Before-and-After Magic: Creating Effective Graphs for Data Presentation
 
By Cybele Elaine Werts, Northeast Regional Resource Center, a project of WestEd  
This ran in the Special Libraries Association (SLA) magazine, Information Outlook, March 2004

Read the Acrobat version

First, some before and after graphics to give you an idea of how improved visuals can really change a chart:

Before Bar Chart

After Bar Chart

Before Trend Graph

After Trend Graph

Graphics reveal data.
~ Edward Tufte, The Visual Display of Quantitative Information

Two people arrive at a job interview for a public relations position at an education foundation. Both are highly qualified for the job. Both are wearing suits that are freshly pressed, but one suit is lined and tailored, and the color and style are flattering. The other is far less attractive. Employers may not be able to say what makes for a well-made suit, but they will still be affected on a subconscious level and give the job to the person who appears more competent. We are all affected by how things look, even if we don't know why. When all else is equal, visual presentation can make the difference between landing a great job and standing in the unemployment line.

Data presentation is similar to a well-made suit in that it is a critical part of reporting, particularly in the current climate of data-driven decisionmaking. Unfortunately, graphs can be very confusing to readers and can muddy your message. This article is about creating charts that showcase your data, focusing on the visual or graphic aspects of chart design. This is not about choosing the right chart for your data or an analysis of manipulating statistics. We assume that readers have a good sense of the data they are presenting, as well as some skills in using a spreadsheet software application such as Excel. For the sake of brevity, only bar charts and trend graphs will be reviewed, but these guidelines apply to most types of charts.

General Chart and Graph Recommendations

The purpose of charts and graphs is to express the information you're writing about in a visual way. The sample charts show how small changes in layout and color can make your data clearer. You may laugh at the "crabby" examples, but each of those ineffective elements has been observed in real-life reports. If you're going to use charts, it's a good idea to include them in the body of your document because putting them in an appendix creates an unnecessary step for the reader and might mean they won't be read at all. If you have multiple graphs on a page, help the reader distinguish among them by using color; for example, use green for the education charts and blue for the government charts. Finally, if there is a large amount of data, a chart is optimal because it's difficult for the eye to read a huge table full of numbers. In contrast, if the data set is small, use a table, which saves space and is easier to read. Here are more tips for well-designed charts:

Overall Style

• Titles should clearly state what the graph is and avoid acronyms. You may think everyone knows what NERRC is, but your chart may go to readers who are unfamiliar with the name. (NERRC is the Northeast Regional Resource Center .)
• Include a total of items measured if it's relevant. For example, N = 98 indicates that your sample included 98 responses.
• Use a legend only when necessary because it requires the reader to take another mental step to translate the information. Notice that placing the identifier (Boca the cat) under the bar takes less time to read.
• Keep the angle of the whole chart front and center. Skewing, or twisting the bars to odd angles, may add visual interest, but can change the meaning of your data.

Clean, Simple Visuals

• Use easy-to-read fonts like Times or Helvetica. Complex fonts or tiny point sizes (under 12 points) decrease readability.
• Avoid cute clip art unless it's directly related to the data you are presenting.
• Avoid lurid colors like lime green or hot pink; they are hard on the eyes and distract from the message. Do use colors that can be copied easily in black and white. For example, multiple, similar shades of green will merge into an indiscriminate gray on a copy.
• Consider using colors that are more accessible to people with visual impairments or color blindness. See the article in the references section for more information.
• Remove the gray "walls" and background lines unless they are needed to make the data more readable.
• Make borders thinner and dark gray instead of a heavy and black. This puts the emphasis on the content of the chart rather than the border.
• Remove the default x-axis "1" in favor of an actual x-axis title, or leave it blank. Adjust the angle of the text for optimum readability.
• Use larger increments in the y-axis, such as 25, 50, 75, instead of smaller increments, such as 10, 20, 30. Make the maximum number only slightly higher than the tallest bar to optimize use of space. Choose a point/font size that is easy to read and doesn't obscure the numbers above and below.

Bar Charts

Bar charts are one of the most common types of graphs. A bar chart shows data at a "point in time" not unlike a snapshot. In contrast, if you have data that shows something over time you'll want to use what's called a trend graph. Trend graphs are far more useful because they show how things change over time, and use the space of the chart more effectively.  

One of the challenges of developing charts in Excel is that the default options make for a disorderly chart. You can see how this looks in the sample "crabby" bar graph. The problem is that there is too much information, both in the bars and in the legend. In contrast, look at the sample "happy" bar graph, whose information is easy to read and clearly delineated. Here are some guidelines for creating a sharp-looking bar chart:

• For bar charts that stand alone (aren't being compared to another bar chart), sort the data in either ascending or descending order, because readers want to know the highest and lowest values. If you are comparing one bar chart to another, keep the order of the bars consistent so the reader can compare graphs without having to transpose the bars.
• Avoid three-dimensional bars (they look like rectangular boxes) unless there's a specific reason to use them. They take up more space and make it hard to read the value of the bar.
• Use solid colors or gradations inside the bars and in the background of the chart. Patterns like checks or plaid make it difficult to read the text.
• Put bar totals inside the top of the bar instead of allowing them to float around in the graph. It looks neater and helps prevent confusion about which number goes with which bar.

Trend Graphs

• Limit the number of trend lines so they don't overlap and you can't tell which is which. A full-page chart can include a lot more lines than a small one, which should have only a few.
• Make the lines heavier, with larger markers (the little triangles or boxes along the lines).
• Use colors that are very different from each other. Excel often defaults the trend lines to multiple shades of blue, which makes them very hard to tell apart. Choose colors that contrast well with the background color of the chart. For example, don't use a pale green series line when your graph background is pale blue.
• Legends are more useful in trend graphs, but color-coded text can be easier to read (see sample).

Your Ideas and Suggestions Are Welcome

Please send your ideas and comments on creating effective charts to CybeleW@aol.com  

How to Change Default Chart Settings in Excel for Windows

First, create your chart in the basic style that you want, such as bar, trend, or pie.

This report includes samples of well-designed charts:

Outcome Based Planning: State Partners and Local Communities Working Together to Improve the Well-being of All Vermonters, by Vermont 's Department of Developmental and Mental Health Services,

http://www.ahs.state.vt.us/pdfFiles/OutcomeBasedPlanning03.pdf

Additional Reading on Data and Presentation

The Visual Display of Quantitative Information, by Edward Tufte

A Guide to Effective Accountability Reporting, by the Council of Chief State School Officers (see Chapter Three in particular), http://www.ccsso.org/content/pdfs/GEAR.pdf 

How to Lie with Statistics, by Darrell Huff

Graphical Data Presentation, a product of Deakin University , http://www.deakin.edu.au/~agoodman/sci101/chap12.php

A Pie or a Slice? Graphing, Charting, and Presenting Data, by Dr. Madhukar Pai MD, DNB, http://www.sunmed.org/graphing.html

Considering the Color-Blind, by Chuck Newman, http://webtechniques.com/archives/2000/08/newman/  

*This cartoon courtesy of Michael Giangreco, a well known specialist in the area of Special Education. He allows educators to use these cartoons copyright free. Please note that I changed the original names to that of myself and my Director, Kristin Reedy. 

Reprinting Information
Would you like to reprint this column? If so, do ask! I usually allow distribution because spiritually speaking, sharing ideas is an important way of expressing my faith. Please e-mail me at CybeleW@aol.com