BE SURE TO BRING THESE NOTES TO EVERY CLASS

STATISTICS FOR MANAGEMENT: SAMPLING, INTERVAL ESTIMATION AND

TESTING OF HYPOTHESIS
Some results on sampling
Consider a finite population of size N. Let X1,…, XN be the values of a variable for the N units in
the population. Write
1 1 N
µ = ( X 1 + ... + X N ) and σ 2 = Σ ( X i − µ)2
N N i =1
for the population mean and population variance, respectively. A random sample of size n (≥ 2) is
drawn from the population by simple random sampling, i.e., all possible samples of size n have the
same probability of being selected. Let x1,…, xn be the values of the variable under consideration
for the n units in the sample. Write
1 1 n
x = ( x1 + ... + xn ) and s2 = Σ ( xi − x ) 2
n n − 1 i =1
for the sample mean and sample variance, respectively. Clearly, x and s 2 are random variables;
they depend on the particular sample drawn, i.e., their values depend on the outcome of a random
experiment, namely, the very process of drawing the sample. So, it is meaningful to think about the
expectation and variance of x and s 2 . The following results hold in this connection.

Simple random sampling with replacement:

σ2
E (x ) = µ V (x) = E (s 2 ) = σ 2
n
Simple random sampling without replacement:
σ2  N −n Nσ 2
E (x ) = µ V (x) =   E (s 2 ) =
n  N −1  N −1
Inference: Interval estimation and testing of hypothesis
Interval estimation: Statistical inference aims at inferring about a population on the basis of a random
sample drawn from it. A population characteristic, such as the population mean, is called a parameter
and typically denoted by θ. If θ is unknown, then one may wish to estimate it on the basis of sample
data. Such estimation can be done by specifying either a single value as an estimate of θ or an interval
that is highly likely to cover the true value of θ. These two approaches are known as point and interval
estimation, respectively.
We begin with the latter considering, on the basis of the sample data, an interval [L, U] which
includes θ with a high probability, i.e.,
P(L ≤ θ ≤ U) = 1 – α,
whatever be the true value of θ. In the above, typically α is taken as 0.05 or 0.10. The interval [L, U] is
then called a confidence interval for θ with confidence coefficient 1 – α. Moreover, L and U are called
the lower and upper confidence limits, respectively.
Testing of hypothesis: A hypothesis is a tentative idea. With reference to an unknown parameter θ, the
null hypothesis is the hypothesis that we wish to test on the basis of the sample data. Null hypotheses
of the form H0: θ = θ0 will be considered. Such a H0 is tested against an alternative hypothesis which
can be either (i) H: θ > θ0, or (ii) H: θ < θ0, or (iii) H: θ ≠ θ0. Only one of (i)-(iii) will arise in a
specific situation and the appropriate form of H will depend on the context, driven typically by the
direction along which a deviation from H0 is unwarranted.

