Thumbing through the February 13, 2014 issue of Gun Digest magazine, I was excited to see data on a recent survey of first-time gun buyers, attributed to the National Shooting Sports Foundation (NSSF), the trade association for the shooting, hunting and firearms industry.
Two pie charts, shown below, describe (1) the percentage of new shooters who engage in certain activities (target shooting, hunting, plinking, etc.), and (2) how frequently new shooters shoot.
These are both very important pieces of information, but what’s wrong with this picture?
If you answered that the pieces of the pie chart do not add up to a meaningful whole, you are correct. The first pie chart totals 264% and the second totals 80.3%. A meaningful whole in this case would be 100% of new shooters.
Pie charts are appropriate visual displays of information when we want to show the relative sizes or proportions of different phenomena as a part of a fixed whole. If one slice of the pie grows, another slice has to shrink. You can’t just expand the pie (to 264%). If you remove a slice of the pie, the other slices have to grow. You can’t just shrink the pie (to 80.3%).
A pie chart is not appropriate for a situation in which a single respondent can choose more than one category (a new shooter can be a target shooter and a hunter and a plinker, for example). Or when there are categories of responses that are not displayed (19.7% of new shooters shoot less than once a week or did not respond to this particular question, we can infer). In these cases, a bar chart is more appropriate to display the relative sizes of phenomena.
For a really excellent discussion of pie charts and their potential pitfalls, see “Understanding Pie Charts” on the eagereyes blog.