Data visualization tips

by Stephanie Chasteen on June 14, 2017

Are you trying to tell a story with your data?  This is a big part of my job (as an external evaluator), and I recently attended an excellent webinar on data visualization.  Now, I hate webinars that are trying to sell me a book, but this one was so packed full of great ideas that I hadn’t thought about before, that I might just have to go purchase it!  For those of you who do a lot of data visualization, there are many such professional development opportunities advertised through the American Evaluation Association (I subscribe to their email list, and subscribe to their AEA-365 blog).

Stephanie Evergreen (author of Effective Data Visualization: The Right Chart for the Right Data and Presenting Data Effectively: Communicating Your Findings for Maximum Impact) shared several tips both for charting data, and for juxtaposing charts so that the essential story is clear.  Here are some take-home messages from the webinar.

Humans are good at comparing lengths. 

An early cognitive study (Cleveland and McGill, 1984) is the reason we know that pie charts just aren’t a good idea – humans are not able to make accurate visual judgements of area, volume, and curvature.  So we have a hard time with pie charts.  But we are good judges of differences in length.  Hence, researchers often use bar charts.  Those are fine, since we can easily compare the length of adjacent bars.  BUT, bar charts are not always good if (a) there is a lot of data to compare (bar charts get messy), or you need to compare bars that are not adjacent (we are not good at comparing bars that are far away from each other).

Below are some examples of using our skill at judging length and distance to improve data visualization, with images drawn from the webinar.

Try lollipop plots! 

Why have I never seen this before?  In the case that there is a lot of data to compare, Stephanie Evergreen suggests using lollipop plots.  Below is a slide showing the same data as a bar chart (lots of visual clutter) and a lollipop plot. The lollipop plot draws your eye to the most important aspect – the end point of the line – without the visual noise of the big heavy bars.  You can also use vertical lollipops (replacing vertical bar charts), and she gave an example where the percent represented by the lollipop’s length (e.g., “23%”) is placed in the round part of the lollipop – a very nice visual.  (And yes, she says you can do this in Excel, but if you want to learn how, you should buy her book.  😉

 Try dot plots! 

In the case where you are trying to get people to compare non-adjacent data (for example, clusters of bar charts, representing scores across different classes or years), consider dot plots.  Below is the slide she showed us.  You can see that it becomes very easy to compare the scores across the different lines, since the single location of the dot anchors the score for that group clearly.

She also had some great tips for presentations (animate your data graphs in PPT, use handouts for your main points and for detailed charts), and for reports (1 page handout, 3 page executive summary, 25 page report, with detail in appendices.  For reports, she strongly emphasized that people do not print out these reports anymore, and they should be easily skimmable in electronic form – obvious visual markers for separating sections (e.g., a single page with a large graphic), and clear headings for sections.  She also uses headings for graphs which explain the main point (e.g. “Over half of young adults with autism had any social interaction in the last year”) rather than our standard academic titles (“social interaction results”.)

If this sort of thing is of great interest to you, you might consider joining the American Evaluation Association or checking out their offerings. They have a lot of great stuff about the uses of data.  See for example their Coffee Break Webinar series and their eStudy packages.  Later this summer is one on “data dashboard design” focusing on laying out your data on the page in an informative way.

 

{ 1 comment… read it below or add one }

Yuen-ying Carpenter June 14, 2017 at 7:31 pm

Loved this post, Stephanie, and it came at just a time when I was looking for new ways to impactfully display some data. Looks like I’m adding a few more books to my reading list!

Leave a Comment

Previous post:

Next post: