Questions: Likelihood Ratios and Belief Updates

5 questions to test your understanding

Score: 0 / 5
Question 1 Multiple Choice

A test for a rare disease has 95% sensitivity and a 10% false positive rate (LR ≈ 9.5). The disease affects 1% of the population. A patient tests positive. What is the approximate posterior probability the patient has the disease?

AAbout 95%, because the test is highly sensitive
BAbout 86%, because the LR of 9.5 means strong evidence and the prior is close enough to 1%
CAbout 8.7%, because the low prior (1%) dominates despite the moderate LR
DAbout 50%, because a positive test makes the hypothesis and its negation equally likely
Question 2 Multiple Choice

A new blood marker is found in cancer patients 55% of the time and in healthy people 50% of the time, giving a likelihood ratio of 55/50 = 1.1. A clinician says the marker 'provides useful evidence — a positive result makes cancer 10% more likely.' What is the key error in this reasoning?

ANothing — a 10% likelihood ratio boost is a clinically meaningful update
BThe clinician confused the likelihood ratio with a probability boost; LR = 1.1 means the evidence is nearly uninformative because the marker is barely more expected under cancer than without it
CThe LR should be computed as 50/55 (the reciprocal) for a positive result
DThe marker cannot be useful because it appears in healthy people at all
Question 3 True / False

A likelihood ratio of 100 applied to a prior probability of 1 in 10,000 yields a posterior probability of approximately 1%.

TTrue
FFalse
Question 4 True / False

A likelihood ratio of 20 means there is approximately a 95% probability that the hypothesis is true.

TTrue
FFalse
Question 5 Short Answer

Explain why thinking in log-odds makes it easier to combine multiple independent pieces of evidence when updating beliefs.

Think about your answer, then reveal below.