BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

Must Zero Be Included on Scales of Graphs? Another Look at Fox News' Graph and Huff's Gee-Whiz Graph

This article is more than 10 years old.

My August 4th post calls a graph from Fox News misleading since it is a bar graph with a non-zero baseline. The blog Fallacy Files credits my post as a source for its entry on graphs with no zero on the scale.  I take issue, however, with their title.  The post is entitled “A ‘Gee-Whiz’ Graph,” borrowing a term coined by Darrell Huff in his classic little book How to Lie with Statistics.  The problem is that Huff's "Gee-Whiz Graph"--a graph that deceives since it's missing zero on the scale--is generally a line graph, such as in Figures 1 and 2 below, but the graph from Fox News was a bar graph.  This may seem like a minor point, but it is in fact quite significant.  I disagree with Huff that all graphs require a zero baseline; line graphs do not always require one.  (I should mention here that with the exception of the Gee-Whiz chapter, I am in fact a big fan of most of Huff's book!)

Figure 1. Source: Chapter 5 of How to Lie with Statistics by Darrell Huff Click to enlarge

Figure 2. Source: Chapter 5 of How to Lie with Statistics by Darrell Huff Click to enlarge

Why am I saying that the Fox News’ graph requires a zero baseline but the Huff graphs do not? The Fox News’ graph is a bar graph while these graphs from Huff are line graphs. We judge the value of a bar by its length. Any length begins at zero. We judge the value of a point on a line by its position along a scale. This judgment does not depend on zero. Even if you think you judge a bar graph by the position of the top of the bar, you cannot help but see the length of the bar.

Most other graph experts I respect also disagree with Huff on this point. In The Elements of Graphing Data, William S. Cleveland calls the graphs with zero in these figures “a waste of space that shows very little quantitative information beyond what could be conveyed in one sentence.” In Edward Tufte’s one day course on Information Design that I attended, Tufte said “Darrell Huff was wrong.” Note that the graphs with zero hide the varying rates of change in these figures, something that may be important for some audiences to know. My position is that whether zero needs to be included on graphs in which we judge position depends on the audience. For educated audiences who can be expected to read tick labels, zero is not needed. Some graphs for the general public might be misinterpreted without zero.

Not including zero in the baseline is one way to exaggerate a trend for those who don't read labels carefully.  Can you think of others? That will be the topic of my next post.