3 Rules of Data Visualization
We’re always engaging with the wider community of information and visual design, sharing our work with other designers and agencies and doing everything we can to stay fresh. We’ve learned that not all data visualization is created equal. Information design requires a very specific skill set for correct and effective execution; read on to see just 3 of the many data visualization rules we follow and what it looks like when they are broken.
Rule #1: Let Your Data Breathe
One of the biggest temptations many face when first setting out to visualize data is trying to pack too much visual information into a single design. Whether this is due to “jazzing up” a design or is an attempt to convey too much data, the end result will always be the same: Visually cluttered, unintelligible design that raises more questions than it answers.
Take a look at the following visualization from NBC Nightly News:
It’s clear the designer aimed to make this data more visually stimulating by overlaying it on top of a map of the US, presumably because it discusses the US population. Here are some problems this creates:
- For starters, the graph is just plain difficult to read. Too much visual information is crammed in, confusing the viewer.
- Any time red and blue is overlaid on a map of the US, the viewer may immediately assume the visual is related to political divides.
- The X-axis makes it appear that the West Coast is stuck in 1960, while the East Coast has somehow time-traveled to 2060.
- The Y-axis makes it appear as though Asian Americans only live in Maine and Washington State, Black Americans live exclusively in a thin band in the North, etc.
- Not least of all, the scale is incorrect. By not using a uniform shape to represent the whole (which should typically be a rectangle), the percentages do not match up with the amount of space filled.
Here is how the data could have been visualized to avoid these issues and ensure maximum clarity:
Rule #2: Embrace the Visual Vernacular
Most of us grew up creating and reading graphs in school. When it comes to data visualization, we share an established visual language that is integral to ensuring fluency in communication and understanding. For example, pie charts must always add up to 100%, horizontal timelines should always show the future to the right, etc. Deviating from this shared visual vernacular will only cause confusion.
Here is an example of improper application of our visual language from Reuters:
We’ve all been trained to know that for line graphs, up shows increase and down shows decrease; strip away all the labels from a line graph and you can still infer whether it shows growth or decline. Reuters inverted this rule, choosing up to be down, down to be up. Instead of seeing the enaction of the “Stand Your Ground” law as catalyst for a marked increase in gun deaths in the early 2000s, the viewer quickly assumes the law actually decreased gun deaths in the state before seeing them “increase” again in the 2010s.
Here is how we would visualize this data for improved clarity:
Rule #3: Be Fair to Your Data
When visualizing data, it’s very important to give the viewer all relevant context. This allows the viewer to observe unbiased visualization in order to draw his or her own conclusions. Manipulating scales, start points, and layout can lead to incorrect data interpretation. At best, this is an unintentional mistreatment of information; at worst, it can be a deliberate misleading of the audience to further a particular agenda.
Fox News has had difficulty with this in the past — check out the below graph on Obamacare enrollment:
By starting the X-axis at around 5,250,000 and not labeling it as such, the March 27 data roughly appears to be just 30% of the 7,066,000 goal. In reality, 6,000,000 is 85% of the end goal — that’s a significant difference. The disparity here that causes the viewer to think, “There’s a long way to go,” instead of realizing, “It’s almost there.” Cherry-picked data won’t give the full story, so it’s important to always give the full context of your data.
Here’s how the data should have been visualized:
When it comes to visual communication and information design, even the most subtle issue can create large problems. Remember these best practices of information design when you start your next visual communication project.