Saturday, May 2, 2020

Bad Charting Part 4: Following Conventions

This is Part 4 in a series on avoiding the pitfalls of bad charting. In this final post of the series we'll look at the importance of following conventions in charts, including use of color and the use of 3D in charts.

Conventions


Conventions. Without them, useful communication is impossible: humans couldn't converse, publish, collaborate, or have discourse (civil or otherwise) without conventions. And so it is with charts: fail to follow conventions, and you'll be sending the wrong message. If you're digesting a chart prepared by someone else, be on the lookout for flouting conventions: it's one of the ways a chart can be rigged to send a contrary message to the story the data tells. In Part 1, we mentioned how makers of infographics often don't feel constrained to "follow the rules" in their charts. Violating conventions is one of the chief offenses.

Directional Flow

The flow of things on your chart, including arrows, should conform to the culture of your audience. For example, the following are deeply ingrained in Western culture, to the point where few people even bother to think about them consciously: they're just understood.


Time moves left to right, always. Going up means more, gain, or north; down means less, loss, or south. Fail to follow these conventions, and people will struggle to understand your visuals. Targeting a different culture than yours? Do your research and find out what they consider normal.

Here's the most egregious example I know of ignoring conventions. What does this chart make you conclude about the effect of Florida's "Stand Your Ground" law? Gun deaths appear to drop rapidly after 2005.



Now think about what we covered previously about the importance of accurate chart axes, and take a look at the Y axis tick markings: they're upside down! This chart is effectively inverted, because it fails to follow the convention that up means more.

On a Map, Darker Map Shades Mean More

Consider a map that is conveying data, such as which US states make the most revenue. The convention is that darker shades mean more (greater density or higher amounts).

In the map chart below, darker blue shades were used for lower export profits and lighter green for higher. It's the opposite from most maps of this kind. This doesn't make the chart evil, but it is flouting a convention. If the shading scale was reversed, readers would get what the map is trying to convey more readily. Every time you go against a convention, you up the chance someone will get the wrong message.

Use of Color


If you're just trying to select two colors for showing last month's expenses vs. this month's expenses, color choice might not seem very important, but it can be. It helps to understand how your audience perceives color so you can use it to support the story your chart is telling. We'll talk about some of the meanings attached to colors by different groups. First though you should understand that it's a mistake to rely solely on color to communicate something: a substantial number of people are colorblind (and many don't even know it). Always accompany color with other visual cues. Here's how a colored chart might appear to a colorblind person:



Color might appear to be one of those worldwide or east/west cultural matters, as in "Western culture associates red with danger"—and it is, to some degree. Then again, red is often used to communicate other things like excitement or Communism to that same audience. 

Like most things, going extreme in color will make your chart worse, not better. Avoid too many colors or strong saturated hues: it's the color graphics equivalent of shouting. If you practice restraint in your general color palette, then you can highlight something really effectively with a stronger color.In the two charts below, is the color chart on the left easier or harder to perceive than the grayscale version to the right? I find the color chart loud and ugly with colors that distract rather than help. The grayscale chart is effective without the color.


Different groups attach different meanings to color, known as Color Biases. Speaking broadly, western culture audience will attach any of these associations to colors:
  • Blue: trust, security, peace, coolness
  • Green: nature, freshness, luck, environment, wealth, inexperience, jealousy
  • Orange: warmth, harvest, light, heat
  • Purple: power, royalty, ambition, independence
  • Red: warmth, excitement, passion, love, danger
  • Yellow: joy, value, sunlight, caution, cowardice

We can't stop there however. Many industries also have color biases, and failing to know that could really upset how chart colors will hit your audience.You might be inclined to use green for good things and red for bad, but to someone in health care green means infected and red means healthy! It's worth getting to know your audience, and one easy way to do that is to examine how charts in their industry are commonly colored and labelled.
  • In Finance: blue is reliable/subdued, green is profitable, yellow is highlighted/important, red is unprofitable
  • In Health Care: blue is dead, green is infected, yellow is jaundiced, red is healthy
  • To Control Engineers, blue means cold/water, green is safe, yellow is caution, red is danger

3D

There's only one rule for using 3D in your charts and it's very simple: don't do it, ever. Why, you ask? Making that bar chart or pie chart into 3D adds a nice touch, you might argue. Let's see. Look at the two pie charts below. Who sold the most in Q1? And who sold the most in Q2? When I asked this question the last time I presented on charting, many people said Paul (blue) sold the most in Q1 and Bryan (orange) sold the most in Q2. 


In fact, these are two pie charts of the same data (below), simply with a different rotation. Paul and Bryan have exactly the same sales. The 3D effects break the visual contract that normally gives a pie chart its power: your eye and brain interpreting the relative proportions of the visual elements. That's completely demolished by a 3D pie chart. There's no greater evil in the charting world.



In the 3D column chart below, your first impulse is to see the green 1997 value as way smaller than the red 1995 value. You can intellectually understand there's perspective in the graphic, but a reader's first take on the chart is still going to be misleading.

3D column chart

There's simply nothing to be gained by making your charts 3D. Promise me you won't do it.

Resources

This series was inspired by the first resource I came across about deceptive charting, How to Lie with Charts by Gerald Everett Jones. Now in its fourth edition, Gerald has kept his book current with new material on topics like fake news and social media disinformation.

Well, that's it for this series. I hope you now feel well-equipped to detect misleading charts when you encounter them, and to avoid being misleading in the charts you create.

No comments: