Statistics Problem Set — Complete Solutions
Normal Distributions, Sampling Distributions & Confidence Intervals
Question 0
Medical test: P(D) = 0.06, P(+|D) = 0.95, P(+|D′) = 0.05. Using Bayes’ theorem.
a. P(person has disease | test positive)
P(D|+) = [P(+|D)·P(D)] / [P(+|D)·P(D) + P(+|D′)·P(D′)]
= (0.95 × 0.06) / [(0.95 × 0.06) + (0.05 × 0.94)]
= 0.0570 / (0.0570 + 0.0470) = 0.0570 / 0.1040
P(D|+) = 0.548
b. P(person has disease | test negative)
P(−|D) = 1 − 0.95 = 0.05, P(−|D′) = 1 − 0.05 = 0.95
P(D|−) = (0.05 × 0.06) / [(0.05 × 0.06) + (0.95 × 0.94)]
= 0.0030 / (0.0030 + 0.8930) = 0.0030 / 0.8960
P(D|−) = 0.003
Question 1
Area under the standard normal curve to the left of z (i.e. Φ(z)).
a. z = 1.60
Φ(1.60) = 0.945
b. z = 1.83
Φ(1.83) = 0.966
c. z = 0.90
Φ(0.90) = 0.816
d. z = 4.18
Φ(4.18) ≈ 1.000 (essentially all area lies to the left)
Question 2
Probabilities for the standard normal random variable z.
P(0.58 < z < 1.74)
= Φ(1.74) − Φ(0.58) = 0.9591 − 0.7190
= 0.240
P(−1.55 < z < −0.44)
= Φ(−0.44) − Φ(−1.55) = 0.3300 − 0.0606
= 0.269
P(z > 1.34)
= 1 − Φ(1.34) = 1 − 0.9099
= 0.090
P(z < −4.32)
z = −4.32 is far in the extreme left tail.
≈ 0.000
Question 3
Cerebral Blood Flow (CBF): normal, µ = 74, σ = 16.
a. P(60 < X < 80)
z■ = (60 − 74)/16 = −0.88, z■ = (80 − 74)/16 = 0.38
P = Φ(0.38) − Φ(−0.88) = 0.6480 − 0.1894
P(60 < X < 80) = 0.459
b. P(X > 100)
z = (100 − 74)/16 = 1.63
P = 1 − Φ(1.63) = 1 − 0.9484
P(X > 100) = 0.052
c. P(X < 40) (probability of a false “at risk” diagnosis)
z = (40 − 74)/16 = −2.13
P = Φ(−2.13) = 1 − 0.9834
P(X < 40) = 0.017
Question 4
Sunflower stem diameter: normal, µ = 35 mm, σ = 3 mm.
a. P(X > 40)
z = (40 − 35)/3 = 1.67
P = 1 − Φ(1.67) = 1 − 0.9525
P(X > 40) = 0.048
b. P(both of 2 independent plants > 40 mm)
= [P(X > 40)]² = (0.0475)²
≈ 0.00226 (about 0.002)
c. Limits containing the middle 95% of diameters
Middle 95% → z = ±1.96
X = 35 ± 1.96(3) = 35 ± 5.88
Limits: (29.12 mm, 40.88 mm)
d. 90th percentile of the distribution
z0.90 = 1.28
X = 35 + 1.28(3) = 35 + 3.84
90th percentile ≈ 38.84 mm
Question 5
College faculty (2-yr institutions): µ = $64,571, σ = $4,000, n = 60 sampled.
a. Sampling distribution of x■
Since n = 60 ≥ 30, the Central Limit Theorem applies, so x■ is approximately normally distributed with:
µx■ = µ = $64,571
σx■ = σ/√n = 4000/√60 = 4000/7.746
σx■ = $516.40
b. P(x■ > $66,000)
z = (66000 − 64571)/516.40 = 1429/516.40 = 2.77
P = 1 − Φ(2.77) = 1 − 0.9972
P(x■ > 66,000) = 0.003
c. Is x■ = $66,000 unusual?
Because P(x■ > 66,000) ≈ 0.003 is very small (less than 0.05), obtaining a sample mean of $66,000
would be a rare/unusual event if the true mean really were $64,571. This is evidence that the actual
average salary for two-year college professors may be higher than $64,571, rather than this being a
chance fluctuation.
Question 6
p■ = sample proportion of adults who do NOT suffer excess weight; p = 0.20, n = 120.
a. Sampling distribution of p■
Check: np = 120(0.20) = 24 ≥ 5 and n(1−p) = 120(0.80) = 96 ≥ 5, so the normal approximation is
appropriate.
