Exercises to

Probability Theory and Statistics

(sf1901)

These are essentially translations of the Swedish pdf-exercises, but I

have replaced a few of them with – in my opinion – more relevant
ones. Harald Lang, 2016.
1.1 Let A and B be two events. Express the following events in
words, and display them in a Venn diagram:
a) A  B b) A  B c) A  B
d) Prove (using a Venn diagram) the rule
P( A  B )  P( A)  P( B )  P( A  B )
e) Prove (using a Venn diagram) de Morgan’s rules:
( A  B )*  A  B and ( A  B )*  A  B

1.2 At the production of a certain item, two types of defects, A and

B, can occur. We know that P(A) = 0.1, P(B) = 0.2 and
P( A  B )  0.05. Compute the probability that a produced unit
has
a) at least one of the defects
b) defect A but not defect B
c) none of the defects
d) precisely one of the defects A and B
1.3 We take at random three cards (without replacement) from an
ordinary deck of cards (of 52 cards). Compute (using the
classical probability definition) the probabilities for the events
a) all three are hearts
b) none of the cards is hearts
c) all three are aces.
1.4 (Challenging!) Compute the probability that out of 23 randomly
selected individuals, at least two of them have the same
birthday. (Assume that all birthdays are equally common, and
that there are 365 days in a year.)
1.5 Discuss the Monty Hall problem (see Wikipedia). Answer the
following:

1
 a) If you decide in advance to always stick to your originally
chosen door, what is your chance to win the car?
b) If you decide in advance to always change door (when there are
two closed doors,) what is your chance to win the car?

Answers

1.2 a) 0.25 b) 0.05 c) 0.75 d) 0.20

1.3 a) 0.01294 b) 0.4135 c) 0.0001810

1.4 0.5073

1.5 a) 1/3 b) 2/3

2
2.1 The two events A and B are independent with P(A) = 0.1 and
P(B) = 0.05. Compute P( A  B ). What can we say about the
events A and B ?
2.2 The two events A and B have both positive probabilities.
a) If they are disjoint, can they be independent?
b) If they are independent, can they be disjoint?
2.3 The two events A and B have probabilities P(A) = 0.1,
P(B) = 0.05 and P( A  B )  0.8. Determine if A and B are
independent events.

2.4 We have three lottery tickets; one is a winning ticket and two
are blanks. The three individuals Joe, Kim and Mary draw a
ticket each in order, (Joe first, Mary last.)
a) What is the probability that Joe wins?
b) that Kim wins?
c) that Mary wins?
2.5 From a sign with the text MALMO two random letters fall to
the ground. An individual, who is ignorant of the original text,
puts them up again at random. What is the probability that the
original text will be restored?

2.6 Assume that the probability that a born baby is a boy is 0.5, and
that gender is independent between births. A family has four
children. Compute
a) the probability that they have two boys and two girls,
conditional that the first (oldest) is a boy
b) the probability that they have two boys and two girls,
conditional that at least one is a boy.

3
2.7 For the three events A, B and C it holds that
P( A  B  C )  0.1, P( A)  0.5 and P( B | A)  0.4.
Compute the probability P(C | A  B).

Answers
2.1 0.855
2.2 a) no b) no
2.3 dependent
2.4 All three probabilities are 1/3
2.5 0.55
2.6 a) 0.375 b) 0.4
2.7 0.5

4
3.1 At a production process, the produced items are tested for
defects. A defective unit is classified as such with probability
0.9, whereas a correct unit is classified as such with probability
0.85. Furthermore, 10% of the produced units are defective.
Compute the conditional probability that a unit is defective,
given that is has been classified as such.

3.2 A and B are two independent events with probabilities 0.5 and
0.4 respectively. Compute the conditional probability that both
A and B occur, when it is known that at least one of the events A
and B has occurred.

3.3 In a certain risk group, individuals are tested for a certain

disease S. A person who has the disease gets the correct
diagnosis with probability 0.99, whereas a person who does not
suffer from S gets the correct diagnosis with probability 0.95.
Furthermore, it is known that 6% of the individuals in the group
get the diagnosis “suffer from S”.
Determine a) the proportion of individuals in the group who
suffer from S, and b) the probability that a person who gets the
diagnosis “suffer from S” actually carries the disease.

3.4 We have two urns, A and B, who contain black and white balls.
A contains two black and three white balls, whereas B contains
two black and two white balls. We draw one ball at random
from A and put it in B, without noticing its colour. Next we
draw at random one ball from B, and notice that it is white.
Compute the probability that the ball we moved from A to B
was black.

5
Answers
3.1 0.4

3.2 2/7

3.3 a) 1/94 b) 33/188 ≈ 0.1755

3.4 4/13

6
4.1 A library has on average 78 visitors on an ordinary Friday.
Compute the probability that the library has more than 90
visitors on an ordinary Friday. Motivate carefully!

