When Axios published a chart showing left‑ and right‑wing domestic terrorism incidents from 1994 through July 2025, it sparked a minor political scandal. White House Deputy Press Secretary Abigail Jackson must have seen the chart and claimed on social media that left‑wing terrorism had reached a 30‑year high.
In fact, Axios’ story said only that left‑wing attacks have outpaced far‑right attacks for the first time in 30 years, and an editor’s note made that explicit. Now I’m not taking sides in this bru-ha-ha, but the confusion is at least partly the chart’s fault, and it highlights the importance when designers are making data‑visualization choices.
A tidy stack – but a messy takeaway
The original graphic was a stacked bar chart. Each grey bar represents the number of right‑wing incidents in a given year, and each orange bar stacked on top shows the left‑wing count. Simple, right? For journalists and data visualizers who look at these all the time, the answer is yes.
Stacked bar charts have an intuitive appeal: they display totals while breaking out the parts and looking tidy. However, they also come with hidden complexity. The top segments draw the eye, shaping the narrative before a viewer reads any numbers. Readers often assume the top segment is the most important and overlook the rest. ChartExpo does a deep dive into stacked bar charts and warns that stacked bars can bury signals under color (definitely in this case, with the hot orange vs. the neutral grey), making it difficult to compare categories precisely.
Researchers have shown that even simple bar graphs are frequently misunderstood by most people. Remember, most people are not visual and even a simple visualization is confusing and might be skipped entirely.
A Wellesley College study asked people to interpret average values displayed as bars and found that about one in five readers misinterpreted them, sketching most data points below the average. The authors conclude that simplification in graph design can yield more confusion than clarification. When we stack bars and ask readers to mentally subtract segments to find one series, we add to that cognitive load.
How misreading happens
In the Axios case, Jackson appears to have focused on the orange bars at the top of the stacks and assumed that meant overall levels had spiked. Had she examined the grey segments, she would have seen that right‑wing incidents still dwarf left‑wing ones historically. The CSIS data shows that right‑wing attacks peaked at 33 in 1995, while left‑wing attacks have only reached eight incidents in recent years. But stacked bar charts don’t make those comparisons easy, and our brains are wired to grab onto the flashy color or the top of the pile. In a contentious political environment, that can lead to hot‑takes and misstatements.
When you read the CSIS article, they actually use fever lines to plot both datasets on one graph. Now before you lose your you know what, I know there are specific use cases for fever lines vs. bar charts. I’m not going to open that debate for now, only to say there’s still a better way.
Jackson should have read the article before tweeting and she bears responsibility for misrepresenting the numbers. But if a graphic leaves room for misinterpretation, the designer has not done their job. The goal is comprehension, not artistry. Steak over sizzle. If people walk away with the wrong impression or draw the wrong conclusion, the visualization has failed, full stop.
A simpler approach
So how do we fix this mess, Eddie? The simplest fix is to avoid stacking! A grouped bar chart places left‑ and right‑wing counts side by side, sharing a common baseline. That makes it easy to see which category is larger each year and how both series change over time. It sacrifices the tidy total but massively increases clarity.
Even separate bar charts or line charts for each dataset would convey the trend without forcing the viewer to mentally subtract one segment from another. As ChartExpo writes, when precise comparison is needed, a grouped bar chart is the better choice.
Here’s a simple grouped bar chart based on the CSIS data. It visualizes the same story as the stacked chart but avoids the cognitive hurdles. You can immediately see that right‑wing incidents have generally been higher, while left‑wing incidents have risen modestly in recent years.
This design also opens the door to splitting the data into multiple views. For example, you could break down the chart by presidential terms (note how right-wing attacks increase at the start of the Obama presidency). Also, there’s no law that says all the data must be on one graphic. Unlike newspaper days, the internet doesn’t have a space budget; use as much as you need to tell the story clearly, cleanly and simply.
The takeaway: simplicity over style
When it comes to data visualization, the simplest viz is the best viz. If your chart leaves anything more than a 0.00 percent chance of confusion, you risk misinforming your audience. Flashy designs and stacked layouts look “cool” but they are harder to interpret. Most people don’t read legends or methodological notes; they glance at the picture and move on. That means the picture must be self‑explanatory.
Remember, don’t impress your friends, impress your readers.
