Communicating Numbers II - Charitable Charts

Selecting the correct medium and formatting visualizations to ensure accuracy and clarity.

Source: Few, S. (2005). Effectively Communicating Numbers: Selecting the Best Means and Manner of Display, Perceptual Edge.

Tables vs. Graphs:

Tables work best when the display will be used to look up individual values.

Graphs work best when the message you wish to communicate resides in the shape of the data(patterns, trends, and exceptions).

Quantitative vs. Categorical Data

Quantitative data consists of numbers and the corresponding category, like a country (category) and it’s population (number).

Categories come in three scales: nominal, ordinal, and interval.

• nominal → difference only in name (nominally).
• ordinal → have a natural order (“A, B, C”) but do not represent numbers.
• interval → intrinsic order, representing intervals of quantitative series, like $\leq 50$ or $[20-49]$.

Common Relationships in Business Data

The first question that you should always ask about a number is “Compared to what?”

• Time-Series: → Line
• Ranking: → Bars
• Part-to-Whole: → Bars. Avoid pie charts because area is harder to compare than length.
• Deviation: → Bars & Lines or Points to show differences from a target.
• Distribution: → Histograms (vertical bars) or Frequency Polygons (lines).
• Correlation: → Scatter plots with points and trend lines.
• Nominal Comparison: → Bars.

Encoding Quantitative Data

The two most powerful attributes of visual perception are line length and 2-D position. Color, shape, and size, do a less effective job of graphically representing quantitative values. For example the larger circle has 16 times the area of the smaller circle.

Most viewers will underestimate the difference because our ability to accurately compare 2-D areas is not well-developed. In pie and doughnut charts, curved edges and the lack of a base-line lead to a consistent underestimation of the actual percentages (Cleveland & McGill, 1984). Bar charts are much easier to compare.

Points

Unlike lines, points emphasize individual values, rather than the shape of those values as they move up and down.

They have two primary strengths:

1. they can encode values along two quantitative scales simultaneously → scatter plot
2. replace bars when the quantitative scale does not begin at zero

Small Multiples

If you need to display one more variable than you can fit into a single graph. Small multiples, are small multiples are similar graphs with consistent scales, arranged in little space so that they can be seen simultaneously. Keeping the quantitative scale consistent makes it is easy to compare between different categories.

Lines

Connect the individual values in a series, emphasizing the shape of the data. In time-series they show trends, patterns, and exceptions clearly. Use lines for data with inherent connection like prices over time (interval scale). In nominal or ordinal scales the items are often not related sequentially so that you should use bars or points instead.

Bars

Encode quantitative values in two ways:

1. the 2-D position of the bar’s endpoint in relation to the quantitative scale
2. the length of the bar.

Bars encourage focus to and comparison of individual values. They don’t do as well as lines in revealing the overall shape of the data.

Whenever you use bars, your scale must include zero. Otherwise, the relative differences are distorted.

Use horizontal bars when labels are long or there are many items to prevent squeezed text.

Boxes

Are like bars, except that both ends encode quantitative values. They are used to encode a range of values, usually from the highest to the lowest, rather than a single value. → Box plot

Remove Distraction and “Chart Junk”

Anything that does not contribute to the meaning of the data distracts communication. Remove things like bright colors and fancy backgrounds. Subdue necessary grid lines and labels.

It is a good basic practice to use relatively soft colors in graphs, such as lowly saturated, natural colors found in nature, reserving the use of bright, dark, and highly saturated colors for those occasions when you need to make something stand out.

Use only 5 to 10 major tick marks on an axis to avoid clutter.

Hide distracting data series, while keeping the accessible on demand, use filters/slicers to allow users to focus or zoom out.

Readability

• Highlight specific data with contrasting borders, thicker lines, or larger point sizes.
• Label lines directly at their endpoints instead of using a distant legend.
• Position variables you want to compare next to each other in the layout.
• Include notes directly on the graph to describe specific events or provide instructions on how to interpret complex views.
• Arrange legend labels in the same order as the data bars/lines they represent.
• On very large graphs, place scales on both sides to help the eye identify values accurately.
• Let your axis scale to extend slightly below the lowest and above the highest value.
• For scales involving positive and negative numbers, position the axis line at zero.