4.2 Fifteen persons toss coins; two coins each.

a) What distribution has the number of individuals who get the
same result on the two coins?
b) Compute the probability that at most six individuals get the
same result on the two coins.
4.3 Two defective units have accidentally been mixed in among
three flawless units. In order to find the defective units, the
units are tested one at a time, in order, until either the two
defective units have been found, or the three flawless units have
been found.
a) Determine the distribution for X, the number of tested units.
b) Compute the expected value E(X) and the standard deviation
SD(X).
4.4 A certain Friday evening 1% of the car drivers are intoxicated.
a) At a road check 50 drivers are tested. Compute the probability
that at least one intoxicated driver is caught.
b) At a road check drivers are tested. Let X be the number of tested
drivers when the first intoxicated driver is caught. Compute the
expected number E(X) and the standard deviation SD(X).
4.5 Prove that if X  Po( ) and Y  Po(  ) are independent, then
X | ( X  Y  n)  Bin(n,  / (   )).
Answers
4.1 0.08101
4.2 a) Bin(15, 0.5) b) 0.3036
4.3 a) P(X = 2, 3, 4) = 0.1, 0.3, 0.6 b) E(X) = 3.5, SD( X )  0.45

4.4 a) 0.3950 b) E(X) = 100, SD(X) = 30 11

7
5.1a) Determine the constant c such that the function

 cx 0 x6
2
f ( x)  
0 otherwise
becomes a probability density function.
b) Determine the corresponding distribution function.
c) Compute the expected value and variance.
5.2 A random variable X has the distribution function

0 x0

FX ( x)   x 3 0  x 1

1 x 1

Compute E(X) and SD(X).

5.3 The length X of a telephone call can often be approximated as
an exp( ) random variable. If   2 3min 1 , compute
a) the probability P(1  X  10).
b) E(X) and SD(X).

5.4 It is a fact that if X  exp( ), then 2 X   2 (2). Use this fact

to compute the probability in 5.3.a) with your pocket calculator.
5.5 In a coordinate system, the interval [–a, a] on the x-axis
represents a fault fissure. An earth quake will occur such that its
epicentre X on that interval follows a uniform probability
distribution. At (0,d) is a dam. Let Y be the distance from the
dam to the epicentre X. Determine the density function for Y.
5.6 X  exp( ).
a) Determine the density function for Y  e X .
b) Use a calculator to compute E  X .

8
Answers
1 3
5.1 a) c = 1/72 b) FX ( x)  x for 0 < x < 6, = 0 otherwise,
216
c) E(X) = 4.5, V(X) = 1.35

5.2 E(X)=3/4, SD( X )  3 / 80

5.3 a) 0.5121, E(X) = 3/2, SD( X )  3 / 2

y
5.5 fY ( y )  for d  y  a 2  d 2 , = 0 otherwise.
a y d 2 2


5.6a) fY ( y )  1 
, y > 1 (a Pareto distribution)
y

0.8862   
b)   
  2 

9
6.1 X and Y are independent, discrete random variables whose
probability functions are given in the tables below:
k 1 2 3 k 1 2 3
p X (k ) 1 3 1 2 1 6 pY (k ) 2 3 1 6 1 6

a) Compute the probability P(X + Y = 4).

b) Compute the conditional probability P( X  2 | X  Y  4).

6.2 Two friends A and B have decided to meet at a cafe “somewhat

after eight o’clock”. A arrives X minutes past eight, and B
arrives Y minutes past eight.
Assume now that X  U(0, 4) and Y  U(0,6), and that X and
Y are independent. What is the probability that A has to wait for
B?
6.3 X and Y are independent exponentially distributed random
variables with expected values 2. Let U = max(X, Y) and
V = min(X, Y). Compute E(U – V).
6.4 X and Y are as in 6.3. Determine the density function for
Z = X + Y.
6.5 X  exp( ). We draw a random value x of X, and let
Y  exp( x). Compute the probability P(Y > 1).

Answers
6.1 a) 1/4 b) 5/9
6.2 2/3
6.3 2
1 0.5 x
6.4 xe , x  0.
4

6.5
1 

10
7.1 The two random variables X and Y have expected values
E(X) = –1 and E(Y) = 1, and variances V(X) = 2 and V(Y) = 4.
The covariance is Cov(X, Y) = –1. Compute the standard
deviation SD(X – 2Y + 1).

7.2 The random variable X  exp(2.5). Compute E e  X   and

 
V e X .

7.3 X  U(1, 2). We draw a random value x of X, and let

Y  exp( x). Compute the expected value E(Y).

7.4 The random variables X 1 , X 2 and X 3 are independent and have

all the expected value 2 and standard deviation 3. Let
Y  3 X 1  2 X 2  X 3  6.
a) Compute E(Y) and SD(Y).
b) Compute the correlation coefficient  (Y , X 1  X 2 ) .

