Graphical Presentation of Data

Choosing Tables or Graphs

Graphs are preferable for some situations and tables for others. The following list points out the advantages of each:

• Graphs show the big picture: patterns, trends, correlations, and the general shape of the data, while tables show exact values and offer precision.
• It is easy to detect extreme values such as the maximum and the minimum in a graph, while it is more difficult to do so with a table.
• Outliers, which are data points far from the rest of the data, are easier to spot on a graph, while they are more difficult to detect on a table.
• Graphs help discover data errors since problems such as an average value greater than a maximum value can be spotted easily. Tables are less effective in highlighting data errors.
• A great deal of information can be shown in a small space with a graph, while this is not true with tables.
• Graphs are appropriate for paper documents, Web documents, computer screens, or projector screens, while large tables are not appropriate for projector screens since the audience cannot see the details.
• Tables are useful for looking up values, while graphs offer only approximate values.
• Tables can accommodate a number of variables with different units of measures more easily than can graphs.
• The reader can use the data from tables for other purposes such as calculations and drawing other figures. This is more difficult with data from graphs.

Designing Graphs

Good graphs are powerful tools to visualize and understand data. Unfortunately, graphs can also confuse, mislead, or even deceive. This section provides principles of graph design to enable the reader to design effective graphs and to avoid common mistakes.

The data stand out in an effective graph. The designer should emphasize the data and de-emphasize [Page 444]everything else. Grid lines, if used, should be in the background. This can be achieved by making them a pale gray or by using adotted line. The plotting symbols and lines should be visually prominent; care must be taken so that the symbols are not hidden by tick marks, axes, grid lines, other data points, or other graphical elements. Clutter has no place in a graph. Too many tick marks or tick mark labels are a form of clutter. So are too many decimal places. The number of decimal places in labels should be appropriate for the data.

Figure 1 Examples of a Graph, a Table, a Sentence, and a Text Box (Not All the Same Data)

One form of clutter is adding a pseudo third dimension to bars, pies, lines, or other graphical elements. The unnecessary dimension often distorts the data. It always adds clutter. If bars are drawn with depth, the reader does not know how to read the bar. Is the value read from the front where the arrow on the X bar points or from the back where the arrow on the Y bar points? It turns out that the way to read these bars depends on the software that was used to create them, but the reader rarely knows what software was used. Two-dimensional bar charts are unambiguous. If the designer chooses to use a bar chart and knows the [Page 445]categories and values of the data, then a two-dimensional bar chart should always be the choice.

...