Mains PYQs
UPSC IFoS Mains Statistics paper 1 2025 questions
- There are sometimes errors in transmission of coded messages. Morse code uses 'dots' and 'dashes' which generally happen in the proportion $3:4$. Assume that there is interference on transmission line and a received dot is mistakenly taken as a dash with probability $\frac{1}{6}$ and vice versa. Find (i) the probability that a dot has been received and (ii) the probability that a dot was sent given that dot was received. [4+4]
- (i) Check whether
$$F(x)=\begin{cases}\frac{e^x}{2}, & x<0\\ \frac{1}{2}, & 0\le x<1\\ 1-\frac{e^{1-x}}{2}, & x\ge1\end{cases}$$
is the cumulative distribution function of a random variable $X$.
(ii) For $i=1,2,\ldots,n$, let $X_i$'s be independent and identically distributed as $N(\mu,\sigma^2)$. Show that weak law of large numbers holds. [4+4]
- Let $X_1,X_2,\ldots,X_n$ be a random sample from $U(0,\theta)$, $\theta\in\Theta=(0,\infty)$. Let $X_{(n)}=\max\{X_1,X_2,\ldots,X_n\}$ and $Y_n=\frac{2}{n}\sum_{i=1}^{n}X_i$.
(i) Show that $X_{(n)}$ is consistent estimator of $\theta$.
(ii) Check the estimator $Y_n$ for consistency and unbiasedness for the parameter $\theta$. [4+4]
- Let $X_1,X_2,\ldots,X_n$ be a random sample from exponential distribution with mean $\theta$.
(i) Obtain the maximum likelihood estimator of $\theta$.
(ii) Obtain the asymptotic distribution of the maximum likelihood estimator that you have obtained in (i) above. [4+4]
- Let $X$ be a random variable whose p.m.f. under $H_0$ and $H_1$ is given by
| $x$ | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| $f(x\mid H_0)$ | 0.01 | 0.01 | 0.01 | 0.01 | 0.01 | 0.01 | 0.94 |
| $f(x\mid H_1)$ | 0.06 | 0.05 | 0.04 | 0.03 | 0.02 | 0.01 | 0.79 |
(i) Find the most powerful test for $H_0$ versus $H_1$ with size $0.04$. Compute the probability of Type II error for this test.
(ii) Write down the most powerful test for $H_0$ versus $H_1$ with size $0.10$. Compute the power of your test. [4+4]
- Let
$$f(x)=\begin{cases}\frac{x+1}{2}, & -1<x<1\\ 0, & \text{otherwise}\end{cases}$$
be the probability density function (p.d.f.) of a continuous random variable $X$. Let $Y=X^2$. Find (i) the probability density function of $Y$ and check that it is a valid p.d.f., (ii) $E(Y)$ and (iii) $V(Y)$. [4+3+3]
- (i) Let $\{X_n\}$ be a sequence of random variables such that
$P(X_n=0)=1-\left(\frac{1}{2}\right)^n$ and
$P(X_n=1)=\left(\frac{1}{2}\right)^n$, $n=1,2,\ldots$
Then check whether $X_n\to X$ in probability, almost everywhere and in second mean, where $X$ is a random variable taking value '0' with probability 1. [6]
(ii) State Glivenko-Cantelli theorem. [4]
(iii) Let $\{A_n\}$ be an independent sequence of events such that
$$\sum_{n=1}^{\infty}P(A_n)=\infty$$
Then show that $P\left(\limsup_n A_n\right)=P(A_n\ \text{i.o.})=1$. [5]
- (i) Let $X_1,X_2,\ldots,X_n$ be a random sample from normal distribution with mean $\theta$ and variance $\sigma^2$, where $\sigma^2$ is known. Obtain the uniformly most powerful size $\alpha$ test for testing $H_0:\theta\le\theta_0$ versus $H_1:\theta>\theta_0$. Using the test, write down $(1-\alpha)$ level one-sided confidence interval for $\theta$. [8]
(ii) State the purpose of Wilcoxon-Mann-Whitney test and write down the test statistic. State the asymptotic distribution of the test statistic. [7]
- For two continuous random variables $X$ and $Y$, the joint probability density function of $(X,Y)$ is given as
$$f(x,y)=\begin{cases}c(x+2y), & 0<x<2,\ 0<y<1\\ 0, & \text{otherwise}\end{cases}$$
Find (i) the value of $c$, (ii) the marginal probability density functions (p.d.f.) of $X$ and $Y$, (iii) the joint cumulative distribution function (c.d.f.) of $X$ and $Y$, (iv) $\mathrm{cov}(X,Y)$ (covariance of $X$ and $Y$) and (v) the correlation between $X$ and $Y$, and comment upon the type of relationship. [2+4+4+4+6]
- Let $X_1,X_2,\ldots,X_n$ be a random sample from Poisson distribution with mean $\lambda$. Suppose prior distribution of $\lambda$ over $(0,\infty)$ is given by $\pi(\lambda)=\begin{cases}e^{-\lambda}, & \lambda>0\\ 0, & \text{otherwise}\end{cases}$.