1
 For instance, let θ be the unknown proportion of defective items in a lot. If the seller claims that θ is
as small as 0.005 and the buyer is doubtful about this, then it makes sense to test H0: θ = 0.005 against
the alternative H: θ > 0.005, because only the case θ > 0.005 is of real concern to the buyer. As another
example, if the finance minister claims that the per capita annual income, say θ (in US dollars), of a
country is as high as 30000 and the opposition parties express reservation about this claim thereby
implying that θ is not so high in reality, then one needs to test H0: θ = 30000 against H: θ < 30000.
There are two possible errors that one may commit in hypothesis testing:
Type 1 error: Rejecting H0 when it is true Type 2 error: Accepting H0 when it is false
Each of these can arise due to sampling fluctuations. Thus, in the example on defective items, suppose
the lot size is 10000 and a sample of 20 items is drawn for testing H0. If the seller’s claim is true, then
among the 10000 items in the lot, only 50 are defectives, but by chance it may so happen that out of
these 50 defectives, as many as 14 appear in the sample of size 20. Then obviously one will reject H0
though it is true and thus commit type 1 error. On the other hand, it is also possible that the seller’s
claim is false, say as many as 1000 items of the lot are defectives, but still none of these defectives
may appear in the sample of 20 items. This will lead to acceptance of a false H0, that is, type 2 error.
A good test rule aims at controlling the probabilities of the two types of error discussed above.
However, under wide generality, these two probabilities cannot be minimized simultaneously. A
standard practice, therefore, is to keep the probability of type 1 error fixed at a preassigned level, say α,
and subject to this stipulation, find a test rule that minimizes the probability of type 2 error. The level α
at which the probability of type 1 error is held fixed is called the level of significance or, simply, the
level of the test. Commonly, α is taken as 0.05 or 0.10. Incidentally, the power of a test is the
probability of rejecting H0 when it is false. Thus, power = 1 – probability of type 2 error.
The constrained optimization problem indicated in the last paragraph is quite nontrivial. So, we
avoid further details and present only two popular test rules and the associated confidence intervals.
These rules also hint at a connection between interval estimation and testing of hypothesis.
Test rules and confidence intervals
1. Testing a mean: Let x1,…, xn be the observations in a random sample drawn from a population with
mean θ and standard deviation σ , both unknown. Consider H0: θ = θ0 . Obtain z = n ( x − θ 0 ) / s ,
where x = ( x1 + ... + x n ) /n and s 2 = Σ in=1 ( xi − x ) 2 /(n − 1) . Assume that the sample size is large, say
n ≥ 30. Then the following test rules may be employed:
(a) for alternative θ > θ0, reject H0 at level α if z > z(α); [note: if α = 0.05, then
(b) for alternative θ < θ0, reject H0 at level α if z < –z(α); z(α) = 1.645 and z(α/2) = 1.96]
(c) for alternative θ ≠ θ0, reject H0 at level α if | z | > z(α/2);
A confidence interval for θ with confidence coefficient 1 – α is given by the limits x mz (α /2) s / n .
2. Testing a proportion: Consider a population where an unknown proportion θ of units belong to a
specific category. Let p be the proportion of units in this category in a random sample of size n from
the population. Consider H0: θ = θ0 . Obtain z = n ( p − θ 0 ) / θ 0 (1 − θ 0 ) . Assume that n ≥ 30. Then
the same test rules as in (a), (b), (c) above may be employed depending on the alternative hypothesis.
Moreover, a confidence interval for θ with confidence coefficient 1 – α is given by the limits
p mz (α /2) p (1 − p ) / n .

2
Sampling, interval estimation and testing of hypothesis: Questions for practice
1. In a game of chance, each player is given an urn containing one five-rupee coin, two two-rupee
coins and three one-rupee coins; note that the urn contains six coins altogether. The player draws
three coins at random and without replacement. Let X, Y and Z denote the values of these three
coins in rupees. Define
1
M = median ( X, Y , Z), L = min (X, Y, Z), U = max (X, Y, Z), S = (L + U)
2
Find (a) P(M = 1) , (b) P(M = 2), (c) P(S = 1), (d) P(S = 2), (e) P(S = 3), (f) P(S < M) and (g) E(S).
Ans: (a) 0.5, (b) 0.5, (c) 0.05, (d) 0, (e) 0.45, (f) 0.15, (g) 2.25
2. An alchemist visited the court of a medieval warlord and said “Your excellency, here is my tribute
to you. I have six envelopes. One of these contains a single copper coin, another contains two copper
coins, while a third one contains three copper coins. The remaining three envelopes are empty.
Kindly pick up any three of these six envelopes at random and without replacement. I shall convert
all the coins in the selected envelopes to gold coins dating from the period of King Solomon – you
can imagine their value as antiques !”
“But what happens if I end up picking only the three empty envelopes?”, thundered the warlord, “I
shall behead you then.”
“Take it easy, your excellency”, calmly replied the alchemist “I am also a sorcerer – in that extreme
case, I shall make seven gold coins for you, again dating from King Solomon’s era, simply from the
air.”
Assume that all the claims of the alchemist were true and that he kept all his promises (the latter point
is natural given the threat about his head!). Let X be the number of gold coins that the warlord
eventually ended up with. Obtain (a) P(X = 3), (b) P(X = 4), (c) P(X = 5), (d) P(X = 6), (e) P(X = 7),
(f) E(X) and (g) Var(X). Ans: (a) 0.3, (b) 0.15, (c) 0.15, (d) 0.05, (e) 0.05, (f) 3.35, (g) 2.6275
3. A textbook on business statistics contains five chapters. A student, who is not very serious, takes a
simple random sample (without replacement) of three chapters. He studies these three chapters with
some seriousness and completely ignores the remaining two chapters.
In the final examination, the question paper on this subject consists of five questions, one from each
chapter. The questions from Chapters 1 and 2 are compulsory and carry 18 and 12 marks
respectively. The questions from the other three chapters carry 20 marks each and each student is
supposed to answer any one of these three questions (even if a student answers more than one of
these three questions, he/she gets credit for only one of them). Thus the maximum possible score for
any student is 50.
Obviously, the student under consideration gets zero in any question from a chapter that he had
ignored (so he does his best to avoid such a question, if possible). Furthermore, as he is not very
serious with his studies, he gets only 50% of the marks in any question from a chapter that he had
included for study. Let T be his score in the examination.
Obtain the probability distribution of T and hence the expectation and variance of T.
Ans: The possible values of T are 10, 16, 19 and 25 with respective probabilities 0.1, 0.3, 0.3 and 0.3;
E(T) = 19, V(T) = 21.6
4. Five multinationals A,B,C,D and E offer scholarships to the students of a business school. The
scholarship amounts (in appropriate units) are as shown in the following table:
Multinational A B C D E
Scholarship amount 4 2 4 5 2
A simple random sample of three (out of the five) multinationals is drawn without replacement and
the associated scholarship amounts are noted. Let X, Y and Z denote the ordered values of these
amounts in the sense that X ≤ Y ≤ Z. Also let

