Writing in the Harvard Business Review recently, chart designer Scott Berinato proffered a vivid example of the profoundly misleading influence that clever chart visuals can exert. In emails, engineers from the leading Japanese airbag manufacturer Takata allegedly changed the colours or lines in a graphic to “divert attention” from the actual details of test results to try “to dress them up.” The results had actually shown that the airbags can explode when deployed, scattering debris at dangerously high velocity into the vehicle’s cabin.
From a Gestalt perspective, altering the colour and line characteristics in a chart can be a potent and sometimes mendacious way of bringing figures of attention out of the amorphous ground of huge data samples.
A primary question to ask yourself when presented with a chart is “What does the chart designer want me to see?” Charts projected on a screen in large meetings or on C-SPAN will have one thing in common: they make it virtually impossible for the assembled viewers to read the actual, small-print figures the lines are supposed to be representing.
One especially egregious example cited by Berinato appeared in the Boston Globe newspaper last year. Purporting to plot the rate of abortions and cancer screening services year by year from 2006 to 2013, the graphic showed a red line depicting abortions climbing inexorably – and steeply – every year and a soft pink line representing cancer screening descending equally steeply year on year. The lines intersected dramatically in 2010, implying that abortions were surging at the expense of cancer screenings.
However, if the chart makers had used the same y-axis to plot the results instead of different ones that exaggerated the apparent increase in abortions, it would immediately become clear that abortions actually remained virtually steady while cancer screenings did indeed decline. No significant relationship existed between the two trends at all, yet the chart implied that there was a dramatic one.
In Part Three, I’ll explore more about Gestalt visualisations and ways you can immunise yourself against data manipulation.
