This Exam P sample reference tests Conditional Distributions. This is a conditional Bernoulli variance obtained by normalizing the relevant slice of a discrete joint distribution. Within Y=1, the probability that W=1 is 3/8, so the variance is 15/64 and choice B is correct.
These notes identify the calculation error associated with each wrong letter when that error is reproducible.
AThis uses the joint probability P(W=1,Y=1)=3/20 as though it were conditional; the resulting Bernoulli calculation (3/20)(17/20)=51/400 omits normalization by P(Y=1).
CThis ignores the condition on Y. Across all cells P(W=1)=8/20=2/5, whose unconditional variance is (2/5)(3/5)=6/25.
DThis reports the conditional mean P(W=1|Y=1)=3/8. For a Bernoulli variable, that mean must still be multiplied by 1-p to obtain the variance.
EThis reports the unconditional mean P(W=1)=8/20=2/5 and therefore omits both the conditioning and the variance calculation.
Original practice · fully worked
Original variant: variance of a conditionally scored audit flag
A quality audit records binary indicators U and V. Their joint probabilities are P(U=0,V=0)=0.18, P(U=0,V=1)=0.12, P(U=1,V=0)=0.28, and P(U=1,V=1)=0.42. The audit score is S=3U-1. Calculate Var(S | V=1).
A 14/81
B 7/9
C 4/3
D 14/9
E 189/100
Variant answer in brief
Given V=1, the probability that U=1 is 0.42/(0.12+0.42)=7/9. Scaling a Bernoulli variable by 3 multiplies its variance by 9, so Var(S|V=1)=14/9 and choice D is correct.
Setup
Setup
Normalize the V=1 column of the joint table.
Pr(U=1∣V=1)=0.12+0.420.42=97
Pr(U=0∣V=1)=92
Model
Model
A constant shift has no effect on variance, while multiplication by 3 contributes a factor of 9.
Var(S∣V=1)=Var(3U−1∣V=1)=9Var(U∣V=1)
Compute
Compute
Use the conditional Bernoulli variance and apply the scaling factor.
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.