," in
Kolers, Paul A., Wrolstad, Merald E. and Bouma, Herman, eds.,
Processing of Visible Language 2, (New York: Plenum Press,
1980).
Wainer, Howard, "How to Display Data Badly," The American
Statistician, 38:137-147, 2 (1984).
and Francolini, Carl M., "An Empirical Inquiry
Concerning Human Understanding of Two-Variable Color Maps,"
The American Statistician, 34:81-93, 2 (1980).
and Thissen, David, "Graphical Data Analysis," Annual
Review of Psychology, 32:191-241, (1981).
Walston, John, "How Others Get the Facts," Design, pp. 18-19,
27 (1988).
White, Jan V., Using Charts and Graphs (New York: R.R. Bowker
Company, 1984).
Wright, Patricia, "Functional Literacy: Reading and Writing
at Work," Ergonomics, 31:265-290, 3 (1988).
Wurman, Richard Saul, Information Anxiety (New York:
Doubleday, 1989).
Zoglin, Richard, "How Dailies Cover a TV War," Time, February
11, 1991, p. 78.
1 Edward R. Tufte, "Attention to Detail, or Less is a Bore,"
PC Computing, Nov., 1988, p. 110.
2 Ibid.
3 Stuart Silverstone, "Saying It with Images," Aldus
Magazine, May/June, 1991, p 28.
4 Edward R. Tufte, The Visual Display of Quantitative
Information (Cheshire, CT: Graphics Press, 1983) p. 162-166.
5 Jonathan Friendly, "USA Today's Fast Gains Raise Both Hopes
and Doubts," The New York Times, April 11, 1983, A20.
6 James D. Kelly, "The Data-Ink Ratio and Accuracy of
Newspaper Graphs," Journalism Quarterly, 66:632-639, 4
(1989).
7 Tufte, Visual Display, 9.
8 Ibid. 13.
9 Howard Wainer and David Thissen, "Graphical Data Analysis,"
Annual Review of Psychology, 32:196, (1981).
10 Calvin F. Schmid and Stanton E. Schmid, Handbook of
Graphic Presentation (New York: Ronald Press, 1979).
11 Howard Wainer, "How to Display Data Badly," The American
Statistician, 38:138, 2 (1984).
12 Jacques Bertin, Graphics and Graphic Information
Processing (Berlin: Walter de Gruyter, 1981) p. 22.
13 Jacques Bertin, Semiology of Graphics (Madison, WI: The
University of Wisconsin Press, 1983).
14 William S. Cleveland, The Elements of Graphing Data
(Monterey, CA: Wadsworth Advanced Books and Software, 1985).
15 A.S.C. Ehrenberg, Data Reduction (London: John Wiley &
Sons, 1975).
16 Keith Kenney and Stephen Lacy, "Economic Forces Behind
Newspapers' Increasing Use of Color and Graphics," Newspaper
Research Journal, 8:40, 3 (1987).
17 Howard S. Shapiro, "Giving a Graphic Example: The
Increasing Use of Charts and Maps," Nieman Reports, 36:7, 1
(1982).
18 Edward J. Smith and Donna J. Hajash, "Informational
Graphics in 30 Daily Newspapers," Journalism Quarterly,
65:714-718, Fall (1988).
19 Howard Wainer, "Making Newspaper Graphs Fit to Print," in
Paul A. Kolers, Merald E. Wrolstad and Herman Bouma, eds.,
Processing of Visible Language 2, (New York: Plenum Press,
1980).
20 James W. Tankard Jr., "Quantitative Graphics in
Newspapers," Journalism Quarterly, 64:406-415, Summer/Fall
(1987).
21 Steve Pasternack and Sandra H. Utt, "Reader Use &
Understanding of Newspaper Infographics," Newspaper Research
Journal, 11:28-41, 2 (1990).
22 James W. Tankard Jr., "Effects of Chartoons & Three-
Dimensional Graphs on Interest & Information Gain," Newspaper
Research Journal, 10:91-103, 3 (1989).
23Kelly, "Data-Ink Ratio," 639.
24 Daryl Moen, "Graphics Require Careful Editing," Presstime,
August, 1989, pp. 15-17.
25 Silverstone, "Saying It with Images," 28.
26 William S. Cleveland and Robert McGill, "Graphical
Perception: Theory, Experimentation, and Application to the
Development of Graphical Methods," Journal of the American
Statistical Association, 79:531-554, 387 (1984).
27 Michael Macdonald-Ross, "How Numbers are Shown," Audio-
Visual Communication Review, 25:359-409, 4 (1977).
28 Bertin, Semiology of Graphics.
29Tufte, Visual Display, 96.
30 Kelly, "Data-Ink Ratio," 639.
31 Tufte, Visual Display, 139.
32 Wainer and Thissen, "Graphical Data Analysis," 222.
33 Tufte, Visual Display, 57.
34 S.L. Jarvenpaa, "Empirical Investigation of Tufte's "Lie
Factor" with Computer Generated Graphics," presented to the
Third IFAC/IFIP/IEA/IFORS Conference, Oulu, Finland, 1988.
35 Tufte, Visual Display, 13.
36 Ibid. 162-166 .
37Ibid. 168.
38 Ibid.
39 Mario R. Garcia, Contemporary Newspaper Design (Englewood
Cliffs, NJ: Prentice Hall, 1987). Garcia, Contemporary
Newspaper Design; Edward J. Smith and Donna J. Hajash,
"Informational Graphics in 30 Daily Newspapers," Journalism
Quarterly, 65:714-718, Fall (1988). Smith and Hajash,
"Graphics in 30 Daily Newspapers;" Keith Kenney and Stephen
Lacy, "Economic Forces Behind Newspapers' Increasing Use of
Color and Graphics," Newspaper Research Journal, 8:33-41, 3
(1987). Kenney and Lacy, "Increasing Use of Color and
Graphics;" and Kelly, "Data-Ink Ratio".
40 Smith and Hajash, "Graphics in 30 Daily Newspapers," 716.
41Richard Meislin, graphics editor for the New York Times.
Telephone interview, 22 May 1992.
42Richard Curtis, graphics director for USA Today. Telephone
interview, 20 May 1992.
###