6. Carry-over from earlier exams
Exam 2 reused the five-number summary from the first unit. These are the older ideas most likely to return.
Five-number summary
- The five numbers: minimum, first quartile (25th percentile), median (50th), third quartile (75th), maximum.
- From a CDF: solve F(x) = 0.25, F(x) = 0.5 and F(x) = 0.75. The minimum and maximum are the ends of the support.
- For a normal distribution: the median equals the mean. The quartiles are at about mean ± 0.67 × SD. There is no minimum or maximum.
- What the quartiles mean: 25% of values are below the first quartile and 25% are above the third quartile. Those two chances are equal by definition.
Total wait is approximately normal with mean 600 and standard deviation 30. Give its median and quartiles.
Answer
Median = 600, because a normal distribution is symmetric. First quartile ≈ 600 − 0.67 × 30 = 580. Third quartile ≈ 600 + 0.67 × 30 = 620.
Binomial: exactly k of n
- Formula: P(X = k) = C(n, k) × pk × (1 − p)n − k.
- Two-stage questions: first find p for one trial with the normal, then use the binomial for "exactly k of n".
- Quick values: C(5,3) = 10, C(6,2) = 15, C(4,2) = 6, C(5,2) = 10, C(n,1) = n.
A part passes inspection with probability 0.975. Four parts are independent. Write the probability that exactly one fails.
Answer
Failures are binomial with n = 4 and p = 0.025. P(exactly 1) = C(4,1) × 0.025 × 0.975³ = 4 × 0.025 × 0.975³. Leaving it in this form is fine without a calculator.
What would turn a binomial count into a negative binomial one?Exam 2, Q2
Answer
Change the stopping rule. A binomial fixes the number of trials and counts the successes. A negative binomial fixes the number of successes and counts the trials: keep going until the r-th success. With r = 1 it is geometric.
Independent, or only uncorrelated?
- Independent: every cell of the joint table equals row total × column total.
- Zero covariance does not prove independence. Covariance only measures a straight-line relationship.
- Independent implies zero covariance, never the other way round.
X is −1, 0 or 1 with probabilities 1/4, 1/2, 1/4, and Y = X². Find Cov(X, Y) and say whether they are independent.Exam 2, Q1
Answer
E(X) = 0. E(XY) = E(X³) = (−1)(1/4) + 0 + (1)(1/4) = 0. So Cov = 0 − 0 × E(Y) = 0.
They are dependent. P(X = 0 and Y = 1) = 0, but P(X = 0) × P(Y = 1) = 1/2 × 1/2 = 1/4. Knowing X tells you Y exactly. The covariance is zero only because the relationship is a curve, not a line.
Estimators
- Unbiased: an estimator is unbiased when its expected value equals the thing it estimates. The sample mean is unbiased for μ and the sample proportion is unbiased for p.
- Standard error: the standard deviation of the estimator. For a mean it is s/√n. For a proportion it is √(p̂(1 − p̂)/n).
- Margin of error: multiplier × standard error.
A sample of 100 has a standard deviation of 20. What is the standard error of the mean, and about what is the 95% margin of error?
Answer
Standard error = 20/√100 = 2. Margin of error ≈ 2 × 2 = 4, using 1.96 ≈ 2.
Percentiles of a normal
- Value from a percent: x = μ + zσ, where z has that percent to its left.
- Upper-tail wording: "exceeded with probability 0.05" means the 95th percentile.
- Handy z-values: 90th is 1.28, 95th is 1.645, 97.5th is 1.96, 99th is 2.33, 99.9th is 3.09.
A total is normal with mean 60 and standard deviation 5. What value is exceeded only 2.5% of the time?
Answer
That is the 97.5th percentile, 2 standard deviations above the mean: 60 + 2 × 5 = 70. With the exact multiplier it is 60 + 1.96 × 5 = 69.8.