µp■ = p = 0.20
σp■ = √[p(1−p)/n] = √[(0.2)(0.8)/120] = √0.0013333
Approximately normal, mean = 0.20, standard deviation = 0.0365
b. P(p■ > 0.25)
z = (0.25 − 0.20)/0.0365 = 1.37
P = 1 − Φ(1.37) = 1 − 0.9147
P(p■ > 0.25) = 0.085
c. P(0.25 < p■ < 0.30)
z■ = 1.37 (from part b), z■ = (0.30 − 0.20)/0.0365 = 2.74
P = Φ(2.74) − Φ(1.37) = 0.9969 − 0.9147
P(0.25 < p■ < 0.30) = 0.082
d. If p■ exceeded 0.30 …
P(p■ > 0.30) = 1 − 0.9969 = 0.0031, an extremely rare event under p = 0.20. This would cast doubt on
the claim that p = 0.20; it suggests the true proportion of adults who do not suffer excess weight is
actually greater than 20%.
Question 7
Biased coin, P(head) = 0.2, flipped n = 10,000 times. Find P(1950 < X < 2100) using the normal
approximation to the binomial.
µ = np = 10000(0.2) = 2000
σ = √[np(1−p)] = √[10000(0.2)(0.8)] = √1600 = 40
z■ = (1950 − 2000)/40 = −1.25, z■ = (2100 − 2000)/40 = 2.50
P = Φ(2.50) − Φ(−1.25) = 0.9938 − 0.1056
P(1950 < X < 2100) ≈ 0.888
(Using a continuity correction of ±0.5 gives z■ = −1.24, z■ = 2.49 and P ≈ 0.886 — essentially the
same result.)
Question 8
Body temperature: n = 130, x■ = 98.25°, s = 0.73°.
a. 99% confidence interval for µ
x■ ± z0.005(s/√n), z0.005 = 2.576
SE = 0.73/√130 = 0.73/11.402 = 0.0640
Margin of error = 2.576(0.0640) = 0.165
CI = 98.25 ± 0.165
99% CI: (98.085°, 98.415°)
b. Does the interval contain 98.6°?
No. The interval (98.085°, 98.415°) lies entirely below 98.6°. This indicates that 98.6°F is not a
plausible value for the true mean human body temperature based on this sample — the actual average
body temperature of healthy people appears to be lower than the commonly cited 98.6°F.
Question 9
n = 50, x■ = 56.4, s² = 2.6. Find the point estimate for µ and the margin of error.
Point estimate: µ■ = x■ = 56.4
s = √2.6 = 1.6125
SE = s/√n = 1.6125/√50 = 1.6125/7.071 = 0.2280
Margin of error (95%, z = 1.96) = 1.96(0.2280)
Point estimate = 56.4; Margin of error ≈ 0.447
Question 47
90% confidence intervals for µ (z0.05 = 1.645).
a. n = 125, x■ = 0.84, s² = 0.086
s = √0.086 = 0.2933
SE = 0.2933/√125 = 0.02623
Margin of error = 1.645(0.02623) = 0.0432
CI = 0.84 ± 0.043
90% CI: (0.797, 0.883)
b. n = 50, x■ = 21.9, s² = 3.44
s = √3.44 = 1.8547
SE = 1.8547/√50 = 0.2623
Margin of error = 1.645(0.2623) = 0.4315
CI = 21.9 ± 0.432
90% CI: (21.468, 22.332)
c. Interpretation
For part (a): we are 90% confident that the true population mean lies between 0.797 and 0.883. For
part (b): we are 90% confident that the true population mean lies between 21.468 and 22.332.
(Formally: if we repeated this sampling process many times and built a CI each time, about 90% of
those intervals would contain the true population mean µ.)
Question 10
Binomial sample: n = 300, x = 263 successes. Find a 90% confidence interval for p.
p■ = x/n = 263/300 = 0.8767
SE = √[p■(1−p■)/n] = √[(0.8767)(0.1233)/300] = √0.0003603 = 0.01898
Margin of error = 1.645(0.01898) = 0.0312
CI = 0.8767 ± 0.0312
90% CI: (0.845, 0.908)
Interpretation: we are 90% confident that the true population proportion of successes p lies between
0.845 and 0.908.
Question 11
Independent samples — 95% CI for (µ■ − µ■).
Population 1 Population 2
Sample size 35 49
Sample mean 12.7 7.4
Sample variance 1.38 4.14
(x■■ − x■■) ± z0.025√[s■²/n■ + s■²/n■], z0.025 = 1.96
Difference = 12.7 − 7.4 = 5.3
SE = √[1.38/35 + 4.14/49] = √[0.03943 + 0.08449] = √0.12392 = 0.3520
Margin of error = 1.96(0.3520) = 0.690
CI = 5.3 ± 0.690
95% CI for (µ■ − µ■): (4.610, 5.990)
Since the entire interval is positive, we are 95% confident that µ■ is greater than µ■, with the
difference likely between about 4.61 and 5.99.