This Exam P sample reference tests Conditional Probability. Split the callers into attendant, same-day return, and next-day return groups. Their joint application probabilities are 0.48, 0.18, and 0.04, so Bayes' theorem gives 0.48/0.70=24/35=0.6857, choice E.
These notes identify the calculation error associated with each wrong letter when that error is reproducible.
AThe value 0.06 is 0.10(0.60), the joint probability of a next-day return caller who does not apply. It is neither the requested branch nor conditioned on applying.
BThe value 0.26 is approximately 0.18/0.70, the posterior probability of the same-day return group among applicants.
CThe value 0.48 is the joint probability of attendant contact and application. It must still be divided by the total application probability.
DThe value 0.60 is the prior probability of attendant contact and ignores the information that the caller applied.
Original practice · fully worked
Original variant: source inference after split test results
A laboratory receives sealed samples from sources A, B, and C in proportions 0.40, 0.35, and 0.25. Conditional on the source, each of two screening checks passes independently. The pass probabilities per check are 0.80 for A, 0.50 for B, and 0.20 for C. A selected sample passes exactly one of the two checks. Calculate the probability that it came from source B.
A 0.320
B 0.350
C 0.383
D 0.457
E 0.500
Variant answer in brief
For a source with per-check pass probability p, exactly one pass has probability 2p(1-p). The prior-weighted likelihoods are 0.128, 0.175, and 0.080, so the posterior for source B is 0.175/0.383=0.4569, choice D.
Setup
Setup
Let E be the event that exactly one screening check passes.
Pr(E∣p)=2p(1−p)
Model
Model
Compute the likelihood of the observed split result for each possible source.
Pr(E∣A)=2(0.80)(0.20)=0.32
Pr(E∣B)=2(0.50)(0.50)=0.50
Pr(E∣C)=2(0.20)(0.80)=0.32
Compute
Compute
Multiply by the source proportions and normalize the weight for source B.
Pr(A∩E)=0.40(0.32)=0.128
Pr(B∩E)=0.35(0.50)=0.175
Pr(C∩E)=0.25(0.32)=0.080
Pr(B∣E)=0.128+0.175+0.0800.175=0.4569191
Answer
Answer
Among samples with one passing check, about 45.7% came from source B.
The 3108-page Probability Proof Manual reorganizes 718 verified Exam P solutions by syllabus skill and adds formula proofs, error patterns, and original worked practice.