reporting with numbers

Economics

Using graphs and other forms of data visualization

1) When using graphs, make sure these have titles, and that all axes are clearly labeled.

DO THIS

A graph (see below) displaying information on changing food prices between 2021 and 2022 breaks this up into six-month segments, with separate markers along the X axis for January 2021, June 2021, December 2021, June 2022, and December 2022.

A line graph titled "Food price increases are slowing," with the subtitle "Percent change in food prices compared to the month prior." The x-axis ranges from January '21 to December '22, and the y-axis from +0 to +1.0%. The source is listed as Bureau of Labor Statistics.

DON’T DO THIS

A graph (see below) placed in an article about rising food prices lacks a title as well as what data points mean.

A bar graph without a title. "12-month % change" is shown for each of 11 types of food products. Change appears to be between 7% and 18% for all categories.

DISCUSSION

Data visuals need to be accompanied by text. Titles are especially important, as they cue people into the kinds of data being presented. Titles can also be used to contextualize this data, and to draw attention to the key takeaway points within a given graph or chart. The first example does this quite effectively, as the title “Food price increases are slowing” directs readers’ eyes to the major trend in the data. By contrast, the graph in the second example does not even include a title, which means that readers are left for themselves to try to determine what is being tracked and what all of these individual data points mean. To avoid this kind of confusion, add titles, and make sure all axes are clearly labeled.

2) Make sure that captions, legends, and footnotes are all clear, comprehensible, and appropriately located.

DO THIS

A map (see below) displaying data on unemployment benefits around the U.S. includes a key that clearly explains the colors used to illustrate differences between states.

A map titled "how long do your state's unemployment benefits last?" States are colored in shades of blue corresponding to "fewer than 26 weeks," "26 weeks," and "more than 26 weeks."

DON’T DO THIS

A chart (see below) depicting U.S. job growth since the “Great Recession” of 2008 fails to explain what the difference between red- and blue-shaded bars is, and lacks anything resembling a legend.

A bar graph titled "U.S. Job Growth by Year: 2008 to 2022". The x-axis shows years, the y-axis is not labeled and ranges from -12000 to 8000. The bars for 2008 and 2017 to 2020 are colored red, the others are blue. The source is given as Bureau of Labor Statistics.

DISCUSSION

Data visuals need to be accompanied by text, including titles, captions, legends, and keys. Construct each of these elements in ways that help audiences understand the data you are depicting. The first example displays this information prominently and accessibly, while the second contains nothing to help readers understand the difference between blue- and red-shaded bars (which viewers might interpret politically, as in GOP = red and Democrats = blue). Without this key information, audiences will not be able to understand or interpret data visuals.

3) Keep things simple. Avoid overloading graphics with too much data.

DO THIS

A chart (see below) lists changes in the cost of over two dozen goods and services over the course of 2022.

An untitled bar graph displaying percent of price change for a wide range of products and services, as well as (bolded) Consumer Price Index categories. The source is given as the Bureau of Labor Statistics.

DON’T DO THIS

A graph (see below) displaying German stock market trends is jam-packed with text, arrows, and a series of unnecessary monthly chronological markers along the X axis.

An untitled line graph, with the x-axis showing time from June 2021 to January 2023 and the y-axis showing "MSCI Germany Index Level (With Net Dividends). A large amount of text is on the graph, with each brief description (such as "Russia restricts gas flow to Germany") connected to a point on the graph by an arrow. There is very little whitespace on the graph. The source is given as FactSet.

DISCUSSION

Whenever presenting economic data in graphical form, simpler is generally better. Although the first example contains lots of data, devoting one line to each data point makes the visualization both comprehensive and comprehensible. By contrast, the second example attempts to package a dizzying amount of data into a single graph, and is rather unreadable. Moreover, the accompanying text does little to explain these data points.

Provide feedback.