Skip to content
CSIR NET Part C

Counterexample bank

Part C is won by knowing which tempting claims are false. 149 counterexamples; 86 free to read right now. Pick your subject below.

Knowing the claim is false is half of it. The mistakes that cost marks shows the wrong options candidates actually pick, and the reasoning slip behind each one.

← All subjects

Probability, Statistics & Linear Programming

“Convergence in probability ⇒ almost sure convergence”
“Uncorrelated ⇒ independent”
“Pairwise independent events are mutually independent”

False — Two fair coin tosses: A = first is heads, B = second is heads, C = the two agree

“Every random variable has a moment generating function”

False — The standard Cauchy distribution

“A mixture of two distributions is a linear combination of the variables”
“The maximum of independent exponentials is exponential”

False — max⁡(X1,X2)\max(X_{1}, X_{2}) with XiX_{i} ~ Exp(1)

“If X and Y are each normal and uncorrelated then they are independent”

False — X ~ N(0,1),ε=±1N(0,1), \varepsilon = \pm1 with probability ½ independent of X,Y=εXX, Y = \varepsilon{}X

“Uncorrelated implies independent”
“Convergence in probability implies almost sure convergence”

False — The typewriter sequence on [0,1]

“Xn→0X_{n} \to 0 almost surely implies E[Xn]→0E[X_{n}] \to 0”
“An irreducible chain with a stationary distribution converges to it”

False — The two-state chain that swaps deterministically (period 2)

“A recurrent chain has a stationary distribution”
“A sufficient statistic is complete”

False — X(n)(X_{(n)} (or (X(1),X(n)))(X_{(1)}, X_{(n)})) for Uniform(−θ,θ)(-\theta, \theta)

“The Cramér–Rao bound is attained by the UMVUE”
“The MLE is unbiased”

False — σ\sigmâ2=(1/n)∑(Xi−Xˉ)2^{2} = (1/n)\sum(X_{i} - {\bar{X}})^{2} for N(μ,σ2)N(\mu, \sigma^{2}), or X(n)X_{(n)} for Uniform(0,θ)(0,\theta)

“The MLE is unique”
“A UMP test exists for every testing problem”

False — H0:μ=0H_{0}: \mu = 0 vs H1:μ≠0H_{1}: \mu \ne 0 for N(μ,1)N(\mu, 1)

“Any interval of the form [Xˉ−(S/n)t,∞)[{\bar{X}} - (S/\sqrt{n})t, \infty) with a 90% quantile t is a 90% confidence interval”

False — t = tn−1,0.9t_{n-1,0.9} under the convention tm,αt_{m,\alpha} =(1−α)−= (1-\alpha)-th quantile

“OLS is the BLUE in every linear model”

False — Yi=βxi+εiY_{i} = \beta{}x_{i} + \varepsilon_{i} with Var(εi)=σ2xi2Var(\varepsilon_{i}) = \sigma^{2}x_{i}^{2}

“Every parameter in the one-way ANOVA model is estimable”
“If every marginal is normal then the vector is multivariate normal”

False — X ~ N(0,1) and Y=εXY = \varepsilon{}X with ε=±1\varepsilon = \pm1 independent

“Systematic sampling is always at least as efficient as SRS”

False — A population with a periodic pattern of period k, sampled every k-th unit