Biostatistics: multiple comparisons: practice question

A PSM question in the style of the NEET PG paper. Answer it first, then read why the right option is right and why each of the others is wrong.

Try it

PSM · Biostatistics: multiple comparisons

Sample item

A trial reports five secondary outcomes, each tested at an unadjusted 5% level. One outcome has p = 0.03. After Bonferroni correction for the five tests, how should this result be read?

AEach test is judged at 0.01, so p = 0.03 is significantBEach test is judged at 0.05, so p = 0.03 is significantCEach test is judged at 0.01, so p = 0.03 is not significantDEach test is judged at 0.25, so p = 0.03 is significant

Original practice item, not a recalled exam question. Nothing here is recorded.

The answer and the reasoning

C. Each test is judged at 0.01, so p = 0.03 is not significant

Repeated testing inflates the overall type I error. The Bonferroni correction divides the significance level by the number of tests, 0.05 / 5 = 0.01. A p-value of 0.03 is larger than 0.01 and so is no longer considered significant.

Option by option

A. Each test is judged at 0.01, so p = 0.03 is significant
The threshold is right but the conclusion is reversed; 0.03 is greater than 0.01.
B. Each test is judged at 0.05, so p = 0.03 is significant
An unadjusted level of 0.05 ignores the multiple comparisons and inflates false positives.
C. Each test is judged at 0.01, so p = 0.03 is not significant (correct)
Correct. The adjusted threshold is 0.01 and 0.03 exceeds it.
D. Each test is judged at 0.25, so p = 0.03 is significant
Multiplying alpha by the number of tests would make the false positive problem worse, not correct it.

Source note: Original NextPrep sample item — pending expert review. This item is original practice material awaiting independent expert review, so confirm important facts in a standard textbook.

Questions about this PSM question

What is the answer to this PSM question?

The best answer is C, Each test is judged at 0.01, so p = 0.03 is not significant. Repeated testing inflates the overall type I error.

Why is "Each test is judged at 0.01, so p = 0.03 is significant" not the answer?

The threshold is right but the conclusion is reversed; 0.03 is greater than 0.01.

Why is "Each test is judged at 0.05, so p = 0.03 is significant" not the answer?

An unadjusted level of 0.05 ignores the multiple comparisons and inflates false positives.

Why is "Each test is judged at 0.25, so p = 0.03 is significant" not the answer?

Multiplying alpha by the number of tests would make the false positive problem worse, not correct it.

What concept does this question test?

It tests Biostatistics: multiple comparisons in PSM. The Bonferroni correction divides the significance level by the number of tests, 0.05 / 5 = 0.01. A p-value of 0.03 is larger than 0.01 and so is no longer considered significant.

How hard is this question?

We tag it as a multi-step question: it asks you to combine more than one idea. The tag is our own editorial judgement, not an official difficulty rating, so read the reason for every option, not only the answer.

How should I study Biostatistics: multiple comparisons?

Practise biostatistics with small calculations until sensitivity, predictive values and risk measures are automatic. Then use the flashcards and the other PSM questions on this site to test what stuck.

Is this a real NEET PG question?

No. It is an original practice item (editorial status: sample, version 1). NextPrep does not publish recalled or leaked questions, and independent faculty review of this item has not been completed yet, so confirm important facts in a standard textbook.