STP Data Analysis 4 — Questions and Answers
Question 1: The Westgard 1:3s rule in quality control means a run is rejected when:
- One control value falls outside ±2 SD
- One control value falls outside ±3 SD (Correct answer)
- Two consecutive controls fall outside ±2 SD
- The range between controls exceeds 4 SD
Correct answer: One control value falls outside ±3 SD
The 1:3s rule triggers rejection when a single QC result exceeds 3 standard deviations from the mean, indicating a probable error.
Question 2: In a 2×2 contingency table, which cell value is used to calculate sensitivity?
- True positives only
- True positives divided by (true positives + false negatives) (Correct answer)
- True negatives divided by (true negatives + false positives)
- True positives divided by (true positives + false positives)
Correct answer: True positives divided by (true positives + false negatives)
Sensitivity = TP/(TP+FN), representing the proportion of true disease cases correctly identified by the test.
Question 3: Log transformation of laboratory data is commonly applied to:
- Normally distributed data to improve symmetry
- Positively skewed data to approximate normality (Correct answer)
- Categorical data to enable parametric testing
- Data with no variability to increase spread
Correct answer: Positively skewed data to approximate normality
Log transformation compresses the right tail of positively skewed distributions, making them more approximately normal.
Question 4: Which term describes the smallest concentration of analyte that can be reliably distinguished from zero (blank)?
- Limit of quantitation (LOQ)
- Analytical range
- Limit of detection (LOD) (Correct answer)
- Reference interval
Correct answer: Limit of detection (LOD)
The limit of detection (LOD) is the lowest analyte concentration statistically distinguishable from a blank sample with defined confidence.
Question 5: In a Passing-Bablok regression, the 95% CI for the intercept includes 0 and the slope CI includes 1. This indicates:
- Significant systematic and proportional bias
- No significant systematic or proportional bias (Correct answer)
- Significant systematic bias only
- Significant proportional bias only
Correct answer: No significant systematic or proportional bias
When the CI for intercept includes 0 and the CI for slope includes 1, neither systematic nor proportional bias is statistically significant.
Question 6: Which measure is most resistant to the effect of outliers in a dataset?
- Mean
- Variance
- Standard deviation
- Median (Correct answer)
Correct answer: Median
The median, as the middle value of ranked data, is unaffected by extreme outliers unlike the mean or variance.
Question 7: In a laboratory context, 'imprecision' is best quantified by:
- Bias from the reference method
- Within-run and between-run coefficients of variation (Correct answer)
- The analytical measurement range
- Positive predictive value of the test
Correct answer: Within-run and between-run coefficients of variation
Imprecision is measured using coefficients of variation from within-run (repeatability) and between-run (reproducibility) studies.
STP Data Analysis 4
The Westgard 1:3s rule in quality control means a run is rejected when: