Course contentsShow
← Data Science
Lesson 1946 of 2,14545. Communicating InsightsFree lesson

Supporting Your Claims with Evidence

Organize data, visualizations, and statistical tests as supporting evidence for each key point.

Supporting Your Claims with Evidence

You'll learn how to systematically organize data, visualizations, and statistical tests to back up every key point in your data story.

The Courtroom Analogy

Think of your data presentation like a legal case. You wouldn't just tell a jury "trust me, the defendant is guilty"—you'd present evidence: witness testimonies, documents, forensic data. Similarly, every claim you make about your data needs its own supporting evidence trail.

What Counts as Evidence?

When you make a statement like "Customer churn increased 15% in Q3," you need three types of evidence stacked together:

  1. The raw data: Show the actual numbers or aggregated metrics
  2. Visual proof: A chart that makes the trend immediately visible
  3. Statistical validation: A test confirming the change isn't random noise

Each piece reinforces the others. The data provides precision, the visualization provides intuition, and the statistical test provides confidence.

Organizing Evidence Around Claims

Structure your argument claim-by-claim:

  • Claim: "Mobile users convert at lower rates than desktop users"
  • Evidence package:
    • Table showing conversion rates by platform (data)
    • Bar chart comparing the two (visualization)
    • Chi-square test showing p < 0.001 (statistical test)

Don't dump all your evidence upfront. Match each piece to the specific point it supports, appearing right when your audience needs it to evaluate your reasoning.

The One-to-One Rule

Every major claim needs at least one piece of hard evidence. If you can't find supporting data for a statement, either rephrase it as a hypothesis rather than a conclusion, or remove it entirely.

Key Takeaway: Treat each claim in your data story as a mini-thesis requiring its own evidence package of data, visualization, and statistical validation—organized to appear exactly when your audience needs to verify your reasoning.