While writing your findings, you may decide to include data visualization to graphically represent your data. Useful data visualizations accurately and effectively convey study findings in a visually appealing and accessible format.
Examples of bad data visualization are easy to find. Upfront work, technical know-how, and creativity are all needed to create a good data visualization.
Regardless of whether you include data visualizations in the final report, there are a few items to consider to report findings responsibly. Consider reporting the survey response rate, how it is defined, and whose voices are included in the survey data, while not including information that might identify any single person.
How to create an effective data visualization
1. Decide the key point(s) that you want the audience to understand by examining the visualization.
If you include too many points in a visualization, you will sacrifice clarity and increase the likelihood that the point(s) will be misinterpreted or missed altogether.
2. Look for visualizations examples that best suit your intended point and the type of data you used.
MIT’s MASSVIS dataset is an excellent resource for all types of visualization examples, as is the New York Times, The Chronicle, and The Economist. Alternatively, Stephanie Evergreen’s blog provides detailed guidance on the best visualizations to support your intended point.
3. Select a visualization tool and begin building.
Excel is a handy tool for building many visualizations, whereas Tableau or R is better in other instances (see a ranking of tools by the type of visualization). For guidance in Excel, see Evergreens’ blog.
Tableau Public provides limited details on how to build visualizations. However, YouTube offers step-by-step instructions for Tableau on how to create many popular visualizations (e.g., a Sankey diagram, doughnut charts).
When you create a data visualization, consider:
- Is color used effectively?
- Is the font used easily legible?
- Are there unnecessary elements in the visualization?
- Does the title highlight the primary point of the visualization, or is it purely descriptive?
How to report responsibly
If you plan to report findings based on survey data, you will want to report the survey response rate and how it is defined. Most people think of a response rate as being the number of people you actually surveyed divided by the number of people you attempted to survey. This would mean that if you sent out a survey to 100 people, and 40 people responded, your response rate would be 40 percent (i.e., 40 respondents / 100 people).
But, what exactly do you mean by respond? Do you consider respondents to be those who completed:
- the whole survey?
- half of the survey?
- one question?
There is no one way to report a response rate. However, it should be reported. For even more detailed information on response rate calculations, see Do Response Rates Matter? by the American Association for Public Opinion Research.
One aim of data reporting is to provide insight into trends, patterns, and themes among respondents. This is true of data collected via surveys, focus groups, and face-to-face interviews. It is extremely important that you never identify respondents. Even a few pieces of information can potentially reveal a respondent’s identity.
Finally, if your report includes survey data, provide an overview of who is represented in your data file, specifically for key interest groups that are represented in the survey data. For example, are male respondents underrepresented compared to female respondents? Did a certain class year respond more so than another class year (e.g., first year students versus seniors)?
Clarifying who is represented within your survey data enables readers to determine whether some voices are missing or not.