TEST CODE: PSA (Objective type) and PSB (Short answer type) 2025

SYLLABUS

Mathematics

Arithmetic, geometric and harmonic progressions. Trigonometry. Two dimensional coordinate

geometry: Straight lines, circles, parabolas, ellipses and hyperbolas.

Elementary set theory. Functions and relations. Elementary combinatorics: Permutations and
combinations, Binomial and multinomial theorem.

Theory of equations.

Complex numbers and De Moivre’s theorem.

Vector spaces. Determinant, rank, trace and inverse of a matrix. System of linear equations.
Eigenvalues and eigenvectors of matrices.

Limit and continuity of functions of one variable. Differentiation and integration. Applications of
differential calculus, maxima and minima.

Statistics and Probability

Notions of sample space and probability. Combinatorial probability. Conditional probability and
independence. Bayes Theorem. Random variables and expectations. Moments and moment
generating functions. Standard univariate discrete and continuous distributions. Distribution of
functions of a random variable. Distribution of order statistics. Joint probability distributions.
Marginal and conditional probability distributions. Multinomial distribution. Bivariate normal
and multivariate normal distributions.

Sampling distributions of statistics. Statement and applications of Weak law of large numbers and
Central limit theorem.

Descriptive statistical measures. Pearson product-moment correlation and Spearman’s rank

correlation. Simple and multiple linear regression.

Elementary theory of estimation (unbiasedness, minimum variance, sufficiency). Methods of

estimation (maximum likelihood method, method of moments). Tests of hypotheses (basic
concepts and simple applications of Neyman-Pearson Lemma). Confidence intervals. Inference
related to regression.

Basic experimental designs such as CRD, RBD, LSD and their analyses. ANOVA. Elements of
factorial designs. Conventional sampling techniques (SRSWR/SRSWOR) including stratification.
 PSA-2024
Notations and Abbreviations

The following are used throughout the question paper.

R Set of real numbers

E Expectation
V Variance
Cov Covariance
⌊x⌋ Largest integer less than or equal to x
iid independent and identically distributed
pmf probability mass function
cdf cumulative distribution function
MLE Maximum likelihood estimator

1. Let n be an odd positive integer. For any permutation σ of

{1, 2, 3, . . . , n}, consider the product
n
∏
P (σ) = (j − σ(j)) = (1 − σ(1))(2 − σ(2)) · · · (n − σ(n)).
j=1

Then

(A) P (σ) = 0 for all σ.

(B) P (σ) is odd for all σ.
(C) P (σ) is even for all σ.
(D) None of the above.

1
2. Let a, b, c be three distinct non-zero real numbers. The expres-
sion
(x − b)(x − c) (x − c)(x − a) (x − a)(x − b)
a +b +c −x
(a − b)(a − c) (b − c)(b − a) (c − a)(c − b)

is equal to zero for

(A) exactly 2 values of x. (B) exactly 3 values of x.

(C) infinitely many values of x. (D) no value of x.

3. Equation of the straight line passing through the centres of the

two circles x2 + y 2 − 4x = 0 and 2x2 + 2y 2 + 4x − 5y = 0 is

(A) (x − 2)y + (x − 1)(y + 5/4) = 0.

(B) 4(x + 1)y − (x − 2)(4y − 5) = 0.
(C) (x + 1)y − (x − 2)(y − 5) = 0.
(D) 2x + y = 3.

4. Let p be a positive real and T be the set of all triangles that

have perimeter p. A triangle in T has maximum area if its
angles are

(A) 60◦ , 60◦ and 60◦ (B) 90◦ , 45◦ and 45◦

(C) 120◦ , 30◦ and 30◦ (D) 50◦ , 60◦ and 70◦

5. For x ≥ 0, let f (x) denote the non-negative square root of x.

The expression
( ) ( )
f sin4 x + 4 cos2 x − f cos4 x + 4 sin2 x

equals

(A) sin 2x (B) cos x (C) sin x (D) cos 2x

2
6. Let S be the set of all 3 × 3 matrices A such that A2 = 0 and
the entries of A belong to {0, 1}. Then the number of elements
in the set S is

(A) 9 (B) 12 (C) 15 (D) 6

7. Let A be an m × n matrix and let P be an n × n idempotent

matrix. If P ̸= I, then which of the following statements is true?

(A) rank (AP ) < m

(B) rank (AP ) < min{m, n}
(C) rank (AP ) = min{m, n}
(D) rank (AP ) < n

8. Let a1 , a2 , a3 , a4 be four distinct positive numbers. Consider the

4 × 4 matrix B = ((bij )) where bij = ai + aj for i, j = 1, 2, 3, 4.
Then rank (B) is

(A) 1 (B) 4 (C) 2 (D) 3

9. Let S be the set of vectors x = (x1 , x2 , x3 ) in R3 such that

xi ∈ {1, 2, 3, 4, 5} for all i and not all xi ’s are equal. For how
many ordered pairs (x, y) ∈ S×S are x and y linearly dependent
vectors in R3 ?

(A) 137 (B) 125 (C) 126 (D) 132

3
10. Let f : [0, 1] → [0, 1] be a function such that

|f (x) − f (y)| < |x − y|

for all x ̸= y ∈ [0, 1]. Consider the following two statements:

(i) There exists at least one point a ∈ [0, 1] such that f (a) = a.
(ii) There exists at most one point a ∈ [0, 1] such that f (a) = a.

Choose the correct option.

(A) Both (i) and (ii) are true.

(B) (ii) is true but (i) is false.
(C) (i) is true but (ii) is false.
(D) Both (i) and (ii) are false.

11. The number of subsets {a, b, c} of {1, 2, . . . , 24} such that a, b

and c are in arithmetic progression is

(A) 66 (B) 132 (C) 276 (D) 138

12. Let f and g be two real-valued differentiable functions on R. Let

h : R → R be defined as h(x) = max{f (x), g(x)}. Consider the
following two statements:

(i) If h is differentiable then f ≥ g or f ≤ g.

(ii) If f ≥ g or f ≤ g then h is differentiable.

Choose the correct option.

(A) (ii) is true but (i) is false.

(B) h is always differentiable.
(C) Both (i) and (ii) are true.
(D) (i) is true but (ii) is false.

4
13. Let X(1) ≤ · · · ≤ X(n) denote the order statistics from a random
sample of size n, drawn from an exponential distribution with
mean equal to 1. For z > y > 0, the conditional density of X(n)
given X(1) = y is

(A) n exp{−(z − y)} (1 − exp{−(z − y)})n−1

(B) (n − 1) exp{−(z − y)} (exp{−(z − y)})n−2
(C) (n − 1) exp{−(z − y)} (1 − exp{−(z − y)})n−2
(D) exp{−(z − y)}

14. Let F and G be two cdfs and let α, β > 0. Which of the following
is not a cdf?
(A) (αF (αx) + βG(βx)) /(α + β) (B) F (αx)G(βx)

(C) (F (αx) + G(βx)) /(α + β) (D) (F (αx))α (G(βx))β

15. Suppose X1 , X2 and X3 are three random variables such that

X1 , (X1 + X2 )/2 and (X1 + X2 + X3 )/3 are iid standard normal
random variables. Based on the above information choose the
correct statement from the options given below.

(A) This is not possible.

(B) X1 and X3 are independent normal random variables.
(C) X1 , X2 and X3 are identically distributed normal random
variables.
(D) X1 , X2 and X3 are independent normal random variables.

5
16. Suppose E1 , . . . , Em are disjoint events satisfying 0 < P(Ej ) < 1,
m
∪
for each j = 1, . . . , m. Let E = Ej . Assume 0 < P(E) < 1.
j=1
Let A be any other event and define p = P(A | E). Which of
the following statements is true?
∑
m
(A) p = P(A | Ej ) P(E)
j=1
∑
m
(B) p = P(A | Ej ) P (Ej | E)
j=1
∑
m
(C) p = P(A | Ej )
j=1
∑
m
(D) p = P(A | Ej ) P(Ej )
j=1

17. Suppose A is the set of all 6-digit numbers formed using each
of the digits 1, 2, 3, . . . , 6 exactly once. If a number X is chosen
uniformly at random from A, then what is the probability that
X is divisible by 6 but not by 9 or 11?

(A) 1/2 (B) 1/6 (C) 1/11 (D) 2/33

18. Suppose X is a random variable taking values an with probability

(1 − p) pn for n ≥ 0, where a ̸= 0 is a real number and 0 < p < 1.
Consider the following statements.

(i) E[X] is finite for all values of a and p.

√
(ii) V(X) is finite whenever |a| < 1/ p.

Which of the following statements is correct?

(A) Only (ii) is true.

(B) Both (i) and (ii) are true.
(C) Only (i) is true.
(D) Neither (i) nor (ii) is true.

6
19. A coin with success probability 1/4 is independently tossed
N times and the outcomes are recorded. Given that the total
number of successes is m, what is the probability that the first
k tosses are successes, where k ≤ m?

k
(A) (1/4)k (3/4)N −k (B)
m )
(
m(m − 1) · · · (m − k + 1) m
(C) (D) (1/4)k (3/4)m−k
N (N − 1) · · · (N − k + 1) k

20. Suppose X and Y are two independent random variables both

following Poisson distribution with parameter 10. What is the
value of E(X − Y )2 ?

(A) 30 (B) 10 (C) 40 (D) 20

21. Let M denote the region bounded by the parallelogram with

vertices at (0, 1), (2, 1), (0, −1) and (−2, −1). Let (X, Y ) denote
the x and y co-ordinates of a randomly chosen point from M .
Consider the folowing statements.

(i) The marginal distribution of X is uniform on (−2, 2).

(ii) E(XY ) = 0.
(iii) X and Y are independent.

Which of the following statements is correct?

(A) Only (ii) and (iii) are true.

(B) Only (ii) is true.
(C) Only (i) is true.
(D) None of the statements is true.

7
22. Suppose {Xi : i ≥ 1} are iid Poisson random variables with
standard deviation 2. Then as n → ∞
n
1∑
Xi (Xi − 2)
n i=1

converges in probability to

(A) 16 (B) 8 (C) 12 (D) 2

23. For x1 , x2 , . . . , x2n ∈ [−1, 1], where the xi ’s are not necessarily
distinct, define
2n 2n
1 ∑ 2 1 ∑
m= xi and s = (xi − m)2 .
2n i=1 2n i=1

Let s2max denote the maximum possible value that s2 can take
over all such x1 , x2 , . . . , x2n .
Based on the above information, choose the FALSE statement
from the options given below.

(A) s2 ∈ [0, 1] for all such x1 , x2 , . . . , x2n .

(B) m ∈ [−1, 1] for all such x1 , x2 , . . . , x2n .
(C) If m = 0 then the corresponding s2 = s2max .
(D) If s2 = s2max then the corresponding m = 0.

8
24. Consider a random variable with pmf given by


 p if x = 1,




3p if x = 2,
P (X = x) =

 2p if x = 3,




1 − 6p if x = 4,

where the unknown parameter p ∈ [1/20, 1/10]. A random

sample of size 4 from this distribution yielded the observations
x1 = 2, x2 = 3, x3 = 2, x4 = 1. What is the maximum likelihood
estimate of p based on these observations?
√
(A) 1/10 (B) 1/ 4 18 (C) 1/18 (D) 1/20

25. Suppose that we have observations (xi , yi , zi ), i = 1, . . . , n. Let

βbx be the estimated coefficient obtained by least-squares linear
bx , γ
regression of z on x, without intercept. Let γ by be the esti-
mated coefficients obtained by similarly regressing z on x and
y jointly, also without intercept. Which of the following state-
ments is true?
∑
n ∑
n
(A) (zi − βbx xi )2 > (zi − γ
bx xi )2 .
i=1 i=1
∑n ∑n
(B) (zi − βbx xi )2 < (zi − γ by yi )2 .
bx xi − γ
i=1 i=1
∑n ∑n
(C) (zi − βbx xi )2 ≤ by yi )2 .
(zi − γ
i=1 i=1
∑n ∑n
(D) (zi − βbx xi )2 ≥ (zi − γ by yi )2 .
bx xi − γ
i=1 i=1

9
26. Suppose that we have observations (x1 , y1 ), . . . , (xn , yn ). We fit
a simple linear regression of y on x, with intercept, using these
observations via the least-squares method. Let ei , i = 1, . . . , n,
be the residuals of the fitted regression model. Consider the
following statements.

1
∑
n
(i) n
ei = 0.
i=1

1
∑n
1
∑
n
1
∑
n
(ii) n
(ei − ē)(xi − x̄) = 0 where x̄ = n
xi and ē = n
ei .
i=1 i=1 i=1

1
∑n
(iii) n
ei xi = 0.
i=1

Which of the following statements is correct?

(A) (i), (ii) and (iii) are all true.

(B) Only (ii) and (iii) are true.
(C) Only (i) and (ii) are true.
(D) Only (i) and (iii) are true.

27. Let X1 , . . . , Xn be iid N (µ, σ 2 ) random variables, where µ ∈ R

and σ > 0 are unknown parameters. Let Φ(·) and Φ−1 (·) denote
the cdf of N (0, 1) and its inverse respectively. Define
n n
1∑ 1∑ 2
X̄n = Xi and Sn2 = (Xi − X̄n ) .
n i=1 n i=1

For α ∈ (0, 1), let ξα (µ, σ) be a parametric function satisfying

P(X1 ≤ ξα (µ, σ)) = α.

What is the MLE of ξα (µ, σ)?

( )
(A) Φ X̄n + αSn .
(B) X̄n + αSn .
(C) X(⌊nα⌋) , the ⌊nα⌋-th order statistic of X1 , . . . , Xn .
(D) X̄n + Φ−1 (α) Sn .

10
28. A random sample is obtained from a N (µ, 1) distribution, where
µ ∈ R is an unknown parameter. Using this sample, a most
powerful (MP) test ϕ of size α = 1/2 is constructed for testing
H0 : µ = 0 against H1 : µ = 1. Let β denote the power of this
test. Which of the following statements is correct?

(A) ϕ is an MP test of size < 1/2 with power β for testing

H0′ : µ = −1 against H1′ : µ = 1.
(B) ϕ is an MP test of size 1/2 with power > β for testing
H0′ : µ = 0 against H1′ : µ = 1/2.
(C) ψ = (1 − ϕ) is not an MP test of size (1 − β) for testing
H0′ : µ = 1 against H1′ : µ = 0.
(D) ϕ is an MP test of size 1/2 for testing H0′ : µ = 0 against
H1′ : µ = a for some a ∈ (−∞, 0).

29. Let λ > 0. Let X1 , X2 , X3 be independent Poisson random vari-

ables with parameters λ, 2λ and λ/2 respectively. Which of the
following statistics is sufficient for λ?

(A) T = 2X1 + 4X2 + X3

(B) T = 2X1 + X2 + 4X3
(C) T = max{2X1 , 4X2 , X3 }
(D) T = X1 + X2 + X3

11
30. Let N ≥ 3 be an integer. A box contains N balls numbered
{1, 2, . . . , N }. A simple random sample s of size n (1 < n < N )
is selected with replacement from the box. For i = 1, . . . , N ,
define 
1 if the i-th ball is selected in s,
Wi =
0 otherwise.

Which of the following statements is true?

(A) Cov(W1 , W2 ) < 0.

∑
N
(B) Wi = n.
i=1

(C) E(Wi ) = n/N for each i = 1, . . . , N .

(D) Cov(W1 , W2 ) = 0.

12
 PSB-2024
Notations and Abbreviations
The following are used throughout the question paper.

R Set of real numbers

R n
n - dimensional Euclidean space
Q Set of rational numbers
E Expectation
V Variance
P Probability
Cov Covariance
T
A Transpose of the matrix A
i.i.d. independent and identically distributed
p.d.f. probability density function

1. Let m ̸= n and let A = ((aij )) be an m × n real matrix such that

AT A = I. Let b1 , . . . , bm ∈ R. Prove that the system of linear
equations in the unknowns x1 , . . . , xn

a11 x1 + a12 x2 + ··· + a1n xn = b1 ,

a21 x1 + a22 x2 + ··· + a2n xn = b2 ,
.. .. .. ..
. . . .
am1 x1 + am2 x2 + · · · + amn xn = bm

has at most one solution.

2. Use the identity n2 + (2n + 1) = (n + 1)2 to prove that the set

S = {(x, y) ∈ Q × Q : x2 + y 2 = 1} has infinitely many elements.

1
3. A box contains 7 indistinguishable green balls, 5 indistinguish-
able white balls and 6 indistinguishable black balls. Suppose
balls are drawn using simple random sampling with replacement
until balls of all colours are obtained. Let N denote the mini-
mum number of draws needed to achieve this.

(a) For each nonnegative integer n, calculate P(N > n).

(b) Using (a) or otherwise, compute E(N ).

4. Suppose (X, Y ) is uniformly distributed over the region

{ }
(x, y) ∈ R2 : 0 < y < 1, |x| < 1 − y .

Calculate the p.d.f. of |X| + Y .

5. Let X1 , X2 , X3 be i.i.d. N (0, 1) random variables. Define, for

1 ≤ i ̸= j ≤ 3, 
1 if X > X ,
i j
Wij =
0 otherwise.

(a) Calculate E(Wij ) and V(Wij ) for all 1 ≤ i ̸= j ≤ 3.

(b) Calculate Cov(W12 , Wij ) for all 1 ≤ i ̸= j ≤ 3.

6. Assume that the length, in minutes, of a phone-call of an in-

dividual follows an exponential distribution with an unknown
parameter λ > 0 with density function f (x) = λ e−λx , x > 0.
However, when the phone company calculates the length of a
phone-call, it always considers the nearest integer greater than
or equal to the actual length. For example, a 22.09 minutes long
phone-call will have a call-length of 23 minutes in the phone
company records. Suppose you have the data on the lengths
of n independent phone-calls T1 , T2 , . . . , Tn of that individual as
reported by the phone company. Based on this data, compute
the maximum likelihood estimator of λ.

2
7. Suppose that X1 , X2 are i.i.d. U (0, θ) random variables. We
want to test H0 : θ = 1 against H1 : θ = θ1 where θ1 > 1 is fixed.
For this, we adopt two testing strategies.

Test 1: Reject H0 if X1 > 0.95.

Test 2: Reject H0 if max{X1 , X2 } > κ.

(a) Find κ such that both the tests have the same size.
(b) Find the powers of the two tests. For Test 2, use the value
of κ obtained in part (a).
(c) Which of the two tests would you prefer and why?

