This Exam P sample reference tests Bayes' Theorem. Prior-weighting the three conditional rates gives total left-handed probability 221/2400. The no-left-handed-parent contribution is 117/2400, so the requested posterior is 117/221=9/17, approximately 0.529 and choice C.
These notes identify the calculation error associated with each wrong letter when that error is reproducible.
AThe value 0.09 is the marginal probability that a child is left-handed: 221/2400=0.09208. That is the Bayes denominator, not the posterior share of one family category.
BWith the correct denominator 221/2400, a posterior of 0.42 would require a no-parent joint weight of only 92.82/2400. The stated prior and conditional rate instead give 117/2400, so this choice underweights the target category.
DThe value 0.78 is the prior probability 39/50 of having no left-handed parents. Observing that the child is left-handed changes this category probability.
EThe value 0.91 is the complement of the marginal left-handed probability: 1-221/2400=0.90792. It concerns whether the child is left-handed, not the requested posterior family category.
Original practice · fully worked
Original variant: update from disagreeing message filters
One quarter of incoming messages are malicious and the rest are benign. Two filters operate independently after the message type is fixed. Filter 1 alerts on 80% of malicious and 20% of benign messages; Filter 2 alerts on 50% of malicious and 10% of benign messages. A message triggered Filter 1 but not Filter 2. Calculate the probability that it was malicious.
A 0.1000
B 0.1800
C 0.2500
D 0.4000
E 0.4255
Variant answer in brief
Conditional independence makes the observed alert pattern have likelihood 0.80(0.50)=0.40 for malicious messages and 0.20(0.90)=0.18 for benign messages. Prior-weighting gives masses 0.10 and 0.135, so the malicious posterior is 0.10/0.235=20/47, approximately 0.4255 and choice E.
Setup
Setup
Let R denote the mixed result in which the first filter alerts and the second does not.
Pr(M)=0.25,Pr(B)=0.75
Model
Model
Multiply the two filter factors within each message type using their stated conditional independence.
Pr(R∣M)=0.80(1−0.50)=0.40
Pr(R∣B)=0.20(1−0.10)=0.18
Compute
Compute
Weight the two likelihoods by their priors and normalize the malicious route.
Pr(M∩R)=0.25(0.40)=0.10
Pr(B∩R)=0.75(0.18)=0.135
Pr(M∣R)=0.10+0.1350.10=4720=0.4255319…
Answer
Answer
The posterior malicious probability rounds to 0.4255.
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.