Posts Tagged ‘deception’

Reality Inverted

Reuters graph implies the opposite of what’s really happening

Graphics — especially simple graphics— should take into account reader’s expectations, not require careful study to correctly interpret their meaning. This Reuters graph inverts the vertical axis, creating the impression that deaths went down when in fact deaths went up. This graph runs counter to a well established convention that y-values increase from the bottom to the top of a graph. Breaking convention produces a graph that seriously misleads. A redrawn version of the graph instantly delivers the correct impression.

florida gun deaths    florida gun deaths-2


Consistency is Essential

Pew Research Center’s Project for Excellence in Journalism produces a less than excellent graph

Readers are often in a hurry and they appreciate the use of graphics to communicate quickly. Graphic artists must be careful to not mislead a reader who does not have the time to carefully scrutinize and interpret a simple graph. The problem with this pie chart is that a quick glance creates the impression that the Kindle Fire has a market share of 21%. A more careful inspection reveals that the correct value is about 10%.

Most readers will assume that all the callouts in a single pie chart will refer to the whole pie (100%). In this pie chart the 21% callout refers to the 48% slice. This misleads the reader. Don’t do this!

Linear Extrapolation

December 28, 2011 Leave a comment

Google makes a very optimistic projection

Drawing a straight line through the existing data, I get 150 million, Google projects 400 million.Who’s better at line drawing?

Dueling Graphics

Wall Street Journal and others

The WSJ’s attempt to show that the rich are not really all that rich set off a series of dueling graphs.The graph was promptly redrawn by Mother Jones and redrawn once again by blogger Gregg Hilferding. This aptly demonstrates that the bins used for aggregating data into bars should have some basis in reality. Viz., defining anyone earning less than $100,000 as “poor” is certainly disingenuous. If the policy being debated affects all those earning more than $200,000, dividing that group into many small bins leads us away from the issue at hand. A graph with just three bins may indeed provide the most insight, but only if the bins were properly constructed.