8. Suppose we have paired observations (X1 , Y1 ), . . . , (Xn , Yn ). A

statistician thinks that the following parametric model may be
appropriate for this data:
For unknown odd integers α, β and unknown ρ ∈ (−1, 1),
the pairs (Xiα , Yiβ ) are i.i.d. bivariate normal with parameters
(0, 0, 1, 1, ρ).
Write down the likelihood function for this model.

9. Let Y1 , . . . , Y2n be i.i.d. Bernoulli(p) random variables, where

p ∈ (0, 1) is unknown. Consider the estimators
2n−1
∑ n
1 1∑
T1,n = Yi Yi+1 and T2,n = Y2i−1 Y2i .
2n − 1 i=1 n i=1

(a) Show that both estimators are consistent for p2 .

(b) Find lim (V(T1,n )/V(T2,n )).
n→∞
(c) For large n, which estimator would you prefer and why?

3
10. A city has N households numbered {1, . . . , N }. Let yi denote the
size of the i-th household. A simple random sample s1 of size
m is drawn without replacement from households {1, . . . , M }.
Another independent simple random sample s2 , also of size m, is
drawn without replacement from households {N −M +1, . . . , N }.
Assume m < M and M > N/2.

(a) Compute πi = P(i ∈ s1 ∪ s2 ), the probability that the

i-th household is included in either of the two samples, for
i = 1, . . . , N .
(b) Show that T is an unbiased estimator of the average house-
hold size in the city, where
( )
1 ∑ yi
T = .
N i∈s ∪s πi
1 2

4
 PSA-2023
1. A sequence of real numbers {an }n≥1 has a peak at n if an ≥ ak
for all k ≥ n. Consider the following statements.

(I) No sequence of real numbers can have only finitely many

peaks.
(II) No sequence of real numbers can have infinitely many
peaks.
(III) Any sequence of real numbers having finitely many peaks
must have the property that an ≥ 0 for all n greater than
some k.

Then

(A) only (I) is true

(B) none of (I), (II) and (III) are true
(C) both (II) and (III) are true
(D) only (III) is true

2. Let C denote the set of complex numbers and let Im(z) denote
the imaginary part of z ∈ C. Consider the set

S = {s ∈ R : there exists z ∈ C such that Im(z) ̸= 0 and s = z 2 +2z−1}.

Then,

(A) S ̸= R, but contains infinitely many elements

(B) S is a non-empty finite set
(C) S = R
(D) S is the empty set

1
3. For a set S, let S c denote the complement of S. Also, for two
sets P and Q, let P \ Q = P ∩ Qc . Let A, B1 , B2 and B3 be four
sets. Which of the following statements is NOT true?

(A) (A ∪ B1 ∪ B2 ∪ B3 )c = Ac ∩ B1c ∩ B2c ∩ B3c

(B) (A \ B1 ) \ (B2 ∪ B3 ) = A \ (B1 ∪ B2 ∪ B3 )
(C) A \ (B1 ∪ B2 ∪ B3 ) = (A \ B1 ) ∪ (A \ B2 ) ∪ (A \ B3 )
(D) A ∩ (B1 ∪ B2 ∪ B3 ) = (A ∩ B1 ) ∪ (A ∩ B2 ) ∪ (A ∩ B3 )

4. For a complex number z, let z̄ be its complex conjugate. Then

the equation
z z̄ 2 + z 2 z̄ = 0

has

(A) exactly three roots

(B) exactly two roots
(C) infinitely many roots
(D) only real roots

5. Let f (x) = x2 + (2a + 1)x + (a2 + 2). The number of values of a

for which one of the roots of the equation f (x) = 0 is twice the
other root is

(A) more than 2 (B) 2 (C) 1 (D) 0

2
6. Let A be a finite set of real numbers having m (≥ 2) elements.
Define a function f : R → R, given by

f (x) = min{|a − x| : a ∈ A}.

Then,

(A) f is continuous everywhere

(B) f is continuous only at finitely many points
(C) f is discontinuous everywhere
(D) f has m discontinuities

7. Let A be the set of functions f : R → R for which |f (x)−f (y)| ≤

2|x − y|2 for all x, y ∈ R and f (0) = 0. Then, for any f ∈ A,

(A) the functions g(x) = P (f (x)) ∈ A for every polynomial P

(B) the function g(x) = x + f (x) ∈ A
(C) the function g(x) = xf (x) ∈ A
(D) the function g(x) = ef (x) ∈ A

8. The number of values of a for which the three lines

2x + y − 1 = 0, ax + 3y − 3 = 0, 3x + 2y − 2 = 0

are concurrent is

(A) more than 2 (B) 1 (C) 0 (D) 2

3
 9. For a non-constant geometric progression for which the second
term is 2 and the common ratio is an integer, the 10th, 20th and
30th terms are in arithmetic progression. Then, the fourth term
is

(A) −2 (B) −4 (C) 4 (D) 2

√ √
x + 8 − 8x + 1
10. lim √ √ equals
x→1 5 − x − 7x − 3

2 1 7
(A) does not exist (B) (C) (D)
3 2 12

11. The rank of the matrix

 
0 1 0 0 0
 
0 1 2 0 0
 
0 0 0 3 0
 .
 
0 1 0 a 0
1 0 0 0 b

(A) depends on the values of both a and b

(B) is independent of the values of both a and b
(C) depends on the value of a but not on the value of b
(D) depends on the value of b but not on the value of a

4
 !
a 1
12. If the matrix A = has 1 as an eigenvalue, then the
2 3
determinant of A is

(A) 5 (B) 2 (C) 4 (D) 3

13. Let a < 500 be a positive integer. Consider a box containing

balls numbered a, a+1, . . . , 500. Suppose that the ball numbered
x is picked with probability
2xa
for x = a, a + 1, . . . , 500.
(500 + a)(500 − a + 1)

Then the value of a is

(A) 251 (B) 1 (C) 499 (D) 2

14. Let f (x) = ax + b for some a, b ∈ R. Define fn (x) inductively

by setting
f1 (x) = f (x)

and
fn+1 (x) = f (fn (x)) for n > 1.

If f7 (x) = 128x + 381, then ab equals

1 1
(A) (B) 32 (C) (D) 8
8 32

5
15. Let n = aaaaaaaaabcd be a 12-digited number divisible by 45
where the digits a, b, c, d are not necessarily distinct and a ̸= 0.
How many such numbers are there?

(A) 216 (B) 207 (C) 189 (D) 198

16. Let a, b and c be the sides of a triangle such that c2 = a2 +b2 −ab.
Then which of the following is always true?

(A) a ≤ c and b ≤ c
(B) a ≤ c ≤ b or b ≤ c ≤ a
(C) c ≤ a and c ≤ b
(D) None of the above

17. Let X be a discrete random variable and Y be a continuous

random variable which is independent of X. Let U = X + Y
and V = XY . Choose the correct statement from the options
given below.

(A) Both U and V are continuous random variables

(B) U is a continuous random variable but V need not be
(C) U is a discrete random variable but V need not be a con-
tinuous random variable
(D) U is a discrete random variable and V is a continuous ran-
dom variable

6
18. Suppose that the sample mean and sample standard deviation
for a set of n observations x1 , x2 , . . . , xn are m and s (> 0),
respectively. These values are updated to m1 and s1 after one
more observation xn+1 is added to the data set.
Based on the above information, choose the correct statement
from the options given below.

(A) If m1 = m then s1 < s

(B) If m1 = m then s1 = s
(C) If m1 < m then s1 = s
(D) If m1 < m then s1 < s

19. Suppose that X1 , X2 , . . . , Xn are independent and identically

distributed random variables with probability density function

λe−λx if x ≥ 0,
fλ (x) =
0 otherwise,

where λ > 0. Let Y = X1 + X2 + · · · + Xn . Then the conditional

distribution of Xn given Y = 1 is

(A) uniform on (0, 1)

(B) exponential with mean 1
(C) beta with parameters n − 1 and 1
(D) beta with parameters 1 and n − 1

7
20. Let X1 , X2 , . . . be a sequence of random variables such that
1
E(Xi ) = 1, Var(Xi ) = 1 for all i and Cov(Xi , Xj ) = for
n
2
1X
all i ̸= j. Let Zn = Xi . Then, lim Var(Zn ) equals
n i=1 n→∞

1 1
(A) (B) (C) 0 (D) 1
2 4

21. Suppose that P(A|B) = 0.4 and P(Ac |B c ) = 0.6. Then, the two
equations are sufficient to find

(A) neither P(A) nor P(B)

(B) both P(A) and P(B)
(C) P(B) but not P(A)
(D) P(A) but not P(B)

22. Let X1 , . . . , Xn be independent and identically distributed nor-

mal random
√ variables with mean 0 and variance σ 2 > 0. Define
n Xn P P
Tn = where X n = n1 ni=1 Xi and Sn2 = n1 ni=1 Xi2 .
Sn
Then, the distribution of Tn is

(A) Student’s t with n degrees of freedom

(B) Student’s t with (n − 1) degrees of freedom
(C) normal with mean 0 and variance 1
(D) None of the above

8
 !
X −Y
23. Consider a matrix M = where X and Y are indepen-
Y X
dent standard normal random variables. Then the probability
that M is a non-singular matrix is

1 1
(A) 0 (B) 1 (C) √ (D)
2 2

24. Let (U, V ) be a point chosen uniformly at random from the unit
circle {(u, v) ∈ R2 : u2 + v 2 = 1}. Then Var(U ) is

1 1 1
(A) (B) (C) 1 (D)
3 2 4

25. Suppose that we choose 2 cards simultaneously at random from

a deck of 20 cards numbered 1, 2, . . . , 20. What is the probability
that the smaller of the two numbers divides the larger?

36 46 56 66
(A) (B) (C) (D)
190 190 190 190

9
26. Let X be a random variable with probability density function
displayed in the following graph.

0.20

0.15
Density f(x)

0.10

0.05

0.00

−10 −5 0 5 10

Match the following random variables with their respective prob-

ability density functions
X
(i) X + 3 (ii) X − 3 (iii) 2X (iv) .
2

(a) (b)
0.4

0.3

0.2

0.1
Density f(x)

0.0

(c) (d)
0.4

0.3

0.2

0.1

0.0

−10 −5 0 5 10 −10 −5 0 5 10

(A) (i)–(d), (ii)–(c), (iii)–(b), (iv)–(a)

(B) (i)–(d), (ii)–(c), (iii)–(a), (iv)–(b)
(C) (i)–(c), (ii)–(d), (iii)–(a), (iv)–(b)
(D) (i)–(c), (ii)–(d), (iii)–(b), (iv)–(a)

10
27. Suppose that we want to fit the regression model

y = β1 x + β2 x2 + ϵ

to 10 pairs of observations (x1 , y1 ), . . . , (x10 , y10 ) where xi ’s take

two values, 0 and 1. Which of the following can be estimated
using the method of least squares?

(A) Both β1 and β2

(B) β1 but not β2
(C) β2 but not β1
(D) β1 + β2

28. Consider a bivariate sample (X1 , Y1 ), . . . , (X9 , Y9 ) where Xi = i

for i = 1, 2, . . . , 9. The least squares regression line for this
dataset is obtained as y = 3 + 2x. Later it turns out that Y5 was
recorded wrongly. When the revised regression line is obtained
which of the following are possible?

(A) Intercept can change but slope cannot

(B) Slope can change but intercept cannot
(C) Both intercept and slope can change
(D) Neither intercept nor slope can change

11
29. Assume X1 , . . . , Xn are independent and identically distributed
N (µ, 1) random variables with µ ∈ R. We want to test H0 :
µ = 0 versus H1 : µ ̸= 0. Consider the following two one-sided
testing problems

H0,A : µ = 0 versus H1,A : µ > 0

and H0,B : µ = 0 versus H1,B : µ < 0.

Let ϕA,η (x) and ϕB,η (x) denote the most powerful tests of size
η ∈ (0, 1) for H0,A and H0,B , respectively. Then, for testing H0
versus H1 ,

(A) ϕ(x) = ϕA,η (x) + ϕB,η (x) is a test of size η

(B) ϕ(x) = ϕA,η (x) ϕB,η (x) is a test of size 2η
1 η
(C) ϕ(x) = max{ϕA,η (x), ϕB,η (x)} is a test of size 2
2
(D) ϕ(x) = ϕA,η (x) + ϕB,η (x) is a test of size 2η

12
30. Let ϕ denote the probability density function of the standard
normal distribution. Let fθ , for θ ∈ {0, 1}, be defined as

ϕ(x) if θ = 0,
fθ (x) =

 1 ϕ x−1 if θ = 1.
2 2

Assume that X1 , . . . , Xn are independent and identically dis-

tributed from the density fθ (x). Which of the following is a
sufficient statistic for θ?
n
X
(A) Xi
i=1
n n
!
X X
(B) Xi , Xi2
i=1 i=1
n
X
(C) Xi2
i=1
n n
!
X X
(D) Xi , 1(|Xi | ≥ 2)
i=1 i=1

13
 PSB-2023
GROUP A

1. Let An = ((aij )) be the n × n matrix defined by



 0 if |i − j| > 1,


aij = 1 if |i − j| = 1,



2 if i = j.

Find the determinant of An for n ≥ 1.

2. Consider a random permutation of the eight numbers 1, 2, . . . , 8.

Compute the probability that no two adjacent numbers in this
permutation have a product which is odd. Give justification for
your computations.

3. Identify, with justification, all cumulative distribution functions

F that satisfy F (x) = F (x2023 ) for every x ∈ R.

GROUP B

4. A six-faced fair die is rolled repeatedly till 1 appears. Let X

be the total number of rolls and Y be the number of times 6
appeared in these X rolls.

(a) Find E[Y |X = x].

(b) Find E[Y ].

1
5. Suppose X1 , . . . , Xn (n ≥ 2) are independent and identically
distributed observations from a distribution having probability
density function

e−(x−θ) if x ≥ θ,
fθ (x) =
0 if x < θ,

where θ ∈ R. Let
Z ∞
ψ(θ) = fθ (x)dx.
1

Define θbn = min{X1 , . . . , Xn }. Consider ψ(θbn ) as an estimator

of ψ(θ) and let Bn (θ) denote the associated bias.

(a) Show that Bn (θ) > 0 for every θ < 1 and Bn (θ) = 0 for
every θ ≥ 1.
(b) Show that lim Bn (θ) = 0 for every θ < 1.
n→∞

6. Suppose that two observations X1 and X2 are drawn at ran-

dom from a distribution with the following probability density
function

 1

 if 0 ≤ x ≤ θ or 2θ ≤ x ≤ 3θ,
 2θ
fθ (x) =



0 otherwise,

where θ > 0. Determine the maximum likelihood estimator of θ

for each of the following observed values of X1 and X2 .

(a) X1 = 7 and X2 = 9.
(b) X1 = 4 and X2 = 9.
(c) X1 = 5 and X2 = 9.

2
7. Mr. X wants to compare five different varieties of wheat. He has
access to a piece of land which has been divided into 25 smaller
plots as shown in the figure below.

Mr. X has asked for your help in his venture.

(a) Give an allocation of the wheat varieties to the plots so

that you are able to compare them. Justify your allocation
rule. Clearly state all the assumptions you make.
(b) Once the crop is harvested, how would you analyze the data
and compare the different varieties of wheat?

3
8. Let X1 , . . . , Xn be independent and identically distributed
Bernoulli(θ) random variables where θ ∈ (0, 1). The maximum
likelihood estimator of θ is the sample mean X n . However, a
statistician feels that the sample size n is too small, and decides
to increase the sample size. In order to do so, he records the
observed values of the data points, {x1 , . . . , xn }, and then se-
lects a random sample of size m = kn with replacement from
{x1 , . . . , xn }, where k is a positive integer. The values in the
observed sample are recorded as {Y1 , . . . , Ym }. The statistician
proposes the new estimator T = 12 (Y m + X n ), where Y m is the
sample mean of Y1 , . . . , Ym .

(a) Show that T is unbiased for θ.

(b) Is T a better estimator of θ than X n ? Justify your answer.

4
9. To test whether the heights of siblings are correlated, a re-
searcher devised the following plan: She identified a random
sample of n families with at least two adult male children.
For the ith family, suppose that Xi and Yi are the heights
of the first and second male child, respectively. Assume that
(X1 , Y1 ), . . . , (Xn , Yn ) are independent bivariate normal random
vectors with parameters (µ, µ, σ 2 , σ 2 , ρ), where µ and σ 2 are
known from previous studies. She is interested in testing the
null hypothesis H0 : ρ = 0 against the alternative H1 : ρ = 0.5.
Unfortunately, due to a mistake in the questionnaire, she was
only able to observe (Ui , Vi ) for each i, where Ui = max(Xi , Yi )
and Vi = min(Xi , Yi ).

(a) Based on the observed sample, obtain the test statistic cor-
responding to the most powerful test of H0 against H1 .
(b) Find a critical value so that the size of the test converges
to 0.05 as n → ∞.

5
 PSA-2022
1. Let {xj }j≥1 be a sequence of positive numbers in geometric
progression. Let Pn denote the product of the first n terms
of {xj }. Which of the following statements is true for all n ≥ 1?

3
(A) P3n = P2n /Pn3
2
(B) P3n = P2n /Pn
1/2
(C) P3n = P2n Pn2
(D) P3n = Pn P2n

2. Suppose that f : R → R is a differentiable non-decreasing

function whose derivative f � is continuous. Fix a < b, and let

S = {x ∈ (a, b) : f � (x) = 0} .

Then, f (a) < f (b) if and only if

(A) S is a nonempty finite set.

(B) S is a countably infinite set.
(C) S is the empty set.
(D) S is a proper subset of (a, b).

1
 nx
3. Let x > 1. Then lim x−n n(−1)
n→∞

(A) equals ex .
(B) does not exist.
(C) equals +∞.
(D) equals 0.

4. Let f and g be monotonic functions on a closed interval [ a, b ].

Consider the following statements.

(I) f + g is monotonic on [ a, b ].
(II) f g is monotonic on [ a, b ].
(III) The maxima of f + g is attained at either a or b.
(IV) The maxima of f g is attained at either a or b.

Which of the above statements are correct?

(A) Only (I) and (III).

(B) Only (III) and (IV).
(C) Only (II) and (IV).
(D) None of (I), (II), (III), (IV).

