4/12/2012

Consider Your Message When Choosing What Chart to Use

 

Originally published by Naomi Robbins on blogs.forbes.com. Republished with permission.

 

Different chart types emphasize different aspects of the data. I show a simple data set and plot it three simple ways to show that each emphasizes a different message. All use bar charts.

The data consist of  the number of singers and musicians as well as the number of dancers employed in several states. The data come from the May 2011 State Occupational Employment Data from the Bureau of Labor Statistics.

Figure 1. Stacked bar charts emphasize the totals.

Figure 1 plots the data using a stacked bar chart. I don’t usually use stacked bar charts since it is too difficult to compare categories that are not at the baseline; in this case, the dancers. However, here our primary interest is in the total and showing the breakdown between musicians and dancers is of incidental interest. The message is conveyed in the chart title: that NY and CA have more combined musicians, singers and dancers than the other states shown.

Now suppose we are interested in comparing musicians and singers to dancers. A grouped bar chart is much better at showing this comparison. Again, the title emphasizes the message.
 

Figure 2. Grouped bars are far superior for comparing categories.

Finally, suppose we want to compare musicians and dancers separately. Since grouped bar charts have extraneous information between the comparisons of interest, a side-by-side bar chart might be more effective.

Figure 3. Side-by-side bar charts ease comparisons within a category.

Figure 3 facilitates comparisons within a category. It is much easier to compare the number of dancers in these states with the side-by-side bar chart. Plotting dancers and musicians in separate panels is another option here.

In all three figures we have the same three variables: the state, the number of musician and singers, and the number of dancers. Each of the figures emphasizes a different message. Think about the story you want to tell before picking a chart type.

Comments