Spreadsheet-based hypothesis testing

Learning objectives

• Run t and chi-square tests in a free spreadsheet.
• Compute a manual two-proportion z-statistic.
• State conclusions in both statistical and practical language.

Core idea

Hypothesis testing measures how compatible data are with H0 while controlling a prespecified Type I error rate alpha.

Spreadsheet lab

• Two independent means: =T.TEST(A2:A31, B2:B31, 2, 3) (2-tailed, Welch).
• Paired means: =T.TEST(A2:A21, B2:B21, 2, 1).
• Association: =CHISQ.TEST(observed_range, expected_range).
• Two proportions manually: z = (p1−p2)/SQRT(p̄(1−p̄)(1/n1+1/n2)); p = 2*(1-NORM.S.DIST(ABS(z),TRUE)).

Garment-factory example

Two sewing lines had first-pass yields 82% (n=200) and 88% (n=200). p̄ = 0.85, z ≈ −1.78, two-tailed p ≈ 0.075 → not conclusive; capture more data or examine practical relevance.

Method

1. State parameter, H0/H1, alpha.
2. Plot data before running the test.
3. Report effect, CI, p and practical meaning.

Common mistakes

• Reading p as 'probability H0 is true'.
• Inferring causation from an uncontrolled comparison.