2
5. What is the range of the function
� �
2 1 −x2
f (x) = x + e , x ∈ (−∞, ∞) ?
2
�1 � � � � � � �
(A) 2
, e−1/2 (B) 0, 12 (C) 0, e−1/2 (D) 0, e−1/2

6. Let (a, b) be a pair of real numbers. Consider the function



 −ax − b, if x ≤ −1,


f (x) = a2 x2 + b, if − 1 < x ≤ 1,



5x2 + 1, if x > 1.

For how many distinct pairs (a, b) is the function f continuous

everywhere?

(A) Exactly 1 (B) ∞ (C) Exactly 2 (D) 0

7. Consider a continuous function f : [−1, 1] → R which is

differentiable everywhere in the interval (−1, 1). Further,
suppose that f (−1) = − 12 �
and f (x) ≤ 1 for all x ∈ (−1, 1).
Which of the following statements is false?

(A) f (1) can be greater than 1 but not greater than 2.

(B) f (1) can be greater than 2.
(C) f (1) can be less than −2.
(D) f (1) can be less than −1 but not less than −2.

3
 8. For θ ∈ R with θ �= 0, what is the value of
� ∞
x 2 2
√ eθx−θ x /2 dx ?
−∞ 2π
√ √
1 e e 1
(A) (B) (C) (D)
θ|θ| θ2 θ|θ| θ2

9. Let S� be the� set of all real numbers a such that the matrix
1 −2
M= has no real eigenvalue. Then,
8 a

(A) S = (−9, 7).

(B) S = (−7, 9).
(C) The interval (−9, 7) is a proper subset of S.
(D) S is a proper subset of the interval (−7, 9).

10. Let M be the set of all n × n real matrices. Which of the

following statements is false?

(A) rank(AB) = rank(BA) for all A, B ∈ M.

(B) AB = I if and only if BA = I for all A, B ∈ M.
(C) det(AB) = det(BA) for all A, B ∈ M.
(D) trace(AB) = trace(BA) for all A, B ∈ M.

4
11. Let M (x) denote the matrix
 
x 0 1
 
0 1 0  ,
x 0 −x

where x is a real number. Consider the function

� �
F (x) = trace M (x) M (x)T

where AT denotes the transpose of the matrix A. Which of the

following statements is true?

(A) The range of F contains 1.

(B) The range of F has a finite lower bound.
(C) The range of F contains 0.
(D) The range of F has a finite upper bound.

12. Saurabh went to an amusement park with 9 tokens. There are

4 different rides available, with 3 of them costing 2 tokens each
and one costing 5 tokens. Saurabh can take each ride as many
times as he likes as long as he has tokens available. In how many
different ways can he choose the rides so that he has no tokens
left, if the order in which the rides were taken is ignored?

(A) 15 (B) 12 (C) 7 (D) 6

5
13. A six digit number is chosen at random. What is the probability
that at least two consecutive digits in the chosen number are
equal?

(A) 1 − (0.9)6 (B) (0.9)5 (C) (0.9)6 (D) 1 − (0.9)5

14. A multiple-choice test consists of 20 questions. Each question

has four choices, exactly one of which is correct. For each
question, a student is able to correctly identify one of the choices
as wrong, and chooses one answer at random from the remaining
three choices. The student will get a scholarship if she answers
at least 18 questions correctly. What is the probability that she
gets the scholarship?

(A) 1771/320 (B) 801/320 (C) 801/420 (D) 1771/420

15. The probability that a lie detector correctly determines whether

a person is lying or telling the truth is 0.8, independently of
whether the person is actually lying. Assume that people tell
the truth 95% of the times. Given that for a particular person
the lie detector says that the person is telling the truth, what is
the conditional probability that he is actually telling the truth?

36 76
(A) 37
(B) 77
(C) 0.95 (D) 0.8

6
16. Suppose that X has a uniform distribution on {1, 2, . . . , n}.
Further suppose that conditioned on the event {X = x}, Y
has a uniform distribution on {1, 2, ..., x}. Let

py (x) = P (X = x | Y = y).

Then for each y = 1, 2, . . . , n,

(A) py (x) > py (x + 1) for x = y, y + 1, . . . , n − 1.

(B) py (x) < py (x + 1) for x = y, y + 1, . . . , n − 1.
(C) py (x) = 1/n for x = 1, 2, . . . , n.
(D) py (x) = 1/(n − y + 1) for x = y, y + 1, . . . , n.

17. For n ≥ 3, let (X1 , Y1 ), (X2 , Y2 ), . . . , (Xn , Yn ) be independently

and identically distributed bivariate normal random variables
with parameters (0, 0, 1, 1, ρ). Suppose (i1 , . . . , in ) is a
random permutation of 1, 2, . . . , n such that all n! permutations
�
n
are equally likely. What is the expected value of T = X k Yi k ?
k=1

ρ √
(A) ρ (B) 1−ρ
(C) 0 (D) nρ

18. Suppose X follows a Poisson distribution with probability mass

function f satisfying f (3) = f (4). What is E[X(X − 1)(X − 2)]?

(A) 27 (B) 3 (C) 4 (D) 64

7
19. Let X be a random variable with the Student’s t-distribution
with 1 degree of freedom. Let M (t) be its moment generating
function. Which of the following statements is correct?

(A) M (t) is defined only on (−a, a) for some a ∈ (0, ∞).

(B) E(X α ) is defined for all α ∈ (0, 1) but not for α = 1.
(C) M (t) is defined for all real numbers t.
(D) E(X α ) is defined for all α ∈ (0, 2) but not for α = 2.

20. Which of the following distributions has a unique mode at 12 ?

1
(A) Beta distribution on (0, 1) with parameters 2
and 12 .
(B) Gamma distribution with scale parameter 1 and shape
parameter 12 .
(C) Gamma distribution with scale parameter 1 and shape
parameter 32 .
(D) Beta distribution on (0, 1) with parameters 1 and 1.

8
21. Suppose X1 , X2 , . . . , X100 are independent and identically
100
�
distributed as Bernoulli(p), where p ∈ (0, 1). If Y = Xi and
i=1
Z = min{X1 , X2 , . . . , X100 }, then

(A) Y and Z do not have the same mean but have the same
variance.
(B) Y and Z have the same mean but not the same variance.
(C) Y and Z neither have the same mean nor the same variance.
(D) Y and Z have the same distribution.

22. Suppose that X1 , X2 , . . . are independent and identically

distributed bounded random variables with common mean 3 and
common variance 5. For n ≥ 1, define

Yn = (X1 − X2 )2 + (X3 − X4 )2 + . . . + (X2n−1 − X2n )2 .

1
Then Y
2n n
converges in probability to

(A) 5. (B) 0. (C) 14. (D) 20.

9
23. Let X be a random variable. Define the function
P(X < x) + P(X ≤ x)
g(x) = , for all x ∈ R.
2
Which of the following statements is always true?

(A) g is right continuous at all x ∈ R.

(B) g is not a monotone function.
(C) g is left continuous at all x ∈ R.
(D) lim g(x) = 1.
x→∞

24. Consider positive numbers x1 , x2 , . . . , xn . Let θ̂1 be the value of

�n
θ that minimizes [log(xi /θ)]2 , and θ̂2 be the value of θ that
i=1
�
n
minimizes [(xi − θ)2 /xi ]. If g and h are the geometric mean
i=1
and the harmonic mean of x1 , x2 , . . . , xn , respectively, then,

(A) θ̂1 = g and θ̂2 = h.

(B) θ̂1 �= g but θ̂2 = h.
(C) θ̂1 �= g and θ̂2 �= h.
(D) θ̂1 = g but θ̂2 �= h.

25. Let xi = i and yi = i2 for i = 1, 2, 3, 4. Which value of β̂ satisfies

4
� 4
�
|yi − β̂xi | ≤ |yi − βxi | for all β ∈ R?
i=1 i=1

(A) 2 (B) 10/3 (C) 5/2 (D) 3

10
26. Suppose X1 , X2 , . . . , Xn is a random sample from the uniform
distribution on (0, θ), and let Yi = log Xi for i = 1, 2, . . . , n.
Which of the following statistics is an unbiased estimator of
log θ?

(A) Y − 1
(B) max(Y1 , Y2 , . . . , Yn )
(C) Y + 1
(D) log(2X)

27. Suppose X1 , X2 , . . . , X5 is a random sample from a Poisson dis-

tribution with mean λ. Consider a test for H0 : λ = 0.2 vs
H1 : λ = 0.4 that rejects H0 if and only if X > 0.4. What is
the probability of type 2 error of this test?

(A) 3 e−2 (B) 5 e−2 (C) 1.4 e−0.4 (D) 1.44 e−0.4

28. Suppose a single observation X is obtained from a distribution

with a probability density function f , which is either f0 or f1 ,
where 
2x if x ∈ (0, 1),
f0 (x) =
0 if x ∈
/ (0, 1),
and 
5x4 if x ∈ (0, 1),
f1 (x) =
0 if x ∈
/ (0, 1).
What is the maximum possible power of a level α test, based on
this observation, for H0 : f = f0 against H1 : f = f1 ?

� � √ √
(A) 1 − (1 − α)5 (B) (1 − α)5 (C) 1 − α5 (D) α5

11
29. Suppose Y1 , Y2 , . . . , Y5 is a random sample from the exponential
distribution with mean µ. Which of the following intervals
contains µ with 100(1−α)% confidence? Here xk,p is the number
�
5
such that P (χ2k > xk,p ) = p, and T = Yi .
i=1
� �
T
(A) 0,
x5,1−α
� �
2x10,α
(B) 0,
T
� �
5T
(C) 0,
− log(1 − α)
� �
2T
(D) 0,
x10,1−α

30. Let X be normally distributed with mean θ and variance 1,

where θ ∈ R. Assume that φ(X) is a most powerful (MP) test
of size α (with 0 < α < 1) for testing H0 : θ = 0 against
H1 : θ = 2. Which of the following statements is correct?

� 0 : θ = −1 against H1 : θ = 2, φ is a test of
(A) For testing H
size α.
� 0 : θ = −1 against H1 : θ = 2, φ is a MP test
(B) For testing H
of size α� , where α� < α.
� 1 : θ = −2, φ is a MP test
(C) For testing H0 : θ = 0 against H
of size α.
� 0 : θ = 1/2 against H1 : θ = 2, φ is a test of
(D) For testing H
size α� , where α� < α.

12
 PSB-2022
GROUP A

1. Consider a sequence {rk }k≥1 of rational numbers lying in the

interval (0, 1). Further, assume that rk converges to an irrational
mk
number as k → ∞. Suppose rk = nk
, for all k ≥ 1, where mk
and nk are positive integers with no common divisors. Show
that the set of integers {nk : k ≥ 1} is not bounded.

2. Consider a 4 × 4 real matrix A which has positive trace and

negative determinant.

(a) Show that A must have at least two real eigenvalues.

(b) Show that A can have non-real eigenvalues.

3. Let P be a regular polygon with 24 sides. Consider all the

triangles whose vertices are also vertices of P . Find the number
of such triangles that are neither isosceles nor equilateral.

GROUP B

4. Suppose U and V are independent and identically distributed

random variables following a binomial distribution with
1
parameters n and 2
. Let the random variable T denote the
number of distinct real roots of the quadratic equation

x2 + 2U x + V 2 = 0.

Find E(T ) and Var(T ).

1
5. Suppose r ≥ 1 distinct books are distributed at random among
n ≥ 3 children.

(a) For each j ∈ {0, 1, 2, . . . , r}, compute the probability that

the first child gets exactly j books.
(b) Let X be the number of children who do not get any book,
and Y be the number of children who get exactly one book.
Show that
r(n − 1)(n − 2)r−1 r(n − 1)2r−1
Cov(X, Y ) = − .
nr−1 n2r−2

6. Consider two random variables (X, Y ) distributed as bivariate

normal with parameters (µ1 , µ2 , σ12 , σ22 , ρ). Based on a random
sample from this bivariate distribution, the fitted least squares
regression line of Y on X and that of X on Y were as follows:

Y = 22 − 3X

X = 5.84 − 0.12Y

(a) Compute the maximum likelihood estimates of the

following parametric functions:
σ1
(i) min{µ1 , µ2 } (ii) (iii) ρ
σ2

You are not required to derive the expressions for the

maximum likelihood estimators of µ1 , µ2 , σ12 , σ22 and ρ.
(b) If a new observation (5, 5) is included in the above set
of observations, determine whether each of the maximum
likelihood estimates obtained in (a) will increase, decrease
or remain unchanged.

2
 7. Suppose N students arriving at a college are all equally likely
to have a particular disease with an unknown probability p.
The disease status (affected / not affected) of all students are
independent. Blood samples are collected from all N students.
In order to estimate p, two strategies are proposed.

Strategy 1 Test all samples separately to obtain the status for all N
students.

Strategy 2 Randomly partition the students into m disjoint groups,

each comprising K = N/m students (with K ≥ 2 being
an integer). For each group, pool (mix) the blood samples
from all K students within the group and test the pooled
sample. If the pooled sample tests positive, then at least
one student within that group is affected. If the pooled
sample tests negative, all students within that group are
unaffected.

(a) Based on the data obtained from Strategy 2, find a real-

valued sufficient statistic for p and the maximum likelihood
estimator of p.

(b) If a group tests positive, then all students within that group
are further tested individually. Suppose that each test (for
individual sample or pooled sample) has equal cost. Then,
which of the two strategies would you prefer to identify
all students affected with the disease when the underlying
p = 0.5, N = 200, and m = 20?

3
8. Based on historical data, a positive random variable X arising
from an unknown distribution is believed to have the chi-square
distribution with 1 degree of freedom. A new theory suggests
√
that it may be better to model X as exponentially distributed
with mean λ, where λ is such that E(X) for this model is the
same as that for the earlier model.

(a) Compute λ.
(b) Suppose X1 , X2 , . . . , Xn are independent observations from
this unknown distribution. For α ∈ (0, 1), consider the
most powerful level α test for the null hypothesis that the
earlier model is correct against the alternative that the new
model is correct. Show that the rejection region of this test
is
� �
� √ �√
(x1 , x2 , . . . , xn ) : xi − 2 2 xi > c
i i

where c is a constant that depends on α.

4
9. The following ANOVA table for three factors A, B, and C was
obtained (under a suitable model) from some data, but several
values were illegible and marked ‘∗’. It is known that A had two
levels and there were 24 observations in all.

Source df Sum of squares Mean Square F ratio

A ∗ ∗ ∗ ∗
B ∗ ∗ ∗ 0.62
C ∗ ∗ 25.4 ∗
AB 2 ∗ 5.4 0.54
BC 2 ∗ 4.2 ∗
Error ∗ ∗ ∗
Total ∗ 257.4

(a) Fill in the missing values, providing suitable justification.

(b) Suppose that a student has access only to tables for t distri-
butions. Can she test the hypothesis of equality of effects
of levels of A and that of equality of effects of levels of B,
both at 5% level of significance? Justify your answer.

5
 PSA-2021
1. Let f be a di↵erentiable function on R such that the number of ele-
ments in the set {x : f (x) = 0} is 50. Then the minimum number of
elements in the set {x : f 0 (x) = 0} is

(A) 50 (B) 51 (C) 49 (D) 0

2. Let P3 be the vector space of all polynomials with real coefficients and
of degree less than or equal to 3 over the field of real numbers. The
dimension for the subspace of polynomials p(x) such that
p(1) = p(2) = 0 is

(A) 0 (B) 1 (C) 3 (D) 2

3. Let k denote the number of 2 ⇥ 2 real matrices X with

" #
10 9
X2 = .
4 10

Then

(A) k = 2 (B) k = 0 (C) k = 3 (D) k = 1

1
 sin x4
4. lim is
x!0 (sin x)3

4
(A) 1 (B) 1 (C) 0 (D) 3

5. Let p and q be two non-zero real numbers and pq 6= 1. Suppose the

roots of the equation
✓ ◆ ✓ ◆ ✓ ◆
1 2 1 1 1
p+ x p+ +q+ x+ q+ =0
q p q p

are distinct and rational. Then

(A) p = q
(B) p and q MUST be rational numbers
(C) p + q MUST be a rational number
p
(D) MUST be a rational number
q

6. Let
[x3 ] [x]3
↵ = lim and = lim ,
x!1 x3 x!1 x3

where [x] denotes the greatest integer less than or equal to x. Then

(A) ↵ = 1, =0
(B) ↵ = 1, =1
(C) ↵ = 1, does not exist
(D) Neither ↵ nor exists

2
7. For every natural number n, let
Z n
In = ({x}n + { x}n ) dx
0

where {x} = x [x] where [x] denotes the greatest integer less than
or equal to x. Then In equals

2n 2nn+1
(A) (B) 0 (C) (D) n
n+1 n+1

8. The lengths of the sides of a right-angled triangle are numerically

equal to the arithmetic mean, the geometric mean and the harmonic
a
mean of two distinct positive real numbers a, b, a > b. Then is
b
p
(A) 5 + 2
p
(B) 2 + 3
p
(C) 2 + 5
p
(D) 3 + 2

9. The number of non-real roots of the polynomial equation

x12 + 2x6 = 5 is

(A) 10 (B) 0 (C) 2 (D) 6

3
10. A matrix is chosen at random from the set of all 2 ⇥ 2 matrices with
elements 0 or 1 only. The probability that the determinant of the
chosen matrix is non-zero is

3 3 1 1
(A) 16
(B) 8
(C) 8
(D) 4

11. If ( n C0 + n C1 )( n C1 + n C2 ) . . . ( n Cn 1 + n Cn ) = k n C0 n C1 . . . n Cn 1 ,
then k is equal to
nn
(A) n!
(n+1)n
(B) n!
(n+1)n
(C) nn!
(n+1)n+1
(D) n!

12. Let ⇡ = (a1 , a2 , · · · , a2021 ) be a permutation of (1, 2, · · · , 2021). For

every such permutation ⇡,
2021
Y
P (⇡) = (aj j).
j=1

Then

(A) P (⇡) = 0 for all ⇡

(B) P (⇡) is odd for all ⇡
(C) P (⇡) is even for all ⇡
(D) There exists ⇡1 and ⇡2 for which P (⇡1 ) is even and P (⇡2 ) is odd.

