FPGEE Biostatistics 4 — Questions and Answers
Question 1: Berkson's bias is a specific form of selection bias most commonly associated with which study design?
- Randomized controlled trials
- Cohort studies
- Hospital-based case-control studies (Correct answer)
- Cross-sectional studies
Correct answer: Hospital-based case-control studies
Berkson's bias occurs when both cases and controls are drawn from hospitalized patients, distorting the true association because hospitalization rates differ by exposure and disease.
Question 2: A study of a new antibiotic uses intention-to-treat (ITT) analysis. Which statement best describes this approach?
- Only patients who completed the full course are analyzed
- All randomized patients are analyzed in their originally assigned groups regardless of adherence (Correct answer)
- Patients are analyzed based on the treatment they actually received
- Dropouts are excluded from the final analysis
Correct answer: All randomized patients are analyzed in their originally assigned groups regardless of adherence
ITT analysis preserves the benefits of randomization and gives a conservative, real-world estimate of treatment effect by including all randomized subjects.
Question 3: Which measure of central tendency is most resistant to outliers in a skewed pharmacokinetic dataset?
- Mean
- Variance
- Median (Correct answer)
- Standard deviation
Correct answer: Median
The median is the middle value and is not influenced by extreme values (outliers), making it robust for skewed distributions.
Question 4: A diagnostic test for a rare disease (prevalence 1%) has sensitivity 99% and specificity 99%. What is the approximate positive predictive value?
- 99%
- 50%
- 1%
- Approximately 50% (Correct answer)
Correct answer: Approximately 50%
PPV ≈ TP/(TP+FP); with 1% prevalence and 99% specificity, false positives (~1% of 9,900 = 99) nearly equal true positives (~99), giving PPV ≈ 50%.
Question 5: In multiple linear regression, what does the R² value represent?
- The correlation between each predictor and the outcome
- The proportion of variance in the outcome explained by all predictors combined (Correct answer)
- The statistical significance of each regression coefficient
- The number of predictors in the model
Correct answer: The proportion of variance in the outcome explained by all predictors combined
R² (coefficient of determination) represents the proportion of total variance in the dependent variable that is explained by the regression model.
Question 6: A pharmacy researcher performs 20 independent hypothesis tests simultaneously without adjustment. Approximately how many false positives are expected at α = 0.05?
- 0
- 1 (Correct answer)
- 5
- 10
Correct answer: 1
With 20 tests at α = 0.05, approximately 20 × 0.05 = 1 false positive is expected by chance alone, illustrating the multiple comparisons problem.
Question 7: Which nonparametric test is the equivalent of a one-way ANOVA when the normality assumption is violated?
- Wilcoxon signed-rank test
- Mann-Whitney U test
- Kruskal-Wallis test (Correct answer)
- Friedman test
Correct answer: Kruskal-Wallis test
The Kruskal-Wallis test compares median ranks across three or more independent groups and is the nonparametric alternative to one-way ANOVA.
FPGEE Biostatistics 4
Berkson's bias is a specific form of selection bias most commonly associated with which study design?