3
 1
T= (X + 3Y + Z).
5
Obtain (a) P(X = 2), (b) P(Y = 4), (c) P(Z = 5), (d) P(3 < T < 4), (e) P(X = 2 & Z = 4)
(b) the joint probability distribution of X and Z.
Ans: (a) 0.9, (b) 0.7, (c) 0.6, (d) 0.6, (e) 0.4.
5. The manager of a casino plans the following game. Each player will be given a box containing
five coins of which one is “gold” (valued Rs 100), one is “silver” (valued Rs 25) and three are
“ordinary” (valued Re 1 each). The player draws two coins at random and without replacement. Let
A and B be the values (in rupees) of the two coins so drawn. Define X = |A − B|. The player
receives a payoff (X + 25) if none of the two selected coins is ordinary; the payoff is zero
otherwise. Obtain (a) the probability distribution of X, (b) E(X) and (c) the expected payoff.
Ans: (a) Values 0, 24, 75, 99 with probabilities 0.3, 0.3, 0.1, 0.3, (b) 44.4, (c) 10.
6. On the desk of an executive, there are five letters with page lengths 2,4,6,8 and 10. Three letters are
picked up at random and without replacement. Let A, B and C be the page lengths of the selected
letters, X = minimum(A,B,C) and Y = maximum (A,B,C). Define M as the arithmetic mean of X and
Y. Obtain (a) the probability distribution of M, (b) E(M), (c) P(Y−X > 4) and (d) P(M−X > 2).
Ans: (a) The possible values of M are 4, 5, 6, 7, 8 with respective probabilities 0.1, 0.2, 0.4, 0.2, 0.1,
(b) 6, (c) 0.7, (d) 0.7.
7. From a batch of five students, with scores 4, 6, 2, 8 and 10 in a quiz, two are selected by simple
random sampling without replacement. Let X and Y be the scores of these two students. Find
(a) P( | X – Y | > 2), (b) P((X+Y)/2 is even), and (c) E(X 2 +Y 2). Ans: (a) 0.6, (b) 0.4, (c) 88
8. A box contains five balls of weights (in kg) 3, 3, 7, 9 and 9. A simple random sample of three balls
is drawn without replacement. Let P, Q, R be the weights (in kg) of the three balls in the sample, and
Z = min(P, Q, R), Y = max(P, Q, R). Obtain (i) P(Y − Z > 2) and (ii) E (Y − 2 Z ) . Ans: (i) 0.9, (ii) 2
9. The probability distribution of a certain psychological trait influencing consumer behavior is
specified by the probability density function f (x) , where
2(θ − x)
f (x) = , if 0 ≤ x ≤ θ ,
θ
2

