🎯 Learning Goals
- Understand how to structure an effective Tableau dashboard
- Learn best practices for incorporating text and media into a dashboard
- Identify the most effective visualization types for different types of data
- Explore the impact of interactivity in dashboards and how to implement it
- Gain inspiration from public Tableau dashboards
Why Storytelling with Data Matters
Great data visualizations do more than just present numbers, they tell a story. A well-structured dashboard guides the audience through key insights and makes data-driven conclusions clear and compelling. As data scientists, we are ethically and morally obligated to not misrepresent data. Misleading visualizations can distort reality, spread misinformation, and lead to poor decision-making.
Identifying Misleading Dashboards
Before we dive into some awesome dashboards, we’re going to look into some..well..less than good dashboards.
Try-It | Misleading/Ineffective Data Visualizations
For each of the data visualizations, think about:
- What is misleading/ineffective about this dashboard?
- How could it be fixed to present accurate insights?
- What are some real-world consequences that could’ve come out of this?
Misleading/Ineffective Data Visualization #1
Misleading/Ineffective Data Visualization #2
Misleading/Ineffective Data Visualization #3
Source: Spotify
Misleading/Ineffective Data Visualization #4
Misleading/Ineffective Data Visualization #5
Common Data Misrepresentation Tactics to Watch Out For
Truncated Y-Axis: when the vertical axis (y-axis) of a chart does not start at zero, it can exaggerate differences between data points and make small changes look much larger than they really are
On the left side is the truncated y-axis chart and on the left is the chart starting at 0 on the y-axis. The both show the same difference but the visualized bar on the left look much more different than the one on the right
Cherry-Picked Data: when only certain data points are selected to support a particular narrative and ignoring other relevant data that may provide a fuller picture, it can create a false impression of trends, effectiveness, or causality by excluding contradictory or neutral data
In the example shown above, a user shares COVID case curves for the US for the periods of early July through mid August for the years 2020 (before the vaccination campaign) and 2021 (during the vaccination campaign). Because the number of cases in August 2021 is higher than in August 2020, the user suggests that the vaccination campaign failed. Besides the fact that the implication is based on a single county, this example also carefully selects the time frame that most effectively supports the argument, omitting a large drop in cases in Spring 2021.
Misleading Color Scales: colors can be chosen in a way that exaggerates, hides, or distorts differences in data
Color Palettes that can be used in data visualization
Source: Of Colors and Color Legends
In these two data visualizations of temperature anomalies, the use of the rainbow/jet color palette is not as effective as using the vik color palette. In the bottom data visualization, the anomalies are equally represented and it is color-vision deficiency friendly
Inconsistent Intervals: if time intervals on an axis are not equally spaced, it can distort trends and mislead viewers
The years 1999 and 2002 are omitted from the graph to create a data visualization that shows a specific trend. This however is misleading because you need to represent a consistent interval to accurately capture trends in your data.
Dashboards on Tableau Public! Data Visualization of the Day!
Sometimes, the best way to improve in an industry or skill is by seeing how the best do it. Professional athletes watch other players compete, chefs dine at other restaurants for inspiration, and artists visit galleries to see what’s possible. Similarly, we're going to check out some of the top data visualizations on Tableau Public to see what’s possible in this space!
As you know, Tableau Public is a free online visual data analytics platform that allows people to learn and practice data skills! Most weekdays, the Tableau Public team selects one of the millions of data visualizations published on Tableau Public to feature as the “Viz of the Day”
These visualizations are created by individuals, companies, and data enthusiasts and represent some of the most compelling and innovative uses of Tableau.
Viz of the Day visualizations are designed to:
- Spark meaningful data conversations by showcasing innovative and insightful visualizations.
- Demonstrate what’s possible in Tableau through creative storytelling and advanced techniques.
- Elevate inspirational and informational data visualizations from authors all over the world.
They also give data enthusiasts the opportunity to:
- Grow. Explore new chart types, design styles, or visualization methods. Many Tableau Public workbooks can even be downloaded to “peek under the hood” and see how they were built.
- Connect. Find and connect with rising talent and established data rockstars in the Tableau Community.
- Be inspired. Thousands of people around the globe see Viz of the Day and gain new ideas for their own work.
If your dashboards don’t look like these just yet, that’s completely okay! These dashboards are created by experienced professionals and dedicated data enthusiasts who have spent years honing their craft. The goal here isn’t to compare but to learn, get inspired, and see what’s possible!
Try-It | Explore Data Visualizations of the Day
Spend 10 minutes exploring the Tableau Public Visualization of the Day.
1. Pick your favorite visualization.
2. Be ready to share why you chose it. What made it stand out?
3. What storytelling techniques or design choices were particularly effective?
Spend 10 minutes exploring the Tableau Public Visualization of the Day.
1. Pick your favorite visualization.
2. Be ready to share why you chose it. What made it stand out?
3. What storytelling techniques or design choices were particularly effective?
📊 Structuring an Effective Dashboard for Storytelling
What Makes a Dashboard Work?
A great dashboard is one that effectively communicates insights and does not mislead the audience. It should prioritize:
- Clarity: Focus on essential insights and avoid unnecessary complexity.
- Logical Structure: Arrange elements in a way that tells a coherent story. Include titles and axis labels.
- Honest Representation: Ensure data is not distorted or manipulated through misleading visuals, improper scaling, or missing context.
- Accessibility: Use readable text, appropriate colors, and clear labels to make information easy to interpret.
Throughout this lesson, we've explored common pitfalls in data visualization, such as truncated Y-axes, cherry-picked data, and misleading color scales. Keeping these principles in mind will help you build dashboards that are not only visually appealing but also responsible and trustworthy.
💭 Incorporating Text and Media in Dashboards
1. Using Text for Context
- Titles: Clearly describe what’s being shown.
- Annotations: Use tooltips and labels to highlight key insights.
- Summary Sections: A short paragraph can help users interpret the data.
2. Enhancing Visuals with Media
- Logos and Icons: Improve branding and readability.
- Background Images: Add depth but avoid distractions.
- Custom Shapes and Buttons: Help guide users through interactive elements.
🎨 Choosing the Right Visualization for Your Data
The same dataset can tell very different stories depending on how it is presented. A well-chosen visualization clarifies trends, comparisons, and insights!
Type of Data | Best Visualization Type | Why? |
Comparisons | Bar Chart, Bullet Graph | Easy to compare multiple categories |
Trends Over Time | Line Chart, Area Chart | Shows patterns across dates |
Parts of a Whole | Pie Chart, Treemap | Illustrates proportions |
Relationships | Scatter Plot, Bubble Chart | Shows correlations between two variables |
Distributions | Histogram, Box Plot | Highlights frequency and variability |
Practice | Fixing the Misleading Data Visualizations
Now that you’ve seen great data visualizations and learned a bit about what makes a great data visualizations, go back to one of the misleading/ineffective dashboards from earlier (or find one yourself) and redesign it! It does not have to be a formal Tableau workbook and you can use any platform (Canva, Google Slides, etc.) you’d like.
- Apply best practices in structure, layout, and storytelling.
- Incorporate interactive elements like filters or parameters.
- Improve color choices, branding, and design for clarity.
- Extra credit: is there anything you would add to it to go above and beyond?
- Be ready to explain your changes and why they improve the visualization.
For a summary of this lesson, check out the 12. Storytelling With Data Visualizations in Tableau One-Pager!
Made with Bullet