Course contentsShow
← Data Science
Lesson 420 of 2,14511. Hypothesis Tests for Proportions and CategoriesFree lesson

Interpreting Chi-Squared Test Results

Using the test statistic and p-value to decide whether observed data fits the hypothesized distribution.

Interpreting Chi-Squared Test Results

What You'll Learn

You'll learn how to use the chi-squared test statistic and p-value to determine whether your observed data matches the distribution you hypothesized.

The Decision Point

After calculating your chi-squared test statistic and degrees of freedom (from earlier lessons), you're ready to make a decision. Just like other hypothesis tests, you compare your p-value against your chosen significance level (α, typically 0.05).

The interpretation follows this logic:

  • Low p-value (< α): Your observed frequencies differ significantly from expected frequencies. Reject the null hypothesis—your data doesn't fit the hypothesized distribution.
  • High p-value (≥ α): The observed frequencies are reasonably close to expected frequencies. Fail to reject the null hypothesis—your data is consistent with the hypothesized distribution.

What the Test Statistic Tells You

The chi-squared statistic itself measures total "distance" between observed and expected frequencies. A larger χ² means bigger discrepancies. Think of it like a thermometer for mismatch: the hotter it gets, the less likely your data fits your hypothesis.

However, don't interpret χ² in isolation—always use the p-value, which accounts for your degrees of freedom and tells you if that level of mismatch could reasonably happen by chance.

Putting It Together

If you're testing whether a die is fair and get χ² = 12.5 with p = 0.028:

  • Since 0.028 < 0.05, you reject the null hypothesis
  • Conclusion: The die doesn't appear to follow a uniform distribution; it's likely biased

If instead p = 0.15:

  • Since 0.15 ≥ 0.05, you fail to reject the null hypothesis
  • Conclusion: The observed rolls are consistent with a fair die

Key Takeaway: Interpret chi-squared tests by comparing the p-value to your significance level α—reject the null hypothesis if p < α, indicating your observed data doesn't fit the hypothesized distribution well enough to be explained by chance alone.