Linear Combinations of Independent Random Variables
Standardization
This Exam P sample reference tests Normal Distribution. The daily cost is the weighted normal sum 0.2X+0.1Y, with mean 7.5 and variance 0.45. Standardizing the cost threshold gives z=-0.745356 and probability 0.228028, which rounds to choice B.
These notes identify the calculation error associated with each wrong letter when that error is reproducible.
AUsing the variance 0.45 directly as the standard deviation gives Φ((7-7.5)/0.45)=0.1333, which rounds to 0.13.
CAdding the weighted standard deviations gives 0.2(3)+0.1(3)=0.9. Using that as the aggregate standard deviation gives Φ(-0.5/0.9)=0.2893.
DWeighting variances linearly instead of squaring the weights gives 0.2(9)+0.1(9)=2.7. Standardizing with √(2.7) gives 0.3805.
EDividing the deviation -0.5 by the mean cost 7.5 instead of by the standard deviation gives Φ(-0.5/7.5)=0.4734.
Original practice · fully worked
Original variant: comparing two scores with a shared normal input
Three independent normal calibration inputs U, V, and W have means 40, 25, and 10 and standard deviations 4, 3, and 2, respectively. Algorithm A reports U+V, while Algorithm B reports 1.5U+W. Calculate the probability that Algorithm B reports the larger value.
A 0.733
B 0.763
C 0.862
D 0.887
E 0.994
Variant answer in brief
Subtracting the two scores leaves D=0.5U+W-V, a normal variable with mean 5 and variance 17. Therefore P(D>0)=Φ(5/√(17))=0.887374, so choice D.
Setup
Setup
Express the score comparison as one linear contrast of the independent inputs.
D=B−A=(1.5U+W)−(U+V)=0.5U+W−V
Model
Model
Compute the contrast mean and add its independent variance contributions.
E[D]=0.5(40)+10−25=5
Var(D)=0.52(42)+22+32=17
Compute
Compute
Standardize zero under the contrast distribution.
Pr(B>A)=Pr(D>0)=Φ(175)
=0.8873735468…
Answer
Answer
Algorithm B reports the larger value with probability approximately 0.887.
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.