Find the Bayes estimator of $\lambda$ under quadratic error loss function. Is the prior distribution conjugate? Justify your answer.
- A random sample of size 10 from a continuous distribution over $[0,1]$ is below :
$0.73,\ 0.25,\ 0.48,\ 0.68,\ 0.79,\ 0.89,\ 0.61,\ 0.96,\ 0.54,\ 0.41$
Check whether the data were randomly drawn from uniform distribution over $[0,1]$, using Kolmogorov-Smirnov goodness of fit test. (The critical value for $\alpha=0.05$ is $0.409$)
- (i) Let $X$ and $Y$ be discrete random variables with the probability mass function given as
| $Y \backslash X$ | 1 | 2 | 3 |
|---|---|---|---|
| 2 | $\frac{1}{12}$ | $\frac{1}{6}$ | $\frac{1}{12}$ |
| 3 | $\frac{1}{6}$ | 0 | $\frac{1}{6}$ |
| 4 | 0 | $\frac{1}{3}$ | 0 |
Check whether $X$ and $Y$ are independent random variables. [5]
(ii) Let $X$ follow binomial distribution with parameters $n$ and $p$. Find the characteristic function (c.f.) of $X$. Using the c.f., find $E(X)$ and $V(X)$. [5]
- (i) Let $\{X_n\mid n\ge0\}$ be a Markov Chain (MC) with state space $\{0,1,2\}$. The transition probability matrix (t.p.m.) is given as
$$P=\begin{bmatrix}0&1&0\\ \frac{1}{2}&0&\frac{1}{2}\\ 0&1&0\end{bmatrix}$$
(rows and columns indexed 0, 1, 2)
Check whether the given MC is irreducible. Find the period of each state. Identify the recurrent states. [1+2+2]
(ii) Let $X$ be the number of ice blocks in a drink with probability mass function (p.m.f.)
| $x$ | 3 | 4 | 5 | 6 |
|---|---|---|---|---|
| $P(X=x)$ | 0.4 | 0.3 | 0.1 | 0.2 |
For a sample of 49 drinks, find $P(\bar{X}<4.5)$.
(Refer the tables enclosed) [5]
- Define one-parameter exponential family of distributions and verify whether the following distributions belong to the one-parameter exponential family of distributions :
(i) $P(X=x\mid\lambda)=\begin{cases}\frac{e^{-\lambda}\lambda^x}{x!(1-e^{-\lambda})}, & x=1,2,\ldots,\ \lambda>0\\ 0, & \text{otherwise}\end{cases}$ [5]
(ii) $f_X(x\mid\mu)=\begin{cases}\frac{1}{2}e^{-(x-\mu)/2}, & x>\mu\\ 0, & \text{otherwise}\end{cases}$ [5]
(iii) $f_X(x\mid\theta)=\begin{cases}1 & \text{if }\theta<x<\theta+1\\ 0, & \text{otherwise}\end{cases}$ [5]
(iv) $f_X(x\mid\sigma)=\begin{cases}\frac{1}{\sqrt{2\pi}\sigma}e^{-\frac{1}{2}(x-5)^2/\sigma^2}, & -\infty<x<\infty,\ \sigma>0\\ 0, & \text{otherwise}\end{cases}$ [5]
- Compute the Mahalanobis distance of a point $(3,5)$ from the sample mean vector $\begin{bmatrix}4\\6.4\end{bmatrix}$ with the sample variance matrix $\begin{bmatrix}2.5&3.25\\3.25&4.7\end{bmatrix}$.
- For the purpose of simple linear regression, the following data set $(X,Y)$ is taken :
| $X$ | 8 | 6 | 9 | 10 | 12 |
|---|---|---|---|---|---|
| $Y$ | 10 | 5 | 18 | 7 | 10 |
Study the sensitivity of the outlier $(9,18)$ in the given data and interpret.
- A sample of 50 students is to be drawn from a population of 500 students from two colleges $A$ and $B$. The means and standard deviations of their scores are given below :
| College | Total number of students | Mean | Standard deviation |
|---|---|---|---|
| $A$ | 300 | 40 | 10 |
| $B$ | 200 | 50 | 40 |
How would you draw the samples using proportional allocation? Hence, obtain the variance of estimate of the population mean and compare its efficiency with simple random sampling without replacement.
- The following data were obtained from an experiment involving four tropical feeds $A$, $B$, $C$ and $D$ tried on 20 chicks. All the 20 chicks were treated alike in all respects except the feeding treatments :
| Feed | Gain in weight (in gm) | | | | |
|---|---|---|---|---|---|
| $A$ | 55 | 49 | 42 | 21 | 52 |
| $B$ | 61 | 112 | 30 | 89 | 63 |
| $C$ | 42 | 97 | 81 | 95 | 92 |
| $D$ | 169 | 137 | 169 | 85 | 154 |
Test whether all the feeds are equally effective at $\alpha=0.05$ level of significance. (Refer the table enclosed at the end)
- Suppose the random variables $X_1,X_2,X_3$ have covariance matrix
$$\Sigma=\begin{bmatrix}2&0.5&-1\\0.5&1&0.5\\-1&0.5&2\end{bmatrix}$$
Find the first principal component vector and its variance.
