Module 06 Knowledge Check
5 questions. Pass mark 4 out of 5. Answer every question before checking the answer key below, then retry after reading the feedback.
1. stats.ttest_ind(a, b) in scipy defaults to…
- A. Welch's test
- B. the equal-variance (Student's) test
- C. a paired test
- D. a rank test
2. Before-and-after measurements on the same people should use…
- A. ttest_ind
- B. ttest_rel
- C. chi2_contingency
- D. f_oneway
3. A significant chi2_contingency tells you…
- A. which cells drive the association
- B. that an association exists, and nothing about its direction
- C. the effect size
- D. that one variable causes the other
4. After a significant f_oneway, the pairwise follow-up should be…
- A. plain ttest_ind on each pair
- B. pairwise_tukeyhsd from statsmodels
- C. skipped entirely
- D. run only on the largest groups
5. A non-significant result from 10 per group…
- A. shows there is no effect
- B. is close to uninformative, since a real 5-point effect would be missed 88% of the time
- C. proves the effect is small
- D. should be reported as a trend
---
Answer Key and Explanations
Check these only after attempting every question.
1. B — the equal-variance (Student's) test
That is the opposite of R's default. On unequal spreads and sizes it gave p = 0.0019 where Welch gave p = 0.1086.
2. B — ttest_rel
In the lesson the independent test gave p = 0.706 and the paired test p = 0.012 on identical numbers.
3. B — that an association exists, and nothing about its direction
The standardised residuals give the direction: premium at +3.04 on churned, basic at −2.72.
4. B — pairwise_tukeyhsd from statsmodels
scipy offers no post-hoc, and unadjusted pairwise tests reintroduce the multiplicity the F-test controlled.
5. B — is close to uninformative, since a real 5-point effect would be missed 88% of the time
Power at n = 10 was 12%. The study was never capable of finding it.
Practical Check
Apply this module to your own work: complete the module activity for *Comparing Groups and Categorical Data*, then write one sentence naming what your result shows and one naming what it does not.
Strong Answer Pattern
A strong answer names the task, the evidence used, the check performed, and the remaining limitation. It avoids "proved", "guaranteed", or "always" unless the evidence genuinely supports it.