Try a New Way to Graph Survey Data in 2015

If you’re like me (and you are), as I draw the curtains on 2014, I’m thinking ahead to want I want to accomplish next year. If you’re like me (and you are), the overarching theme of my 2015 resolutions is basically to be more rockstar*. Do this with your data by outfitting it in some rockstar gear. In other words, try a new graph type that conveys with way more impact.

In my last NTEN blog post, I showed how to turn a table into my favorite new graph type – the amazing and useful dot plot. That’s totally rockstar.

This time, let’s talk about Likert-type survey data. You know, the kind that runs Strongly Disagree to Strongly Agree. Probably the most common way people show this data is through a regular ol’ stacked bar chart, like so:


Stacked bar charts often get knocked down a bit because they can be hard to interpret — especially those middle values, because they don’t have a common baseline to measure against. For example, see the light blue Agree bars in the bottom two rows? Are they the same? It’s kind of hard to tell, right? The strong color coding certainly helps counter that criticism, and of course a good title can go a long way in making a graph more interpretable. So this isn’t terrible, but it could be better. Let’s look at some alternative chart types for Likert-type data.

Alternative #1: The Aggregated Bar

If the level of detail isn’t important, aggregate the positive and negative responses so you have just two groups. Most of the time, those we are reporting to really don’t care, anyway (you know I’m right).

Back in your table, you would simply sum the positive and negative values into a new table and then graph that baby so it would look like this:


Now, it doesn’t matter if those middle chunks of the bar are not comparable because we don’t have any more middle chunks. Agree and Strongly Agree have been grouped together and renamed “Agree” for simplicity’s sake. Check out how one nonprofit revised right along these lines. Way to be a rockstar!

Alternative #2: Diverging Stacked Bar
What if you do need to keep all of the categories? You may have a stronger impact with your communication if you convert the standard stacked bar chart into a diverging stacked bar chart. It looks like this:


Essentially, the positive categories are on one side of a line while the negative categories are on the other. The key here is that the two main sides of the Likert scale have a common baseline (in the middle) so it’s easier for the eye to interpret the data and see just how positive the positive responses are, for example, while still keeping all 4 categories.

Don’t worry if you struggle to immediately love this one. I find that those of us data nerds who graph all day long in Excel hesitate just a little with a really new graph type like this one simply because it isn’t something we are used to seeing. Try it out with your less data nerdy audiences and gauge their reactions.

How do you do it? No worries, mate, I’ve got step-by-step instructions with Excel screenshots for your handy reference. And yeah, you can still make a diverging stacked bar even if you have a neutral category.

You’ll also notice that I changed the font here to something condensed (tall and skinny letters). This is my super helpful go-to strategy for squeezing labels into skinny chunks of a stacked bar. (Add this to your resolution list, rockstar). I also eliminated the legend and retyped the labels in textboxes so I could both color code to match the chunk color and reposition exactly where I wanted.

Both of these alternatives can help you really rock out your Likert-type data. But even if you aren’t working with data of that nature, branch out a bit in 2015 and try on a new graph type, like dumbbell dot plots, slopegraphs, or span charts. You’ve got this, rockstar!

*rockstar: n. a person who demonstrates great data visualization skills. adj. Having or pertaining to the qualities of a rockstar.

Stephanie Evergreen
Evergreen Data
Stephanie Evergreen, PhD, is CEO of Evergreen Data. She is a sought-after speaker, designer, and researcher. She regularly keynotes and provides hands-on workshops and design services to help people rock out their communications about data. She writes a popular blog. Her book, Presenting Data Effectively, was published by Sage in October 2013. You can find Stephanie on Twitter @evergreendata.