There’s more than one way to visualize those agree-disagree survey scales. Today, let’s look at seven of your choices.
Last month I spoke to Harvard University graduate students about visualizing survey results. I made up some fake numbers to accompany their real survey questions.
Stacked Bar Chart with Diverging Color Palette
Option A is a stacked bar chart. I chose horizontal bars over vertical bars because I arrange ordinal variables from left to right across the page. I also chose to display numbers of people instead of percentages. I use numbers when we’ve got fewer than 100 responses and I convert those numbers into percentages when we’ve got more than 100 responses. Agree/disagree survey scales are diverging variables so we need a diverging color palette. I selected two hues (blue and magenta). The most saturated version of each color goes on the outer poles. The only problem with this traditional approach is that the graph is a bit crowded. Where are the viewers supposed to look? What’s most important? When everything all at once is competing for attention, it’s easy to lose viewers.
Draw Attention to Strongly Agree or Strongly Disagree
Option B draws attention to the strongly agree responses. This is a good option for personalities who like to see the world through rose-colored glasses.
Option C draws attention to strongly disagree responses. This is a good option for internal audiences or for audiences with whom you’ve already built some trust and rapport. Very few viewers have an appetite for bad news.
I often use a combination of these two styles when presenting slideshows to live audiences. On the first slide, I show the good news. A few moments later, I show the bad news on the next slide. Then, we pause and discuss what might’ve led to these results.
Collapse the Categories
Option D combines the strongly agree and agree responses and the disagree and strongly disagree responses. This approach can declutter a crowded graph; focus viewers on big-picture patterns; whet their appetite for additional details when they’ve got time to spare; or satisfy their appetite for just a bit of information if they’re in a hurry.
Diverging Stacked Bar Chart
Option E is a diverging stacked bar chart in which the agrees slide over to the left and the disagrees slide over to the right.
Option F draws our attention to the poles.
Finally, I built four square pies that draw attention to the strongly agree and agree responses. Square pies are a good option for small n‘s because each shape subtly reminds us that we’re talking about a small group of real human beings. You could take this concept a step further with illustrated icons instead of squares. But good luck finding an icon to represent feeling connected to a museum or having your voice valued…
This is a great post! I could see using all options for projects I’ve worked on, but do tend to more often use options similar to A, B or C. I’m a big fan of collapsing categories where possible (shout out to Option D) but I find most people receiving this information like seeing the breakouts by actual categories (i.e., including the “Strongly…” buckets). We tend to use a 5-point scale that also includes a “Neither Agrees nor Disagrees” category. For those, I’ll either include them in a format like Option A and gray them out or I’ll put their response counts off to the right for awareness (unless they’re relatively large counts). I’ve never used square pies before but completely agree that numbers are usually looked at as numbers and we often forget that they represent real humans/employees.
It’s an impressive article and suggests very interesting way to draw attention to specific results of Likert scale survey. In one of my evaluation studies, I used Likert scale and then boiled down the results to a single number. Please find it here and feel free to comment:
The link to my use of Likert scale survey results is here: https://goo.gl/AELJtU
Love this! Ann – when you buy the template – for which iteration (or all?) are you getting the template?
Thanks Lauren. The template includes all three versions (the stacked bar charts, the clustered stacked bar charts, and the dot plot), plus the Word documents showing how you could arrange the charts on the page, plus a PDF’d version of the Word doc.
[…] patterns. They’re kind of like square versions of pie charts. I introduced them in an earlier post about visualizing survey results. Look […]
[…] There are several ways to visualize agree/disagree scales, like stacked bar charts, diverging stacked bar charts, or even waffle charts. The existing bar charts would be easiest to automate across dozens of workshop evaluation surveys, so we’ll keep them. […]
How do you handle “Neither Agree Nor Disagree” answers in the diverging stacked bar chart?
Good question! I don’t recommend using diverging stacked charts for odd-numbered survey questions. In other words, if you’ve got a 3-point, 5-point, or 7-point scale with a “neither agree nor disagree” midpoint, I’d suggest a regular stacked bar chart instead of a diverging chart. Diverging charts shine with even-number scales.
Love these ideas for making charts so much more effective. Thank you!
It’s so good to hear from you, Bernadette!!!!!