10 Common Data Visualization Pitfalls to Avoid

“The purpose of visualization is insight, not pictures.”

― Ben Schneiderman

Data is powerful. All industries rely on data to make informed decisions about virtually everything — product development, marketing, recruiting, etc. However, just like a treasure trove, we must dig deep to find quality data.

In Agoda, data is integral to each of our work processes. Therefore, it is necessary that we accurately present our data. This will ensure that we make decisions that provide suitable solutions and reduce potential data handling risks.

When creating data visualizations, it can be easy to make mistakes that lead to wrong interpretation. In this article, we will look at bad data visualization and how to avoid it.

What are the signs of bad data visualization?

“Data visualization is the language of decision making. Good charts effectively convey information. Great charts enable, inform, and improve decision making.” — Dante Vitagliano.

Data visualization is a great technique to portray large amounts of information in a simple way. All data visualizations have the same goal: to make information easy to understand so that users can make quick insights or decisions.

Creating a good data visualization is more than just simply placing some data into colorful charts. It is critical that our visualizations are not overdone and instead reach the perfect balance of being engaging, instructive, and simple to navigate.

In most cases, poorly constructed visualizations are more confusing than helpful. The following are signs of bad data visualization:

• Hiding the relevant data
• Presenting too much data
• Distorting the presentation of data
• Describing the data inaccurately.
• Visualization that confuses your audiences

10 common data visualization mistakes and how to avoid them

Here are ten typical mistakes that result in poor data visualization. Try to avoid these to get the most out of your data visualizations.

1. Misleading color contrast

Although using various colors aids in interpreting data visualizations, too much color can confuse the user. It’s crucial to stick to a limited number of unique colors.

Source: Common Pitfalls of Color Use

Impact of this type of visualization

• The user misunderstands which value is more significant.
• When there is too much color in a visualization, it might take the user longer to understand the information.

How to solve this

• Use colors to show which value is higher or lower than the others. Colors with a high contrast cause viewers to perceive more data value.
• The simplest way to determine contrast value is to compare contrast colors on greyscale to check if the color you choose displays the difference.

• Use hot tone — cool tone color.
  Colors with a hot/cool tone help indicate a significant difference in values and positive/negative emotions.

2. Overwhelming charts with too much data

There’s nothing wrong with having a lot of data to offer you more depth, but having too much data to show it all at once might cause the audience to get overwhelmed.

Source: Excessive data in a single visualization immediately overwhelms viewers.

Impact of this type of visualization

• The user is unable to understand all of the visualization details.
• The user has no idea where to focus their attention.
• It will be hard to decipher the message in a short amount of time.

Key takeaway

• Start by determining only what users need to focus on, so you can limit the data to only those most relevant to the message you want to convey.
• Do not put all of your insights into a chart. Multiple visualizations can help you communicate data more effectively.
• It is recommended that no more than 5–6 colors be used in a single visualization.

3. Omitting baseline and truncating scale

This data visualization problem is widespread in politics and sports, and it might indicate false patterns or even trends that do not exist.

Source: When Data Visualization Really Is not Useful (and When It Is)

Impact of this type of visualization

• The audience does not fully comprehend the data, which leads to misconceptions and, in some instances, social trends.
• When the audience realizes the visualization is displaying erroneous information, they will lose trust in the organization.

Key takeaway

• Concentrate on creating data visualizations with a zero-baseline y-axis.
• If removing the zero makes sense, add a zero-break to communicate that * If the minor adjustments are truly significant, not beginning from zero is acceptable.

4. Biased text modifications

In data visualization, not only does the chart itself communicate with the audience, but the title, label, notation, and description also aid users in comprehending the message.

If, on the other hand, these changes present a story that differs somewhat from the data, the user may become confused.

Source: https://ecampusontario.pressbooks.pub/bio16610w18/chapter/how-graph-misrepresents-data/

This graph, for example, depicts the percentage of children that suffer from orthopedic injuries. If the user reads the headline without reading the description, they may believe that 5.2% of normal children suffer from spinal cord damage, resulting in serious misunderstandings.

The impact of this type of visualization

Even if the data is correct, audience interpretations might be affected if the text modification is misleading.

Key takeaway:

• Written descriptions should only be used if they are required to clarify what is being displayed.
• Be sure that the title, label, and description convey the intended meaning without bias.

