What is bad about non-parametric statistics tests?

What is bad about non-parametric statistics tests?

• A. They are always more powerful than parametric tests.
• B. Not as powerful as parametric tests.
• C. They require large sample sizes.
• D. They are only for interval data.

Answer: (B)

The main idea is about power and distribution assumptions. Non-parametric tests don’t assume a specific distribution and are robust to outliers or non-normal data, which makes them useful when those assumptions are violated or when data are ordinal. The trade-off is that they typically have lower power than parametric tests when the parametric assumptions (like normality and equal variances) hold. That means they may need larger samples to detect the same effect, or miss smaller effects that a parametric test would pick up. They’re not inherently better or worse in all cases; they’re just less powerful under ideal parametric conditions, while avoiding assumption-related problems when those conditions aren’t met. They aren’t reserved for interval data, and many non-parametric tests work with ordinal or nominal data, so the idea that they require large samples or are only for interval data isn’t correct.

The main idea is about power and distribution assumptions. Non-parametric tests don’t assume a specific distribution and are robust to outliers or non-normal data, which makes them useful when those assumptions are violated or when data are ordinal. The trade-off is that they typically have lower power than parametric tests when the parametric assumptions (like normality and equal variances) hold. That means they may need larger samples to detect the same effect, or miss smaller effects that a parametric test would pick up. They’re not inherently better or worse in all cases; they’re just less powerful under ideal parametric conditions, while avoiding assumption-related problems when those conditions aren’t met. They aren’t reserved for interval data, and many non-parametric tests work with ordinal or nominal data, so the idea that they require large samples or are only for interval data isn’t correct.