4
 " # " # " #
1 3 2 0 4 0
13. Let A = ,B= and C = . Then the sum
4 2 0 3 0 1

BAC C 2 AB 2 B 2n 1 AC 2n 1
C 2n AB 2n
trA + tr + tr + · · · + tr + tr +···
9 92 92n 1 92n
(A) is equal to 14
(B) is equal to 9
(C) is 1
(D) is equal to 12

14. Let F be the set of all bijections f from the set S = {1, 2, 3, 4, 5, 6}
onto itself such that f (f (i)) 6= i for any i 2 S. The number of elements
in F are

(A) 160 (B) 265 (C) 120 (D) 200

15. Let f : [0, 1] ! R be a monotonic function such that the number of its
R1
discontinuity points is finite. For t 0 let mt = 0 xt f (x)dx. Then
which of the following statements is true?

(A) There exists a t 0 for which mt is not defined.

(B) mt < 1 for all t 0.
(C) mt is defined and is equal to 1 for all t 0.
(D) mt is defined for all t 0 and there exists t1 , t2 0 such that
mt1 is equal to 1 and mt2 < 1.

5
16. Let F (x) and G(x) be cumulative distribution functions of random
variables X and Y , respectively. Then which of the following would
NOT be a cumulative distribution function?

(A) H(x) = (F (x))2

(B) H(x) = (1 F ( x) + F (x)) /2

(C) H(x) = (F (x) + G(x))/2

p
(D) H(x) = G(x)

17. Suppose that X and Y are independent and identically distributed

exponential random variables with mean 1/ where > 0. Then
P (X > 4Y |X < 5Y ) equals

(A) 1/25

(B) 4/5
5 4
(C) e /e
4 5 4
(D) e e / 1 e

18. Suppose X1 and X2 are independent and identically distributed

exponential random variables with mean 2. The expectation of
max{X1 , X2 } is

(A) 4 (B) 3.5 (C) 3 (D) 2.5

6
19. Let X be a random variable. Let F (t) denote its cumulative dis-
tribution function and M (t) denote its moment generating function.
Consider the following two statements.

(I) F (t) = 1 F ( t) for t 2 R

(II) M (t) = M ( t) for t 2 R

Which of the above statements are individually equivalent to random

variable X being symmetric about zero?

(A) Neither (I) nor (II)

(B) (II) but not (I)
(C) (I) and (II)
(D) (I) but not (II)

20. Suppose that X1 , X2 , . . . are independent and identically distributed

exponential random variables with mean > 0. For n 1, define
Yn = X12 + X22 + · · · + Xn2 . Then

V (Yn )
lim =
n!1 n
(A) 2 2 (2 2
1)
2
(B) 2
4
(C) 20
4
(D)

7
21. The mean and the variance of a set of observations are 10 and 25
respectively. When two new observations are included in the set, both
the mean and the variance remain the same. Which of the following
is a correct statement?

(A) the two new observations are both 10

(B) the two new observations are 20 and 0
(C) the two new observations cannot be uniquely determined unless
the number of observations are known
(D) the two new observations are 15 and 5

22. Consider data on three variables X, Y and Z. Suppose least squares

regressions of Y on X, Y on Z and X on Z are performed. If the slope
coefficient in each of the three regressions is 2, what can be said about
the slope coefficient of the least squares regression of Y on (X + Z)?

(A) It is greater than 2

(B) It is less than 2
(C) It is equal to 2
(D) It cannot be determined whether it is more than 2 or less than 2
but it cannot be equal to 2

8
23. Suppose that the probability of a diagnostic test detecting the presence
of an infection is 0.9 for a diabetic individual and 0.8 for a non-diabetic
individual. An individual is randomly selected from the population.
What should be the ratio of the proportions of diabetic to non-diabetic
individuals in the population so that the probability of the selected
individual being diabetic given that the above diagnostic test produces
a negative result is the same as that of the individual being non-
diabetic given that the test produces a positive result?

(A) 4 : 3 (B) 2 : 1 (C) 3 : 2 (D) 5 : 3

24. Consider independent random samples X1 , . . . , Xm and Y1 , . . . , Yn

2 2
(m, n 2) from N (µ1 , ) and N (µ2 , ), respectively. Suppose we
want to test H0 : µ1 = µ2 vs H1 : µ1 6= µ2 at level of significance
↵ (0 < ↵ < 1). Then

(A) the two sample t-test statistic yields more power than the
ANOVA statistic for any non-zero value of µ1 µ2 and for any
m, n.

(B) the two sample t-test statistic and the ANOVA statistic yield the
same power for any non-zero value of µ1 µ2 and for any m, n.

(C) the two sample t-test statistic yields less power than the ANOVA
statistic for any non-zero value of µ1 µ2 and for any m, n.

(D) the two sample t-test statistic and the ANOVA statistic will yield
the same power for any non-zero value of µ1 µ2 if m = n but
will not necessarily yield the same power if m 6= n.

9
25. In order to compare between t( 2) treatments, one carries out t
replications of an experiment using a randomized block design with
t blocks. The error degrees of freedom for testing the equality of
treatment e↵ects in such a design is

(A) t + 1
(B) 3(t 1)
(C) (t 1)(t2 + t 1)
(D) t(t 1)(t + 1)

26. Let Xi be independently distributed Poisson(i ) random variables,

n
X
i = 1, . . . , n and > 0, is an unknown parameter. Assume Xi > 0.
i=1
Then the MLE of is
Pn
i=1 iXi
(A)
(n + 1)
Pn
i=1 Xi
(B)
n(n + 1)
P
2 ni=1 Xi
(C)
n(n + 1)
Pn
i=1 iXi
(D)
n

10
27. Suppose X1 , X2 , X3 are independent normally distributed random
2
variables with mean µ and variance . However, instead of
X1 , X2 , X3 , we only observe Y1 = X2 X1 and Y2 = X3 X2 . Which
2
of the following statistics is sufficient for ?

(A) Y12 + Y22 Y1 Y2

(B) Y12 + Y22 + 2Y1 Y2
(C) Y12 + Y22
(D) Y12 + Y22 + Y1 Y2

28. Consider data on body mass index (BMI) and fasting blood sugar
levels (fgl) for two groups of individuals. Suppose that the correlation
coefficient between BMI and fgl is equal to 0.5 in each of the two
groups. If the two groups are combined, the correlation coefficient
between BMI and fgl:

(A) can be negative

(B) will be equal to 0.5
(C) will be positive but not necessarily be equal to 0.5
(D) can be zero but cannot be negative

11
29. A random variable Z is obtained as follows. Let X ⇠ U (0, 1), and
Y given X = x be Bernoulli with probability of success x. If Y = 1,
Z is defined to be X. Otherwise, the experiment is repeated until a
pair (X, Y ) with Y = 1 is obtained. Then, the probability density
function of Z on (0, 1) is

(A) 2(1 z)
(B) 2z
(C) 12z 2 (1 z)
(D) 6z(1 z)

30. Suppose X1 , X2 , . . . , X9 are independent variables such that Xi is

distributed as Poisson with mean i, i = 1, 2 . . . , 9.
5
X 9
X
The conditional variance of Xi given Xi = 60 is
i=1 i=1

(A) 20 (B) 100/3 (C) 400/27 (D) 40/3

12
 PSB-2021
Group A

1. Define f : R → R as

cos(2x) if x is rational,
f (x) =
sin2 (x) if x is irrational.

Find all the real numbers where

(a) f is continuous,
(b) f is differentiable.

2. Consider the matrix  

5 3
P= 3 2  .
 

8 5 3×2

Find a matrix G such that AGA = A where A = PPT .

3. Consider the set of all five digit integers formed by permuting the
digits 1, 2, 4, 6 and 7. Let X denote a randomly chosen integer from
this set and let Y denote the position, from the right, of the digit 4
in the randomly chosen integer. For example, if the integer chosen is
12647, then Y is 2 and for 41276, Y is 5.

(a) Find E(X) and E(Y ).

(b) Find E[X|Y = y] for all possible values of y.
(c) Show that X and Y are uncorrelated but not independent.

1
 Group B

4. Let U1 , U2 , U3 be i.i.d. random variables which are uniformly dis-

tributed on (0, 1). Let X = min(U1 , U2 ) and Y = max(U2 , U3 ).

(a) Find P (X ≤ x, Y ≤ y) for all x, y ∈ R.

(b) Find P (X = Y ).
(c) Find E[XI{X=Y } ] where IA is the indicator function of A.

5. Consider a game with six states 1, 2, 3, 4, 5, 6. Initially a player starts

either in state 1 or in state 6. At each step the player jumps from one
state to another as per the following rules.
A perfectly balanced die is tossed at each step.

(i) When the player is in state 1 or 6: If the roll of the die results
in k then the player moves to state k, for k = 1, . . . , 6.
(ii) When the player is in state 2 or 3: If the roll of the die results in
1, 2 or 3 then the player moves to state 4. Otherwise the player
moves to state 5.
(iii) When the player is in state 4 or 5: If the roll of the die results in
4, 5 or 6 then the player moves to state 2. Otherwise the player
moves to state 3.

The player wins when s/he visits 2 more states, besides the starting
one.

(a) Calculate the probability that the player will eventually move
out of states 1 and 6.
(b) Calculate the expected time the player will remain within states
1 and 6.
(c) Calculate the expected time for a player to win, i.e., to visit 2
more states, besides the starting one.

2
6. Let X1 , . . . , Xn be i.i.d. random variables which are uniformly dis-
tributed on (θ, 2θ), θ > 0.
X(n)
(a) Show that 2
is the maximum likelihood estimator (MLE) of θ
where X(n) = max(X1 , . . . , Xn ).
(b) Find an unbiased estimator for θ based on the MLE.
(c) Given any  > 0, show that
 
X(n)
lim P − θ >  = 0.
n→∞ 2

7. There are two urns each contains N balls numbered from 1 to N .

From each urn a sample of size n is selected without replacement.
Denote the set of numbers appearing in the first and second samples
by s1 = {i1 , . . . , in } and s2 = {j1 , . . . , jn } respectively. Let

X = |s1 ∩ s2 | = number of common elements in s1 and s2 .

(a) Find the probability distribution of X.

(b) Suppose n = 6 and the observed value of X is 4. Obtain a
method of moments estimate of N .

8. Let X1 , X2 , X3 be i.i.d. random variables from N (µ, σ 2 ). Let

3 3 3
1X X 1X
X̄ = Xi , T1 = Xi 2 , T2 = (Xi − X̄)2 .
3 i=1 i=1
3 i=1

(a) Compute E[T1 |X̄] and E[T1 |T2 ]

(b) Obtain the exact critical region of a level α (0 < α < 1) test for
X̄ 2
H0 : µ = 0 vs H1 : µ 6= 0 that rejects H0 if and only if is
T1
sufficiently large.

3
9. Consider a linear regression model:

yi = α + βxi + ei , i = 1, 2, . . . , n

where xi ’s are fixed and ei ’s are i.i.d. random errors with mean 0 and
variance σ 2 .
Define two estimators of β as follows
Pn Pn
i=1 y i xi y i
βb1 = Pn and βb2 = Pi=1
n 2
.
i=1 xi i=1 xi

(a) Obtain an unbiased estimator of β as a linear combination of βb1

and βb2 .
(b) Find mean squared errors of βb1 and βb2 . Which, between βb1 and
βb2 , has lower mean squared error?

4
 PSA-2020
1. How many distinct straight lines can one form that are given by an
equation ax + by = 0, where a and b are numbers from the set
{0, 1, 2, 3, 4, 5, 6, 7}?

(A) 63.
(B) 57.
(C) 37.
(D) 49.

2. Which of the following sequences is not monotone?

(A) an = n2 − 3n, n ≥ 1.
(B) an = 3n2 − n, n ≥ 1.
(C) an = log( 34 )n , n ≥ 1.
(D) an = 7n − n2 , n ≥ 1.

3. Let A = (1, −1), B = (−2, 0), C = (1, 2) and D be the vertices of a

parallelogram in the X-Y plane listed clockwise. Then the point D is:

(A) (4, 1).

(B) (−2, −3).
(C) (3, 0).
(D) (−2, 1).

1
ROUGH WORK
4. Let f be the function on the set of real numbers defined by

ex sin x cos x
f (x) = e2x sin 2x cos 2x .
e3x sin 3x cos 3x

Then f 0 (0) is

(A) 1. (B) 0. (C) −1. (D) 2.

5. The period of the function f given by f (x) = sin( x3 − π2 ) , where x is

real, is:

(A) 6π. (B) 9 π2 . (C) 2π. (D) 13 π2 .

6. Let √ ! !
1 3 1 2 −1
X= √ and A = .
2 −1 3 3 −2
Let X t denote the transpose of the matrix X. If B = XAX t , then
X t B 105 X equals

(A) I. (B) A. (C) X t AX. (D) B.

2
ROUGH WORK
7. Suppose
n
X
aj (n) = kj , j = 0, 1, 2, 3, n = 1, 2, . . . .
k=1

a1 (n)a2 (n)
Then limn→∞ a0 (n)a3 (n) equals

(A) 0. (B) 13 . (C) 32 . (D) 1.

8. Let S be the set of all 3 × 3 real matrices A = ((aij )) such that the matrix
((a3ij )) has rank one. Let R be the set

R = {rank(A) : A ∈ S}.

Then R is equal to

(A) {1}. (B) {1, 2}. (C) {1, 3}. (D) {1, 2, 3}.

9. Consider the circles

S1 : x2 + y 2 + 4x + 2y − 4 = 0,
S2 : x2 + y 2 − 4x − 4y + 4 = 0.

Then the number of common tangents of S1 and S2 is

(A) 1. (B) 2. (C) 3. (D) 4.

3
ROUGH WORK
10. Let A be the 3 × 3 matrix
 
1 2 4
A = 0 0 3 
 

0 0 −1

Then the determinant of the matrix A17 + A10 − I is

(A) 1. (B) 2. (C) −1. (D) 0.

11. Let m and n be nonzero integers. Define

Am,n = x ∈ R : n2 x3 + 2020x2 + mx = 0 .


Then the number of pairs (m, n) for which Am,n has exactly two points
is

(A) 0. (B) 10. (C) 16. (D) ∞.

12. The set of all real solutions of the inequality 2|x| > |x − 1| is

x : x < − 13 ∪ {x : x > 1}.


(A)
1

(B) x:x> 3 .
(C) {x : −1 < x < 1}.
1

(D) {x : x < −1} ∪ x : x > 3 .

4
ROUGH WORK
13. Let S and T be two non-empty sets and f : S → T be a function such
that for all subsets A and B of S, we have f (A ∩ B) = f (A) ∩ f (B).
Then

(A) There exists an A ⊆ S such that f −1 (f (A)) 6= A.

(B) There exist disjoint subsets A and B of S such that
f (A) ∩ f (B) 6= φ.
(C) f is injective.
(D) None of the above statements is true.

14. Let S be the set of all 3 × 3 matrices A such that among the 9 entries
of A, there are exactly three 0’s, exactly three 1’s and exactly three 2’s.
The number of matrices in S that have trace divisible by 3 is

(A) 580. (B) 600. (C) 150. (D) 120.

15. Let x1 , . . . , xn ∈ R be distinct reals. Define the set

n
o
A = f1 (t), . . . , fn (t) : t ∈ R ,

where for 1 ≤ k ≤ n

1 if xk ≤ t,
fk (t) =
0 otherwise.

Then A contains

(A) exactly n distinct elements.

(B) exactly (n + 1) distinct elements.
(C) exactly 2n distinct elements.
(D) infinitely many distinct elements.

5
ROUGH WORK
16. Let X, Y be independent and identically distributed random variables
with
P (X = k) = (0.75)(0.25)k , k = 0, 1, 2, . . . .

Then P (X > Y ) is equal to

(A) 0.5 (B) 0.4 (C) 0.2 (D) 0.6

17. Let X1 , X2 , . . . be independent and identically distributed Bin(n, p) ran-

dom variables. Define

Sk = X12 + X22 + · · · + Xk2 , k = 1, 2, . . . ,

and let ε > 0. Then as k → ∞,

 
(A) P Skk − n2 p2 > ε → 0 for all p ∈ (0, 1).
 
(B) P Skk − np2 > ε → 0 for all p ∈ (0, 1).
 
(C) P Skk − n4 > ε → 0 when p = 1/2.
 
(D) P Skk − n(n+1)
4 > ε → 0 when p = 1/2.

18. Let X, Y have joint probability density function


 1 ye−xy if x > 0, 2 ≤ y ≤ 4,
f (x, y) = 2
0 otherwise.

Then E(XY ) is given by

(A) 3. (B) 1. (C) 2. (D) 1.5

6
ROUGH WORK
19. Let X and Y be random variables with V (X) = 9 and V (Y ) = 4. Then
which of the following is true for non negative reals a and b?

(A) V (aX + bY ) = 0 if and only if a = b = 0.

(B) V (aX + bY ) lies between (3a − 2b)2 and (3a + 2b)2 .
(C) V (aX + bY ) lies between 3a and 2b.
(D) V (aX + bY ) lies between (3a)2 and (2b)2 .

20. The moment generating function of a continuous random variable X

is given by M (t) = et(t+1) , −∞ < t < ∞. Let Φ denote the cumula-
tive distribution function of a standard normal random variable. Then
P (X ≥ 0) is

(A) Φ(0). (B) Φ(− √12 ). (C) Φ( √12 ). (D) Φ( 12 ).

21. A multiple choice test uses the following scoring procedure to discourage
students from guessing. For each correct response the score is 5, for each
incorrect response the score is 0 and for no response the score is 2. If
there are 4 choices for each question, what is the minimum number of in-
correct choices that the student must eliminate before it is advantageous
to guess among the rest than leave the question unanswered?

(A) 0. (B) 1. (C) 2. (D) 3.

7
ROUGH WORK
22. Each of 100 laboratory rats has available plain water and a mixture of
water and caffeine in their cages. After 24 hours, two measurements
were recorded for each rat: the amount of caffeine consumed (X) and
the blood pressure (Y ). Based on the data, the correlation coefficient
between X and Y was found to be 0.428. Which of the following con-
clusions is justified on the basis of the study?

(A) The correlation between caffeine consumed and blood pressure in

the population of rats is 0.428.
(B) If rats stop drinking the mixture of water and caffeine, their blood
presure will go down.
(C) Rats with low blood pressure do not consume the mixture of water
and caffeine as much as rats with high blood pressure.
(D) About 18% of variation in blood pressure can be explained by a
linear relationship between blood pressure and caffeine consumed.

23. Suppose X is distributed as Bin(3, 23 ) and the conditional distribution

of Y given X is N (X, X 2 + 1). Then the variance of Y is

19 17 13 11
(A) 3 . (B) 3 . (C) 3 . (D) 3 .