and f (x) = 0, otherwise. Here θ is a positive-valued unknown parameter. A single observation,

say X, is drawn at random from this distribution. Let I = [kX , cX ] be a confidence interval for θ ,
where k (≤ 1) and c (> 1) are positive constants. What values of c, if any, will ensure a confidence
coefficient 0.95 for I when (i) k = 0.5 and (ii) k = 1 ? Ans: (i) 39.4936, (ii) 39.4936
10. A certain quality characteristic in a production process has the following probability density
function: f ( x) = 12 exp(− | x − θ | ) , − ∞ < x < ∞ . Here θ is a real-valued unknown parameter. A
single observation, X , is drawn at random from this distribution. Consider a confidence interval of
the form [ X − 2c, X + c] for θ , where c is a suitable positive constant. What is the value of c if the
confidence interval has confidence coefficient 95% ? Ans: 2.3902
11. The distribution of a psychological trait X is specified by the probability density function f (x) ,
1
where f (x) = ( x − θ ) if θ ≤ x ≤ θ + 4 , and f (x) = 0 otherwise. Find a constant d, if any, such
8
that [ X − d , X + 2] is a confidence interval for θ with confidence coefficient 0.95. Ans: 3.8987
12. The distribution of a certain quality characteristic has a probability density function
θ2
f (x) = , if | x | ≥ θ ,
| x |3
= 0, otherwise.
4
Here θ is a positive-valued unknown parameter. A single observation, X , is drawn at random from
this distribution. Find a constant h, if any, such that the [ h | X | , | X | ] is a confidence interval for θ
with confidence coefficient 0.90. Ans: 0.3162
13. In a production process, the distribution of a quality characteristic, X, is continuous uniform over
the range [0, θ ] . According to expert A, θ = 10 , while according to expert B, θ = 12 . To resolve this
conflict, a random sample of three items is drawn from the production process and their X-values are
noted. If at least two of these three X-values exceed 9, then the opinion of expert B is accepted. Else
the opinion of expert A is accepted. (i) Find the probability that the opinion of expert B is accepted
when that of expert A is actually true. (ii) Find the probability that the opinion of expert A is accepted
when that of expert B is actually true. Ans: (i) 0.028, (ii) 0.8438
14. The distribution of a certain quality characteristic is continuous uniform over the range [−2θ , θ ] ,
where θ ( > 0) is an unknown parameter. For testing the null hypothesis H 0 : θ = 5 against the
alternative H : θ > 5 , the value, say X, of the quality characteristic is observed for a single randomly
chosen item. Under a test rule which rejects the null hypothesis when X is outside the range [−9, 7] ,
obtain the probability of type II error at θ = 6 . Also, find the probability of type I error.
Ans: 5/6, 1/15
15. The life (in hours) of an electrical component is exponentially distributed with mean θ , where
θ ( > 0) is an unknown parameter. For testing the null hypothesis H 0 : θ = 7 against the alternative
H : θ < 7 , four such components are chosen at random. Under a test rule which rejects the null
hypothesis when three or more of these four survive for less than seven hours, what is the probability
of type I error? Also, find the probability of type II error when θ = 4. Ans: 0.5313, 0.1419

16. In a production process, the number of defects per item follows the Poisson distribution with
unknown mean θ . In order to test the null hypothesis H 0 : θ = 1 against the alternative H : θ > 1 ,
the following procedure is suggested:
Draw one item at random and observe the number of defects, say X, therein. If X ≤ 1 then accept
H 0 . On the other hand, if X ≥ 4 then reject H 0 .
If X equals 2 or 3, then draw one more item at random and observe the number of defects, say Y,
therein. If X + Y ≥ 4 then reject H 0 ; otherwise, accept H 0 .
(a) What is the probability of type I error with the above procedure?
(b) What is the probability of type II error with the above procedure when θ = 2 ?
Ans: (a) 0.1063, (b) 0.5403

