Choose the Right Viz for the Job

When you think of visualizing data, your mind probably goes to bar graphs or maybe pie charts. However, there are many more species of visualizations. Ever heard of a waterfall or a circular area chart? Your first decision when visualizing data is what type of chart or graph to choose and that depends on what you want to show and what type of data you have.

I highly recommend Andrew Abela’s simple decision tree called “Chart Suggestions—A Thought-Starter” (see image below).  It’s based on Gene Zelazny's classic work Saying It With Charts. The decision tree starts with the basic question: “What would you like to show?” And provides four options:

Comparison. You have two or more groups of things or people and you want to see which group is largest or smallest (or somewhere in between) on some measure. You also may want to see how these groups compare on the measure over time.

Distribution. You have a bunch of data points (e.g. the ages of participants in a program or test scores of students in a class) and you want to know how spread out or bunched up they are. Are most of the ages, test scores (whatever) near the average? Or is there a wide range? Are there some extreme outliers?

Composition. You want to understand who or what makes up a larger group such as how many of the participants in a program are in different age brackets or how many have been in the program for different lengths of time.

Relationship. You want to know if one thing is related to another, either at one point in time or over time. Do participants in a mental health program report less distress over time? Do those with lower incomes have higher heart rates?

Once you answer this basic question, the decision tree helps you to choose a specific chart based on the type of data you have. Abela’s chart chooser includes the types of charts you are most likely to select. But there are more rare species out there. To learn more about the wide array of ways to visualize data, check out the Data Visualization Catalogue.

See other data tips in this series for more information on how to effectively visualize and make good use of your organization's data.

What Averages Obscure

Nonprofits (and everyone else) are addicted to averages. We like to talk about how participants do on average. We might describe how many visitors we have in an average week. But how much are we missing when we focus solely on averages? Short answer: it depends, but it could be a lot. If I only showed you the average sized guy in the picture, would you appreciate the full range of sizes?

To figure out what and how much we are missing, we need to calculate—or better yet show—how spread out our data points are. Understanding the spread gives us an idea of how well the average or the median represents the data. When the spread of values in the data set is large, the average obscures the real picture more than when the spread is small.

Spread measures include range, quartiles, absolute deviation, variance and standard deviation. For more on these measures, check this out. 

A great way to quickly grasp the spread of your data is to make a box plot. A box plot (aka. box and whisker diagram) shows the distribution of data including the minimum, first quartile, median, third quartile, and maximum. The box plots below show the affordability of neighborhoods in five cities. Each red circle represents a zip code area. The gray boxes show where 50 percent of the zip code areas fall on the affordability scale. And the median is where the dark gray meets the light gray. You can see that, in general (i.e according to the median), New York is more affordable than Los Angeles. However, New York has some zip code areas that are much less affordable than the median seems to suggest.

So when looking at your data, don’t just look at averages, also consider the spread.

See other data tips in this series for more information on how to effectively visualize and make good use of your organization's data.

Image created by Moxilla for Noun Project.

Pies are for eating

There are better ways to show your data.

Pies are delicious but often inscrutable when applied to data. Humans are pretty good at deciphering some visual cues and pretty bad at others. For example, we do well when comparing lengths along a common scale. So looking at this image, we can confidently proclaim the E bar as the tallest. But we would be hard pressed to pick out which pie slice is largest. That’s because we don’t do so well with angles. So when comparing the quantities of several things, bar charts are almost always better than pie charts. The only exception is when you want to compare a part to a whole. In this case, a pie chart does a good job of showing that girls, for example, represent only a sliver of all the participants in a program or that 30 to 40 year olds are the majority of visitors to an event. But once you get beyond 2 (or maybe 3) slices, skip the pie and dust off the trusty bar chart.