Storytelling with Data

by Cole Nussbaumer Knaflic

  Graphing applications (like Excel) typically have conditional formatting functionality built in that allows you to apply formatting like that shown in Figure 2.5 with ease. Be sure when you leverage this to always include a legend to help the reader interpret the data (in this case, the LOW-HIGH subtitle on the heatmap with color corresponding to the conditional formatting color serves this purpose).

  Next, let’s shift our discussion to the visuals we tend to think of first when it comes to communicating with data: graphs.

  Graphs

  While tables interact with our verbal system, graphs interact with our visual system, which is faster at processing information. This means that a well-designed graph will typically get the information across more quickly than a well-designed table. As I mentioned at the onset of this chapter, there are a plethora of graph types out there. The good news is that a handful of them will meet most of your everyday needs.

  The types of graphs I frequently use fall into four categories: points, lines, bars, and area. We will examine these more closely and discuss the subtypes that I find myself using on a regular basis, with specific use cases and examples for each.

  Chart or graph?

  Some draw a distinction between charts and graphs. Typically, “chart” is the broader category, with “graphs” being one of the subtypes (other chart types include maps and diagrams). I don’t tend to draw this distinction, since nearly all of the charts I deal with on a regular basis are graphs. Throughout this book, I use the words chart and graph interchangeably.

  Points

  Scatterplot

  Scatterplots can be useful for showing the relationship between two things, because they allow you to encode data simultaneously on a horizontal x-axis and vertical y-axis to see whether and what relationship exists. They tend to be more frequently used in scientific fields (and perhaps, because of this, are sometimes viewed as complicated to understand by those less familiar with them). Though infrequent, there are use cases for scatterplots in the business world as well.

  For example, let’s say that we manage a bus fleet and want to understand the relationship between miles driven and cost per mile. The scatterplot may look something like Figure 2.6.

  Figure 2.6 Scatterplot

  If we want to focus primarily on those cases where cost per mile is above average, a slightly modified scatterplot designed to draw our eye there more quickly might look something like what is shown in Figure 2.7.

  Figure 2.7 Modified scatterplot

  We can use Figure 2.7 to make observations such as cost per mile is higher than average when less than about 1,700 miles or more than about 3,300 miles were driven for the sample observed. We’ll talk more about the design choices made here and reasons for them in upcoming chapters.

  Lines

  Line graphs are most commonly used to plot continuous data. Because the points are physically connected via the line, it implies a connection between the points that may not make sense for categorical data (a set of data that is sorted or divided into different categories). Often, our continuous data is in some unit of time: days, months, quarters, or years.

  Within the line graph category, there are two types of charts that I frequently find myself using: the standard line graph and the slopegraph.

  Line graph

  The line graph can show a single series of data, two series of data, or multiple series, as illustrated in Figure 2.8.

  Figure 2.8 Line graphs

  Showing average within a range in a line graph

  In some cases, the line in your line graph may represent a summary statistic, like the average, or the point estimate of a forecast. If you also want to give a sense of the range (or confidence level, depending on the situation), you can do that directly on the graph by also visualizing this range. For example, the graph in Figure 2.9 shows the minimum, average, and maximum wait times at passport control for an airport over a 13-month period.

  Figure 2.9 Showing average within a range in a line graph

  Note that when you’re graphing time on the horizontal x-axis of a line graph, the data plotted must be in consistent intervals. I recently saw a graph where the units on the x-axis were decades from 1900 forward (1910, 1920, 1930, etc.) and then switched to yearly after 2010 (2011, 2012, 2013, 2014). This meant that the distance between the decade points and annual points looked the same. This is a misleading way to show the data. Be consistent in the time points you plot.

  Slopegraph

  Slopegraphs can be useful when you have two time periods or points of comparison and want to quickly show relative increases and decreases or differences across various categories between the two data points.

  The best way to explain the value of and use case for slopegraphs is through a specific example. Imagine that you are analyzing and communicating data from a recent employee feedback survey. To show the relative change in survey categories from 2014 to 2015, the slopegraph might look something like Figure 2.10.

  Figure 2.10 Slopegraph

  Slopegraphs pack in a lot of information. In addition to the absolute values (the points), the lines that connect them give you the visual increase or decrease in rate of change (via the slope or direction) without ever having to explain that’s what they are doing, or what exactly a “rate of change” is—rather, it’s intuitive.

  Slopegraph template

  Slopegraphs can take a bit of patience to set up because they often aren’t one of the standard graphs included in graphing applications. An Excel template with an example slopegraph and instructions for customized use can be downloaded here: storytellingwithdata.com/slopegraph-template.