5
BE SURE TO BRING THESE NOTES TO EVERY CLASS
STATISTICS FOR MANAGEMENT: POINT ESTIMATION
Statistical inference aims at inferring about a population on the basis of a random sample drawn
therefrom. A population characteristic, such as the population mean, is called a parameter and
typically denoted by θ. If θ is unknown, then one may wish to estimate it on the basis of sample data.
For instance, the estimator of an unknown population mean can be the corresponding sample mean.
This is a case of point estimation where an unknown parameter is estimated by a single value.
Unbiasedness and minimum variance: Let T be an estimator of an unknown population parameter θ.
The numerical value of T in any particular sample is an estimate of θ. We call T unbiased for θ if the
average of T over all possible samples equals θ, i.e., if E(T) = θ, whatever be the true value of θ. Thus,
under simple random sampling, the sample mean is unbiased for the population mean.
An unbiased estimator is, however, by no means unique, e.g., for a normal population, both the
sample mean and the sample median are unbiased estimators of the population mean. In order to make
a choice from amongst rival unbiased estimators, the criterion of minimum variance is invoked. It calls
for minimizing the scatter of an unbiased estimator of T around θ. Thus, T is the minimum variance
unbiased estimator (MVUE) of θ if (a) T is unbiased for θ, and (b) T has the smallest variance among
all unbiased estimators of θ, whatever be the true value of θ. It can be shown that an MVUE, if it
exists, is unique. In particular, for the Poisson, exponential or normal populations, sample mean is the
MVUE of the population mean. Similarly, in the context of the binomial distribution, the sample
proportion of successes is the MVUE of the probability of success in any trial.
Maximum likelihood estimation: While the aforesaid approach based on unbiasedness and minimum
variance is intuitively appealing, it becomes mathematically intractable beyond simple situations.
Therefore, we consider a broad spectrum method of estimation, namely, that of maximum likelihood.
Under wide generality, this method leads to estimators with attractive properties especially when the
sample size is moderate to large. Let x1,…, xn be observations in a random sample from a population
given by a probability mass or density function f ( x ;θ ) , involving an unknown parameter θ. Then the
likelihood function for θ is defined as the product
L(θ) = f ( x1 ;θ ).... f ( x n ;θ ) ,
and the maximum likelihood estimator (MLE) of θ is the maximizer of L(θ). Maximization of L(θ) is
equivalent to that of log L(θ) and, in many situations, it is indeed convenient to work with log L(θ)
rather than L(θ) itself. Moreover, standard tools of calculus may be employed for such maximization if
appropriate differentiability conditions hold.
Example. Let x1,…, xn (> 0) be observations in a random sample from an exponential population with
unknown mean θ. Then f ( x ;θ ) = (1 / θ )e − x / θ for x > 0, so that
L(θ) = f ( x1 ;θ ).... f ( x n ;θ ) = (1 / θ ) n exp{− ( x1 + ... + x n ) / θ } = (1 / θ ) n exp(−n x / θ ) ,
where x = ( x1 + ... + x n ) /n . Hence
log L(θ) = − n log θ − n x / θ , and d log L(θ ) /dθ = − n /θ + n x /θ 2 .
The equation d log L(θ ) /dθ = 0 has unique root x , which is a maximizer because d 2 log L(θ ) /dθ 2
= n /θ 2 − 2n x /θ 3 is negative at θ = x . Hence the MLE of the population mean θ of the exponential
distribution is the sample mean. ■
For the Poisson or normal populations too, it can be shown similarly that the sample mean is the
MLE of the population mean. Also, for the binomial distribution, the sample proportion of successes is
the MLE of the probability of success in any trial. Thus in all these situations the MLE coincides with
the MVUE. In general, however, this may not always be the case; the MLE is not even guaranteed to
be unbiased, despite its having excellent large sample properties.
6
 Before concluding this section, we mention another attractive property of the MLE, namely, that of
invariance, i.e., if T is the MLE of θ, then g(T) is the MLE of g(θ) for any one-to-one function g.

