A Design Problem

I found this graph on the Web site of SAS, a leader in the business intelligence software market. It served as an example of the data presentation capabilities of SAS's software, which are elaborate but not always well used.

My Analysis

Notice the following problems in the example above:

• The attempt to use a graph to display both the shape of the data and provide precise numeric measures results in a cluttered presentation that is hard to read.
• The use of a legend requires that you bounce back and forth between the lines and the legend to determine which lines belong to which cities.
• More visual emphasis has been given to the area surrounding the data region of the graph, rather than to the data itself.

A Solution

It took me 10 minutes to enter the data and produce the following alternative:

Assuming that the shape of the data and precise monthly temperatures are both needed by readers, this graph provides a simple but elegant solution.

• A graph and a table have been integrated to do what each does best: the graph shows the shape of the data and the table gives the exact temperatures.
• The inconvenience of the legend has been removed by placing the city labels next to the lines themselves.
• It is now easy to focus on the shape of the data in the graph without distraction from the numbers.

SAS' Response

SAS has set a new precedent that I would love to see other software vendors follow. A talented member of the SAS team, Robert Allison, agreed that the original graph had problems and took it upon himself to demonstrate that SAS's software could be used to not only duplicate my solution, but surpass it. Here's the improved solution that Robert sent to me:

Notice how Robert's addition of the seasons and a reference line to mark the freezing point have enriched the graph with meaningful context.

I appreciate SAS's thoughtful response and excellent attitude.