I’m a data visualization guy. I work with data, I interpret data, and I help other people understand the data that I’ve interpreted by accurately representing that data visually. A clear, accurate graph can be a powerful tool for displaying complex information in an understandable way. But, there’s danger in graphs as well. Graphs and statistics can be easily manipulated in ways to prove points that the data might not back up.

Lots of data about health gets graphed every day. Naomi Robbins, writing for Forbes, gives us a good real-world example of a misleading graph that attempts to communicate the quality of different hospitals. The graph shows the results of post-visit survey questions on individual hospital quality. At first blush, the example graph show generally good survey outcomes at this hospital, and one you might be comfortable visiting. Upon further inspection, the individual graphs don’t have a consistent scale, which visually overstates the results for several survey questions, so you can’t really compare across the different graphs visually.

So what is a misleading graph, and how can you spot them? A misleading graph is any graph that distracts the reader in some way from fully understanding what the data actually say. There are a lot of ways that misleading graphs get created. Some of the ways include graphs with:

• No title
• No axes
• No labels on the axes
• Intervals on the axes that are unequal
• Vertical axes that don’t start at zero
• Pictures or icons to represent data
• A series of graphs that are not on the same scale
• Three dimensions, where the third dimension does not add information
• The wrong kind of graph

Many times, misleading graphs are created entirely unintentionally. So, how do you avoid creating a misleading graph when trying to communicate your message to your audience?

First, use the right kind of graph. Graphing data has never been easier—all sorts of options are available, which means that many incorrect options are available as well. Pay attention to the type of data you have: for example, if you’re comparing the health outcomes from different groups, you’d use a bar chart. If you have data that varies over time, you’d probably use line a line graph. If your data describes different sectors that are a proportion of the whole, use a pie chart.

This may seem like overly simple advice, but I can’t tell you how many line graphs I’ve seen that compare different groups, which makes for a meaningless graph. Or how many pie charts have individual components that add up to more than 100 percent. Unlabeled axes are astoundingly common.

Next, avoid misleading techniques. Unfortunately, many of the most visually striking graphs are misleading; omitting axes, for example, makes for a clean, uncluttered look. But remember that your first goal is to convey meaningful information, not impress your audience with aesthetics. A good communicator can make accurate, aesthetically pleasing graphs.

If you’re looking for good information on avoiding creating misleading graphs in your own communication efforts, a first stop might be the Wikipedia article for misleading graphs to learn what the different types of misleading graphs look like. If you want to learn even more, check out “How to Lie with Statistics” by Darrell Huff or “The Wall Street Journal Guide to Information Graphics: The Dos and Don’ts of Presenting Data, Facts, and Figures” by Dona Wong.

A final word of warning in communicating health data: be sure to carefully examine any graphs you want to use in your communication, whether you generate them or they come from someone else, and see if they include any of the danger signs discussed here. If your message is worth telling, it’s worth telling correctly. Visualizing data correctly is the single best way to determine whether your message is correct. And if you have to use a misleading graph to get your message across, the message, not the data is likely the one that is incorrect.

Gregory L. Torell is a 4th-year PhD candidate in economics at the University of Wyoming, where he studies environmental and natural resource economics, the economics of electricity and rangeland economics. He previously studied economics and German at New Mexico State University, receiving B.A.s in Economics and Foreign Language: German, as well as an M.S. in Agricultural Economics with a Master’s Minor in Experimental Statistics. He also studied macroeconomics and game theory at the University of Bern in Switzerland.  He is always up for a good discussion of music, soccer, the latest book you read or your favorite podcasts.