Mains PYQs
UPSC Mains Statistics paper 1 2026 questions
- (i) Show that for $n$ events $A_1, A_2, \cdots, A_n$
$$P\left(\bigcap_{i=1}^{n} A_i\right) \geq \sum_{i=1}^{n} P(A_i) - (n-1)$$ [7]
(ii) For two mutually exclusive events $A$ and $B$, show that
$$P(A \mid A \cup B) = \frac{P(A)}{P(A)+P(B)};\ P(A),\ P(B) > 0$$ [3]
- Let the joint probability mass function of $X$ and $Y$ be
$$p(x,y) = \frac{x^2+2y}{80},\ x=0,1,2;\ y=1,2,3,4$$
Find-
(i) the marginal probability mass functions of $X$ and $Y$; [6]
(ii) $P(X \leq 1, Y > 2)$. [4]
- A random variable $X$ has the probability density function
$$f(x) = \frac{e^{-x}x^{\lambda}}{\lambda !};\ x > 0$$
where $\lambda$ is a positive integer. Using Chebyshev's inequality, show that
$$P[0 < X < 2(\lambda+1)] > \frac{\lambda}{\lambda+1}$$
- (i) Describe how Rao-Blackwell theorem enables one to find the UMVUE. [5]
(ii) Let $X_1, X_2, \cdots, X_n$ be a random sample from a population with p.d.f.
$$f(x,\theta) = \theta x^{\theta-1};\ 0 < x < 1,\ \theta > 0$$
Show that $\prod_{i=1}^{n} X_i$ is sufficient statistic of $\theta$. [5]
- Obtain the likelihood ratio test (LRT) for testing $H_0: \theta \leq \theta_0$ against $H_1: \theta > \theta_0$, using a random sample of size $n$ taken from $P(\theta)$.
- (i) Find the moment generating function (m.g.f.) of a random variable $X$ with probability density function (p.d.f.)
$$f(x) = \begin{cases} x, & 0 \leq x < 1 \\ 2-x, & 1 \leq x < 2 \\ 0, & \text{otherwise} \end{cases}$$
Using this m.g.f., find $E(X)$ and $V(X)$. [10]
(ii) Let $\phi_X(t) = e^{-t^2/2}$ be the characteristic function of a random variable $X$. Using the inversion formula, find the probability density function of $X$. [10]
- (i) The time required, in hours, to repair a car at a garage is a random variable $X$ with probability density function (p.d.f.)
$$f(x) = c(4x - x^2),\ 0 \leq x \leq 4,\ c \text{ is a constant}$$
Find the value of $c$. Obtain the probabilities that for a car which arrives now at the garage-
(1) $X$ is at least one but less than three;
(2) $X$ is more than two. [9]
(ii) Let
$$f(x) = \frac{3x}{4}(2-x);\ 0 < x < 2$$
be the probability density function of a continuous random variable $X$. Write the cumulative distribution function of $X$ and find
$$P\left(\frac{1}{2} < X < \frac{3}{2}\right)$$ [6]
- (i) Let $X$ follow uniform distribution on $(0, 1)$ and let $Y = -\log_e(1-X)$. Find the distribution of $Y$. [5]
(ii) Let $X_1, X_2, \cdots$ be a sequence of independent random variables such that
$$P[X_k = n] = \frac{1}{n} \text{ and } P[X_k = 0] = 1 - \frac{1}{n},\ k = 1, 2, \cdots$$
Check whether the weak law of large numbers (WLLN) holds for this sequence or not. [5]
(iii) A manufacturer produces items with mean weight $100\,\mathrm{g}$ and variance $16\,\mathrm{g}^2$. For a sample of 64 items, find $P(99 < \overline{X} < 101)$, using central limit theorem (CLT). $\overline{X}$ is the sample mean of 64 items. [5]
- Let the joint probability density function of $X$ and $Y$ be
$$f(x,y) = \frac{3}{4}\left(xy + \frac{x^2}{2}\right);\ 0 < x < 1,\ 0 < y < 2$$
Find-
(i) the marginal probability density functions of $X$ and $Y$;
(ii) $E(X)$ and $V(X)$;
(iii) $E(Y)$ and $V(Y)$;
(iv) covariance between $X$ and $Y$;
(v) correlation coefficient between $X$ and $Y$.
