Graphical Competence: Why Organizations Produce Bad Graphics

Overview

Every other skill in this pack fixes a display; this one fixes the shop that made it. Tufte's diagnosis in VDQI Chapter 3 ("Sources of Graphical Integrity and Sophistication," pp. 79–87) is that lying and inept graphics are not primarily technical accidents — bias and stereotyping account for some distortions, but the primary causes are the prevailing skills, attitudes, and organizational structure of the people who design and edit statistical graphics (p. 79). Two working doctrines — statistics are boring and graphics are for the unsophisticated — hand control of data displays to a single expertise (the artistic), and the predictable products are graphics that lie, that use only the simplest designs, and that miss the real news in the data (p. 87). The remedy is structural: graphical competence demands three different skills at once, and the substantive and statistical two must actually participate.

Boundaries with siblings: the mechanics of graphical lies (Lie Factor, design variation, the six integrity principles of p. 77) live in tufte-graphical-integrity; single-artifact corruption detection lives in tufte-evidence-corruption (whose cross-book note defers to this chapter for the causal account); the institutional critique of slideware is tufte-cognitive-style. This skill carries the organizational diagnosis, the sophistication metric, and the staffing remedy.

§1. The Diagnosis — Three Organizational Causes (pp. 79–81)

1. Lack of quantitative skills among professional artists (p. 79). What lurks behind the inept graphic, in Tufte's diagnosis, is missing judgment about quantitative evidence. Nearly all who produce graphics for mass publication trained exclusively in the fine arts, with little experience analyzing data — and even graphical-design textbooks are silent on how to think about numbers. The trade vocabulary ("creative," "concept," "style" in every permutation) is Big Think jargon for the small task of drawing a time-series a few data points long, and advancement goes to those who beautify data, never mind statistical integrity. Tufte's footnote sharpens the historical point: Playfair's original work did not contain the errors now routine, and evidence gathered by the American Statistical Association's graphics committee showed formal training in graphic presentation in marked decline (p. 79, fn. 1, citing Wainer) — the incompetence is a regression, not a baseline.

2. The doctrine that statistical data are boring (pp. 79–80). Exhibits: Time's first full-time chart specialist describing the challenge as presenting statistics as a visual idea rather than a tedious parade of numbers; a design handbook opening its statistical-charts chapter with "Why are statistics so boring?" The doctrine's real function is political — it serves bureaucratic struggles over a publication's resources. If the numbers are dull, then an artist, indeed an Art Department and an Art Director, are required to animate them; the doctrine thereby moves data graphics under the control of artists rather than of the people who write the words and know the substance. As the art bureaucracy grows, style replaces content — and the word people, having lost the space, console themselves that statistics really are tedious (p. 80).

"If the statistics are boring, then you've got the wrong numbers." — Tufte, VDQI, p. 80

Finding the right numbers takes as much specialized skill — statistical skill — and hard work as producing a beautiful design or covering a complex story (p. 80). Boredom is a sourcing failure, not a decoration deficit.

Historical cost of the doctrines (p. 53). Tufte dates the damage: the assumption that graphics were mainly devices for showing the obvious to the ignorant governed the graphically barren years from 1930 to 1970, sending the field down two fruitless paths — overdecorating charts so the supposed dullards would stay awake, and treating lie-detection as the field's main analytic task. It ended only when Tukey's work in the late 1960s made graphics respectable as instruments for reasoning about quantitative information. The doctrines this chapter names are not harmless attitudes; they stalled a discipline for four decades.

3. The doctrine that graphics are only for the unsophisticated reader (pp. 80–81). Exhibits: Consumer Reports producing the first test issue of its children's magazine with a sugar article longer on graphics than on information, for fear children might be overwhelmed by too many facts; an art director responsible for ~3,000 graphics a year (some 2.5 billion printed images) explaining that graphics exist to lure the reader's attention away from the advertising rather than to explain the news; a network TV news director's rule, "If you have to explain it, don't use it" (p. 81) — a line of thinking that terminates in the company-cafeteria pictograph Mary Eleanor Spear had already catalogued in 1952 as an unnecessary chart.