Which condition is the main reason for choosing a non-parametric test instead of a parametric test?
Free Applied Statistics in Healthcare practice question on Non-parametric tests and chi-square for BSN and Post-RN nursing students. Choose your answer, then open the answer and the explanation.
AThe outcome variable is measured on a continuous scale with normal distribution
BThe outcome variable is measured on an ordinal scale with a small sample
CThe outcome variable is measured on a ratio scale with equal variances
DThe outcome variable is measured on an interval scale with a large sample
Show answer and explanation
✓ Correct answerB. The outcome variable is measured on an ordinal scale with a small sample
Explanation
Parametric tests such as the t-test and ANOVA assume that the outcome is measured on an interval or ratio scale and that it is roughly normally distributed in the population. Non-parametric tests make far fewer assumptions: they work on ranks, so they suit ordinal data, very small samples, and distributions that are clearly skewed or non-normal. Choosing a test therefore depends mainly on the measurement scale and the shape of the distribution, not on how large the effect is.
Why each option is right or wrong
✗A. The outcome variable is measured on a continuous scale with normal distribution — A continuous, normally distributed outcome is exactly the situation in which parametric tests are preferred.
✓B. The outcome variable is measured on an ordinal scale with a small sample — Ordinal measurement and a small sample break the assumptions of parametric tests, so a rank-based test is the appropriate choice.
✗C. The outcome variable is measured on a ratio scale with equal variances — Equal variances on a ratio scale satisfy the assumptions of parametric tests, so there is no need to abandon them.
✗D. The outcome variable is measured on an interval scale with a large sample — A large sample with interval-scale data meets the assumptions of parametric tests, so a rank-based test would waste information.
Feature
Parametric test
Non-parametric test
Measurement scale
Interval or ratio
Ordinal or nominal (ranks)
Distribution assumption
Approximately normal
No normality needed
Typical sample size
Moderate to large
Small samples acceptable
Example
Independent t-test, ANOVA
Mann-Whitney U, Wilcoxon, Kruskal-Wallis
EXAM HINT Ranks for ranks: ordinal or odd-shaped data go to rank-based tests.
اردو وضاحت
Non-parametric test کیا ہوتا ہے؟
Non-parametric test ایک ایسا statistical test ہے جو data کی normal distribution کی شرط نہیں رکھتا۔ یہ numbers کی بجائے ranks پر کام کرتا ہے، اس لیے ordinal data یا بہت چھوٹے sample میں بھی چل جاتا ہے۔
Non-parametric test کب چننا چاہیے؟
جب outcome variable ordinal scale پر ہو یا sample بہت چھوٹا ہو اور data normal نہ لگے، تو parametric test کے assumptions پورے نہیں ہوتے، اس لیے rank-based test زیادہ مناسب ہے۔ اگر data interval یا ratio scale پر ہو اور normal distribution ہو تو t-test یا ANOVA بہتر ہے کیونکہ وہ data کی پوری معلومات استعمال کرتے ہیں۔
Key terms
ordinal scale — A measurement scale where items are put in order (mild, moderate, severe) but the gaps between steps are not equal. (ایسا scale جس میں درجہ بندی تو ہوتی ہے مگر درجوں کے درمیان فاصلہ برابر نہیں ہوتا، جیسے (mild, moderate, severe))