Exam 2 review

Where each question came from, and the answers.

Question 1: joint table with Y = X²

X is −1, 0 or 1 with probabilities 1/4, 1/2, 1/4. Y = X².

Y = 0Y = 1total
X = −101/41/4
X = 01/201/2
X = 101/41/4
total1/21/21

Covariance, and independent or dependent?

Answer

E(X) = 0 and E(XY) = (−1)(1)(1/4) + (1)(1)(1/4) = 0, so Cov = 0.

Dependent. The cell for X = 0 and Y = 1 is 0, but the product of its row and column totals is 1/2 × 1/2 = 1/4. Physically, the power is fixed once you know the voltage. Covariance only detects a straight-line relationship and this one is a curve.

Question 2: name the distribution

20 components, each defective with probability 0.05, independently. X counts the defectives.

Type of random variable, and the single change that makes it negative binomial.

Answer

Binomial with n = 20 and p = 0.05.

Change the stopping rule: do not fix the sample at 20. Keep inspecting until a set number r of defectives has been found, and let X be the number of components inspected.

Question 3: crack length

f(x) = 8x⁻³ for x ≥ 2.

The CDF.

Answer

F(x) = ∫ from 2 to x of 8t⁻³ dt = [−4t⁻²] from 2 to x = 1 − 4/x² for x ≥ 2. The constant is fixed by F(2) = 0.

The five-number summary.

Answer

Minimum: 2. First quartile: 1 − 4/x² = 0.25 gives x² = 16/3, so x = 4/√3. Median: 1 − 4/x² = 0.5 gives x² = 8, so x = 2√2. Third quartile: 1 − 4/x² = 0.75 gives x² = 16, so x = 4. Maximum: none, the distribution runs to infinity.

More likely to be less than 4/√3 or at least 4?

Answer

Equally likely. 4/√3 is the first quartile, so 25% of cracks are shorter than it. 4 is the third quartile, so 25% are at least that long.