Questions Statistics
➢ Prove that the sum of squared deviations from the mean is minimized:
∑(Xᵢ − X̄)² ≤ ∑(Xᵢ − a)² for any a ≠ X̄.
➢ A dataset has μ = 100, σ = 15, and skewness = +1.2. Estimate the
median and mode. Justify your reasoning.
➢ Derive the condition under which the harmonic mean equals the
geometric mean for two positive numbers.
➢ Given a lognormal distribution with μ = 3, σ = 0.5 (parameters of the
underlying normal), find the exact coefficient of variation.
➢ Design a robust alternative to the Empirical Rule for a distribution with
kurtosis = 5.2.
➢ Prove Chebyshev’s Theorem for k = 1.5 using probability axioms.
➢ A manufacturing process has μ = 50mm, σ = 0.2mm. If specifications
are 50 ± 0.5mm, calculate the expected defect rate without assuming
normality.
➢ For a bimodal distribution with modes at 10 and 20, can the mean equal
15? Provide a counterexample if false.
➢ Given X with E[X] = 10 and E[X²] = 120, find the minimum and
maximum possible values of σ_X.
➢ If P(X ≤ 5) = 0.3 and P(X ≥ 15) = 0.2 for non-negative X, what bounds
can you place on μ?
➢ Given P(A)=0.4, P(B|A)=0.3, P(C|B)=0.6, and P(C|¬B)=0.1, compute
P(A|C).
➢ In a clinical trial, Treatment X succeeds for 60% of Group M and 30%
of Group F, while Treatment Y succeeds for 50% of both groups. Yet
overall, Treatment X has higher success. Explain this mathematically.
➢ For independent variables X and Y, show that Cov(X,XY) =
E[X²]E[Y] - E[X]²E[Y].
➢ Given P(U|V)=P(U|W)=0.7 and P(U|V∩W)=0.49, determine if V and
W are independent.
➢ Derive the decomposition Var(Z) = E[Var(Z|X)] + Var(E[Z|X]) for
discrete X.
➢ A system transitions between two states with retention probabilities 0.7
and 0.4. Calculate its long-term state distribution.
➢ Starting with 10 and winning 1 per round with probability 0.48, find the
chance of reaching 20 before 0.
➢ For trials continuing until r successes with success probability p, prove
the expected trial count is r(1-p)/p.
➢ When X|λ follows Poisson(λ) and λ follows Gamma(a,b), demonstrate
that X's marginal distribution is NegBinom(a, b/(b+1)).
➢ For a variable following 0.3N(0,1) + 0.7N(5,2), compute P(X > 3).
➢ For a standard normal variable constrained to positive values, derive its
expected value.
➢ Given count data with n = 10 and success probability p ~ Beta(2,5),
determine P(exactly 3 successes).
➢ For two independent exponential variables with rate λ, find the PDF of
their maximum.
➢ If U is uniform on [0,1], identify the distribution of -ln(1-U).
➢ For bivariate normal data with correlation 0.6, compute P(X > μ_X | Y
= μ_Y).
➢ Derive the CDF for the maximum of n independent exponential
variables.
➢ Define and characterize a dependence structure exhibiting asymmetric
tail relationships.
➢ A process has μ = 100, σ = 5, and skewness +1.2. Estimate its median
and justify.
➢ For positive numbers, establish the condition where harmonic mean
equals geometric mean.
➢ Given lognormal parameters μ = 3, σ = 0.5, compute its exact coefficient
of variation.
➢ Develop a dispersion rule for distributions with kurtosis = 5.2,
analogous to the Empirical Rule.
➢ Prove Chebyshev's inequality for k = 1.5 using probability axioms.
➢ For a manufacturing process (μ = 50mm, σ = 0.2mm) with specs 50 ±
0.5mm, calculate defect probability without normality assumptions.
➢ Construct an example where a bimodal distribution (modes at 10, 20)
has mean 15.
➢ Given E[X] = 10 and E[X²] = 120, determine σ_X's possible range.
➢ If P(V ≤ 5) = 0.3 and P(V ≥ 15) = 0.2 for non-negative V, bound μ_V.
➢ For a particle moving ±1 per step (probabilities 0.6/0.4), find P(total
movement ≥ +2 after 10 steps).
➢ Compute the probability of each outcome appearing exactly twice in 12
independent trials with 6 possible outcomes.
➢ Demonstrate the convergence of a discrete counting process to a
limiting distribution as trial frequency increases while maintaining
constant expected rate.
➢ For a diagnostic test with 90% sensitivity, 85% specificity, and 2%
prevalence, calculate its positive predictive value.
Application Based Questions
Finance: A stock’s daily return has skewness -0.8 and
kurtosis 5. Compute P(return <−2σ) using Cornish-Fisher
expansion.
Healthcare: For a diagnostic test with sensitivity 90%,
specificity 85%, and prevalence 2%, find the positive
predictive value.
Quality Control: Design a CUSUM control chart for a
process with μ0=100, σ=5, and shift detection threshold
δ=1.5σ.
Sports Analytics: Using Poisson processes, model the
probability a soccer team scores exactly 2 goals given
λ=1.4/game.
Climate Science: Fit a GEV distribution to annual
maximum temperatures and estimate the 100-year return
level.
