I’m a data nerd and I’m proud of that. I have a doctorate in research and evaluation, for heaven’s sake. I geek out in conversations about Excel with anybody who can handle it on a pretty regular basis. Evaluation is amazing. Evaluation is useful. Evaluation is course-altering. But generally, data visualizations in evaluation tend to suck. They often look like this (something I really really made a while ago before I knew what I was doing):
How did I ever expect anyone to make any useful decisions with this table??? Turning this into a rockstar data visualization would have communicated impact so much more effectively.
This is not the time to go to a graphic designer for help. If you’re like me, you’re crunching numbers on your impact metrics at the last minute, pulling together a report on the fly. There’s no time to meet with a designer. And more importantly, the vast majority of graphic designers do not know enough about data to make a decent graph. Edward Tufte, grandfather of data visualization, wrote, “Nearly all those who produce graphics for mass publication are trained exclusively in the fine arts and have had little experience with the analysis of data.” The opposite is also true — most researchers and evaluators have little familiarity or skill with graphic design and art. So I’m going to teach you four basic tweaks you can make to graphs right inside Excel that will elevate them to rockstar status — tweaks that mix data and design — to best communicate your work’s impact.
Tweak #1: Choose the right type of graph
The data in my table came from a survey given before and after participants took part in a parenting program. Pre-post data can be displayed in lots of ways, but I’m going to choose a dot plot for this one. Dot plots are a new-ish graph type that can totally be made in Excel with a little ninja skill (see a step-by-step tutorial here). They are really powerful and easy to interpret for non-data nerd audiences. It’s just the thing for communicating impact in this case. Here’s my program data in a dot plot:
Tweak #2: Reorder
Ok, so that last tweak was kind of a doozy, but I promise the rest will be a snap. This graph is going to make more sense if I reorder the data. Right now, the order of the questions is just the order of the questions on the survey. But who really cares about the order of the questions on the survey? I’m going to make this more useful by reordering according to the most growth. This involves another calculation in my table, subtracting pretest scores from posttest scores — no problem. Then I sort the data on that new column representing the growth:
And now it is much easier to see that stress reduction was a huge impact for participants, more so than understanding what the baby wants, which is more closely related to the program curriculum. Woo hoo! Amazing beneficial side impact identified!
Tweak #3: Turn the title into a sentence
Why make the audience guess at the point of the data being visualized? When we use generic titles, we put a lot of burden on our readers to summarize the data into the takeaway points we have in our heads. Unfortunately, ESP is not in the skill set for most readers. Instead of a generic title, I’m going to replace it with a full, declarative sentence. I’ll add a subtitle too, for extra mileage.
Now readers will make no mistake about the point of the graph! Ever look at a data visualization and think “… so what?” Yeah, a declarative sentence for a title answers that question!
Also, you probably noticed that I snuck the legend, or key, right into the title. This way readers know which dot represents what. This leads me to the last tweak.
Tweak #4: Use an emphasis color
Blue and orange are okay but I can push the ability to communicate impact a little further by swapping them out for gray and an emphasis color.
This program has a lovely tomato-y red color in their logo, so I used it on the posttest scores to really bring attention to the final impact in each area. I think it also better shows that the average participant rating was at the maximum on two of the questions. Nice!
Four tweaks later, and now we have a rockstar data visualization that really communicates impact. These tweaks and other best practices are summarized in my data visualization checklist. Handy, huh? All yours (and your boss too — just slip it quietly under her door).