  Whether a slopegraph will work in your specific situation depends on the data itself. If many of the lines are overlapping, a slopegraph may not work, though in some cases you can still emphasize a single series at a time with success. For example, we can draw attention to the single category that decreased over time from the preceding example.

  In Figure 2.11, our attention is drawn immediately to the decrease in “Career development,” while the rest of the data is preserved for context without competing for attention. We will talk about the strategy behind this when we discuss preattentive attributes in Chapter 4.

  Figure 2.11 Modified slopegraph

  While lines work well to show data over time, bars tend to be my go-to graph type for plotting categorical data, where information is organized into groups.

  Bars

  Sometimes bar charts are avoided because they are common. This is a mistake. Rather, bar charts should be leveraged because they are common, as this means less of a learning curve for your audience. Instead of using their brain power to try to understand how to read the graph, your audience spends it figuring out what information to take away from the visual.

  Bar charts are easy for our eyes to read. Our eyes compare the end points of the bars, so it is easy to see quickly which category is the biggest, which is the smallest, and also the incremental difference between categories. Note that, because of how our eyes compare the relative end points of the bars, it is important that bar charts always have a zero baseline (where the x-axis crosses the y-axis at zero), otherwise you get a false visual comparison.

  Consider Figure 2.12 from Fox News.

  Figure 2.12 Fox News bar chart

  For this example, let’s imagine we are back in the fall of 2012. We are wondering what will happen if the Bush tax cuts expire. On the left-hand side, we have what the top tax rate is currently, 35%, and on the right-hand side what it will be as of January 1, at 39.6%.

  When you look at this graph, how does it make you feel about the potential expiration of the tax cuts? Perhaps worried about the huge increase? Let’s take a closer look.

  Note that the bottom number on the vertical axis (shown at the far right) is not zero, but rather 34. This means that the bars, in theory, should continue down through the bottom of the page. In fact, the way this is graphed, the visual increase is 460% (the heights of the bars ar
e 35 – 34 = 1 and 39.6 – 34 = 5.6, so (5.6 – 1) / 1 = 460%). If we graph the bars with a zero baseline so that the heights are accurately represented (35 and 39.6), we get an actual visual increase of 13% ((39.6 – 35) / 35). Let’s look at a side-by-side comparison in Figure 2.13.

  Figure 2.13 Bar charts must have a zero baseline

  In Figure 2.13, what looked like a huge increase on the left is reduced considerably when plotted appropriately. Perhaps the tax increase isn’t so worrisome, or at least not as severe as originally depicted. Because of the way our eyes compare the relative end points of the bars, it’s important to have the context of the entire bar there in order to make an accurate comparison.

  You’ll note that a couple of other design changes were made in the remake of this visual as well. The y-axis labels that were placed on the right-hand side of the original visual were moved to the left (so we see how to interpret the data before we get to the actual data). The data labels that were originally outside of the bars were pulled inside to reduce clutter. If I were plotting this data outside of this specific lesson, I might omit the y-axis entirely and show only the data labels within the bars to reduce redundant information. However, in this case, I preserved the axis to make it clear that it begins at zero.

  Graph axis vs. data labels

  When graphing data, a common decision to make is whether to preserve the axis labels or eliminate the axis and instead label the data points directly. In making this decision, consider the level of specificity needed. If you want your audience to focus on big-picture trends, think about preserving the axis but deemphasizing it by making it grey. If the specific numerical values are important, it may be better to label the data points directly. In this latter case, it’s usually best to omit the axis to avoid the inclusion of redundant information. Always consider how you want your audience to use the visual and construct it accordingly.

  The rule we’ve illustrated here is that bar charts must have a zero baseline. Note that this rule does not apply to line graphs. With line graphs, since the focus is on the relative position in space (rather than the length from the baseline or axis), you can get away with a nonzero baseline. Still, you should approach with caution—make it clear to your audience that you are using a nonzero baseline and take context into account so you don’t overzoom and make minor changes or differences appear significant.

  Ethics and data visualization

  But what if changing the scale on a bar chart or otherwise manipulating the data better reinforces the point you want to make? Misleading in this manner by inaccurately visualizing data is not OK. Beyond ethical concerns, it is risky territory. All it takes is one discerning audience member to notice the issue (for example, the y-axis of a bar chart beginning at something other than zero) and your entire argument will be thrown out the window, along with your credibility.

