Examkin

Question

Difficulty: HardInferential Statistical Tests

A team of pulmonologists is analyzing data from a prospective cohort study of 350 patients with severe persistent asthma. The study aims to evaluate clinical factors associated with asthma exacerbations and lung function decline over a 12-month follow-up period. The investigators collected the following baseline and outcome variables:

- Decline in forced expiratory volume in 1 second (FEV1), measured as a continuous variable with a normal distribution.
- Total number of asthma exacerbations during the 12-month period, which is heavily right-skewed count data.
- Smoking history, categorized into three independent groups: current smokers, former smokers, and never smokers.
- Requirement for maintenance oral corticosteroids (OCS), categorized as an independent binary variable (yes or no).

The research team is finalizing their statistical analysis plan. Based on the characteristics of the collected variables, which of the following statistical tests are appropriate for hypothesis testing in this study? (Select all that apply)

  1. An independent two-sample t-test to compare the mean decline in FEV1 between patients requiring maintenance OCS and those who do not.Answer
  2. B
    A one-way analysis of variance (ANOVA) to compare the mean number of asthma exacerbations across the three smoking history groups.
  3. The Kruskal-Wallis test to compare the median number of asthma exacerbations across the three smoking history groups.Answer
  4. D
    The McNemar test to evaluate the association between smoking history (three groups) and the requirement for maintenance OCS (binary).
  5. A Chi-square test of independence to assess the relationship between smoking history and the requirement for maintenance OCS.Answer

Answer

The appropriate analytical choices are using an independent two-sample t-test for FEV1 decline by OCS use, the Kruskal-Wallis test for exacerbations by smoking history, and a Chi-square test for the association between smoking history and OCS use.
Selecting the correct statistical test requires mapping the study's variables to their mathematical properties. An independent two-sample t-test is appropriately selected for comparing FEV1 decline (a normally distributed continuous variable) across two independent groups (OCS requirement). Because asthma exacerbation counts are highly skewed, the Kruskal-Wallis test is correctly chosen as the non-parametric method to compare this outcome across the three independent smoking groups. Finally, the Chi-square test is the appropriate method to test for an association between two distinct categorical variables (smoking history and OCS requirement).

Step-by-Step Solution

1
Analyze the variable types and distributions.
FEV1 decline is continuous/normal. Exacerbations are continuous/skewed. Smoking history is categorical (3 groups). OCS requirement is categorical (2 groups).
Statistical test selection depends fundamentally on whether variables are categorical or continuous, how many groups exist, whether samples are independent or paired, and the distribution of the data.
2
Determine the test for FEV1 decline (continuous, normal) vs. OCS requirement (categorical, 2 independent groups).
Independent two-sample t-test.
The t-test compares the means of a normally distributed continuous variable across exactly two independent groups.
3
Determine the test for asthma exacerbations (continuous, skewed) vs. smoking history (categorical, 3 independent groups).
Kruskal-Wallis test.
Because the data is skewed, parametric tests (like ANOVA) are invalid. The non-parametric equivalent for comparing three or more independent groups is Kruskal-Wallis.
4
Determine the test for smoking history (categorical) vs. OCS requirement (categorical).
Chi-square test of independence.
Both variables are categorical and represent independent observations, making Chi-square the correct choice to test for an association between them.

Key Concept

Selecting the correct inferential statistical test based on variable types (categorical vs. continuous), number of groups, dependency (independent vs. paired), and underlying distribution (normal vs. skewed).
Rate this question

Topics

Inferential Statistical Tests

More questions from this topic