There are numerous benefits to communicating important information through visualizations, whether it be rapidly processing information or gaining better understanding of trends in your data. Nowadays, you don’t even need a degree in Data Science to create compelling data visualizations. BI software has made it easy for anyone to turn raw numbers into visually-appealing charts and graphs.
However, alongside the freedom and flexibility of democratized data, there are guidelines that must be followed to create an effective and informative data viz. If the rules are ignored, your “helpful” charts may end up causing more harm than good. By unintentionally (or intentionally) misrepresenting, manipulating, or altering data, you are compromising the integrity and validity of your visualization. So, how can you know if you are committing a data viz sin? Read on to learn about three of the most common, easily avoidable data viz downfalls so you can to spot misleading charts and avoid making the same data viz design mistakes.
How Data Visualization Can Be Used to Deceive Viewers
Describing Data Inaccurately
Data visualizations can mislead if you mislabel data or use text that inaccurately explains the chart or graph. A mislabeled pie chart or inaccurate axis can quickly ruin your visualization and destroy the viewer’s trust in the information you are providing.
When adding text to a data visualization, some DBAs and data designers forget that correlation does not necessarily equal causation. When you are presenting data that appears to show a correlation, be mindful not to imply or state that there is a causal relationship. There may be a third, unseen factor, often referred to as a “confounding factor,” that is influencing the variables you are analyzing and visualizing.
How to prevent: Don’t shy away from incorporating text into your data visualizations out of fear that you may mislead. Adding context can actually enhance your viewer’s experience and help them better understand the data.
In some cases, you may notice that an annotation is necessary to stop the viewer from making inaccurate assumptions based on absolute numbers. Context enables you to provide a valid interpretation of the data, guide the reader through your process, and explain an outlier or trend in the data. You just need to be careful not to overstep your analysis or embellish to make a point.
To avoid mislabeling, make time to get an extra set of eyes on your chart or graph before it gets published and go through this checklist:
- Does everything that needs to be labeled have a label? (If precise values aren’t important, it may make for a cleaner visualization to simply leave them out).
- Are the labels visible and easy to read?
- Are all of the labels matching up with the correct numbers?
Showing Too Much Data
The goal of data visualization is to take a large amount of data and make it easier to understand by putting it in a visual format. Your audience should be able to look at your visualization and quickly find what they are looking for. If you incorporate too many data points in your chart or graph, you aren’t accomplishing this goal.
When a chart is too busy, it can be hard to decipher the main points. The viewer may not know where to focus their attention or why the chart was created in the first place. While effective data visualizations inspire action, cluttered data visualizations lead to analysis paralysis where the viewer is so overloaded with information that they are unable to make a decision.
How to prevent: Less is more, so keep it simple. Make sure everything that you include in your visualization has a purpose. Figure out what information is essential to help your data viz viewers understand your main point.
Your chart or graph should have a narrow focus that highlights an important trend, relationship, or outlier in the data without omitting or changing any of the relevant data points. If you think you need to provide more information or detail, you can always create more than one chart or graph to present the most accurate picture of the data.
Read next: Fighting Cognitive Bias with Data
Good data visualization is all about transparency – it uncovers and exposes your data. When data is distorted, it undermines the whole purpose of a data visualization. Some common ways that data gets distorted are:
- Inappropriate chart type: Not all chart types will accomplish the same goal, so it’s important to understand which chart type is the best option to convey your data’s story.
- Not to scale: When a graph is not to scale, it is no longer possible to quickly gauge what it is telling you. It can also make the data appear inaccurate and lead to greater confusion.
- Omitted data: When data is omitted, trends that don’t exist may be created and trends that do exist may be hidden.
- Truncated y-axis: A non-zero baseline can skew data in a way that is especially harmful when comparing data points.
How to prevent: Be mindful of chart choice. For example, pie charts are used to compare parts of a whole, while bar charts are used to show the differences between groups. Whether you are looking to present the distribution, relationship, composition, or comparison of your data, this infographic can help you choose the right chart type. Also, it is best to avoid 3D charts because they make it difficult to actually tell what the numbers are.
The safest way to organize the y-axis is to start at zero and then go up to the highest data point in your set. That way, small differences do not appear larger than they actually are. Discontinuities in axes can sometimes be necessary, but it is important to remember that introducing them means that you are giving up the power of comparison. If you absolutely have to use a y-axis break, clearly indicate that you have done so and make sure that the scale is proportionate to the data.