December 14, 2011 Off

Using indexing to avoid the lie factor – Fox News Chart redesign

By
in Experiments in Info Design, Not So Great Information Design

Ran across an interesting article on a graphic shown on Fox News regarding unemployment under President Obama.  Here’s the original chart:

Now, it’s probably just a data input mistake, but notice how the 8.6% on the far right isn’t exactly where it should be.  Here’s a couple horizontal lines to help you out:

It’s not really that this bothers me much; mistakes happen.  What I didn’t like was moveon.org’s redesign:

which they claim is more “honest”.  Unfortunately, they forgot about a little thing called the Lie Factor (a la Tufte).  In this particular chart, the most recent drop in unemployment appears to be about 33% in the chart, when in fact it’s only about 4.5% (0.4% change on a base of 9%).

This is a problem I run into frequently – when the chart creator is most interested in showing the change in a variable that starts at a high base. Another common example is a quality rate that varies between 99% and 100% – how do you determine where to set the Y-axis start and end points without skewing the data?

My suggestion is to think about the question you’re really trying to answer.  In the chart above, the author is most interested in the change in unemployment during the Obama administration.  The key word there is change – the relative start and end points aren’t nearly as important as whether it went up or down.  This is a great time to use indexing to eliminate the lie factor and tell an honest story.  Here’s a quick example using the data above, but indexing the unemployment rate back to January of 2009 when Obama was inaugurated:

This chart tells a slightly different story than either the original or the redesign.  Obama had a rough couple of years, but in 2011 it looks like there has actually been some progress.  Still, the unemployment rate is 10% higher now than when he took office (note:  I’m not arguing causality here – my economics professors can rest easy).

This redesign isn’t perfect, and it would be difficult to show only this chart in a larger conversation about unemployment, but if you’re looking to show the change in the unemployment rate during Obama’s presidency, I would argue that this is a more effective way to do so than either the flawed original or the deceitful redesign.

Comments

comments

Powered by Facebook Comments

Comments are closed.