Solutions to the second partial examination for Statistics

Bachelor in Aerospace Engineering, Grs. 45, 46, 47

December 13, 2023

P1. (4 points) A government agency is concerned about the efficiency of security screening at an
international airport with two terminals. In the first place, they want to assess whether the mean
time taken for security checks in Terminal A is shorter than 6 minutes. In order to do so, a sample
of nA = 26 observations is taken, rendering a sample mean of xA = 5.4 minutes and a sample
standard deviation sA = 1.2 minutes.

a) (0.25 pts.) The Shapiro-Wilk test on the sample of Terminal A results in a p-value of 0.656.
What is the interpretation?
b) (1.25 pts.) Test whether the mean time for security checks in Terminal A is shorter than 6
minutes (use α = 0.05)
c) (0.5 pts.) Provide some relevant (and tight) bound to the p-value of the test in part b).

Another sample of nB = 26 observations of Terminal B is taken. It renders a sample mean

xB = 6.2 min and a sample standard deviation sB = 1.5 min (with p-value of 0.583 in the
Shapiro-Wilk test).

d) (0.25 pts.) A test was conducted to compare the variability of the check times. Specifically,
2 2 2 2
H0 : σA = σB vs. H1 : σA ̸= σB is tested, with a p-value of 0.271. Interpret this number.
e) (1.25 pts.) Test whether there is a significant difference in the mean time for security checks
between Terminal A and Terminal B (use α = 0.05)
f) (0.5 pts.) Build a 99% confidence interval on the difference in mean times of security check
of the two terminals.

Solution. a) The p-value of the Shapiro-Wilk tests is greater than 0.1 (or any other reasonable significance
level) so, from hereafter, we can assume that the security check time in Terminal A is normally
distributed. We denote the random variable as XA ∼ N (µA , σA ), which is given in minutes.
b) Test H0 : µA = 6 vs. H1 : µA < 6 with sinificance level α = 0.05. The test statistic is
t = sAx/A√−6
nA = −2.55, and H0 is rejected if t < −tnA −1,α = −t25,0.05 = −1.708. Since in
our sample t = −2.55 > −1.708, we REJECT H0 and, with significance level α = 0.05, we
conclude that the mean time for security checks in Terminal A is shorter than 6 min.
c) p-value= P (T < −2.55) < 0.01 for T ∼ t25 (and p-value> 0.005).
d) The security check time in Terminal B is normally distributed as XB ∼ N (µB , σB ), in min.
The p-value corresponds to a variance comparison test. For any reasonable sig. level, the null
2 2
hypothesis is not rejected, and we can assume that both variances are identical, σA = σB .
e) Test H0 : µA = µB vs. H1 : µA ̸= µB with sig. level α = 0.05. The test statistic is t =
xA −xB (n −1)s2 +(nB −1)s2B
q
1 1
with s2p = A nAA+nB −2 = 1.845, and H0 is rejected if |t| > tnA +nB −2,α/2 =
sp nA +n
B
−t50,0.025 = 2.009. Since in our samples t = −2.124 < −2.009, we REJECT H0 and conclude
that, with sig. level 0.05, the mean time for security checks between terminals A and B differ.
h q i
f) The 99% confidence interval on µA − µb is xA − xB ∓ tnA +nB −2,α/2 sp n1A + n1B , where
√
nA = 26, nB = 26, α = 0.01, t50,0.005 = 2.678, xA = 5.4, xB = 6.2, and sp = 1.845, so the
CI is [−1.8089 , 0.2089] .

1
 P2. (4 points) An aircraft manufacturing company is concerned about the consistency of the diameter
measurements, assumed to be normally distributed, of a specific component in aircraft engines.
They have collected diameter measurements (in millimeters) from several trial samples taken with
the process in the in-control state obtaining a grand mean µ̂ = x = 75.6mm and an estimate of
the standard deviation σ̂ = 1.2mm.

a) (0.5 pts.) Set up an X-chart to monitor the process based on samples of size n = 4.
b) (1 pt.) If the process mean is shifted to 78mm while the std. dev. remains constant, what is
the probability that such a shift is detected with the chart in a)?, and the ARL?
c) (0.5 pts.) The specifications for the diameter are 75.5 ± 2 = [73.5 , 77.5], what is the proba-
bility of producing a nonconforming component?
d) (0.75 pts.) Use the proportion of nonconforming units computed in part c) to set up a p-chart
for samples of n = 100 components. How many nonconforming units are allowed for a sample
in the in-control state?
e) (0.75 pts.) What is the probability that at most one out of ten components of the given kind
fails to meet the specifications?
f) (0.5 pts.) Determine the diameter exceeded by 85% of such components.

