Overview
This article reviews how graphs were misused to mislead the public during the Covid-19 pandemic, highlighting specific examples and best practices for accurate data visualization.
The Power and Risk of Graphs
- Graphs are effective tools for conveying data but can be manipulated to misinform.
- Common misleading tactics include manipulating axes, using deceptive scales, and omitting important information.
Example 1: Omission and Logarithmic Scales
- A Fox News Covid-19 graph used an uneven, quasi-logarithmic y-axis to flatten the curve visually.
- The y-axis started at 30 instead of zero, minimizing the appearance of case growth.
- Logarithmic scales should only be used when comparing data with large order-of-magnitude differences and must be labeled clearly.
Example 2: Misleading Data Aggregation
- A Florida Covid-19 graph appeared to show cases declining, but underlying data was flawed.
- Negative test results were counted multiple times if one person tested negative repeatedly, inflating negative percentages.
- Positive results were only counted once, further distorting the true positive rate.
- Both official and whistleblower data sources presented conflicting interpretations, complicating public understanding.
Example 3: Manipulating Time in Graphs
- A Georgia Covid-19 graph rearranged x-axis dates out of chronological order, creating a false trend.
- The reordered dates and inconsistent color coding misrepresented case progression as improvement.
- Correct ordering of data removed the misleading “staircase” effect.
Best Practices for Accurate Graphs
- Use appropriate vertical scales and start axes at zero when possible.
- Do not skip numbers or rearrange axis values.
- Include clear, informative labels and use all available relevant data.
- Verify the underlying data source before drawing conclusions from a graph.
Key Terms & Definitions
- Logarithmic Scale — a nonlinear scale used for data with wide-ranging values, where each unit increase represents a multiplication.
- Omission — leaving out key data or labels, resulting in misinterpretation.
- Data Aggregation — combining data points, which can mislead if not done consistently.
- Chronological Order — arranging data points by actual sequence in time to accurately reflect trends.
Action Items / Next Steps
- Review examples of misleading graphs and practice identifying deceptive techniques.
- Double-check data sources and axis labels when interpreting or creating graphs.
- Apply best practices when presenting or analyzing data visualizations.