Point estimation: Questions for practice

1. The probability distribution of a certain quality characteristic is specified by the density function
fθ (x) = ( x − θ ) e − ( x −θ ) , if x ≥ θ ,
= 0, otherwise.
Here θ (−∞ < θ < ∞) is an unknown parameter. A random sample of size n is drawn from this
distribution. Let X 1 ,..., X n denote these sample observations and M be their arithmetic mean.
(a) Find a constant c, if any, such that cM is an unbiased estimator of θ .
(b) Find a constant d, if any, such that M + d is an unbiased estimator of θ .
(c) Let n = 4 and consider the estimator Z = 2 M + 1 of θ . What is the mean squared error of this
estimator at θ = 0 ?
(d) For general n, which of the following are consistent estimators of θ ?
2 2 2 2
(i) M − , (ii) M + 4 − , (iii) M − 4 + , (iv) M − 2 + .
n n n n
(e) Let n = 1 and suppose the single observation equals zero. What is the MLE of θ ?
Ans: (a) None, (b) − 2 , (c) 27, (d) only (iv), (e) − 1
2. An electrical unit consists of three identical components and the unit functions if and only if at least
one of the components function. The components are known to function independently. Out of 500
randomly chosen units of this type 468 were found to be functioning for 10 hours or more. On the
basis of this information, find the maximum likelihood estimate of the probability for an individual
component to function for 10 hours or more. Ans: 0.6
3. The probability density function of the payoff (in Rupees) from a single trial with a gambling
instrument is given by
7cx 6 + d
fθ (x) = , − 2θ ≤ x ≤ θ ,
3(43cθ 7 + dθ )
where c and d are known positive constants and θ (> 0) is an unknown parameter. Let X 1 ,…, X n be
the payoffs (in Rupees) from n independent trials of the gambling instrument.
(a) If n = 7 and the payoffs from the seven trials are found to be 40, − 56 , 23, 29, − 86 , − 64 and
− 56 , then what is the maximum likelihood estimate of θ ?
(b) Suppose c = 1 and d = 0, and consider the case of general n. If M be the arithmetic mean of
X 1 ,…, X n , then for what value(s) of the constant k, if any, will kM be an unbiased estimator of θ ?
(c) Let c = 0, d = 1, and M be the arithmetic mean of X 1 ,…, X n . Consider an unbiased estimator of
θ of the form hM, where h is a constant. For what value(s) of n, if any, will such an unbiased
estimator have variance θ 2 ? Ans: (a) 43, (b) – 344/595, (c) 3
4. An unknown parameter θ (0 ≤ θ ≤ 8) indicates the condition of a machine in the beginning of a
working day. The probability for the machine to break down during the day is (1 − 18 θ ) . Define a
random variable X such that
X = 2, if the machine breaks down during the day,
= 0, otherwise.
(a) Find constants a and b, if any, such that T = a + bX is an unbiased estimator of θ .
(b) Consider the estimator W = c − 3 X for θ , where c is a constant. For what value(s) of c, if any,
will the mean squared error of W at θ = 2 be equal to that at θ = 6 ? Ans: (a) a = 8, b = – 4, (b) 7

