What is wheat and what is chaff? Here is a list to help you take sides:
- If you want to fit the data into the shape of real-world objects, that’s not data visualization;
- If you use more than one dimension to represent a data point, that’s not data visualization;
- If your project breaks basic perception laws, that’s not data visualization;
- If there is a perceptually simpler way to represent the data, what you are doing is not data visualization;
- If you use color when color is not needed, that’s not data visualization;
- If you want to grab attention, that’s not data visualization;
- If you reject a chart type because the audience may not be familiar with it, that’s not data visualization;
- If your audience don’t know what the point of the chart is, that’s not data visualization;
- Alphabetical sort is not data visualization;
- If your chart is nothing more than a glorified table, that’s not data visualization;
- No variation, no data visualization;
- If a simple formula can be a better answer, there is not place for data visualization;
- If all you can do with a chart is to compare data points, that’s not data visualization;
- If you have three or fewer data points, that’s not data visualization;
- If you have to scroll, that’s not data visualization;
- If you have to compare slides, that’s not data visualization;
- …
If your blood is boiling with so many enormities, cool down and don’t call me that. It may not be data visualization, but it may be important. It can be aesthetics, it can be literacy, it can be user interaction. They are important too.
Like Michelangelo said, “I saw the angel in the marble and carved until I set him free.” If we want to talk about “data visualization” we must find its essence, its form (in the platonic sense) to be able to compare them to the shadows of everyday life.
Actually, we don’t have to. We can always decide that the goals, the tools, the audiences and the processes are too dissimilar to accept a single concept of “data visualization”. Let’s decide this quickly so that we can return to our many other byzantine matters, like “should we use pie charts?”.
So, what does your angel look like? Are you including the feathers?
A lovely quote from Michelangelo – something we should all aspire for when creating charts –
Fantastic list of guidelines.
Great discussion starter, Jorge! I agree with the majority of your points, though would add the descriptor “effective” (e.g. that’s not *effective* data visualization), as it’s possible to visualize data in many sub-optimal ways – personally, I’d still consider it data viz, just not effective data viz.
My personal take is that, while the majority of best practices hold across different audiences and use cases (choose a chart type that accurately represents the data, draw attention where you want it with preattentive attributes like size and color, identify and cut out visual clutter, tell a story), the most effective data viz may (and possibly should) vary as audience and use case does.
Beautiful use of the Michelangelo quote!
A great preacher I listen to from Seattle often uses this approach to explain their approach to church doctrine…..
Quote
“We hold something called a closed and an open-hand approach. We have two hands, metaphorically speaking, in the closed hand we put doctrines that we will fight for and fight over. They’re very important. In the open hand, we’ll discuss and debate but not divide”
I think we could approach this discussion and academic purism around data visualisation (also applicable to dashboarding)in the same way.
These are my thoughts but
Closed Hand
If you want to fit the data into the shape of real-world objects, that’s not data visualization
If you use color when color is not needed, that’s not data visualization;
Alphabetical sort is not data visualization
If you want to grab attention, that’s not data visualization;
Open Hand
If you use more than one dimension to represent a data point, that’s not data visualization;
If your project breaks basic perception laws, that’s not data visualization;
If there is a perceptually simpler way to represent the data, what you are doing is not data visualization
If you reject a chart type because the audience may not be familiar with it, that’s not data visualization;
If your audience don’t know what the point of the chart is, that’s not data visualization
If your chart is nothing more than a glorified table, that’s not data visualization;
No variation, no data visualization;
If a simple formula can be a better answer, there is not place for data visualization;
If all you can do with a chart is to compare data points, that’s not data visualization;
If you have three or fewer data points, that’s not data visualization;
If you have to scroll, that’s not data visualization;
If you have to compare slides, that’s not data visualization;
What would you put in each hand ???
Thank you for putting this list together,
Can you expand on what you mean when you say “If all you can do with a chart is to compare data points”?
If you are looking at data, and visualize it, what is your intention if not to enable comparisons?
What would be an example of a visualization where the sole purpose was to compare data points, and that attribute made it a poor visualization?
Joe: Jacques Bertin says that a chart must provide three levels of reading, local, global and an intermediate level. In line chart you can see the overall trend, small/sub patterns and you can compare points A and B. In a pie chart, on the other hand, you compare a slice to the others. Not terribly interesting.
An obvious example of poor visualization is a bar chart sorted alphabetically. What can you do with it? You can compare Maryland to Kansas, but where to they fit in the overall pattern?
Thank you for the clarification, I see your point, and agree that a chart that supports non-useful comparisons would be an ineffective data visualization. That just enabling comparisons is not enough, that they should be meaningful.