Test whether a population’s variability matches a target, for example whether a filling machine’s standard deviation exceeds its 2 g specification.
χ² = (n − 1)s² ÷ σ₀², df = n − 1
Example
A filling machine is specified to have an SD of at most 2 g. A sample of 25 bottles has s = 2.8 g. Then χ² = 24 × 2.8² ÷ 2² = 47.04 with 24 df. The right-tailed p = 0.003, so the machine is more variable than specified.
Assumptions
The test is valid only for normally distributed data. Unlike tests of means, it does not become robust with large samples, so check normality first with the Shapiro–Wilk test. For a range of plausible values instead of a yes/no answer, use the variance confidence interval.