One and Two Sample Proportion

Many real‑world processes produce categorical outcomes rather than continuous measurements. Defects are counted as yes/no. Transactions pass or fail. Customers respond or do not respond. In these situations, proportion tests provide the statistical backbone for evaluating differences in performance. 

A 1‑sample proportion test evaluates whether the proportion of “successes” (or failures) in a sample differs from a known or target proportion. For example, if the target defect rate is 2%, the test can determine whether the current process is statistically higher or lower than that benchmark. 

A 2‑sample proportion test compares the proportions of two independent groups. This is useful when comparing defect rates across suppliers, pass/fail rates across shifts, or conversion rates across customer segments. The test evaluates whether the observed difference in proportions is large enough to conclude that the underlying populations differ. 

Both tests rely on the binomial distribution and use normal approximations when sample sizes are sufficiently large. When sample sizes are small, exact methods such as Fisher’s exact test may be more appropriate. 

Interpreting proportion tests requires attention to both statistical and practical significance. A small difference in defect rates may be statistically significant with a large sample but operationally irrelevant. Conversely, a practically meaningful difference may fail to reach statistical significance if the sample size is too small or the process is highly variable. 

Proportion tests are essential in the Analyze phase because they address some of the most common questions in quality improvement: Are we producing fewer defects? Is one supplier performing better than another? Did the improvement reduce the failure rate? When used thoughtfully, these tests provide clear, actionable insight into categorical performance.