  While we’re considering lengths of bars, let’s also spend a moment on the width of bars. There’s no hard-and-fast rule here, but in general the bars should be wider than the white space between the bars. You don’t want the bars to be so wide, however, that your audience wants to compare areas instead of lengths. Consider the following “Goldilocks” of bar charts: too thin, too thick, and just right.

  We’ve discussed some best practices when it comes to bar charts in general. Next let’s take a look at some different varieties. Having a number of bar charts at your disposal gives you flexibility when facing different data visualization challenges. We’ll look at the ones I think you should be familiar with here.

  Vertical bar chart

  The plain vanilla bar chart is the vertical bar chart, or column chart. Like line graphs, vertical bar charts can be single series, two series, or multiple series. Note that as you add more series of data, it becomes more difficult to focus on one at a time and pull out insight, so use multiple series bar charts with caution. Be aware also that there is visual grouping that happens as a result of the spacing in bar charts having more than one data series. This makes the relative order of the categorization important. Consider what you want your audience to be able to compare, and structure your categorization hierarchy to make that as easy as possible.

  Stacked vertical bar chart

  Use cases for stacked vertical bar charts are more limited. They are meant to allow you to compare totals across categories and also see the subcomponent pieces within a given category. This can quickly become visually overwhelming, however—especially given the varied default color schemes in most graphing applications (more to come on that). It is hard to compare the subcomponents across the various categories once you get beyond the bottom series (the one directly next to the x-axis) because you no longer have a consistent baseline to use to compare. This makes it a harder comparison for our eyes to make, as illustrated in Figure 2.16.

  Figure 2.14 Bar width

  Figure 2.15 Bar charts

  Figure 2.16 Comparing series with stacked bar charts

  The stacked vertical bar chart can be structured as absolute numbers (where you plot the numbers directly, as shown in Figure 2.16), or with each column summing to 100% (where you plot the percent of total for each vertical segment; we’ll look at a specific example of this in Chapter 9). Which you choose depends on what you are trying to communicate to your audience. When you use the 100% stacked bar, think about whether it makes sense to also include the absolute numbers for each category total (either in an unobtrusive way in the graph directly, or possibly in a footnote), which may aid in the interpretation of the data.

  Waterfall chart

  The waterfall chart can be used to pull apart the pieces of a stacked bar chart to focus on one at a time, or to show a starting point, increases and decreases, and the resulting ending point.

  The best way to illustrate the use case for a waterfall chart is through a specific example. Imagine that you are an HR business partner and want to understand and communicate how employee headcount has changed over the past year for the client group you support.

  A waterfall chart showing this breakdown might look something like Figure 2.17.

  Figure 2.17 Waterfall chart

  On the left-hand side, we see what the employee headcount for the given team was at the beginning of the year. As we move to the right, first we encounter the incremental additions: new hires and employees transferring into the team from other parts of the organization. This is followed by the deductions: transfers out of the team to other parts of the organization and attrition. The final column represents employee headcount at the end of the year, after the additions and deductions have been applied to the beginning of year headcount.

  Brute-force waterfall charts

  If your graphing application doesn’t have waterfall chart functionality built in, fret not. The secret is to leverage the stacked bar chart and make the first series (the one that appears closest to the x-axis) invisible. It takes a bit of math to set up correctly, but it works great. A blog post on this topic, along with an example Excel version of the above chart and instructions on how to set one up for your own purposes can be downloaded at storytellingwithdata.com/waterfall-chart.

  Horizontal bar chart

  If I had to pick a single go-to graph for categorical data, it would be the horizontal bar chart, which flips the vertical version on its side. Why? Because it is extremely easy to read. The horizontal bar chart is especially useful if your category names are long, as the text is written from left to right, as most audiences read, making your graph legible for your audience. Also, because of the way we typically process information—starting at top left and making z’s with our eyes across the screen or page—the structure of the horizontal bar chart is such that our eyes hit the category names before the actual data. This means by the time we get to the data, we already know what it represents (instead of the darting back and forth our eyes do between the data and category names with vertical bar charts).

  Like the vertical bar chart, the horizontal bar chart can be single series, two series
, or multiple series (Figure 2.18).

  Figure 2.18 Horizontal bar charts

  The logical ordering of categories

  When designing any graph showing categorical data, be thoughtful about how your categories are ordered. If there is a natural ordering to your categories, it may make sense to leverage that. For example, if your categories are age groups—0–10 years old, 11–20 years old, and so on—keep the categories in numerical order. If, however, there isn’t a natural ordering in your categories that makes sense to leverage, think about what ordering of your data will make the most sense. Being thoughtful here can mean providing a construct for your audience, easing the interpretation process.