Solution. Variable X stands for the diameter of the aircraft engine component (in mm). We know X ∼ N .

a) X-chart for samples of size n = 4

s
UCL = x + 3 √ = 77.4
c4 n
centerline = x = 75.6
s
LCL = x − 3 √ = 73.8 .
c4 n

b) After the shift, the diameter is distributed as Y ∼ N (78, 1.2), so in a sample of size n = 4,
Y ∼ N (78, 0.6). The probability of detection is 1 − P (LCL < Y < UCL) = 1 − P (−7 < Z <
−1) = 0.8413. Finally, the ARL is 1/0.8413 = 1.2 .
c) X ∼ N (75.6, 1.2), P (“nonconforming”) = 1 − P (73.5 < X < 77.5) = 1 − P (−1.75 < Z <
1.58) = 2 − 0.9429 − 0.9599 = 0.0972. .
d) p-chart for samples of size n = 100
r
p(1 − p)
UCL = p + 3 = 0.1861
n
centerline = p = 0.0972
r
p(1 − p)
LCL = p + 3 = 0.0083 .
n
A sample in the in-control state contains between 1 and 18 nonconforming units.
e) Denote by T the number of defective components in a group of 10. Clearly enough T ∼
B(n = 10, p = 0.0972) and the probability that at most one of the components is defective is
P (T ≤ 1) = P (T = 0) + P (T = 1) = 0.7469 .
f) Determine w such that P (X > w) = 0.85, so (w − 75.6)/1.2 = −1.035 and finally w =
74.358mm.

2
 P3. (2 points) A company is analyzing the factors that affect the fuel efficiency of commercial aircraft.
They gathered data on variables fuel (fuel efficiency measured in gallons per hour), thrust
(engine thrust in pounds), weight (aircraft weight in pounds), and altitude (cruising altitude
in feet). They performed a multiple linear regression analysis obtaining the table below.

> summary(lm(fuel~thurst+weight+altitude))
Estimate Std.Error t value Pr(>|t|)
(Intercept) 128.45 12.36 10.4 7.1e-17
thrust -0.032 0.008 -4.0 1e-4
weight -0.015 0.004 -3.6 0.001
altitude 0.001 0.001 1.8 0.075
Residual standard error: 537.2 on 86 degrees of freedom

a) (0.75 pts.) Determine dependent variable, independent variables, and sample size?
b) (0.75 pts.) Is there evidence, at 5% level, that each of the regressors affects the dependent
variable?, why? (write H0 and H1 , test-statistics, p-values, and explain the conclusions)
c) (0.5 pts.) Write the fitted regression equation (with the estimates of the coefficients of the
influential regressors).

Solution. a) The dependent variable is the fuel efficiency (y = fuel), while the independent variables are
the engine thrust (x1 = thrust), the aircraft weight (x2 = weight), and the cruising altitude
(x3 = altitude) . In a multiple lin. reg. model, the number of degrees of freedom is n−k−1,
where k is the number of independent variables, so the sample size is n = 86 + 4 = 90.
b) Test H0 : β1 = 0 vs. H1 : β1 ̸= 0 with α = 0.05. The test statistic is β̂1 /S(β̂1 ) = −4.0,
the p-value is 10−4 < 0.05 and thus, the null hypothesis is rejected, so we conclude that the
engine thrust influences the fuel efficiency.
Test H0 : β2 = 0 vs. H1 : β2 ̸= 0 with α = 0.05. The test statistic is β̂2 /S(β̂2 ) = −3.6,
the p-value is 10−3 < 0.05 and thus, the null hypothesis is rejected, so we conclude that the
aircraft weight influences the fuel efficiency.
Test H0 : β3 = 0 vs. H1 : β3 ̸= 0 with α = 0.05. The test statistic is β̂3 /S(β̂3 ) = 1.8, the
p-value is 0.075 > 0.05 and thus, the null hypothesis is not rejected, so we conclude that
there is not significance evidence that the cruising altitude influences the fuel efficiency.
c) We must discard the cruising altitude, since it does not significantly contributed to the model,
ˆ = 128.45 − 0.032thrust − 0.015weight .
fuel