Supporting Claims with Evidence
You'll learn how to anchor every conclusion in your analysis with concrete references to tables, figures, and statistical tests.
Why Evidence Matters
Imagine a lawyer presenting a closing argument without citing any witnesses or physical evidence—the jury would be skeptical. The same applies to data science communication. Every claim you make needs a visible, verifiable anchor point that your audience can examine.
The Three Pillars of Evidence
When you state a finding, immediately connect it to one of these:
1. Tables: Reference specific numbers or comparisons
"Customer retention increased 23% year-over-year (Table 3)."
2. Figures: Point to visual patterns
"The seasonal spike in Q4 is visible in Figure 2, where holiday purchases dominate."
3. Statistical Tests: Cite significance or uncertainty
"This difference is statistically significant (p < 0.01, t-test), ruling out random chance."
Integration Techniques
Don't just say results—show where they live. Use phrases like:
- "As shown in Figure 4..."
- "Table 2 reveals that..."
- "A chi-square test (χ² = 15.3, p = 0.002) confirms..."
Number your tables and figures sequentially and reference them by name. This creates a traceable chain from claim → evidence → source, building credibility and allowing skeptical readers to verify your reasoning independently.
Avoiding Naked Claims
Never write: "The model performs well."
Instead: "The model achieves 94% accuracy on the test set (Table 5), significantly outperforming the baseline (Figure 6)."
This gives your audience two places to look for confirmation and shows you're not cherry-picking—you're systematically documenting your findings.
Key Takeaway: Every analytical claim should be explicitly tied to a numbered table, labeled figure, or cited statistical test—treat evidence references as mandatory footnotes for your conclusions.