Original practice · fully worked
Original variant: calibrate a Gaussian gauge from a central tolerance
The measurement error of a thickness gauge is normally distributed with mean zero and an unknown standard deviation. Ninety percent of its errors lie between -4 and 4 micrometers. Calculate the probability that its error exceeds 6 micrometers.
- A
0.0068 - B
0.0136 - C
0.0273 - D
0.0500 - E
0.9932
Answer and calculation
Symmetry makes 4 micrometers the 95th percentile, so the standard deviation is 4/z_0.95. The 6-micrometer threshold has z-score 1.5z_0.95, and its upper tail is 0.00681, choice A.
Setup
Setup
Let E denote the centered Gaussian measurement error.
E∼N(0,σ2) Pr(−4≤E≤4)=0.90 Model
Model
The omitted probability is split equally between the two tails, so the upper endpoint is the 95th percentile.
Pr(E≤4)=0.95 σ4=z0.95=1.6448536 Compute
Compute
Calibrate the scale, standardize 6, and evaluate the one-sided upper tail.
σ=1.64485364=2.4318273 Pr(E>6)=1−Φ(σ6)=1−Φ(2.4672804)=0.0068072 Answer
Answer
The probability of an error above 6 micrometers is approximately 0.0068.
0.0068(A)