Mann–Whitney U test is a nonparametric statistical method widely used in B.Pharm research to compare two independent groups when data are ordinal or not normally distributed. This rank-sum test (also called Wilcoxon rank-sum) assesses differences in central tendency or distribution without assuming normality, making it useful for pharmacology, clinical trials, dissolution studies, bioequivalence, and adverse event comparisons. Key concepts include U statistic, ranks, ties, exact versus asymptotic p-values, continuity correction, and effect size (r). Understanding assumptions, calculation steps, interpretation, and software implementation (R, SPSS, Python) helps B.Pharm students analyze real-world drug data robustly. Now let’s test your knowledge with 30 MCQs on this topic.
Q1. What is the primary purpose of the Mann–Whitney U test?
- Compare two independent samples when assumptions of t-test are not met
- Compare more than two groups
- Test for correlation between two variables
- Analyze paired data
Correct Answer: Compare two independent samples when assumptions of t-test are not met
Q2. Which alternative name is commonly used for the Mann–Whitney U test?
- Kruskal–Wallis test
- Wilcoxon signed-rank test
- Wilcoxon rank-sum test
- Kolmogorov–Smirnov test
Correct Answer: Wilcoxon rank-sum test
Q3. What types of data are appropriate for the Mann–Whitney U test?
- Nominal data only
- Ordinal or continuous non-normal data
- Only normally distributed continuous data
- Paired repeated measures
Correct Answer: Ordinal or continuous non-normal data
Q4. Which of the following is an important assumption of the Mann–Whitney U test?
- Normal distribution of raw data in each group
- Equal variances in raw data are required
- Observations are independent and distributions have similar shape
- Samples must be paired
Correct Answer: Observations are independent and distributions have similar shape
Q5. When should a B.Pharm student prefer Mann–Whitney U over independent t-test?
- When sample sizes are extremely large
- When data are binary
- When data are non-normal or ordinal
- When groups are paired
Correct Answer: When data are non-normal or ordinal
Q6. What is the null hypothesis in a two-sample Mann–Whitney U test?
- Both groups have equal sample sizes
- The two groups have identical distributions
- The two groups have equal variances
- The mean of group 1 is greater than mean of group 2
Correct Answer: The two groups have identical distributions
Q7. The U statistic can take values in which range?
- 0 to n1 + n2
- 0 to n1 * n2
- -∞ to +∞
- 1 to n1
Correct Answer: 0 to n1 * n2
Q8. How are tied values handled when ranking data for the Mann–Whitney U test?
- Discard all tied observations
- Assign the highest rank to all ties
- Assign average ranks to tied values
- Randomly assign distinct ranks
Correct Answer: Assign average ranks to tied values
Q9. What does a two-sided Mann–Whitney U test evaluate?
- Whether group 1 is strictly greater than group 2
- Whether there is any difference in distributions without direction
- Only differences in variances
- Only differences in medians ignoring distribution shape
Correct Answer: Whether there is any difference in distributions without direction
Q10. Which effect size measure is commonly used with Mann–Whitney U results?
- Cohen’s d
- Pearson’s r from raw scores
- Rank-biserial or r = Z / √N
- Odds ratio
Correct Answer: Rank-biserial or r = Z / √N
Q11. For large sample sizes, Mann–Whitney U is commonly approximated by which distribution?
- Chi-square distribution
- Student’s t-distribution
- Normal distribution (Z) with possible continuity correction
- F-distribution
Correct Answer: Normal distribution (Z) with possible continuity correction
Q12. If two groups have identical medians but different distribution shapes, Mann–Whitney may:
- Always fail to detect any difference
- Detect a significant difference because it compares distributions
- Only detect differences in variances
- Be invalid and return no result
Correct Answer: Detect a significant difference because it compares distributions
Q13. Which software tools commonly implement the Mann–Whitney U test?
- Only Excel
- R, SPSS, and Python libraries
- Only proprietary clinical software
- None; it must be computed by hand
Correct Answer: R, SPSS, and Python libraries
Q14. Which test should be used for paired nonparametric two-sample comparisons?
- Mann–Whitney U test
- Wilcoxon signed-rank test
- Kruskal–Wallis test
- Chi-square test
Correct Answer: Wilcoxon signed-rank test
Q15. For very small sample sizes in Mann–Whitney U, which p-value approach is preferred?
