Accessible data visualization
Data Science courses involve lots of charts and graphs. It's critical to make these visualizations accessible.
- Always include an alt text or caption for graphs that states the key takeaway: Bar chart showing that Category A outperforms Category B by 20%.
- For detailed analysis, provide the underlying data table or at least summarize important data points in text: Category A: 50 units, Category B: 40 units, Category C: 30 units, etc.
- Long descriptions can be used if the graph is complex, for instance, describing the shape of a distribution in a histogram: Bell-shaped curve, centered around 0, with a tail extending to the right indicating positive skew.
- Remember that tools like Ally might auto-generate a rudimentary HTML table from a chart image; however, it’s best if you manually ensure the data is available in a clean format.
Top
Color and contrast in charts
Use color-blind-friendly palettes for any charts. Many statistical tools have such presets or you can choose patterns/hatching in bar graphs.
- Never rely on color alone to differentiate data series; use direct labeling on the chart or distinct markers. For example, in a line chart, use different line styles like solid or dashed, and include a legend that a screen reader can parse. Legends in software like Excel should be properly tagged in the exported PDF.
- Keep contrast high – e.g., if you have text or grid lines on a chart, ensure they meet contrast guidelines against the background.
Top
Tool accessibility (Excel, Tableau, etc.)
Many Data Science tasks use tools like Excel, R, Python notebooks, or Tableau. As a quick win, if you demonstrate data analysis through code, provide commentary in the code or outside it explaining each step’s result, so even if a plot isn’t seen, the student knows what it demonstrated.
- Excel has an accessibility checker and allows you to add alt text to visuals – when sharing Excel files, run the checker and fix issues such as missing alt text on charts or images.
- Jupyter notebooks or other coding notebooks, advise students to format outputs accessibly, for example, if they include an image plot in a notebook, they can add a markdown cell with a description.
- If using Tableau or Power BI for interactive dashboards, be aware that their accessibility is improving but still limited – you may need to provide alternate descriptions of any interactive dashboard elements, or export static accessible versions for screen reader users.
Top
Math and formulas in statistics
Data Science often involves statistical formulas. Treat these like math content – use proper notation and explain in text.
- If you show the formula for standard deviation, also write in words what each symbol means. For example: sigma equals the square root of the variance, where variance is the average of squared deviations from the mean.
- Ensure any statistical notation, like Greek letters α or β, are accessible – use Unicode characters or equation editor so screen readers can identify them. Some screen readers will say alpha when they encounter the α symbol if properly coded.
Top
Interactive data tools
If students are asked to use interactive tools like an online data visualization tool or a polling tool to explore data trends, ensure that those are accessible or have a backup method.
- If an interactive chart is not screen reader-friendly, provide the data and ask the student to conclude by analyzing the numbers instead.
- If using something like Google Colab notebooks, which are mostly accessible, make sure students know how to turn on any accessibility settings. Colab has a high contrast mode and some screen reader optimizations.
- In teaching, consider demonstrating a screen reader or keyboard navigation on a data table, it raises awareness and ensures your content is navigable in multiple ways.
Top
Guides and links
NOTE: This content was adapted from the UMBC Office of Accessibility and Disability Services.
Top