A Quick Journal Graph Revamp

Recently, a colleague asked if I could provide examples of good data visualizations to share with an academic research center. Thanks to widespread education and evangelism, there are many good (or, at least, not terrible) examples of dataviz in modern data reports. However, charts and graphs are still hurtin’ in academic journals. It’s really hard to find examples that convey a meaningful message in a clear way.

Why are journal graphs so rough?

I have my theories.

One is that journal articles typically follow a traditional design that dates back to when researchers submitted type-written manuscripts by snail mail. There hasn’t been the same push for disruption with journals as there has been with other forms of written and visual communication. Beyond general tidiness, journal articles aren’t expected to be visually attractive/impactful.

Second, I think there’s an unstated assumption that if an academic writes an article, it will automatically be accessible for an academic journal reader. Why would we need to change how we write and design articles if they’re already great…or fine…or…whatever? The problem is that most articles aren’t fine; as I’ve mentioned previously, even scientists struggle to understand scientists.

Third, creating ineffective graphs using software defaults is relatively quick. In a world where most scientists and academics are juggling an untenable amount of job expectations and responsibilities, and in light of the other conditions above, it honestly doesn’t make much sense for a scientist to spend time designing better graphs for journal articles.

Unfortunately, when we publish not-great graphs in any outlet, journal or otherwise, we miss opportunities to get our message across and run the risk of readers misunderstanding or misinterpreting the data.

How can we make journal graphs more effective?

1. Tweak graphing software defaults to align with dataviz best practices.*

2. Manually add elements that will boost understanding.

Here’s an example.

I pulled this graph from the journal Prevention Science. Its inclusion here is no way a knock on the importance or quality of the underlying science, nor does this graph stand out from others in its issues. I picked this as an example solely because it’s the first recent open-access journal graph I found that could benefit from some design love.

Here’s the original graph, from Greve et al. (2025) :

This graph shows three groups of Danish families in a research study and how their income changed over time.

Let’s talk about tweaking the (STATA?**) defaults used in this graph, starting with the y (vertical) axis.

It’s really hard to read text that’s turned sideways. It’s really hard to mentally process a string of (…counting…) five consecutive zeroes. And, like in many situations, starting the y axis at something different from zero skews our perception of magnitude and differences.

So, let’s swivel that y-axis label text, add commas to the value labels, and start the axis at zero. I’m also going to remove the gray line from the axis, because it doesn’t contribute anything.

Next, let’s move on to the x (horizontal) axis.

I had to go into the article text to figure out what the numbers along the bottom meant. Turns out that the zero represents a pre-test, and the other number signify the number of years before or after that pre-test. Let’s encode that information directly in the graph by (1) changing the value labels and (2) manually adding some brackets and text boxes.***

Now let’s look at the actual lines within the graph.

I like what the authors did to improve visual accessibility by using a different marker shape for each line. I also like that the lines are different colors; however, we could push that further to make the lines more distinct. And then, whenever possible, I advocate for directly labeling lines (typically manually) rather than making the reader jump back and forth between the data and a legend.

Finally, let’s talk about the title.

The original caption for this graph was “Mean annual disposable family income in intervention, control, and background population families across 8 years.” OK, but what’s important about what this graph shows? What insight should the reader walk away with? Let’s add something a little more informative.

This may not be the most thrilling title, and it may not be exactly what the study authors were going for, so this should definitely be edited to reflect the intended message. Admittedly, the content and format of this title also violate conventions of APA and other journal styles, but I’m willing to take the risk that a manuscript won’t be rejected because of that. Clearly communicating insights should be the priority.

So, there you go! By putting in a little effort to tweak default graph settings and manually add helpful graphical bits, we can make it easier for audiences to read, interpret, and apply data from the millions of academic papers published each year.

Want to boost the impact of YOUR academic publications? Contact us!

* Need to learn more about dataviz best practices? My favorite resource for this is Stephanie Evergreen’s Data Visualization Checklist.

** No, I’m not yelling a random IKEA product name. STATA is a software package for statistical analysis. The authors of this study used STATA for their analyses, but it’s unclear whether they also used it to generate this graph.

*** The software in which you make these manual additions depends on the graphing software you’re using and the final type of product or file you need to output. Personally, I often lay my graphs out on PowerPoint slides and export them as image files.