24. Suppose X and Y are independent observations from N (0, 1). For t > 0,
define g(t) = P (|X − Y | > t | (X + Y ) > t). Then

(A) g is a strictly increasing function.

(B) g is a strictly decreasing function.
(C) g is a constant function.
(D) g is not a monotonic function.

8
ROUGH WORK
25. Let X1 , X2 , . . . , Xn be independent and identically distributed N (1, σ 2 )
random variables. Let σ c2 denote the maximum likelihood estimator of
σ 2 . Then
1 Pn
c2 =
(A) σ n i=1 (Xi − X̄)2 and it is biased.
1 Pn
c2 =
(B) σ n−1 i=1 (Xi − X̄)2 and it is unbiased.
1 Pn
c2 =
(C) σ n−1 i=1 (Xi − 1)2 and it is biased.
1 Pn
c2 =
(D) σ n i=1 (Xi − 1)2 and it is unbiased.

26. Let X1 , X2 , . . . , Xn be independent and identically distributed exponen-

tial random variables with mean λ. Then maximum likelihood estimator
of the median of the distribution is

(A) Sample median. (B) (log 2)X̄. (C) log 2/X̄. (D) X̄/ log 2.

27. Suppose X is a discrete random variable taking values −1, 0, 1, each with
probability 31 . Let Y = |X|. Which one of the following statements is
correct?

(A) Correlation between X and Y is ±1.

(B) X and Y are positively correlated.
(C) X and Y are negatively correlated.
(D) X and Y are uncorrelated.

9
ROUGH WORK
28. Let X1 , X2 , . . . , Xn be independent and identically distributed random
variables with probability density function
1 −| x−1 |
f (x) = e λ , −∞ < x < ∞,
2λ
where λ > 0 is an unknown parameter. Which one of the following is
a form of the most powerful test of its size for testing H0 : λ = 2 vs
H1 : λ = 1?
Pn
(A) Reject H0 if i=1 (Xi − 1)2 > C.
Pn
(B) Reject H0 if i=1 |Xi − 1| > C.
Pn
(C) Reject H0 if i=1 (Xi − 1)2 < C.
Pn
(D) Reject H0 if i=1 |Xi − 1| < C.

29. Suppose X and Y are independent and identically distributed

continuous random variables and are independent of Z. Assume P (Z =
±1) = 1/2. Then which of the following statements is always true?

(A) X + Y has the same distribution as Z(X + Y ).

(B) X − Y has the same distribution as Z(X − Y ).
(C) X/Y has the same distribution as Z(X/Y ).
(D) XY has the same distribution as Z(XY ).

30. Suppose T1 and T2 are two different real valued statistics each of which
is sufficient for an unknown parameter θ. Then, which of the following
is always true?

(A) T1 + T2 is sufficient for θ.

(B) max{T1 , T2 } is sufficient for θ.
(C) (T1 , T2 ) is sufficient for θ.
(D) min{T1 , T2 } is sufficient for θ.

10
ROUGH WORK
ROUGH WORK
ROUGH WORK
 PSB-2020
GROUP A

1. Let f (x) = x2 − 2x + 2. Let L1 and L2 be the tangents to its graph at

x = 0 and x = 2 respectively. Find the area of the region enclosed by
the graph of f and the two lines L1 and L2 .

2. Find the number of 3 × 3 matrices A such that the entries of A belong

to the set Z of all integers, and such that the trace of At A is 6.
(At denotes the transpose of the matrix A).

3. Consider n independent and identically distributed positive random vari-

ables X1 , X2 , . . . , Xn . Suppose S is a fixed subset of {1, 2, . . . , n} con-
sisting of k distinct elements where 1 ≤ k < n.

(a) Compute P 
i∈S Xi
E Pn .
i=1 Xi

(b) Assume that Xi ’s have mean µ and variance σ 2 , 0 < σ 2 < ∞. If

P P
j 6∈ S, show that the correlation between ( i∈S Xi )Xj and i∈S Xi
1 √1 .
lies between − √k+1 and k+1

GROUP B

4. Let X1 , X2 , . . . , Xn be independent and identically distributed random

variables. Let Sn = X1 + · · · + Xn . For each of the following statements,
determine whether they are true or false. Give reasons in each case.

(a) If Sn ∼ Exp with mean n, then each Xi ∼ Exp with mean 1.

(b) If Sn ∼ Bin(nk, p), then each Xi ∼ Bin(k, p).

1
5. Let U1 , U2 , . . . , Un be independent and identically distributed random
variables each having a uniform distribution on (0, 1). Let

X = min{U1 , U2 , . . . , Un }, Y = max{U1 , U2 , . . . , Un }.

Evaluate E [X|Y = y] and E [Y |X = x].

6. Suppose individuals are classified into three categories C1 , C2 and C3 .

Let p2 , (1 − p)2 and 2p(1 − p) be the respective population proportions,
where p ∈ (0, 1). A random sample of N individuals is selected from
the population and the category of each selected individual recorded.
For i = 1, 2, 3, let Xi denote the number of individuals in the sample
X3
belonging to category Ci . Define U = X1 + 2 .

(a) Is U sufficient for p? Justify your answer.

U p(1−p)
(b) Show that the mean squared error of N is 2N .

7. Consider the following model:

yi = βxi + εi xi , i = 1, 2, . . . , n,

where yi , i = 1, 2, . . . , n are observed; xi , i = 1, 2, . . . , n are known posi-

tive constants and β is an unknown parameter. The errors ε1 , ε2 , . . . , εn
are independent and identically distributed random variables having the
probability density function
 
1 |u|
f (u) = exp − , −∞ < u < ∞,
2λ λ

and λ is an unknown parameter.

(a) Find the least squares estimator of β.

(b) Find the maximum likelihood estimator of β.

2
8. Assume that X1 , . . . , Xn is a random sample from N (µ, 1), with µ ∈ R.
We want to test H0 : µ = 0 against H1 : µ = 1. For a fixed integer
m ∈ {1, . . . , n}, the following statistics are defined:

T1 = (X1 + . . . + Xm )/m,
T2 = (X2 + . . . + Xm+1 )/m,
.. ..
.= .
Tn−m+1 = (Xn−m+1 + . . . + Xn )/m.

Fix α ∈ (0, 1). Consider the test

reject H0 if max {Ti : 1 ≤ i ≤ n − m + 1} > cm,α .

Find a choice of cm,α ∈ R in terms of the standard normal distribution

function Φ that ensures that the size of the test is at most α.

9. A finite population has N units, with xi being the value associated

with the ith unit, i = 1, 2, . . . , N . Let x̄N be the population mean.
A statistician carries out the following experiment.

• Step 1: Draw a SRSWOR of size n (< N ) from the population.

Call this sample S1 and denote the sample mean by X̄n .
• Step 2: Draw a SRSWR of size m from S1 . The x-values of the
sampled units are denoted by {Y1 , . . . , Ym }.

An estimator of the population mean is defined as,

m
1 X
Tbm = Yi .
m
i=1

(a) Show that Tbm is an unbiased estimator of the population mean.

(b) Which of the following has lower variance: Tbm or X̄n ?

3
 PSA

PSA-2019 2019

1. Let A = ((aij )) be an m×n matrix with all non-zero real entries. Let B
be obtained from A by replacing a11 by 0 and keeping all other entries
unchanged. If r is the rank of A, then what is the set of possible values
for the rank of B?

(A) {r} (B) {r − 1, r, r + 1} (C) {r, r + 1} (D) {r − 1, r}

2. What is the period of the function g(x) = | cos x| + | sin x|?

(A) π (B) π/2 (C) 2π (D) π/4

3. If 3(cos 100◦ + i sin 100◦ )(cos 110◦ + i sin 110◦ ) = x + iy, where x and y
are real numbers, then
√
(A) x = − 3 2 3 , y = − 32 .
√
3 3
(B) x = 2 , y = 32 .
√
3 3
(C) x = 2 , y = − 32 .
√
(D) x = −323, y = 32 .

1
ROUGH WORK
4. What is the number of 6 digit positive integers in which the sum of
the digits is at least 52?

(A) 66 (B) 24 (C) 28 (D) 120

5. Let the sum

3 + 33 + 333 + · · · + �33 ��
. . . 3�
200 times

be . . . zyx in the decimal system, i.e., x is the unit’s digit, y the ten’s
digit, and so on. What is z?

(A) 0 (B) 9 (C) 7 (D) 3

6. How many times does the digit ‘2’ appear in the set of integers
{1, 2, ..., 1000}?

(A) 590 (B) 600 (C) 300 (D) 299

2
ROUGH WORK
  
0 1 t
 
7. Let t be a real number. Then the rank of 2 t −1 equals
2 2 0

(A) 2 if t = −1, and 3 if t �= −1.

(B) 2 if t = 1, and 3 if t �= 1.
(C) 2 if t = ±1, and 3 if |t| �= 1.
(D) 3 for all t.

�200�
8. The number 100 /4100 lies in

(A) [ 34 , 1) (B) (0, 12 ) (C) [1, ∞) (D) [ 12 , 34 )

9. Let P (x) = x4 + 4x3 − 8x2 − 1. Which of the following is false?

(A) P (x) has a real root in (−4, 1)

(B) P (x) has a real root < −4
(C) P (x) has a real root > 1
(D) P (x) has at least two real roots.

3
ROUGH WORK
10. Two friends, one from Kolkata and one from Delhi, start driving to-
wards each other at the same time. It is given that the distance between
Kolkata and Delhi is 1455 km. One of them drives at a constant speed
of 80 kmph (km per hour), while the other drives at a speed of 50 kmph
during the first hour, 55 kmph during the second hour, 60 kmph dur-
ing the third hour, and so on (i.e., his speeds over successive hours are
in an arithmetic progression). How long will it take for them to meet
each other?

(A) 8 hrs 56 mins

(B) 8 hrs 46 mins
(C) 10 hrs 26 mins
(D) 9 hrs 36 mins

11. How many positive divisors of 25 53 114 are perfect squares?

(A) 60 (B) 18 (C) 120 (D) 4

12. What is the set of numbers x in (0, 2π) such that log log(sin x + cos x)
is well-defined?

(A) [ π8 , 3π
8 ] (B) (0, π2 ) (C) (0, π4 ] (D) (0, π) ∪ ( 3π
2 , 2π)

4
ROUGH WORK
13. Let C1 and C2 be concentric circles with centre at O and radii r1 and
r2 respectively. The line OA2 intersects C1 at A1 . The line A1 D is
tangent to C1 at A1 . What is the length of the line segment A2 D?

� � �
(A) 2r1 (r2 − r1 ) (B) r22 − r12 (C) r2 (D) 2r2 (r2 − r1 )

C2 D

C1

O A1 A2
f

14. The reflection of the point (1, 2) with respect to the line x + 2y = 15 is

(A) (3, 6). (B) (6, 3). (C) (10, 5). (D) (5, 10).

15. How many solutions does the equation cos2 x + 3 sin x cos x + 1 = 0
have for x ∈ [0, 2π)?

(A) 1 (B) 3 (C) 4 (D) 2

5
ROUGH WORK
16. The functions f, g : [0, 1] → [0, 1] are given by f (x) = 12 x(x + 1) and
g(x) = 12 x2 (x + 1). What is the area enclosed between the graphs of
f −1 and g −1 ?

(A) 1/8 (B) 1/4 (C) 5/12 (D) 7/24

17. If f (a) = 2, f � (a) = 1, g(a) = −1 and g � (a) = 2, then what is

g(x)f (a) − f (x)g(a)

lim ?
x→a x−a

(A) 5 (B) 3 (C) −3 (D) −5

18. Draw one observation N at random from the set {1, 2, . . . , 100}. What
is the probability that the last digit of N 2 is 1?

(A) 1/20 (B) 1/50 (C) 1/10 (D) 1/5

6
ROUGH WORK
19. Let X be the number of tosses of a fair coin required to get the first
head. If Y | X = n is distributed as Binomial(n, 12 ), then what is
P(Y = 1)?

(A) 4/9 (B) 1/4 (C) 1/3 (D) 5/9

20. Suppose X is distributed uniformly on (−1, 1). For i = 0, 1, 2, 3, let

� �
pi = P X 2 ∈ ( 4i , i+1
4 ) . For which value of i is pi the largest?

(A) 3 (B) 1 (C) 0 (D) 2

21. In a simulation experiment, two independent observations X1 and X2

are generated from the Poisson distribution with mean 1. The experi-
ment is said to be successful if X1 + X2 is odd. What is the expected
value of the number of experiments required to obtain the first success?

(A) 2(1 + e−2 )

(B) 2/(1 − e−2 )
(C) 2/(1 − e−4 )
(D) 2(1 + e−4 )

7
ROUGH WORK
22. A shopkeeper has 12 bulbs of which 3 are defective. She sells the bulbs
by selecting them at random one at a time. What is the probability
that the seventh bulb sold is the last defective one?

(A) 3/44 (B) 9/44 (C) 13/44 (D) 7/44

23. The chances of Smitha getting admitted to colleges A and B are 60%
and 40% respectively. Assume that colleges admit students indepen-
dently of each other. If Smitha is told that she has been admitted to
at least one college, what is the probability that she got admitted to
college A?

(A) 3/5 (B) 15/19 (C) 10/13 (D) 5/7

24. Suppose X1 , X2 , . . . , Xn is a random sample from an exponential dis-

tribution with mean λ. If λ̂1 and λ̂2 are, respectively, the maximum
likelihood estimators of the mean and the median of the underlying
distribution, then

(A) λ̂1 < λ̂2 .

(B) λ̂1 = λ̂2 .
(C) λ̂1 < λ̂2 and λ̂1 > λ̂2 are both possible.
(D) λ̂1 > λ̂2 .

8
ROUGH WORK
25. Suppose Y, X1 , X2 , . . . , Xn are i.i.d. N (µ, 1) random variables, and
In = (X̄n − an , X̄n + an ) is a 95% confidence interval for µ. Then
P(Y ∈ In )

(A) converges to 1 as n → ∞.
(B) is greater than 0.95 for all n ≥ 1.
(C) is less than 0.95 for all n ≥ 1.
(D) equals 0.95 for all n ≥ 1.

26. Suppose (X1 , Y1 ), (X2 , Y2 ), . . . , (Xn , Yn ) is an i.i.d. sample from

N2 (0, 0, 1, 1, ρ) where |ρ| ≤ 1. Let (i1 , . . . , in ) be a random permu-
�
n
tation of {1, 2, . . . , n}. Define Tn = n1 Xj Yij . What is E(Tn )?
j=1

(A) 1/n (B) 0 (C) ρ/n (D) ρ

27. Suppose X is a N (µ, σ 2 ) random variable, and Y = Φ(X), where Φ

is the cumulative distribution function of a standard normal random
variable. What is E(Y )?
√
(A) Φ(µ/ 2 + σ 2 )
√
(B) Φ(µ/ 1 + σ 2 )
(C) Φ(µ/σ)
√
(D) Φ(µ/ 4 + σ 2 )

9
ROUGH WORK
28. Let X1 , X2 , . . . , Xn be a random sample from a distribution with prob-
ability density function

(λ + 1) xλ 0 ≤ x ≤ 1
fλ (x) =
0 otherwise,

where λ > −1. What is the maximum likelihood estimator of λ?

�
n
(A) 1 + n/( log Xi )
i=1
�n
(B) −1 − n/( log Xi )
i=1

1 �
n
(C) −1 − n log Xi
i=1
�n
(D) 1 − n/( log Xi )
i=1

29. Suppose the joint distribution of (X1 , X2 ) is N2 (0, 0, 2, 2, −0.5). What

√
is the value of P(2X1 + X2 ≤ 2 2)? Here Φ denotes the cumulative
distribution function of a standard normal random variable.

√ √ √
(A) Φ(2/ 7) (B) Φ(2/ 3) (C) Φ(2/ 5) (D) Φ(1)

30. Suppose the joint probability density function of (X, Y ) is


e−x 0 ≤ y ≤ x < ∞
f (x) =
0 otherwise.

What is E(X)?

(A) 2 (B) 1 (C) 6 (D) 1/2

10
ROUGH WORK
 PSB
PSB-2019 2019

GROUP A

1. Let f (x) = x3 − 3x + k, where k is a real number. For what values of

k will f (x) have three distinct real roots?

2. Let A and B be 4 × 4 matrices. Suppose that A has eigenvalues

x1 , x2 , x3 , x4 and B has eigenvalues 1/x1 , 1/x2 , 1/x3 , 1/x4 , where each
xi > 1.

(a) Prove that A + B has at least one eigenvalue greater than 2.

(b) Prove that A − B has at least one eigenvalue greater than 0.
(c) Give an example of A and B so that 1 is not an eigenvalue of AB.

3. Elections are to be scheduled on any seven days in April and May. In

how many ways can the seven days be chosen such that elections are
not scheduled on two consecutive days?

GROUP B

4. Let X and Y be independent and identically distributed random vari-

ables with mean µ > 0 and taking values in {0, 1, 2, . . . }. Suppose, for
all m ≥ 0,
1
P(X = k | X + Y = m) = , k = 0, 1, . . . , m.
m+1
Find the distribution of X in terms of µ.

1
5. Suppose X1 , X2 , . . . , Xn are independent random variables such that

P(Xi = 1) = pi = 1 − P(Xi = 0),

where p1 , p2 , . . . , pn ∈ (0, 1) are all distinct and unknown. Consider

�n
X = Xi and another random variable Y which is distributed as
i=1
1 �
n
Binomial(n, p̄), where p̄ = n pi . Between X and Y , which is a
i=1
�
n
better estimator of pi in terms of their respective mean squared
i=1
errors?

6. Suppose X1 , X2 , . . . , Xn is a random sample from Uniform(0, θ) for

some unknown θ > 0. Let Yn be the minimum of X1 , X2 , . . . , Xn .

(a) Suppose Fn is the cumulative distribution function (c.d.f.) of nYn .

Show that for any real x, Fn (x) converges to F (x), where F is
the c.d.f. of an exponential distribution with mean θ.
(b) Find lim P(n[Yn ] = k) for k = 0, 1, 2, . . . , where [x] denotes the
n→∞
largest integer less than or equal to x.

7. Suppose an SRSWOR of size n has been drawn from a population

labelled 1, 2, . . . , N , where the population size N is unknown.

� of N .
(a) Find the maximum likelihood estimator N
�.
(b) Find the probability mass function of N
n+1 �
(c) Show that n N − 1 is an unbiased estimator of N .