7
5. (a) The distribution of scores in a public examination is normal with unknown mean µ and known
standard deviation 5. It is, however, known that µ equals 52, 60 or 67. If the scores of three randomly
chosen candidates turn out to be 88, 93, 59, then find the maximum likelihood estimate of µ .
(b) Every morning, a casino manager fixes the value of a parameter θ of a gambling instrument as
either 1 or 3 or 5 or 7. A gambler knows that the payoff to him in any single operation of the
instrument has a distribution given by the probability density function
1
fθ (x) = , −∞ < x < ∞.
2
π [1 + ( x + θ ) ]
However, he does not know the value of θ. In two independent trials with the gambling instrument
on a particular day the payoffs to the gambler turn out to be 1 and −5. On the basis of this
information, find the maximum likelihood estimate of θ for that day. Ans: (a) 67, (b) 5
6. A company conducts an aptitude test for the recruitment of its sales personnel. The following are
known: (I) The test score is distributed normally with mean 50 and standard deviation 5 for people
without any past sales experience. (II) The test score is distributed normally with mean θ and
standard deviation 5 for people with past sales experience; here θ is an unknown parameter. (III)
Among the people taking the test, 80% do not have any past sales experience.
A random sample of n test scores is drawn. Let X be the sample mean.
(a) Find constants c and d such that T = (c X + d) is an unbiased estimator of θ .
(b) What will be the variance of T when (i) θ = 60, (ii) θ = 70 ?.
(c) With c and d as in part (a) above, is T consistent for θ ?
Ans: (a) c = 5, d = –200, (b) 1025/n, 2225/n, (c) Yes
7. It is known that the daily number of road accidents, X, in a locality equals either 1 or 2 with
respective probabilities θ2 + θ3 and 1-θ2-θ3. Here θ is an unknown parameter whose only possible
values are 1/3 and 2/3. On the basis of a single observation X, is it possible to find a T(X) which is
an unbiased estimator of 1/θ ? Justify your answer. Ans: T(1) = 27/32, T(2) = 27/8
8. In a business school, a particular paper is considered tough. However, the students are allowed to
make an unlimited number of attempts to pass the paper. The probability for any student to pass the
paper in the x-th attempt is (1 − θ ) x−1θ , where x = 1, 2, 3, …, and θ ( 0 < θ < 1 ) is an unknown
parameter. For a random sample of 36 students, the arithmetic mean of the number of attempts to
pass the paper is observed to be 4. Find the MLE of θ and the MLE of P(X < 3). Ans: 1/4, 7/16
9. There are two competing brands, A and B of a product. The two brands are essentially alike and the
preference of any consumer for A or B depends on a certain psychological trait X. A consumer, with
positive X, prefers A to B. On the other hand, a consumer, with zero or negative X, prefers B to A. The
distribution of X (in the population of all consumers) is continuous uniform over the range [−1, θ ] ,
θ −1
where θ is an unknown parameter. Both θ and the quantity π = , that measures the “relative
θ +1
preference for A to B” in a certain sense, are of interest to a market researcher. In order to estimate
these quantities, a single consumer is chosen at random and his/her preference for A or B is noted. The
preference is, in fact, quantified as Y, where Y = 1 if this consumer prefers A to B, and Y = 0 if this
consumer prefers B to A.
Direction for parts (a), (b) and (c): The parameter θ is known to be positive. No further information
about θ is available. Thus any positive real number is a possible value of θ .
(a) Find constants c and d, if any, such that c + d Y is an unbiased estimator of π .
(b) Find constants a and b, if any, such that a + bY is an unbiased estimator of θ .
(c) As an estimator of θ , what is the mean squared error of T = Y − 1 at θ = 1 ?
Direction for parts (d) and (e): More detailed information is available about θ . It is known that θ
equals either 1 or 2.

8
(d) Find constants g and h, if any, such that g + hY is an unbiased estimator of θ 2 .
(e) Suppose Y = 0 . Then what is the maximum likelihood estimate of θ ?
Ans: (a) c = – 1, d =2, (b) none, (c) 2.5, (d) g = – 8, h = 18, (e) 1
10. The IQ of the students in a college is known to have the continuous uniform distribution over
the range [130-m, 130+m], where m (> 0) is unknown. A random sample of six students shows the
following IQ figures: 125, 132, 122, 129, 137, 133. On the basis of this information, find the
maximum likelihood estimate of m. Ans: 8

11. (a) The distribution of a certain quality characteristic is continuous uniform over the range
[−2θ , θ ] , where θ (> 0) is unknown. If the values of the quality characteristic in a random sample of
size five are 0.63, 0.27, –1.62, 0.71 and –1.38, then what is the maximum likelihood estimate of θ ?
(b) The life (in hours) of an electrical component is exponentially distributed with mean θ , where
θ ( > 0) is unknown. let π be the probability that a system, consisting of two such components
arranged in series, survives for five hours or more. If five randomly chosen components are seen to
survive for 7, 6, 4, 8 and 5 hours, then what is the maximum likelihood estimate of π ?
Ans: (a) 0.81, (b) 0.1889