Our boss here at ZingChart recently passed us a copy of The Chicago Guide to Writing About Numbers. The subtitle is "The Effective Presentation of Quantitative Information." What may look like a brief read is actually packed with tips for presenting data. Many of the pointers focus on *writing* about data. But read on for recommendations from the book that can be used for visually presenting data in your interactive charts.

## Presenting Data

This book is published by the University of Chicago Press. A primary intention is to help those publishing academic papers to present their data in a way that helps readers understand it better. But it is also designed to help reduce language that may affect the objectivity of data being presented.

Author Jane E. Miller also details some ways in which her book is unique:

- It integrates several aspects of presenting data that are usually taught in separate books.
- It identifies how each facet of presenting data affects the others.
- It uses as many examples as possible.

So with that in mind, we dove in. We found several tips that chart makers would find especially helpful.

### 1. Using Digits and Decimal Places

In chapter four, Miller lays out guidelines on the number of digits and decimal places that are appropriate for presenting data. These tips are for data in text, charts, and tables. How many decimal places are really best?

The recommendation in charts is generally “1 to 2.”

- For rational numbers in charts, one or two is acceptable. For example: 32.1 or -0.71
- For percentages, zero decimal places is best. That is, unless several numbers would round to the same value. Then one place is best. For example, 72% or 6.1%.
- For proportions, two or three is recommended. Use two, unless several numbers would round to the same value. In that case, use three. For example, .36 or .002.
- In monetary values, it is best to use no decimal places for large denominations. Use two places for small amounts. For example, $2 million or $12.34.

Does following these best practices make your interactive chart look cluttered? If so, you could try showing shortened versions on the chart scales, and the full number of decimal places in the tooltips and crosshairs.

### 2. Describing Differences in Data

In chapter five, Miller details some key phrases for describing ratios and percentage differences. We thought these could be useful for chart tooltip text. What are the rules of thumb?

- For ratios under 1.0, the rule is
*[Group] is only x% as _____ as the reference value*. For example: “Males were 80% as likely as females to graduate from the program.” - For ratios close to 1.0, use phrasing to express the similarity. For example: “Average test scores were similar for males and females (ratio = 1.02).
- For ratios greater than 1.0, the rule is
*[Group] is 1.y times as ____ as the reference value*. For example: “On average, males were 1.20 times as tall as females.” - For ratios close to a multiple of 1.0, just use the multiple times. For example, “Males were nearly three times as likely to commit a crime as their female peers.”

### 3. Chart Recommendations

Miller has plenty of recommendations for presenting data when it comes to charts. She also recommends reading Tufte and Zelazny for more chart design resources.

`All charts need good titles, labels, and notes.`

- For chart titles, Miller recommends restating the question you were researching in the first place. The title should also differentiate the chart from others in your app that may look similar.
- For axis titles, labels, and legends, her rule of thumb is to label them like you would row and column labels in a table.
- In bar or column charts, continuous variables are typically on the Y axis, and nominal or ordinal variables are on the X axis.
- Single line charts are best for showing relationships between two continuous variables. For example: transportation costs across time.
- Scatter plots visualize the association of two continuous variables.
- Maps are preferable to tables or charts when showing data with a geographic component.
- If your chart is comparing data against a standard or a pattern, consider using reference lines, guides, marker regions, or plot bands.
- If specific data values are worth pointing out, use annotations on your chart.
- If a simple chart type will do, use it over a more complex alternative.
- Remember: your chart may not always be presented in color: consider patterns and contrast for such cases.

#### Chart Type Recommendations

Data Presentation Task | Type of Chart |
---|---|

Distribution of a variable with 2-5 categories | Pie |

Nominal data with less than 5 categories | Bar/Column |

Continuous data with less than 20 values | Line |

Continuous data with summaries of ranges/variance | Box and whisker |

Relationships between two or more categorical variables | Stacked bar or grouped bar |

Relationships between two continuous variables | Scatter plot |

Relationships between two continuous and one categorical variable | Multiple line or scatter |

### 4. Pie Chart Recommendations

Since pie charts can be a contentious topic in the dataviz community, we wanted to share Miller’s do’s and don’t for good pie charts:

**Do**display one variable per pie chart.**Don’t**use pie charts to display averages across groups or time periods.**Don’t**use pie charts to visualize multiple response variables.**Don’t**use pies with skinny slices (combine categories if necessary).**Do**use pie charts to compare composition across three or fewer groups.**Don’t**be afraid to use bubble pie charts.

### 4. 3D Effects

“Steer clear” is Miller’s general take on 3D effects in charts. The objectives of dataviz can generally be accomplished just as effectively or more so with two dimensional styling.

## Making Data Presentation More Effective

Now you’ve seen the tips we found in The Chicago Guide to Writing About Numbers. What will you change in your charts? Do you have additional best practices you would recommend? Share your thoughts with us in the comments below.