- Asymptotic p-value only
- Exact p-value calculation
- P-value is not needed for small samples
- Use bootstrap exclusively
Correct Answer: Exact p-value calculation
Q16. What is an effect of many tied observations on the Mann–Whitney U test?
- Increases test power substantially
- Has no effect on ranks or p-values
- Reduces power and requires tie-adjusted variance
- Converts test into a t-test automatically
Correct Answer: Reduces power and requires tie-adjusted variance
Q17. Which is an appropriate B.Pharm application of Mann–Whitney U?
- Comparing dissolution times between two formulations when data are skewed
- Comparing placebo and drug in a crossover paired design without accounting for pairing
- Estimating hazard ratios for time-to-event data with censoring
- Testing equality of more than two batch means simultaneously
Correct Answer: Comparing dissolution times between two formulations when data are skewed
Q18. How are ranks assigned when computing Mann–Whitney U?
- Rank each group separately
- Combine both groups and rank all observations together
- Use original data order as ranks
- Rank only the larger group
Correct Answer: Combine both groups and rank all observations together
Q19. When reporting Mann–Whitney U results in a paper, which elements are essential?
- Only p-value is necessary
- U value, sample sizes, p-value, and effect size
- Only group means and standard deviations
- Only test statistic without sample sizes
Correct Answer: U value, sample sizes, p-value, and effect size
Q20. If the Mann–Whitney U test yields p < 0.05, the proper interpretation is:
- There is strong evidence that distributions differ between groups
- The means of the groups are equal
- There is no evidence of any difference
- The result proves clinical significance
Correct Answer: There is strong evidence that distributions differ between groups
Q21. Which of the following is NOT required by the Mann–Whitney U test?
- Independence of observations
- Ordinal or continuous measurement
- Normality of the raw data
- Two independent groups
Correct Answer: Normality of the raw data
Q22. Under what condition does Mann–Whitney U test compare medians specifically?
- Always, regardless of distribution shape
- Only when distributions have similar shape and spread
- Only when sample sizes are equal
- Never; it only compares means
Correct Answer: Only when distributions have similar shape and spread
Q23. Which formula is a correct expression for U based on ranks R1 for group 1?
- U = R1 – n1(n1+1)/2
- U = n1*n2 + n1(n1+1)/2 – R1
- U = R1 + R2
- U = (R1/n1) – (R2/n2)
Correct Answer: U = n1*n2 + n1(n1+1)/2 – R1
Q24. What is the expected mean of the U distribution under the null hypothesis?
- n1 + n2
- n1*n2/2
- (n1 – n2)/2
- Zero
Correct Answer: n1*n2/2
Q25. The variance of U used in normal approximation is given by which base expression (before tie correction)?
- n1*n2*(n1+n2+1)/12
- n1*n2/(n1+n2)
- (n1+n2)^2/12
- n1*n2*(n1-n2)/12
Correct Answer: n1*n2*(n1+n2+1)/12
Q26. When should a continuity correction be applied?
- When using the exact distribution for U
- When approximating U by the normal distribution for discrete data
- Only for paired tests
- Continuity correction is never needed
Correct Answer: When approximating U by the normal distribution for discrete data
Q27. Is the Mann–Whitney U test valid with unequal sample sizes?
- No, sample sizes must be equal
- Yes, it remains valid with unequal sizes
- Only if the larger sample is more than twice the smaller
- Only when variances are equal
Correct Answer: Yes, it remains valid with unequal sizes
Q28. Can Mann–Whitney U test handle censored survival data?
- Yes, it is designed for censored data
- No, use survival analysis methods like Kaplan–Meier or log-rank test
- Yes, but only if censoring is random
- No, use paired t-test instead
Correct Answer: No, use survival analysis methods like Kaplan–Meier or log-rank test
Q29. When is a one-sided Mann–Whitney U test appropriate?
- When the data are normally distributed
- When a prior directional hypothesis exists
- Only when sample sizes are equal
- One-sided tests are never appropriate
Correct Answer: When a prior directional hypothesis exists
Q30. Which alternative test compares entire distributions between two samples (not just central tendency)?
- Kruskal–Wallis test
- Chi-square goodness-of-fit
- Kolmogorov–Smirnov test
- ANOVA
Correct Answer: Kolmogorov–Smirnov test
