From 91997e6a0ad0d0be80d1421c9094789d7e2ac2c2 Mon Sep 17 00:00:00 2001 From: Nico Sarian Date: Tue, 9 Jun 2026 20:33:13 -0400 Subject: [PATCH] Assignment 2 Completed --- 02_activities/assignments/assignment_2.md | 63 ++++++++++++++++++----- 1 file changed, 49 insertions(+), 14 deletions(-) diff --git a/02_activities/assignments/assignment_2.md b/02_activities/assignments/assignment_2.md index 7bb5b2df7..7447f9232 100644 --- a/02_activities/assignments/assignment_2.md +++ b/02_activities/assignments/assignment_2.md @@ -9,27 +9,62 @@ - You can find data visualizations at https://public.tableau.com/app/discover or https://datavizproject.com/, or anywhere else you like! - For each visualization (good and bad): - Explain (with reference to material covered up to date, along with readings and other scholarly sources, as needed) why you classified that visualization the way you did. - ``` - Your answer... - - - - - - + + I classify Minard's flow map of Napoleon's 1812 Russian campaign as a strong + visualization for three reasons. + + First, it achieves an exceptionally high data-ink ratio (Tufte, "The Visual + Display of Quantitative Information," 1983). Six variables — army size, two + spatial coordinates (latitude/longitude), direction of travel, location relative + to dates, and temperature on the retreat — are encoded without decorative + clutter. Nearly every mark carries information, which is the principle Tufte + uses to define graphical excellence. + + Second, it uses pre-attentive visual channels effectively. The width of the + band redundantly encodes the magnitude of the army through position and size, + two of the most accurate perceptual channels in Cleveland and McGill's (1984) + ranking of elementary graphical-perception tasks. The collapse of the band from + ~422,000 to ~10,000 men is legible instantly, before any reading of numbers + occurs. + + Third, it tells a story (a stated learning outcome). The dark outbound flow and + the pale returning flow, anchored to a temperature line below, let the viewer + reconstruct cause and consequence — the catastrophic toll of the winter retreat — + in a single integrated frame. This narrative integration of space, time, and + magnitude is why Tufte called it possibly "the best statistical graphic ever + drawn." + + I classify the Fox News ACA-enrollment bar chart as a poor visualization for three reasons. + + First, it violates the proportional-ink / Lie Factor principle (Tufte 1983). The chart compares a March 27 enrollment figure of about 6 million to a target of 7 million, but the 6-million bar is drawn at roughly one-third the height of the 7-million bar. The bars' visual sizes are wildly disproportionate to the values they represent; Tufte's Lie Factor (size of effect shown ÷ size of effect in data) is far from the ideal of 1, manufacturing a sense of failure not present in the numbers. + + Second, it manipulates the baseline. The y-axis does not begin at zero. For bar charts, where the encoded quantity is bar length/area, a non-zero baseline distorts the very channel the reader is meant to decode (Cleveland & McGill, 1984, identify length/position judgments as the perceptual basis of bar charts). Truncating that baseline breaks the mapping between data and ink. + + Third, it fails the test of integrity and equity. Cairo (*The Functional Art*, 2013; *How Charts Lie*, 2019) argues a chart's first duty is to be truthful and to let readers draw their own conclusions. This graphic instead engineers a predetermined political conclusion, encoding rhetoric rather than data — the opposite of supporting critical, informed consumption. - ``` - How could this data visualization have been improved? - ``` - Your answer... - - + + Minard's flow map of Napoleon's 1812 Russian campaign: + First, the graphic relies on a brown/black versus tan colour contrast to + distinguish the advance from the retreat. This single redundant cue can fail for + viewers with low vision or in poor reproductions; adding a clear directional + label or arrowheads on each band, and testing the palette for sufficient + luminance contrast (WCAG-style guidance), would make the advance/retreat + distinction non-reliant on a single channel. + Second, the temperature scale at the bottom is in Réaumur degrees and the + distances/dates are sparse, demanding prior context most modern readers lack. A + brief integrated legend converting units (Réaumur to Celsius), and a short title + caption stating the question being answered, would lower the barrier to entry and + make the chart self-explanatory for a general audience rather than only for + readers already familiar with the campaign. + First, and most fundamentally, the y-axis should start at zero so that bar heights are proportional to the values (Tufte 1983; Cairo 2019). With a zero baseline, 6 million versus 7 million reads as a modest shortfall — roughly 86% of target — rather than a near-total collapse. This single change restores a Lie Factor close to 1 and lets the bars do honest perceptual work. - ``` + Second, the chart should add context and accurate labelling to tell the real story. Showing the enrollment trajectory over time (a line or column series across the sign-up period) rather than two isolated snapshots would convey momentum an the late surge, and clear value labels plus a neutral, descriptive title would let the audience interpret the data themselves. Choosing an appropriate chart type for a part-to-whole-against-target comparison — for example a labelled progress bar o bullet chart (Few, *Show Me the Numbers*, 2004) — would communicate "progress toward goal" far more truthfully than two mismatched columns. + - Word count should not exceed (as a maximum) 500 words for each visualization (i.e. 300 words for your good example and 500 for your bad example)