2
8. Suppose {(xi , yi , zi ) : i = 1, 2, . . . , n} is a set of trivariate observations
on three variables: X, Y , and Z, where zi = 0 for i = 1, 2, . . . , n − 1
and zn = 1. Suppose the least squares linear regression equation of Y
on X based on the first n − 1 observations is

�0 + α
y=α �1 x

and the least squares linear regression equation of Y on X and Z based

on all n observations is

y = β�0 + β�1 x + β�2 z.

�1 = β�1 .
Show that α

9. Let Z be a random variable with probability density function

1
f (z) = e−|z−µ| , z ∈ R
2
with parameter µ ∈ R. Suppose we observe X = max(0, Z).

(a) Find the constant c such that the test that “rejects when X > c”
has size 0.05 for the null hypothesis H0 : µ = 0.
(b) Find the power of this test against the alternative hypothesis
H1 : µ = 2.

3
 PSA-2018
1. Let A be a 2 × 2 nonzero real matrix. Which of the following is true?

(A) A has a nonzero eigenvalue.

(B) A2 has at least one positive entry.
(C) trace (A2 ) is positive.
(D) All entries of A2 cannot be negative.

2. Let A be a 3 × 3 real matrix with zero diagonal entries. If 1 + i is an

eigenvalue of A, the determinant of A equals

(A) 4. (B) −4. (C) 2. (D) −2.

3. Let A be an n × n matrix and let b be an n × 1 vector such that Ax = b

has a unique solution. Let A0 denote the transpose of A. Then which
of the following statements is false?

(A) A0 x = 0 has a unique solution.

(B) A0 x = c has a unique solution for any non-zero c.
(C) Ax = c has a solution for any c.
(D) A2 x = c is inconsistent for some vector c.

1
ROUGH WORK
4. Let A and B be n × n matrices. Assuming all the inverses exist,

(A−1 − B −1 )−1

equals

(A) (I − AB −1 )−1 B.
(B) A(B − A)−1 B.
(C) B(B − A)−1 A.
(D) B(A − B)−1 A.

5. Let f be a function defined on (−π, π) as

f (x) = (| sin x| + | cos x|) · sin x.

Then f is differentiable at

(A) all points.

(B) all points except at x = −π/2, π/2.
(C) all points except at x = 0.
(D) all points except at x = 0, −π/2, π/2.

6. The equation of the tangent to the curve y = sin2 (πx3 /6) at x = 1 is

√
1 3π
(A) y = 4 + 4 (x − 1).
√ √
3π 1− 3π
(B) y = 4 x+ 4 .
√ √
3π 1− 3π
(C) y = 4 x− 4 .
√
1 3π
(D) y = 4 − 4 (x − 1).

2
ROUGH WORK
7. Let f be a function defined from (0, ∞) to R such that

lim f (x) = 1 and f (x + 1) = f (x) for all x.

x→∞

Then f is

(A) continuous and bounded.

(B) continuous but not necessarily bounded.
(C) bounded but not necessarily continuous.
(D) neither necessarily continuous nor necessarily bounded.

1/x
8. The value of lim log x
x→∞

(A) is e. (B) is 0. (C) is 1. (D) does not exist.

9. The number of real solutions of the equation,

x7 + 5x5 + x3 − 3x2 + 3x − 7 = 0

is

(A) 5. (B) 7. (C) 3. (D) 1.

3
ROUGH WORK
10. Let x be a real number. Then
 
lim lim cos2n (m!πx)
m→∞ n→∞

(A) does not exist for any x.

(B) exists for all x.
(C) exists if and only if x is irrational.
(D) exists if and only if x is rational.

11. Let {an }n≥1 be a sequence such that a1 ≤ a2 ≤ · · · ≤ an ≤ · · · .

Suppose the subsequence {a2n }n≥1 is bounded. Then

(A) {a2n }n≥1 is always convergent but {a2n+1 }n≥1 need not be con-
vergent.
(B) both {a2n }n≥1 and {a2n+1 }n≥1 are always convergent and have
the same limit.
(C) {a3n }n≥1 is not necessarily convergent.
(D) both {a2n }n≥1 and {a2n+1 }n≥1 are always convergent but may
have different limits.

12. Let {an }n≥1 be a sequence of positive numbers such that an+1 ≤ an
for all n, and lim an = a. Let pn (x) be the polynomial
n→∞

pn (x) = x2 + an x + 1,

and suppose pn (x) has no real roots for every n. Let α and β be the
roots of the polynomial p(x) = x2 + ax + 1. Then

(A) α = β, α and β are not real.

(B) α = β, α and β are real.
(C) α 6= β, α and β are real.
(D) α 6= β, α and β are not real.

4
ROUGH WORK
13. Consider the set of all functions from {1, 2, . . . , m} to {1, 2, . . . , n},
where n > m. If a function is chosen from this set at random, what is
the probability that it will be strictly increasing?

n n m+n−1 m+n−1
/mn . /nm . /mn . /nm .





(A) m (B) m (C) m (D) m−1

14. A flag is to be designed with 5 vertical stripes using some or all of the
four colours: green, maroon, red and yellow. In how many ways can
this be done so that no two adjacent stripes have the same colour?

(A) 576. (B) 120. (C) 324. (D) 432.

15. Suppose x1 , . . . , x6 are real numbers which satisfy

Y
xi = xj , for all i = 1, . . . , 6.
j6=i

How many choices of (x1 , . . . , x6 ) are possible?

(A) Infinitely many. (B) 2. (C) 3. (D) 1.

5
ROUGH WORK
16. Suppose X is a random variable with P (X > x) = 1/x2 , for all x > 1.
The variance of Y = 1/X 2 is

(A) 1/4. (B) 1/12. (C) 1. (D) 1/2.

17. Let X ∼ N (0, σ 2 ), where σ > 0, and




−1 if X ≤ −1,

Y = X if X ∈ (−1, 1),



1 if X ≥ 1.


Which of the following statements is correct?

(A) Var(Y ) = Var(X).

(B) Var(Y ) < Var(X).
(C) Var(Y ) > Var(X).
(D) Var(Y ) ≥ Var(X) if σ ≥ 1, and Var(Y ) < Var(X) if σ < 1.

18. If a fair coin is tossed 5 times, what is the probability of obtaining at

least 3 consecutive heads?

(A) 1/8. (B) 5/16. (C) 1/4. (D) 3/16.

6
ROUGH WORK
19. Let X and Y be random variables with mean λ. Define

min(X, Y ) with probability 1 ,
2
Z=
max(X, Y ) with probability 1 .
2

What is E(Z)?

√
(A) λ. (B) 4λ/3. (C) λ2 . (D) 3λ/2.

20. A finite population has N (≥ 10) units marked {U1 , . . . , UN }. The

following sampling scheme was used to obtain a sample s. One unit
is selected at random: if this is the i-th unit, then the sample is s =
{Ui−1 , Ui , Ui+1 }, provided i 6∈ {1, N }. If i = 1 then s = {U1 , U2 } and
if i = N then s = {UN −1 , UN }. The probability of selecting U2 in s is

2 3 1 2 3
(A) N. (B) N. (C) (N −2) + N. (D) (N −2) .

21. Suppose X1 , . . . , Xn are i.i.d. observations from a distribution assum-

ing values −1, 1 and 0 with probabilities p, p and 1 − 2p, respectively,
where 0 < p < 21 . Define Zn = ni=1 Xi and an = P (Zn = 1),
Q

bn = P (Zn = −1), cn = P (Zn = 0). Then as n → ∞,

(A) an → 14 , bn → 12 , cn → 14 .
(B) an → 13 , bn → 13 , cn → 13 .
(C) an → 0, bn → 0, cn → 1.
(D) an → p, bn → p, cn → 1 − 2p.

7
ROUGH WORK
22. Suppose X1 , X2 and X3 are i.i.d. positive valued random variables.
Xi
Define Yi = X1 +X2 +X3 , i = 1, 2, 3. The correlation between Y1 and Y3
is

(A) 0. (B) −1/6. (C) −1/3. (D) −1/2.

23. Assume (yi , xi ) satisfies the linear regression model,

yi = βxi + i , for i = 1, . . . , n,

where, β ∈ R is unknown, {xi : 1 ≤ i ≤ n} are fixed constants and

{i : 1 ≤ i ≤ n} are i.i.d. errors with mean zero and variance σ 2 ∈
(0, ∞). Let βb be the least squares estimate of β and ybi = βx
b i be the
predicted value of yi . For each n ≥ 1, define
n
1 X
an = Cov(yi , ybi ).
σ2
i=1

Then,

(A) an = 1. (B) an ∈ (0, 1). (C) an = n. (D) an = 0.

24. Let X and Y be two random variables with E(X|Y = y) = y 2 , where

Y follows N (θ, θ2 ), with θ ∈ R. Then E(X) equals

(A) θ. (B) θ2 . (C) 2θ2 . (D) θ + θ2 .

8
ROUGH WORK
25. Suppose X is a random variable with finite variance. Define X1 = X,
X2 = αX1 , X3 = αX2 , . . . , Xn = αXn−1 , for 0 < α < 1. Then
Corr(X1 , Xn ) is

(A) αn . (B) 1. (C) 0. (D) αn−1 .

26. Let X be a random variable with P (X = 2) = P (X = −2) = 1/6 and

P (X = 1) = P (X = −1) = 1/3. Define Y = 6X 2 + 3. Then

(A) Var(X − Y ) < Var(X).

(B) Var(X − Y ) < Var(X + Y ).
(C) Var(X + Y ) < Var(X).
(D) Var(X − Y ) = Var(X + Y ).

27. Suppose X is a random variable on {0, 1, 2, . . .} with unknown p.m.f.

p(x). To test the hypothesis H0 : X ∼ Poisson(1/2) against H1 :
p(x) = 2−(x+1) for all x ∈ {0, 1, 2, . . .}, we reject H0 if x > 2. The
probability of type-II error for this test is

13 −1/2
(A) 41 . (B) 1 − 8 e . (C) 1 − 23 e−1/2 . (D) 78 .

9
ROUGH WORK
28. Let X be a random variable with
(1 − θ) 1 θ
Pθ (X = −1) = , Pθ (X = 0) = , and Pθ (X = 1) =
2 2 2
for 0 < θ < 1. In a random sample of size 20, the observed frequencies
of −1, 0 and 1 are 6, 4 and 10, respectively. The maximum likelihood
estimate of θ is

(A) 1/5. (B) 4/5. (C) 5/8. (D) 1/4.

29. Two judges evaluate n individuals, with (Ri , Si ) the ranks assigned to
the i-th individual by the two judges. Suppose there are no ties and
Si = Ri + 1, for i = 1, . . . , (n − 1), and Si = 1 if Ri = n. If the
Spearman’s rank correlation between the two evaluations is 0, what is
the value of n?

(A) 7. (B) 11. (C) 4. (D) 5.

30. Let X1 , X2 , . . . be a sequence of i.i.d. random variables with variance 2.

Then for all x,
n
!
1 X
lim P √ (−1)i Xi ≤ x
n→∞ n
i=1

equals

√ √

(A) Φ(x 2). (B) Φ x/ 2 . (C) Φ(x). (D) Φ(2x).

10
ROUGH WORK
 PSB-2018
GROUP A

1. Find all real solutions (x1 , x2 , x3 , λ) for the system of equations

x2 − 3x3 − x1 λ = 0,
x1 − 3x3 − x2 λ = 0,
x1 + x2 + x3 λ = 0.

2. Let {xn }n≥1 be a sequence defined by x1 = 1 and

 1 1/3
xn+1 = x3n + , n ≥ 1.
n(n + 1)(n + 2)

Show that {xn }n≥1 converges and find its limit.

3. Consider all permutations of the integers 1, 2, . . . , 100. In how many

of these permutations will the 25th number be the minimum of the
first 25 numbers and the 50th number be the minimum of the first 50
numbers?

GROUP B

4. An urn contains r > 0 red balls and b > 0 black balls. A ball is drawn
at random from the urn, its colour noted, and returned to the urn.
Further, c > 0 additional balls of the same colour are added to the
urn. This process of drawing a ball and adding c balls of the same
colour is continued. Define Xi = 1 if at the i-th draw the colour of the
ball drawn is red, and 0 otherwise. Compute E( ni=1 Xi ).
P

5. Suppose X1 and X2 are identically distributed random variables, not

necessarily independent, taking values in {1, 2}. If E(X1 X2 ) = 7/3
and E(X1 ) = 3/2, obtain the joint distribution of (X1 , X2 ).

6. A fair 6-sided die is rolled repeatedly until a 6 is obtained. Find the

expected number of rolls conditioned on the event that none of the
rolls yielded an odd number.

1
7. Suppose {(X1 , Y1 ), . . . , (Xn , Yn )} is a random sample from a bivariate
normal distribution with E(Xi ) = E(Yi ) = 0, Var(Xi ) = Var(Yi ) = 1
and unknown Corr(Xi , Yi ) = ρ ∈ (−1, 1), for all i = 1, . . . , n. Define
Wn = n1 ni=1 Xi Yi .
P

(a) Is Wn an unbiased estimator of ρ? Justify your answer.

(b) For large n, obtain an approximate level (1 − α) two-sided confi-
dence interval for ρ, where 0 < α < 1.

8. Let {X1 , . . . , Xn } be an i.i.d. sample from f (x : θ), θ ∈ {0, 1}, with

 
1 if 0 < x < 1,  √1 if 0 < x < 1,
f (x : 0) = and f (x : 1) = 2 x
0 otherwise, 0 otherwise.

Based on the above sample, obtain the most powerful test for testing
H0 : θ = 0 against H1 : θ = 1, at level α, with 0 < α < 1. Find the
critical region in terms of the quantiles of a standard distribution.

9. Suppose (yi , xi ) satisfies the regression model,

yi = α + βxi + i , for i = 1, . . . , n,

where {xi : 1 ≤ i ≤ n} are fixed constants and {i : 1 ≤ i ≤ n} are i.i.d.

N (0, σ 2 ) errors, where α, β and σ 2 (> 0) are unknown parameters.

(a) Let α
e denote the least squares estimate of α obtained assuming
β = 5. Find the mean squared error (MSE) of α
e in terms of the
model parameters.
(b) Obtain the maximum likelihood estimator of this MSE.

2
 PSA-2017
1. Let f : R → R be a nonconstant function satisfying f (x + 2) = f (x)
for every real x. Then lim f (x)
x→∞

(A) does not exist.

(B) exists and equals +∞ or −∞.
(C) exists and is finite.
(D) may or may not exist depending on f .

2. Let A be a square matrix with real entries such that A2 = A. Then

(A) (A + I)10 = I + 1024A.

(B) (A + I)10 = I + 511A.
(C) (A + I)10 = I + 1023A.
(D) (A + I)10 = I + 512A.

3. Let A be a 2 × 2 matrix such that trace(A) = det(A) = 3. What is

trace(A−1 )?

(A) 1/2 (B) 3 (C) 1 (D) 1/3

1
ROUGH WORK
4. Let A be a nonzero 2 × 2 matrix such that A3 = 0. Then which of the
following statements is always false?

(A) trace(A − A2 ) = 0
(B) trace(A + A2 ) = 0
(C) det(I + A) = 0
(D) A2 = 0

5. The vertices A, B, C, D of a square are to be coloured with one of three

colours red, blue, or green such that adjacent vertices get different
colours. What is the number of such colourings?

(A) 18 (B) 12 (C) 20 (D) 24

6. Let S ⊂ R. Consider the statement: “If f is a continuous function

from S to S, then f (x) = x for some x.” This statement is true if S
equals

(A) [0, 1]. (B) (0, 1]. (C) R. (D) [−3, −2] ∪ [2, 3].

2
ROUGH WORK
7. The coordinates of the point at which the tangent of the curve
y = x2 − 1 makes an angle of 45◦ with the positive X-axis are
√ √
5−1 1− 5
(A) ( 2 , 2 ). (B) ( 23 , − 59 ). (C) ( 34 , − 16
7
). (D) ( 12 , − 43 ).

8. Let f : R → R be defined by

x2 sin 1


x if x 6= 0 ,
f (x) =
0 if x = 0 .

Then f is

(A) continuous everywhere but not differentiable at 0.

(B) differentiable everywhere and its derivative is continuous.
(C) discontinuous at 0.
(D) differentiable everywhere and its derivative is discontinuous at 0.

9. If α, β, and γ are the roots of x3 − px + q = 0, then what is the

determinant of  
α β γ
 β γ α ?
 

γ α β

(A) p2 + 6q (B) 1 (C) p (D) 0

3
ROUGH WORK
10. Suppose the rank of  
1 1 2 2
 1 1 1 3 
 

a b b 1
is 2 for some real numbers a and b. What is the value of b ?

1 1
(A) 3 (B) 3 (C) 1 (D) 2

11. What is the infinite sum x + 2x2 + 3x3 + · · · for |x| < 1 ?

x x 1 x
(A) 1−x2
(B) (1+x)2
(C) (1−x)2
(D) (1−x)2

12. A function f : R → R, such that f (x) = f (−x) for all x, has left
derivative 5 at x = 0. Then, the right derivative of f at x = 0

(A) exists and equals −5.

(B) may or may not exist.
(C) does not exist.
(D) exists and equals 5.

4
ROUGH WORK
13. Let f : R → R be defined by
b|x|c
X 1
f (x) = ,
2k
k=0

where byc denotes the greatest integer less than or equal to y. Then,

(A) f is a left continuous function which is not right continuous.

(B) f is a continuous function.
(C) f is neither left continuous nor right continuous.
(D) f is a right continuous function which is not left continuous.

14. A function f : R → R is differentiable at 0. Suppose that

f (x) < f (0) < f (y) for all x < 0 < y.

Then, what is the set of all possible values that f 0 (0) can take?

(A) {0} (B) [0, ∞) (C) (0, ∞) (D) R

15. Suppose that the events A, B, and C are pairwise independent such
that each of them occurs with probability p. Assume that all three of
them cannot occur simultaneously. What is P (A ∪ B ∪ C) ?

(A) 1 − (1 − p)3 (B) p3 (C) 3p(1 − p) (D) 3p

5
ROUGH WORK
16. Let X1 , X2 , . . . be a sequence of independent random variables dis-
tributed exponentially with mean 1. Suppose that N is a random
variable, independent of the Xi -s, that has a Poisson distribution with
mean λ > 0. What is the expected value of X1 + X2 + · · · + XN 2 ?

