Examkin

Question

Difficulty: HardInferential Statistical Tests

A team of oncology researchers is investigating the efficacy of a novel tyrosine kinase inhibitor in patients with advanced hepatocellular carcinoma. The investigators measure serum alpha-fetoprotein (AFP) levels, a continuous variable, in 22 patients immediately before the initiation of therapy and again at 8 weeks post-treatment. A preliminary analysis of the collected data reveals that the distribution of the AFP levels at both time points is severely right-skewed and does not follow a normal distribution. The researchers want to determine if there is a statistically significant median difference in AFP levels within this patient cohort following the 8-week treatment period. Which of the following is the most appropriate inferential statistical test to analyze the primary outcome data in this study?

  1. Wilcoxon signed-rank testAnswer
  2. B
    Paired Student t-test
  3. C
    Mann-Whitney U test
  4. D
    McNemar test
  5. E
    Kruskal-Wallis test

Answer

Wilcoxon signed-rank test
The correct inferential statistical test is the Wilcoxon signed-rank test. Selecting the appropriate statistical test requires evaluating three main components: the type of dependent variable (continuous, ordinal, categorical), the study design (independent vs. paired/matched groups), and the underlying distribution of the data (normal vs. non-normal). In this study, the dependent variable is serum AFP level, which is a continuous variable. The study design involves measuring the same 22 patients at two different time points (pre-treatment and 8 weeks post-treatment), which constitutes paired or dependent groups. Because the data is severely right-skewed and does not follow a normal distribution, parametric tests (such as the paired Student t-test) cannot be validly used. The Wilcoxon signed-rank test is the correct non-parametric alternative to the paired t-test for evaluating the median difference between paired, non-normally distributed continuous variables.

Step-by-Step Solution

1
Identify the independent variable and study design.
The study compares measurements taken from the same 22 patients at two different time points (before and after treatment), indicating a paired (dependent) study design with two comparison groups.
Determining whether groups are independent or paired is the first critical step in selecting a statistical test.
2
Identify the type of the dependent variable.
The dependent variable is serum alpha-fetoprotein (AFP) level, which is explicitly described as a continuous variable.
Test selection depends heavily on whether the outcome data is continuous, ordinal, or categorical.
3
Evaluate the distribution of the data to verify parametric assumptions.
The data is described as severely right-skewed and not normally distributed, violating the assumptions required for parametric testing.
Parametric tests require normally distributed continuous data; severely non-normal data requires non-parametric alternatives.
4
Select the appropriate test based on the design, variable type, and distribution.
The Wilcoxon signed-rank test is chosen.
It is the appropriate non-parametric equivalent to the paired t-test, designed specifically for comparing two paired groups with continuous, non-normal data.

Key Concept

Selecting the correct statistical test requires identifying the data type (continuous vs. categorical), study design (independent vs. paired groups), and data distribution (normal vs. non-normal). The Wilcoxon signed-rank test is the non-parametric choice for paired, continuous data that violates normality assumptions.
Rate this question

Topics

Inferential Statistical Tests

More questions from this topic