5. Choosing the wrong visualization method

Choosing the appropriate visualization to represent your data is a critical step in data visualization. Several charts may be appropriate for displaying your data, but how do you choose the best one?

Source: https://www.tessellationtech.io/top-five-ways-mislead-data-visualization/

In the figure above, both charts may indicate the percentage of responses for each candidate. Still, when we represent it in a pie chart, the user may be confused since the chart portions seem similar to one another, and the number cannot sum to 100 percent.

Impact of this type of visualization

Using the wrong chart type can confuse viewers or mislead them.

How to choose the best chart for your data?

To choose an appropriate chart for your data, you must first determine what insight your data is attempting to communicate. Once you have that information, here are some simple guidelines to assist you in selecting the best chart for your data.

Source: Venn gage

6. Correlations without causations

Have you ever encountered data that showed similar trends despite vastly different factors? Even if you try to figure out how those factors are related, it appears that there isn’t one. This is another form of data visualization blunder that can lead users to try to find out the cause of something that has nothing to do with each other.

Source: Tyler Vigen

As seen in this graph, the number of suicides is rising in tandem with the amount of investment in science, space, and technology in the United States, which appears to be the same trend. Nonetheless, is it true that if we put more money into science, we would see an increase in suicide cases?

Impact of this type of visualization

• The correlation is misleading since it is not connected.

Key takeaway:

• Always look for correlations between various visualizations that are close together.
• The next time you encounter a collection of correlation data, ask yourself if there is a connection.

7. Zooming on favorable data

There is a technique called ‘Cherry Picking.’ It’s a way of selectively displaying data that supports your point of view while ignoring evidence that contradicts it. Only little insight from actual data will be shown in the visualization.

Source: https://flowingdata.com/2017/02/09/how-to-spot-visualization-lies/

Impact of this type of visualization

• This type of visualization hides important data, giving our users just a little insight.

Key takeaway:

• Compare and contrast a zoomed-in visualization with full visuals. (This isn’t always a viable option.)
• Combine the non-zoomed ones into a single group and aggregate the statistics. E.g., by average or sum.

8. Common human visual associations

Humans, like us, perceive things and use our brains to interpret or understand what we see. The principle of Visual Perception and Cognition is also essential in data visualization since it will assist the audience in understanding our message better and faster.

I’ll give you a quick overview of Visual Perception and Cognition below:

Visual Perception

It is the method through which our brain perceives visuals. It’s similar to when we see something and immediately recognize it.

Cognition

It is the next step after visual perception. This is a mental process where we learn about what we see and develop knowledge and comprehension. We use creative colors, icons, fonts, titles, labels, and other elements in data visualization. With these characteristics, we can help our brain interpret the message quickly and differently.

Impact of this type of visualization

• Understanding how our brain interprets information differently can help the audience focus and grasp the message we want to communicate more effectively.

Key takeaway:

• Color is a great way to categorize and support your main points. Color has a vital role in user decisions.
• Data can be ordered, and various chart sections can be sized to make data easier to read. They all have meanings that influence how people perceive them.

9. Improper use of 3D graphics

We can use different charts to show data in the data visualization world. There are also several discussions on chart use guidelines, with 3D being one of the most popular issues.

Most 3D charts are no longer often used to show common data because they have a significant risk of misrepresenting data since our human eye has difficulty interpreting 3D visuals.

This pie chart, for example, makes the rear half appear smaller than the front half, which has a smaller portion. Another type of 3D chart has several issues with accurately showing data values.

Impact of this type of visualization

• 3D charts can distort the reality of data

Key takeaway:

• Use a 2D chart instead if possible.
• If you need to represent a value that spans three axes, a bubble plot/scatter plot with a color gradient can be a good option.

10. Not every insight needs to be represented in data visualization

Your data may, at times, be able to speak for itself. Some values can indicate important information, and it may not be necessary to display that information in a data visualization.

Source: Spiegel Research Center

Impact of this type of visualization

• Sometimes showing data in a chart or graph might not be necessary.

Key takeaway:

• Data visualization is a means of conveying information. There are occasions when it’s okay to use it and others when another tool is more appropriate.

Conclusion

Data visualization’s goal is to make data digestible. Hence they must constantly convey a story to your audience (never misleading). Selecting your core message and the visualization that best represents it is crucial. Another critical consideration is to be aware of and understand what your audience will get as soon as they view your visualization.