(A) N 2 (B) λ + λ2 (C) λ2 (D) 1/λ2

17. For which value of k does the inequality

Var(X − Y ) ≥ |σX − σY |k

hold for all choices of random variables X and Y with finite variances
2 and σ 2 , respectively?
σX Y

(A) 3 (B) 4 (C) 1 (D) 2

18. Consider a diagnostic test for a disease that gives the correct result with
probability 0.9, independently of whether the subject being diagnosed
has the disease or not. Suppose that 20% of the population has the
disease. For a particular subject, given that the test indicates presence
of the disease, what is the conditional probability that the subject has
the disease?

9 9 1 1
(A) 10 (B) 13 (C) 2 (D) 5

6
ROUGH WORK
19. Let X1 , X2 , X3 , X4 be independent standard normal random variables.
Then what is the distribution of
(X1 + X2 + X3 + X4 )2
?
(X1 − X2 + X3 − X4 )2

(A) F4,4 (B) χ24 (C) F1,1 (D) F2,2

20. Suppose that X ∼ Uniform (0, 1) and Y ∼ Bernoulli (1/4), indepen-

dently of each other. Let Z = X + Y . Then,

(A) the distribution of Z is symmetric about 1.

(B) Z has a probability density function.
(C) E(Z) = 5/4.
(D) P (Z ≤ 1) = 1/4.

21. Assume that the events A and B are such that A ⊂ B and P (B) > 0.
For every event E, define 1E to be the random variable which takes
value 1 or 0 depending on whether E occurs or not, respectively. Then,
Cov(1A , 1B ) equals

(A) P (A|B)P (B c ).
(B) P (A)P (B c ).
(C) P (A)P (B).
(D) P (B)P (Ac ).

7
ROUGH WORK
22. Suppose that X is chosen uniformly from {1, 2, . . . , 100}, and given
X = x, Y is chosen uniformly from {1, 2, . . . , x}. What is P (Y = 30) ?

1
(A) 100
70 1
(B) 100 × 30
1 1 1 1


(C) 100 × 1+ 2 + + ··· +
3 30
1 1 1


(D) 100 × 30 + · · · + 100

23. Three numbers are chosen at random from {1, 2, ..., 10} without re-
placement. What is the probability that the minimum of the chosen
numbers is 3 or their maximum is 7 ?

13 19 3 11
(A) 40 (B) 60 (C) 10 (D) 40

24. Let X be a random variable taking values 1, 2, and 3 with respective

probabilities 73 , 27 , and 27 . Find the set of values of α for which E|X −α|
is minimized.

(A) [1, 2] (B) { 13

7 } (C) {2} (D) {1}

8
ROUGH WORK
25. Suppose that X1 , . . . , Xn are independent and identically distributed
normal random variables with mean θ and variance θ, where θ > 0.
Then,
n
X
(A) Xi is sufficient for θ.
i=1
Xn
(B) Xi2 is sufficient for θ.
i=1
n
X
Xi + Xi2 is sufficient for θ.


(C)
i=1
n
X n
X
(D) neither Xi nor Xi2 is sufficient for θ.
i=1 i=1

26. Let X be normally distributed with mean µ and variance σ 2 > 0.

What is the variance of eX ?
2 2
(A) eµ+σ (eσ − 1)
2 2
(B) e2µ+2σ (e2σ − 1)
2 2
(C) e2µ+2σ (eσ − 1)
2 2
(D) e2µ+σ (eσ − 1)

27. Consider data from a single replication of a 24 factorial experiment.

Assume that all interactions of orders higher than two are negligible.
What is the error degrees of freedom in the ANOVA table?

(A) 16 (B) 2 (C) 12 (D) 5

9
ROUGH WORK
28. For n ≥ 2, let X1 , . . . , Xn be independent and identically distributed
normal random variables with mean 0 and variance σ 2 . Consider the
most powerful test for the null hypothesis H0 : σ = 2 against the
alternative H1 : σ = 1, at significance level α for some 0 < α < 1.
What is the power of this test?
In what follows, Fm denotes the cumulative distribution function of a
χ2 random variable with m degrees of freedom, and χ2m,β satisfies the
equation Fm (χ2m,β ) = 1 − β, for every m ≥ 1 and β ∈ (0, 1).

(A) 1 − Fn−1 ( 41 χ2n−1,α )

(B) Fn (4χ2n,1−α )
(C) Fn−1 (4χ2n−1,1−α )
(D) 1 − Fn ( 14 χ2n,α )

29. Suppose that X1 , . . . , Xn are independent and identically distributed

random variables from a distribution which depends on a parameter θ.
For which of the following distributions is the maximum likelihood
estimator of θ not unbiased?

(A) Normal with mean θ and variance 1 where θ ∈ R

(B) Bernoulli with parameter θ ∈ [0, 1]
(C) Poisson with mean θ ≥ 0
(D) Normal with mean 0 and variance θ2 where θ > 0

30. Suppose that X1 , . . . , Xn are independent observations from a uniform

distribution on [0, θ], where θ ∈ N. Then, the maximum likelihood
estimator of θ

(A) may not exist in some cases.

(B) is bX(n) c, where bac = greatest integer ≤ a.
(C) is dX(n) e, where dae = smallest integer ≥ a.


(D) is bX(n) c, if X(n) − bX(n) c < 1/2, otherwise it is dX(n) e.

10
 PSB-2017
GROUP A

1. Let a and b be real numbers. Show that there exists a unique 2 × 2

real symmetric matrix A with trace(A) = a and det(A) = b if and only
if a2 = 4b.

2. Let f : R → R be an infinitely differentiable function and suppose that

for some n ≥ 1,

f (1) = f (0) = f (1) (0) = f (2) (0) = · · · = f (n) (0) = 0,

where f (k) denotes the k-th derivative of f for k ≥ 1. Prove that there
exists x ∈ (0, 1) such that f (n+1) (x) = 0.

3. Consider an urn containing 5 red, 5 black, and 10 white balls. If balls

are drawn without replacement from the urn, calculate the probability
that in the first 7 draws, at least one ball of each colour is drawn.

GROUP B

4. Let X1 , X2 , . . . , Xn be independent random variables, with Xi having

probability mass function
 k
i 1
P (Xi = k) = , for k = 0, 1, 2, . . .
i+1 i+1
and for all i = 1, . . . , n. Let M = min{Xi : 1 ≤ i ≤ n}. Derive the
probability mass function of M .

5. The lifetime in hours of each bulb manufactured by a particular com-

pany follows an independent exponential distribution with mean λ.
To test the null hypothesis H0 : λ = 1000 against the alternative
H1 : λ = 500, a statistician sets up an experiment with 50 bulbs, with
5 bulbs in each of 10 different locations, to examine their lifetimes.
To get quick preliminary results, the statistician decides to stop the
experiment as soon as one bulb fails at each location. Let Yi denote
the lifetime of the first bulb to fail at location i. Obtain the most
powerful test of H0 against H1 based on Y1 , Y2 , . . . , Y10 , and compute
its power.

1
6. Suppose you have a 4-digit combination lock, but you have forgotten
the correct combination. Consider the following three strategies to find
the correct one:

(i) Try the combinations consecutively from 0000 to 9999.

(ii) Try combinations using simple random sampling with replace-
ment from the set of all possible combinations.
(iii) Try combinations using simple random sampling without replace-
ment from the set of all possible combinations.

Assume that the true combination was chosen uniformly at random

from all possible combinations. Determine the expected number of
attempts needed to find the correct combination in all three cases.

7. Consider independent observations {(yi , x1i , x2i ) : 1 ≤ i ≤ n} from the

regression model

yi = β1 x1i + β2 x2i + i , i = 1, . . . , n ,

where x1i and x2i are scalar covariates, β1 and β2 are unknown scalar
coefficients, and i are uncorrelated errors with mean 0 and variance
σ 2 > 0. Instead of using the correct model, we obtain an estimate β̂1
of β1 by minimizing
n
X
(yi − β1 x1i )2 .
i=1

Find the bias and mean squared error of β̂1 .

8. Let θ > 0 be an unknown parameter, and X1 , X2 , . . . , Xn be a random

sample from the distribution with density

2x/θ2 , 0 ≤ x ≤ θ ,
f (x) =
0 , otherwise.

Find the maximum likelihood estimator of θ and its mean squared

error.

2
TEST CODE: PSA (Objective type) and PSB (Short answer type) 2016

SYLLABUS

Mathematics
Arithmetic, geometric and harmonic progressions. Trigonometry. Two dimensional
coordinate geometry: Straight lines, circles, parabolas, ellipses and hyperbolas.
Elementary set theory. Functions and relations. Elementary combinatorics: Per-
mutations and combinations, Binomial and multinomial theorem.
Theory of equations.
Complex numbers and De Moivre’s theorem.
Vectors and vector spaces. Algebra of matrices. Determinant, rank, trace and
inverse of a matrix. Solutions of linear equations. Eigenvalues and eigenvectors of
matrices.
Limits and continuity of functions of one variable. Differentiation. Leibnitz for-
mula. Applications of differential calculus, maxima and minima. Taylor’s theorem.
Indefinite integral. Fundamental theorem of calculus. Riemann integration and prop-
erties. Improper integrals.

Statistics and Probability

Notions of sample space and probability. Combinatorial probability. Conditional
probability and independence. Bayes Theorem. Random variables and expectations.
Moments and moment generating functions. Standard univariate discrete and con-
tinuous distributions. Distribution of functions of a random variable. Distribution of
order statistics. Joint probability distributions. Marginal and conditional probability
distributions. Multinomial distribution. Bivariate normal and multivariate normal
distributions.
Sampling distributions of statistics. Statement and applications of Weak law of
large numbers and Central limit theorem.
Descriptive statistical measures. Contingency tables and measures of association.
Product moment and other types of correlation. Partial and multiple correlation.
Simple and multiple linear regression.
Elementary theory of estimation (unbiasedness, minimum variance, sufficiency).
Methods of estimation (maximum likelihood method, method of moments). Tests
of hypotheses (basic concepts and simple applications of Neyman-Pearson Lemma).
Confidence intervals. Inference related to regression. ANOVA. Elements of nonpara-
metric inference.

1
 PSA-2016
Basic experimental designs such as CRD, RBD, LSD and their analyses. Ele-
ments of factorial designs. Conventional sampling techniques (SRSWR/SRSWOR)
including stratification. Ratio and regression methods of estimation.

SAMPLE QUESTIONS: PSA

Note: For each question there are four suggested answers of which only one is correct.

1. The number of functions f : {1, 2, . . . , 10} → {1, 2, . . . , 10} such that f (x) 6= x
for all x is

(A) 10! (B) 910 (C) 109 (D) 1010 − 1.

2. The set of all ordered pairs of real numbers (x, y) satisfying satisfying y 2 − 2y −
x2 + 4x = 3 is a

(A) circle (B) point (C) hyperbola (D) pair of straight lines.

3. Let
log(2 + x) − x2n sin x
f (x) = lim for x > 0.
n→∞ 1 + x2n
Then
(A) f is continuous at x = 1
(B) lim f (x) 6= lim f (x)
x→1+ x→1−
(C) lim f (x) = sin 1
x→1+
(D) lim f (x) does not exist.
x→1−

4. Suppose a real matrix A satisfies A3 = A, A 6= I, A 6= 0. If Rank(A) = r and

Trace(A) = t, then
(A) r ≥ t and r + t is odd
(B) r ≥ t and r + t is even
(C) r < t and r + t is odd
(D) r < t and r + t is even.

5. Let
1 1
f (x) = x2 + 2
+x+ , x>0
x x
and let m = min{f (x)}. Then

(A) m = 1 (B) m = 4 (C) m = 27/4 (D) m does not exist.

2
 6. Let f be a convex function, i.e.,

f (tx + (1 − t)y) ≤ tf (x) + (1 − t)f (y)

for all 0 ≤ t ≤ 1 and x, y ∈ R. Then which of the following is necessarily true?

(A) 2f (0) + f (4) ≥ 2f (1) + f (2)
(B) f g is a convex function whenever g is convex
(C) f is nondecreasing
(D) none of these.

7. Suppose A is a 100 × 100 real symmetric matrix whose diagonal entries are all
positive. Then which of the following is necessarily true?
(A) All eigenvalues of A are greater than 0
(B) no eigenvalue of A is greater than 0
(C) at least one eigenvalue of A is greater than 0
(D) none of these.

8. The integral
Z 1
sin x
dx
0 xα

(A) is finite only for α = 0

(B) is finite only for |α| < 1
(C) is finite for all α < 2
(D) is infinite for any value of α.

9. Given θ in the range 0 ≤ θ < π, the equation

2x2 + 2y 2 + 4x cos θ + 8y sin θ + 5 = 0

represents a circle for all θ in the interval

(A) 0 < θ < π/3 (B) π/4 < θ < 3π/4

(C) 0 < θ < π/2 (D) 0 ≤ θ < π.

10. How many 5 × 5 matrices are there such that each entry is 0 or 1 and each row
sum and each column sum is 4?

(A) 64 (B) 32 (C) 120 (D) 96.

3
 π k
11. For n ≥ 1, let Gn be the geometric mean of {sin( · ) : 1 ≤ k ≤ n}. Then
2 n
lim Gn is
n→∞

1
(A) 1/4 (B) log 2 (C) 2 log 2 (D) 1/2.

12. Suppose a, b, x, y are real numbers such that a2 + b2 = 81, x2 + y 2 = 121 and
ax + by = 99. Then the set of all possible values of ay − bx is

     
9 9 9
(A) {0} (B) 0, (C) 0, (D) ,∞ .
11 11 11

13. In a triangle with sides of length a, b, c, suppose b + c = x and bc = y. If also

(x + a)(x − a) = y, then the triangle is necessarily

(A) equilateral (B) right angled

(C) acute angled (D) obtuse.

14. Three distinct squares are selected at random from a 8 × 8 chess board. Then
the probability that they form an L-shaped pattern (looked at from one fixed
side only) as drawn below is

196 49 36
(A) 64

(B) 64

(C) 64

(D) greater than 1/2.
3 3 3

15. Suppose X is distributed as Poisson with mean λ. Then E(1/(X + 1)) is

eλ − 1 1 1 − e−λ 1 − e−λ
(A) (B) (C) (D) .
λ λ+1 λ λ+1

16. A permutation of 1, 2, . . . , 100 is chosen at random. Then the probability that

the numbers 1 and 100 appear next to each other equals

(A) 1/100 (B) 1/50 (C) 1/99 (D) 1/98.

4
17. There are 10 boxes each containing 6 white and 7 red balls. Two different boxes
are chosen at random, one ball is drawn simultaneously at random from each
and transferred to the other box. Now a box is again chosen from the 10 boxes
and a ball is chosen from it. Then the probability that this ball is white is

(A) 6/13 (B) 7/13 (C) 5/13 (D) none of these.

18. The number of cars (X) arriving at a service station per day is a random variable
with mean 4. The service station can provide service to a maximum of 4 cars
per day. Then the expected number of cars per day that do not get serviced
equals

∞
X ∞
X
(A) 4 (B) 0 (C) iP (X = i + 4) (D) iP (X = i − 4).
i=0 i=4

19. Suppose X1 and X2 are independent random variables distributed as Ber(p1 )

and Ber(p2 ) respectively. Then Y = max(X1 , X2 ) is distributed as a Bernoulli
random variable with success probability

(A) 1 − p1 p2 (B) p1 + p2 − p1 p2 (C) max{p1 , p2 } (D) min{1 − p1 , 1 − p2 }.

20. Assume that (X, Y ) is bivariate Normal with E(X) = E(Y ) = 0, Var(X) = σ12 ,
Var(Y ) = σ22 and Cor(X, Y ) = ρ for some ρ ∈ (−1, 1). The probability that X
is larger than Y is

   
σ1 −σ2 σ12 +σ22
(A) 1/2 (B) Φ √ (C) Φ √ (D) none of the above.
1−ρ2 1−ρ2

21. Suppose that the bivariate data (x1 , y1 ), . . . , (xn , yn ) lie on the straight line
y = a + bx for some a, b ∈ R. Assume further that neither all the xi ’s are same,
nor are all the yi ’s. Which of the following values is not a possibility for the
correlation coefficient calculated from the above data?

(A) 1 (B) −1 (C) 0 (D) None of the above.

22. In the randomised block design for ANOVA where k is the number of treatments
and b is the number of blocks, the degrees of freedom for error is given by

(A) bk − 1 (B) kb + 1 (C) (b − 1)(k − 1) (D) k + b − 1.

5
23. Suppose X1 , X2 , . . . , Xn is a random sample from a population with mean µ
and finite variance. Consider the following two estimators of µ2 :

1 XX
E1 = Xi Xj
n(n − 1)
1≤i6=j≤n
1 X 2
E2 = X̄ 2 − (Xi − X̄) .
n(n − 1)
1≤i≤n

Then,

(A) E1 is unbiased but E2 is biased

(B) E1 is biased but E2 is unbiased
(C) both E1 and E2 are unbiased
(D) both E1 and E2 are biased.

24. It is known that the proportion of smokers (p) in a population lies in the interval
[1/3, 2/3]. In a random sample of N individuals selected from the population, it
was found that M were smokers. The maximum likelihood estimate of p based
on the above data is

(A) max{1/3, M/N }

(B) min{M/N, 2/3}
(C) M/N
(D) none of the above.

25. Suppose that the least squares linear regression equation of y on x is y = a + bx

and that of x on y is x = c + dy. If it is known that b 6= 0 and d 6= 0, then the
ratio of the standard deviation of x to the standard deviation of y

(A) is √1
bd
p
(B) is d/b
p
(C) is b/d
(D) cannot be determined from the above information.

26. Let bxc denote the largest integer not larger than x. If X is distributed as
N (0, 1), then E(bXc)

(A) is 0 (B) is 0.5 (C) is −0.5 (D) does not exist.

6
27. X and Y are i.i.d. random variables with finite variances. Then

(A) V ar(XY ) = V ar(X)V ar(Y )

(B) V ar(XY ) ≥ V ar(X)V ar(Y )
(C) V ar(XY ) ≤ V ar(X)V ar(Y )
(D) none of the above is necessarily true.

28. Suppose X1 , X2 , . . . , Xn is a random sample of size n from the probability

density
αp p−1 −αx
f (x) = x e ,x > 0
Γ(p)
where p is a known positive constant and α > 0 is an unknown parameter. Let
α̂ = p/X̄ be a proposed estimator of α. Then,

(A) E(α̂) = α
α
(B) E(α̂) = 1
1− np
α
(C) E(α̂) = 1−np

(D) none of the above statements is true.