- The lifetimes, in hours, of two brands of batteries are given below :
| Brand A | 40 | 30 | 40 | 45 | 55 | 30 |
|---|---|---|---|---|---|---|
| Brand B | 50 | 50 | 45 | 55 | 60 | 40 |
Examine whether the two brands are different with respect to average life using (i) Kolmogorov-Smirnov test and (ii) median test.
$$\left[D_{6,6,0.05} = \frac{4}{6}\right]$$ [8+7]
- Prove that there always exists a sufficient statistic for the one-parameter exponential family of distributions. Using it, show that sample total is sufficient statistic for the parameter for the Poisson model. State whether this estimator is complete.
- Let $X_1, X_2, \cdots, X_n$ be i.i.d. random variables following $N(\mu, \sigma^2)$. Find the maximum likelihood estimator for $\sigma^2$ and examine whether the same is unbiased. If biased, suggest an unbiased estimator for $\sigma^2$ and examine whether it is consistent. Also compare the mean squared errors of the estimators.
- A random sample of size $n$ is drawn from $N(\theta, \sigma^2)$. Let the prior distribution of $\theta$ be $N(\mu, \tau^2)$, where $\sigma^2$, $\mu$ and $\tau^2$ are known. Obtain the Bayes estimator for $\theta$ under (i) squared error loss function and (ii) absolute error loss function.
- Describe sequential probability ratio test (SPRT). Find the SPRT for testing $H_0: \theta = \theta_0$ against $H_1: \theta = \theta_1$, using random observations from Poisson $(\theta)$. Also find the OC function for the test.
- Let $Y_1, Y_2, \cdots, Y_n$ be $n$ independent observations from a population with mean $\mu$ and variance $\sigma^2$. Obtain the best linear unbiased estimator of $\mu$ and an unbiased quadratic estimator for $\sigma^2$.
- Explain the problem of classification in discriminant analysis. In case of two known multivariate normal populations, define Fisher's discriminant function and calculate it for the data given below :
$$\boldsymbol{\mu}^{(1)} = \begin{pmatrix} 24 \\ 12 \end{pmatrix},\ \boldsymbol{\mu}^{(2)} = \begin{pmatrix} 18 \\ 10 \end{pmatrix} \text{ and } \Sigma = \begin{pmatrix} 9 & -5 \\ -5 & 8 \end{pmatrix}$$
- Discussing Gauss-Markov linear model $\underset{n\times 1}{\mathbf{Y}} = \underset{n\times p}{X}\ \underset{p\times 1}{\boldsymbol{\beta}} + \underset{n\times 1}{\mathbf{u}}$ with $E(\mathbf{Y}) = X\boldsymbol{\beta}$ and $D(\mathbf{Y})$ = variance and covariance matrix of $\mathbf{Y} = \sigma^2 I$, show that
$$E[\mathbf{Y}'\{I - X(X'X)^{-1}X'\}\mathbf{Y}] = \sigma^2(n-p)$$
- Derive the variance of the sample mean under simple random sample without replacement (SRSWOR).
- A population is divided into three strata. The following data are taken from the population :
| Stratum $h$ | $N_h$ | $S_h^2$ | $C_h$ |
|---|---|---|---|
| 1 | 200 | 25 | 4 |
| 2 | 300 | 36 | 9 |
| 3 | 500 | 49 | 16 |
Determine the sample size for each stratum for a total sample of size 100 under (i) proportional allocation and (ii) optimum allocation.
- Let $\mathbf{l}'\boldsymbol{\beta}$ be any estimable linear function of parameters $\beta_1, \beta_2, \cdots, \beta_p$. Then prove that-
(i) there exists a unique linear function $\mathbf{c}'\mathbf{Y}$ of random variables $Y_1, Y_2, \cdots, Y_n$ such that $\mathbf{c} \in V(A')$ and $E(\mathbf{c}'\mathbf{Y}) = \mathbf{l}'\boldsymbol{\beta}$, where $V(A')$ is the vector space generated by row vectors of $A'$ and $E(\mathbf{Y}) = A\boldsymbol{\beta}$;
(ii) variance of $\mathbf{c}'\mathbf{Y}$ is less than the variance of any other linear unbiased estimator.
