Data Visualization Lesson 4: The Best Pies Are Desserts

Staring bleary-eyed at a spreadsheet of data and a report that still needs visual content, have you ever felt a bit underwhelmed at the prospect of using a simple graphic?

Sure, a bar graph or histogram or line chart would show the data… but sometimes they seem, um… well… a little boring?

Dr. Dana Griffin

Dr. Dana Griffin

It’s easy to be allured by the visual feast of data presentation options. When choosing visual representations of data, it’s tempting to forgo standard graphic fare for goodies we think will be more satisfying to the aesthetic palate.

Which would you choose – a bowl of raw shredded coconut or a slice of coconut cream pie?

However, in some cases the best pies are desserts, not charts.

In this post, we’ll discuss several examples of ways to represent data – including pie charts, figures, and maps – that can be extremely effective in the right circumstances. But it’s worth understanding in what situations they can be difficult for viewers to digest.

Pie Charts.  Pie charts can be used to great effect when comparing a small number of categories (between 2 and 4) that make up a whole. But beyond four categories, pie charts become difficult for viewers to interpret easily and quickly.  Why?  First, viewers have to ascertain which color goes with which category, then assign a relative percentage to it, and remember this information relative to other categories before the whole picture makes sense. To show comparisons among four or more categories, a bar graph or simple table conveys the same information in a more visually straightforward way.

Pictographs / 1-D, 2-D, 3-D Figures.  Darrell Huff’s classic text How to Lie With Statistics offers an excellent overview of the potential perils of using pictographs and figures to compare data.  Cognitive research shows we are prone to systematically underestimate some visual dimensions.

For example, horizontal lines are usually perceived as shorter than vertical ones, even if they are actually the same length, and the human eye also has a difficult time accurately gauging area and volume from many graphs.

Pictographs and figures can be extremely effective in illustrating relative magnitudes when the vertical and horizontal proportions of the figure are consistent and include a clearly labeled baseline.  For a terrific read on the psychology of data visualization, check out Graph Design for the Eye and Mind by Stephen Kosslyn.

Maps. Geographic data + other data = intense temptation to map.  Plain and simple, the conclusions data consumers will reach largely depend on how specific data categories are within the map.

Take for example the familiar “Red State/Blue State” maps that seem to be everywhere in recent presidential election years.  Based on the simple dichotomous color categories, it would be relatively easy for a casual viewer to infer that states (and their residents) are wholly Republican/conservative or Democratic/liberal.  By contrast, these maps by Dr. Robert Vanderbei use presidential election returns data at a different level of specificity (county-level rather than state-level), which elicits a more nuanced impression.  The best maps make data easy for viewers to interpret but resist overstating or oversimplifying.

In my next post, we’ll discuss how to present the results of fancy statistical tests using clear, straightforward graphics (the culinary equivalent: putting kale into chocolate cake and having everyone ask for the recipe).

Stay tuned – it’s sure to be delicious!

Data Visualization Lesson 1: Examine the Y-Axis
Data Visualization Lesson 2: Think of Grandma
Data Visualization Lesson 3: Abela’s Rubric
Data Visualization Lesson 5: Ninth Grade Algebra Wasn’t Worthless After All
Data Visualization Lesson 6: The Ultimate List of Dos and Don’ts


Image Credit

Related posts:

  1. Data Visualization Lesson 3: Abela’s Rubric
  2. Data Visualization Lesson 1: Examine the Y-Axis
  3. Data Visualization Lesson 2: Think of Grandma
  4. 10 Places for Market Researchers to Learn about Data Visualization
  5. A Hunger for Data Visualization
About Dana Griffin

Dr. Dana Griffin is a consumer insights and decision-making expert based in Seattle. An award-winning researcher and data educator, she earned her Ph.D. in political science from the University of Minnesota with concentrations in research methodology and psychology. With extensive experience in both qualitative and quantitative methods, Dana served as the Director of Survey Research at one of the largest newspapers in Minnesota. Prior to joining the private sector as a research and analytics specialist, she was a faculty member at Furman University and the University of Nebraska-Lincoln.

  • Kevin

    “It’s easy to be allured by the visual feast of data presentation options”

    Hey Dana…I just read through the series you have posted on data visualization and thank you for taking the time to not only consider better dataviz techniques, but also for taking the time to publish your thoughts.

    I agree on the allure that vendors market…they all too often push the glitter and gold instead of true value of visualization. Very few tools aside from Tableau and a couple others are willing to eliminate the useless functions as they can easily use them to woo CIOs who maybe don’t understand good vs bad in this field.

    I see you are familiar with Stephen Few’s work so I wanted to point you to a blog post of his that mildly attacks SAS for bad dataviz options in some of their explorations apps. thought you might find it interesting.

    Thanks again!

  • Pingback: Data Visualization Lesson 5: Ninth Grade Algebra Wasn’t Worthless After All - Research Access

  • Pingback: Data Visualization Lesson 6: The Ultimate List of Dos and Don’ts - Research Access