29. The sign test is a nonparametric procedure for testing

(A) whether two populations have identical means.

(B) whether two populations have identical medians.
(C) whether two populations have identical probability distributions.
(D) whether two populations are independent.

30. Let X1 , . . . , Xn be i.i.d. from N (0, σ 2 ). What is the form of the most powerful
test for testing the null hypothesis

H0 : σ = 1 ,

against the alternative

H1 : σ = 2 ?
Pn
(A) Reject H0 if i=1 Xi2 > c
Pn 2
(B) Reject H0 if i=1 Xi <c
Pn
(C) Reject H0 if i=1 (Xi − X̄)2 > c
Pn
(D) Reject H0 if i=1 (Xi − X̄)2 < c.

7
 PSB-2016
SAMPLE QUESTIONS: PSB

1. In the diagram below, L(x) is a straight line that intersects the graph of a
polynomial P (x) of degree 2 at the points A = (−1, 0) and B = (5, 12). The
area of the shaded region is 36 square units.
Obtain the expression for P (x).

L(x)

A
P (x)

2. Let f : [−1, 1] → R be a continuous function. Suppose that f 0 (x) exists and

f 0 (x) ≤ 1 for all x ∈ (−1, 1). If f (1) = 1 and f (−1) = −1, prove that

f (x) = x for all x ∈ [−1, 1].

3. Suppose A is an n × n real symmetric matrix such that

T r(A2 ) = T r(A) = n.

Show that all the eigenvalues of A are equal to 1.

4. For each c ∈ R, define a function Tc : R4 → R4 by

Tc x1 , x2 , x3 , x4 := (1 + c)x1 , x2 + cx3 , x3 + cx2 , (1 + c)x4 .

For every c ∈ R, find the dimension of the null space of Tc .

5. A box contains 50 red balls, 30 green balls and 20 blue balls. Suppose balls are
drawn successively at random with replacement from the box. Let N denote
the minimum number of draws required to obtain balls of all three colours.
Compute P (N > n) for all positive integers n.

8
 6. Let X1 , . . . , Xn be i.i.d. random variables from a continuous distribution whose
density is symmetric around 0. Suppose E(|X1 |) = 2. Define
n
X n
X
Y = Xi and Z= 1(Xi > 0) .
i=1 i=1

Calculate the covariance between Y and Z.

7. Suppose {(yi , x1i , x2i , . . . , xki ) : i = 1, 2, . . . , n1 + n2 } represents a set of multi-

variate observations. It is found that the least squares linear regression fit of y
on (x1 , . . . , xk ) based on the first n1 observations is the same as that based on
the remaining n2 observations, and is given by
k
X
y = β̂0 + β̂j xj .
j=1

If the regression is now performed using all (n1 + n2 ) observations, will the
regression equation remain the same? Justify your answer.

8. Suppose that (X1 , Y1 ), (X2 , Y2 ), . . . , (Xn , Yn ) are the coordinates of n points

chosen independently and uniformly at random within a circle with centre (0, 0)
and unknown radius r.
(a) Obtain the MLE r̂n of r.
(b) Examine whether r̂n is sufficient for r.
(c) For any ε > 0, show that lim P (|r̂n − r| > ε) = 0.
n→∞

9. Suppose X1 , X2 , . . . , Xn is a random sample from an exponential distribution

with mean λ. Assume that the observed data is available on [X1 ], . . . , [Xn ],
instead of X1 , . . . , Xn , where [x] denotes the largest integer less than or equal
to x. Consider a test for H0 : λ = 1 vs H1 : λ > 1 which rejects H0 when
Xn
[Xi ] > cn . Given α ∈ (0, 1), obtain values of cn such that the size of the test
i=1
converges to α as n → ∞.

10. A cake weighing one kilogram is cut into two pieces, and each piece is weighed
separately. Denote the measured weights of the two pieces by X and Y . Assume
that the errors in obtaining X and Y are independent and normally distributed
with mean zero and the same (unknown) variance. Devise a test for the hy-
pothesis that the true weights of the two pieces are equal.

9
TEST CODE: PSA (Objective type) and PSB (Short answer type) 2016

SYLLABUS

Mathematics
Arithmetic, geometric and harmonic progressions. Trigonometry. Two dimensional
coordinate geometry: Straight lines, circles, parabolas, ellipses and hyperbolas.
Elementary set theory. Functions and relations. Elementary combinatorics: Per-
mutations and combinations, Binomial and multinomial theorem.
Theory of equations.
Complex numbers and De Moivre’s theorem.
Vectors and vector spaces. Algebra of matrices. Determinant, rank, trace and
inverse of a matrix. Solutions of linear equations. Eigenvalues and eigenvectors of
matrices.
Limits and continuity of functions of one variable. Differentiation. Leibnitz for-
mula. Applications of differential calculus, maxima and minima. Taylor’s theorem.
Indefinite integral. Fundamental theorem of calculus. Riemann integration and prop-
erties. Improper integrals.

Statistics and Probability

Notions of sample space and probability. Combinatorial probability. Conditional
probability and independence. Bayes Theorem. Random variables and expectations.
Moments and moment generating functions. Standard univariate discrete and con-
tinuous distributions. Distribution of functions of a random variable. Distribution of
order statistics. Joint probability distributions. Marginal and conditional probability
distributions. Multinomial distribution. Bivariate normal and multivariate normal
distributions.
Sampling distributions of statistics. Statement and applications of Weak law of
large numbers and Central limit theorem.
Descriptive statistical measures. Contingency tables and measures of association.
Product moment and other types of correlation. Partial and multiple correlation.
Simple and multiple linear regression.
Elementary theory of estimation (unbiasedness, minimum variance, sufficiency).
Methods of estimation (maximum likelihood method, method of moments). Tests
of hypotheses (basic concepts and simple applications of Neyman-Pearson Lemma).
Confidence intervals. Inference related to regression. ANOVA. Elements of nonpara-
metric inference.

1
 PSA-2016
Basic experimental designs such as CRD, RBD, LSD and their analyses. Ele-
ments of factorial designs. Conventional sampling techniques (SRSWR/SRSWOR)
including stratification. Ratio and regression methods of estimation.

SAMPLE QUESTIONS: PSA

Note: For each question there are four suggested answers of which only one is correct.

1. The number of functions f : {1, 2, . . . , 10} → {1, 2, . . . , 10} such that f (x) 6= x
for all x is

(A) 10! (B) 910 (C) 109 (D) 1010 − 1.

2. The set of all ordered pairs of real numbers (x, y) satisfying satisfying y 2 − 2y −
x2 + 4x = 3 is a

(A) circle (B) point (C) hyperbola (D) pair of straight lines.

3. Let
log(2 + x) − x2n sin x
f (x) = lim for x > 0.
n→∞ 1 + x2n
Then
(A) f is continuous at x = 1
(B) lim f (x) 6= lim f (x)
x→1+ x→1−
(C) lim f (x) = sin 1
x→1+
(D) lim f (x) does not exist.
x→1−

4. Suppose a real matrix A satisfies A3 = A, A 6= I, A 6= 0. If Rank(A) = r and

Trace(A) = t, then
(A) r ≥ t and r + t is odd
(B) r ≥ t and r + t is even
(C) r < t and r + t is odd
(D) r < t and r + t is even.

5. Let
1 1
f (x) = x2 + 2
+x+ , x>0
x x
and let m = min{f (x)}. Then

(A) m = 1 (B) m = 4 (C) m = 27/4 (D) m does not exist.

2
 6. Let f be a convex function, i.e.,

f (tx + (1 − t)y) ≤ tf (x) + (1 − t)f (y)

for all 0 ≤ t ≤ 1 and x, y ∈ R. Then which of the following is necessarily true?

(A) 2f (0) + f (4) ≥ 2f (1) + f (2)
(B) f g is a convex function whenever g is convex
(C) f is nondecreasing
(D) none of these.

7. Suppose A is a 100 × 100 real symmetric matrix whose diagonal entries are all
positive. Then which of the following is necessarily true?
(A) All eigenvalues of A are greater than 0
(B) no eigenvalue of A is greater than 0
(C) at least one eigenvalue of A is greater than 0
(D) none of these.

8. The integral
Z 1
sin x
dx
0 xα

(A) is finite only for α = 0

(B) is finite only for |α| < 1
(C) is finite for all α < 2
(D) is infinite for any value of α.

9. Given θ in the range 0 ≤ θ < π, the equation

2x2 + 2y 2 + 4x cos θ + 8y sin θ + 5 = 0

represents a circle for all θ in the interval

(A) 0 < θ < π/3 (B) π/4 < θ < 3π/4

(C) 0 < θ < π/2 (D) 0 ≤ θ < π.

10. How many 5 × 5 matrices are there such that each entry is 0 or 1 and each row
sum and each column sum is 4?

(A) 64 (B) 32 (C) 120 (D) 96.

3
 π k
11. For n ≥ 1, let Gn be the geometric mean of {sin( · ) : 1 ≤ k ≤ n}. Then
2 n
lim Gn is
n→∞

1
(A) 1/4 (B) log 2 (C) 2 log 2 (D) 1/2.

12. Suppose a, b, x, y are real numbers such that a2 + b2 = 81, x2 + y 2 = 121 and
ax + by = 99. Then the set of all possible values of ay − bx is

     
9 9 9
(A) {0} (B) 0, (C) 0, (D) ,∞ .
11 11 11

13. In a triangle with sides of length a, b, c, suppose b + c = x and bc = y. If also

(x + a)(x − a) = y, then the triangle is necessarily

(A) equilateral (B) right angled

(C) acute angled (D) obtuse.

14. Three distinct squares are selected at random from a 8 × 8 chess board. Then
the probability that they form an L-shaped pattern (looked at from one fixed
side only) as drawn below is

196 49 36
(A) 64

(B) 64

(C) 64

(D) greater than 1/2.
3 3 3

15. Suppose X is distributed as Poisson with mean λ. Then E(1/(X + 1)) is

eλ − 1 1 1 − e−λ 1 − e−λ
(A) (B) (C) (D) .
λ λ+1 λ λ+1

16. A permutation of 1, 2, . . . , 100 is chosen at random. Then the probability that

the numbers 1 and 100 appear next to each other equals

(A) 1/100 (B) 1/50 (C) 1/99 (D) 1/98.

4
17. There are 10 boxes each containing 6 white and 7 red balls. Two different boxes
are chosen at random, one ball is drawn simultaneously at random from each
and transferred to the other box. Now a box is again chosen from the 10 boxes
and a ball is chosen from it. Then the probability that this ball is white is

(A) 6/13 (B) 7/13 (C) 5/13 (D) none of these.

18. The number of cars (X) arriving at a service station per day is a random variable
with mean 4. The service station can provide service to a maximum of 4 cars
per day. Then the expected number of cars per day that do not get serviced
equals

∞
X ∞
X
(A) 4 (B) 0 (C) iP (X = i + 4) (D) iP (X = i − 4).
i=0 i=4

19. Suppose X1 and X2 are independent random variables distributed as Ber(p1 )

and Ber(p2 ) respectively. Then Y = max(X1 , X2 ) is distributed as a Bernoulli
random variable with success probability

(A) 1 − p1 p2 (B) p1 + p2 − p1 p2 (C) max{p1 , p2 } (D) min{1 − p1 , 1 − p2 }.

20. Assume that (X, Y ) is bivariate Normal with E(X) = E(Y ) = 0, Var(X) = σ12 ,
Var(Y ) = σ22 and Cor(X, Y ) = ρ for some ρ ∈ (−1, 1). The probability that X
is larger than Y is

   
σ1 −σ2 σ12 +σ22
(A) 1/2 (B) Φ √ (C) Φ √ (D) none of the above.
1−ρ2 1−ρ2

21. Suppose that the bivariate data (x1 , y1 ), . . . , (xn , yn ) lie on the straight line
y = a + bx for some a, b ∈ R. Assume further that neither all the xi ’s are same,
nor are all the yi ’s. Which of the following values is not a possibility for the
correlation coefficient calculated from the above data?

(A) 1 (B) −1 (C) 0 (D) None of the above.

22. In the randomised block design for ANOVA where k is the number of treatments
and b is the number of blocks, the degrees of freedom for error is given by

(A) bk − 1 (B) kb + 1 (C) (b − 1)(k − 1) (D) k + b − 1.

5
23. Suppose X1 , X2 , . . . , Xn is a random sample from a population with mean µ
and finite variance. Consider the following two estimators of µ2 :

1 XX
E1 = Xi Xj
n(n − 1)
1≤i6=j≤n
1 X 2
E2 = X̄ 2 − (Xi − X̄) .
n(n − 1)
1≤i≤n

Then,

(A) E1 is unbiased but E2 is biased

(B) E1 is biased but E2 is unbiased
(C) both E1 and E2 are unbiased
(D) both E1 and E2 are biased.

24. It is known that the proportion of smokers (p) in a population lies in the interval
[1/3, 2/3]. In a random sample of N individuals selected from the population, it
was found that M were smokers. The maximum likelihood estimate of p based
on the above data is

(A) max{1/3, M/N }

(B) min{M/N, 2/3}
(C) M/N
(D) none of the above.

25. Suppose that the least squares linear regression equation of y on x is y = a + bx

and that of x on y is x = c + dy. If it is known that b 6= 0 and d 6= 0, then the
ratio of the standard deviation of x to the standard deviation of y

(A) is √1
bd
p
(B) is d/b
p
(C) is b/d
(D) cannot be determined from the above information.

26. Let bxc denote the largest integer not larger than x. If X is distributed as
N (0, 1), then E(bXc)

(A) is 0 (B) is 0.5 (C) is −0.5 (D) does not exist.

6
27. X and Y are i.i.d. random variables with finite variances. Then

(A) V ar(XY ) = V ar(X)V ar(Y )

(B) V ar(XY ) ≥ V ar(X)V ar(Y )
(C) V ar(XY ) ≤ V ar(X)V ar(Y )
(D) none of the above is necessarily true.

28. Suppose X1 , X2 , . . . , Xn is a random sample of size n from the probability

density
αp p−1 −αx
f (x) = x e ,x > 0
Γ(p)
where p is a known positive constant and α > 0 is an unknown parameter. Let
α̂ = p/X̄ be a proposed estimator of α. Then,

(A) E(α̂) = α
α
(B) E(α̂) = 1
1− np
α
(C) E(α̂) = 1−np

(D) none of the above statements is true.

29. The sign test is a nonparametric procedure for testing

(A) whether two populations have identical means.

(B) whether two populations have identical medians.
(C) whether two populations have identical probability distributions.
(D) whether two populations are independent.

30. Let X1 , . . . , Xn be i.i.d. from N (0, σ 2 ). What is the form of the most powerful
test for testing the null hypothesis

H0 : σ = 1 ,

against the alternative

H1 : σ = 2 ?
Pn
(A) Reject H0 if i=1 Xi2 > c
Pn 2
(B) Reject H0 if i=1 Xi <c
Pn
(C) Reject H0 if i=1 (Xi − X̄)2 > c
Pn
(D) Reject H0 if i=1 (Xi − X̄)2 < c.

7
 PSB-2016
SAMPLE QUESTIONS: PSB

1. In the diagram below, L(x) is a straight line that intersects the graph of a
polynomial P (x) of degree 2 at the points A = (−1, 0) and B = (5, 12). The
area of the shaded region is 36 square units.
Obtain the expression for P (x).

L(x)

A
P (x)

2. Let f : [−1, 1] → R be a continuous function. Suppose that f 0 (x) exists and

f 0 (x) ≤ 1 for all x ∈ (−1, 1). If f (1) = 1 and f (−1) = −1, prove that

f (x) = x for all x ∈ [−1, 1].

3. Suppose A is an n × n real symmetric matrix such that

T r(A2 ) = T r(A) = n.

Show that all the eigenvalues of A are equal to 1.

4. For each c ∈ R, define a function Tc : R4 → R4 by

Tc x1 , x2 , x3 , x4 := (1 + c)x1 , x2 + cx3 , x3 + cx2 , (1 + c)x4 .

For every c ∈ R, find the dimension of the null space of Tc .

5. A box contains 50 red balls, 30 green balls and 20 blue balls. Suppose balls are
drawn successively at random with replacement from the box. Let N denote
the minimum number of draws required to obtain balls of all three colours.
Compute P (N > n) for all positive integers n.

8
 6. Let X1 , . . . , Xn be i.i.d. random variables from a continuous distribution whose
density is symmetric around 0. Suppose E(|X1 |) = 2. Define
n
X n
X
Y = Xi and Z= 1(Xi > 0) .
i=1 i=1

Calculate the covariance between Y and Z.

7. Suppose {(yi , x1i , x2i , . . . , xki ) : i = 1, 2, . . . , n1 + n2 } represents a set of multi-

variate observations. It is found that the least squares linear regression fit of y
on (x1 , . . . , xk ) based on the first n1 observations is the same as that based on
the remaining n2 observations, and is given by
k
X
y = β̂0 + β̂j xj .
j=1

If the regression is now performed using all (n1 + n2 ) observations, will the
regression equation remain the same? Justify your answer.

8. Suppose that (X1 , Y1 ), (X2 , Y2 ), . . . , (Xn , Yn ) are the coordinates of n points

chosen independently and uniformly at random within a circle with centre (0, 0)
and unknown radius r.
(a) Obtain the MLE r̂n of r.
(b) Examine whether r̂n is sufficient for r.
(c) For any ε > 0, show that lim P (|r̂n − r| > ε) = 0.
n→∞

9. Suppose X1 , X2 , . . . , Xn is a random sample from an exponential distribution

with mean λ. Assume that the observed data is available on [X1 ], . . . , [Xn ],
instead of X1 , . . . , Xn , where [x] denotes the largest integer less than or equal
to x. Consider a test for H0 : λ = 1 vs H1 : λ > 1 which rejects H0 when
Xn
[Xi ] > cn . Given α ∈ (0, 1), obtain values of cn such that the size of the test
i=1
converges to α as n → ∞.

10. A cake weighing one kilogram is cut into two pieces, and each piece is weighed
separately. Denote the measured weights of the two pieces by X and Y . Assume
that the errors in obtaining X and Y are independent and normally distributed
with mean zero and the same (unknown) variance. Devise a test for the hy-
pothesis that the true weights of the two pieces are equal.