- Let $X_1,X_2,X_3$ be uncorrelated random variables having common variance $\sigma^2$, and $E(X_1)=\theta_1+2\theta_2$, $E(X_2)=2\theta_1+\theta_2$ and $E(X_3)=2\theta_1+4\theta_2$.
(i) Obtain the normal equations and the least square estimates $\hat\theta_1$ and $\hat\theta_2$.
(ii) Find $V(\hat\theta_1)$, $V(\hat\theta_2)$ and $\mathrm{cov}(\hat\theta_1,\hat\theta_2)$.
- (i) Let $X$ be a vector random variable with the following covariance matrix :
$$\begin{bmatrix}1&\theta&\frac{\theta}{2}\\ \theta&1&-\frac{\theta}{2}\\ \frac{\theta}{2}&-\frac{\theta}{2}&1\end{bmatrix}$$
$Y_1=X_1+2X_2+X_3$ and $Y_2=3X_2-X_3$
For what value of $\theta$ are $Y_1$ and $Y_2$ uncorrelated? Also, find the variance of $Y_1$. [5]
(ii) $X$ and $Y$ are random vectors with $\mathrm{cov}(X)=\Sigma_{11}$, $\mathrm{cov}(Y)=\Sigma_{22}$ and $\mathrm{cov}(X,Y)=\Sigma$ such that
$$\Sigma=\begin{pmatrix}\Sigma_{11}&\Sigma_{12}\\ \Sigma_{21}&\Sigma_{22}\end{pmatrix}=\begin{bmatrix}1&0.4&0.5&0.6\\0.4&1&0.3&0.4\\0.5&0.3&1&0.2\\0.6&0.4&0.2&1\end{bmatrix}$$
with $\Sigma_{11}^{-\frac{1}{2}}=\begin{bmatrix}1.0681&-0.2229\\-0.2229&1.0681\end{bmatrix}$ and $\Sigma_{22}^{-1}=\begin{bmatrix}1.0417&-0.2083\\-0.2083&1.0417\end{bmatrix}$. Find the first canonical correlation. [5]
- Suppose $X$ is $p$-variate multivariate normal random vector and $Y$ is also a $p$-variate multivariate normal random vector. $X$ and $Y$ are independent with $E(X)=\mu_x$ and $E(Y)=\mu_y$ and with $V(X)=V(Y)=\Sigma$. If $n_1$ observations are obtained from $X$ and $n_2$ observations are obtained from $Y$, derive the test statistic for testing $H_0:\mu_x=\mu_y$ against $H_1:\mu_x\ne\mu_y$. Mention the distribution of the test statistic under null hypothesis.
- Find out the missing entries denoted by $*$ in the following ANOVA table of a Latin square design :
| Source of variation | d.f. | Sum of squares (SS) | Mean sum of squares | $F_{cal}$ |
|---|---|---|---|---|
| Rows | 3 | 259.54 | * | * |
| Columns | * | * | * | * |
| Treatments | * | 1372.12 | * | * |
| Error | 6 | * | 26.06 | |
| Total | * | 1943.07 | | |
Stating the hypotheses, give your conclusions at $\alpha=0.05$ level of significance. Also, calculate the relative efficiency of the above design (LSD) with respect to RBD and CRD.
(Refer the table enclosed at the end)
- (i) Show that systematic sampling is more precise than simple random sampling if the variance within the systematic samples is more than the total variation in the population. [10]
(ii) When do we make use of ratio estimator? Also, state the conditions under which the ratio estimate is optimum. [5]
- Differentiate between partial and complete confounding, and write the ANOVA of a partially confounded $2^3$-experiment.
- The following is a sample of $(X,Y,Z)$ of size 5 :
$(2,6,3),\ (3,6,4),\ (4,8,7),\ (5,10,9),\ (6,10,7)$
Find the partial correlation $r_{xy\cdot z}$ and multiple correlation coefficient $R_{y\cdot xz}$.
- The surface finish of products produced in a machine shop is suspected to be affected by a factor 'Operator' and another factor 'Shift'. The data of this experiment with two replications in different treatment combinations are given below :
| Operator $(A)$ | | Shift $(B)$ 1 | Shift $(B)$ 2 |
|---|---|---|---|
| 1 | rep 1 | 65 | 20 |
| 1 | rep 2 | 70 | 40 |
| 2 | rep 1 | 30 | 50 |
| 2 | rep 2 | 35 | 40 |
Test whether Operators, Shifts or their joint effects are significantly different at $\alpha=0.05$.
(Refer the table enclosed at the end)
- (i) In which situations is cluster sampling preferred? Obtain the variance of sample mean of simple random sample (without replacement) of $n$ clusters from a population of $N$ clusters each having $M$ elements. [10]
(ii) What are non-sampling errors? Write in brief about the measures to control these errors. [5]