Which term refers to procedures that remain reliable even when normal assumptions are violated?

• A. Residual sum of squares
• B. Residual
• C. Reliability
• D. Robust test

Answer: (D)

When statistics stay trustworthy even if assumptions aren’t perfectly met, that quality is described as robustness. A robust test is built to perform well despite violations like non-normal data, unequal variances, or the presence of outliers, so its conclusions remain reasonable rather than being distorted.

That’s why this term fits best: it explicitly signals that the testing procedure is resistant to common problems that can undermine standard methods, maintaining validity and power under a variety of imperfect conditions.

The other terms don’t capture this idea. A residual sum of squares measures how far observed values are from the fitted model but doesn’t speak to how results hold up under assumption violations. A residual is simply the difference between observed and predicted values, not about the method’s robustness. Reliability refers to consistency or repeatability of a measurement, not the sensitivity of a test to assumption breaches.