Data visualization is something they really should teach in high school–or maybe earlier–because we receive a lot of numerical information visually these days. This essay is a brief introduction to the topic.
Tables and charts are a relatively new tool for visually depicting numerical information. You could go back to Rene Descartes, who created the Cartesian (x-y) coordinate system in the early 1600s for graphing points, lines, and conic-section curves. The perpendicular axes have Western conventions for determining numerical values: specifically, points on the coordinate system increase in value from left to right and from bottom to top (ascending/increasing in value the closer up you get to pointing toward Heaven).
Moving ahead nearly 200 years, we encounter English clergyman Joseph Priestly and Scottish economist named William Playfair, who started the first concerted effort to make economic and other forms of information understandable to the general public. Priestly developed the timeline chart to depict overlapping lifespans over time (one of the examples best known to me is the chart of Robert A. Heinlein’s “Future History”). Playfair, meanwhile, worked on line, bar, area, and pie charts–basic utilities you can find in your Microsoft Excel toolbar.
Line charts use a line, often but not always connecting data points in order to show the trend in a set of values over time. The horizontal (x) axis is usually the time, with dates moving from past to future left to right. The vertical (y) axis is the quantity/value being measured, with values increasing as one moves up the chart. The line chart below, for example, shows the price of shares for The Walt Disney Company over the last five trading sessions at the New York Stock Exchange.
Looking at this chart, it might look like the stock is extremely volatile and is definitely lower than it was five trading days ago. However, if you’re looking at the stock in the long term, say, five years, the overall trend of the stock has very much been on the up side.
Note, too, that the five-day chart does not start its vertical axis at zero, as the five-year chart does. Instead, the five-day chart’s “zero point” is actually just the lowest value the stock has reached in the last five days. That might be deceptive (and it can be, which I’ll explain momentarily), but for the intended user–someone who’s been selling or planning to sell a stock in the near future–the five-day chart’s range of possible values is what matters. Pull back the focus and a lot of that volatility smooths out and a broader trend can be seen. The y axis for the five-year chart does start from zero.
- Good for tracking a single value over time or maybe one or two comparable values over the same length of time. The more lines you add, the more colors you need, and eventually things begin to blur.
- Good for showing trends
- Not as good for tracking values with multiple components
- Not very visually appealing
Bar charts consist of vertical or horizontal bars (usually rectangles, but other shapes work as well comparing multiple items. Unlike line charts, which mostly depict changes over time, bar charts can be used to compare multiple items over the same metric without reference to time. For example, the chart below compares past, current, and future launch vehicles. Note that these aren’t exactly “bars,” but do convey the same information–their height/length.
Here’s another example, a split-bar chart, where each bar has multiple related components. In this case, I’ve set up a chart showing bars depicting the number of launches for each NASA program–broken out by those with crew and those without.* This chart could also be seen as a hybrid because I’ve also set up the programs in chronological order, left to right. This hybrid format can also be used to track things like the value of a financial portfolio with multiple components over time.
- Able to compare multiple linear data sets easily based on similar criteria
- Able to compare data sets across time
- Able to compare data sets with multiple constituents
- Not effective for depicting percentages/proportions
We see and understand pie charts relatively easily. The meme below is perhaps the most direct depiction of one. Pie charts are used mostly to show percentages of a single whole, such as a budget or the political makeup of a body like the United States Senate. The problem with a pie chart is that while it can easily depict “slices” of a single whole, it’s not always effective for conveying the scale of whole itself. Nor–unless the wholes are exactly equal–can you use them very well to convey different “pies.”
- Good for depicting percentages
- Easy to comprehend with a few large “slices”
- Difficult to read with a lot of very small slices
- Difficult for depicting actual values of the whole or the slices
- Impossible to compare two different data sets (wholes) unless the sets are exactly equal
All of the above items are primarily graphical. Yet there are times when it’s easier to organize text items visually in a table. Based on the archeological evidence, we’ve been organizing numbers in things like rows and columns for millennia. Like bar charts, tables can be used to compare multiple data sets across the same criteria. And tables, unlike charts, can easily compare text- and number-based information. Today we see tables for quick-reference applications, such as prices, value conversions, or schedules. While tables can be made more colorful for easier legibility, they aren’t always exciting graphical elements.
- Useful for comparing text and numeric data across multiple data objects
- Easy to understand “axes”
- Not useful for showing changes across time
- Not exciting visually
Graphics can lie
How could a chart be made to distort one’s view of data? The placement of the x axis is one trick. If your livelihood depended on the budget of a specific government agency (federal, state, or local). The current president or mayor is proposing a cut to your specific program from $1 million to $900,000, or maybe the expected rate of increase was not as much as expected or hoped for–say, 5 percent instead of 10 percent. That budget cut (or “cut” in the latter situation) must be made to look to your supporters as dire and and horrific in its impact as possible. You might choose a leading title and shifting the x axis farther upward to make the drop look precipitous, as depicted below.
…when the actual cut, visually depicted from zero point up would look something like this:
Or here’s an interesting one…someone inverted the y axis to give a false impression about gun deaths.
Bottom line: know your graphics!
Graphics are becoming increasingly important to how technical information is conveyed. It is vital that technical communicators of all varieties understand the conventions, pros, and cons of different visual representations. This can involve working with a graphic designer or you might be the one designing. Regardless, the method you use for sharing multiple sets of data will determine the readability and even the trustworthiness of your document. Take the time to present your data well and your readers will appreciate it.