- Let $\mathbf{X}_1, \mathbf{X}_2, \cdots, \mathbf{X}_N$ be a random sample from $N_p(\boldsymbol{\mu}, \Sigma)$. Show that Hotelling's $T^2$-statistic for testing $H_0: \boldsymbol{\mu} = \boldsymbol{\mu}_0$ against $H_1: \boldsymbol{\mu} \neq \boldsymbol{\mu}_0$ is a generalization of $t$-statistic for testing $H_0: \mu = \mu_0$ against $H_1: \mu \neq \mu_0$ in the univariate case. Calculate $T^2$-statistic for testing $H_0: \boldsymbol{\mu} = \begin{pmatrix} 9 \\ 5 \end{pmatrix}$ on the basis of the data taken from a bivariate normal distribution given below :
$$X = \begin{pmatrix} 6 & 9 \\ 10 & 6 \\ 8 & 3 \end{pmatrix}$$
- Discuss orthogonal polynomial regression on a single variable along with its principal advantages.
- (i) Give an estimate for the population mean under systematic sampling and derive its variance in terms of the intraclass correlation coefficient. [10]
(ii) Discuss the effect of intraclass correlation on the efficiency of this estimator. [5]
- The following data come from a two-factor random effect model :
| Factor A \ Factor B | $B_1$ | $B_2$ | $B_3$ |
|---|---|---|---|
| $A_1$ | 6, 8 | 10, 12 | 14, 16 |
| $A_2$ | 12, 14 | 16, 18 | 20, 22 |
(i) Given that the experimentor suspects interaction, write the linear model stating all the assumptions for the above data. [5]
(ii) Clearly stating the hypothesis, carry out the tests for the main effects and interactions giving the complete ANOVA table for the above data. [15]
- A balanced incomplete block design (BIBD) has parameters $v = 4$ (treatments), $b = 6$ (blocks), $r = 3$ (replicates per treatment), $k = 2$ (block size), $\lambda = 1$. The following adjusted treatment totals (after removing block effects) are given as $Q_1 = 12$, $Q_2 = 8$, $Q_3 = 4$ and $Q_4 = 0$. The MSE is given as 2.5.
(i) Verify if this is a valid BIBD. [5]
(ii) Estimate the treatment effects and their variances. [5]
(iii) Compute the efficiency factor of this BIBD relative to a randomized complete block design with the same number of replications. [5]
- Define multivariate normal distribution and write down its important properties. Let $\mathbf{X}_1$, $\mathbf{X}_2$, $\mathbf{X}_3$ and $\mathbf{X}_4$ are mutually independent random vectors following $N_p(\boldsymbol{\mu}, \Sigma)$. Then find the marginal distributions of
(i) $\mathbf{W}_1 = \frac{1}{4}\mathbf{X}_1 - \frac{1}{4}\mathbf{X}_2 + \frac{1}{4}\mathbf{X}_3 - \frac{1}{4}\mathbf{X}_4$
(ii) $\mathbf{W}_2 = \frac{1}{4}\mathbf{X}_1 + \frac{1}{4}\mathbf{X}_2 - \frac{1}{4}\mathbf{X}_3 - \frac{1}{4}\mathbf{X}_4$
Also obtain the joint distribution of random vectors $\mathbf{W}_1$ and $\mathbf{W}_2$.
- A population of $N = 4$ units has values $y_1 = 10$, $y_2 = 20$, $y_3 = 30$ and $y_4 = 40$. Consider a sample of size $n = 2$ without replacement with inclusion probabilities proportional to a size measure $x_i$, where $x_1 = 1$, $x_2 = 2$, $x_3 = 3$, $x_4 = 4$. The inclusion probabilities are taken as
$$\pi_i = \frac{n x_i}{\sum x_i};\ i = 1, 2, 3, 4$$
(i) Suppose the selected sample is $\{2, 4\}$. Using Horvitz-Thompson estimator, estimate the population total. [7]
(ii) Given the joint inclusion probabilities $\pi_{24} = 0.3$ for units 2 and 4, estimate the variance of Horvitz-Thompson estimator and comment on the possibility of negative variance estimate. [8]
- (i) Distinguish between complete confounding and partial confounding. [5]
(ii) Illustrate how you will partially confound a $2^4$-factorial experiment into $2^2$ blocks in 2 replicates with no main effects confounded. State the confounded effects in each replicate. [10]