This Exam P sample reference tests Bayes' Theorem. Convert each Poisson mean to its probability of at least one event, weight those likelihoods by the two policyholder proportions, and apply Bayes' theorem. The resulting posterior is 0.276962, so choice B.
These notes identify the calculation error associated with each wrong letter when that error is reproducible.
AThis is the posterior after observing exactly one claim: using the Poisson masses 0.5 exp(−0.5) and 0.2 exp(−0.2) gives 0.246325. The evidence here includes every positive count.
CThis replaces each at-least-one likelihood by its Poisson mean. The resulting ratio 0.15(0.5)/[0.15(0.5)+0.85(0.2)]=0.306122 is not exact.
DThis is the posterior after observing at least two claims. Complementing the zero and one masses gives 0.476007, which conditions on a different event.
EThis normalizes the two positive-count likelihoods without the 0.15 and 0.85 prior weights, giving 0.684609 as if the two groups were equally common.
Original practice · fully worked
Original variant: next-day silence after one message
A monitoring account is in either quiet mode or busy mode, with prior probabilities 0.70 and 0.30. Conditional on its mode, daily message counts are independent Poisson variables with means 0.40 in quiet mode and 1.60 in busy mode. Exactly one message is observed today. Calculate the probability of no messages tomorrow.
A 0.341
B 0.445
C 0.511
D 0.530
E 0.659
Variant answer in brief
Today's one-message likelihood updates the quiet-mode probability to 0.659486. Averaging tomorrow's zero-count probabilities with the two posterior mode weights gives 0.510815, so choice C.
Setup
Setup
Form the prior-weighted likelihood contributions for today's one observed message.
wQ=0.70(0.40e−0.40)=0.1876896129…
wB=0.30(1.60e−1.60)=0.0969103286…
Model
Model
Normalize the contributions to obtain the posterior mode probabilities.
Pr(Q∣N0=1)=wQ+wBwQ=0.6594857746…
Pr(B∣N0=1)=0.3405142254…
Compute
Compute
Given the mode, tomorrow is independent of today and has its mode-specific Poisson zero probability.
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.