7.5 The random variable X has density function

f X ( x)  0.5 x, 0  x  2. Compute V( X 2 ).
7.6 In certain breakfast cereal packs, plastic toys are placed. Each
pack contains one toy, and there are in total six different toys
that are distributed evenly among packs. You don’t know which
toy the pack contains until you open it.
You buy packs, one at a time, and let X be the number of packs
bought when you have just collected at least one of each toy.
Compute E(X).
Answers
7.1 22
 
7.2 E e  X  5 7  
V e X  20 441
7.3 ln(2)
7.4 a) E(Y )  2, SD(Y )  3 14 b)  (Y , X 1  X 2 )  1 14
7.5 4/3
7.6 14.7

11
8.1 X  N(0,1) . Compute
a) P( X  1.8)
b) P( X  1.35)
c) P( 1.2  X  0.5)
d) a, such that P( X  a )  0.05
e) a, such that P( X  a )  0.95

8.2 X  N(5, 2). Compute

a) P( X  6)
b) P(1.8  X  7.2)
c) a, such that P( X  a )  0.05
8.3 X, Y and Z are independent normally distributed random
variables with expected values 2, 1 and 0, respectively. They all
have the sane variance 2. Compute P(4 X  3Y  5Z ).

8.4 A lift in a department store has a sign “a maximum 10 persons,

or at most 800 kg.” Assume that a random person’s weight is
N(70, 10) kg. What is the probability that 10 customers will
overload the lift?
8.5 A new residential area is planned for 1’000 families. The
probability that a family owns 0, 1, 2 and 3 cars is, 0.2, 0.4, 0.3
and 0.1, respectively. How many parking places have to be
planned for, in order that with 90% probability all cars have a
parking place?
8.6 X  Bin(2300, 0.25) and Y  Po(640). X and Y are
independent. Compute (approximately) the probability
P ( X  Y ).

12
Answers
8.1 a) 0.9641 b) 0.0885 c) 0.3849 d) 1.6449 e) 1.9600

8.2 a) 0.6914 b) 0.8095 c) 1.710

8.3 0.6915
8.4 7.83  104

8.5 1'337
8.6 0.0219 [0.02235]

13
9.1 A measuring instrument uses a special kind of battery. Five
batteries of this type have been used to measure the time spell
they deliver enough power. The result was 5, 4, 6, 4, 7 hours.
These values can be regarded as observations of a random
variable X with expected value μ and variance σ2.
a) Estimate μ and σ2 in some suitable way.
b) Show that estimate of μ is unbiased.
c) compute the standard deviation and standard error of the
estimate of μ.

9.2 We have a random sample x1, … , xn (n > 2) of an N(μ ,σ)

random variable. Consider the two estimates
1 1
 *  ( x1    xn ) and ˆ  ( x1  xn ).
n 2
a) Show that both estimates are unbiased.
b) Which of the two estimates is most efficient?
9.3 x1, … , xn are independent observations of a Maxwell distributed
random variable, i.e. its density function is
2 x2
f ( x)  e x /(2 )
, x0,
2

  3/ 2

where α is a parameter. This distribution has expected value

8  and variance  (3  8  ).
a) Compute the ML-estimate ˆ ML of  .
b) Is ˆ ML an unbiased estimate of α?

9.4 We have three independent measurements x1, x2, x3 of the angle

AOC, and two independent measurements x4, x5 of AOB (see
picture below.) The measurements are unbiased with a random
error with standard deviation σ.
a) Compute the LS estimates of 1 and  2 .
b) Are the estimates unbiased?

14
 c) Compute the variances of the estimates.

9.5 The discrete random variable X has probability function

p( j )   (1   ) j 1 , j  1, 2, 3,  where 0    1. We have six
independent observations of X: 4, 5, 4, 6, 4, 1.
a) Compute the ML-estimate of θ.
b) Compute the LS-estimate of θ.
Answers
 ˆ
9.1 a) ˆ  5.2, ˆ 2  1.7 b) SD( ˆ )  , d  ( ˆ )   0.5831
5 5
2 2
9.2 Var(   )  , Var( ˆ )  .   is more efficient.
n 2
  x 2j
1 n
9.3 a) ˆ ML b) yes
3n j 1
1 1 1
9.4 a) ˆ1  ( x 4  x5 ) ˆ2  ( x1  x2  x3 )  ( x 4  x5 )
2 3 2
9.5 a) ˆML  1 4 b) ˆLS  1 4

15
10.1 Four independent measurements of a distance have been
performed with an instrument which gives an unbiased result,
but with a random error which is normally distributed with
standard deviation   0.005 mm. The measurements were
(mm)
1132.155 1132.158 1132.145 1132.163
a) Compute a 95% confidence interval for the true distance.
b) As a) but assume that σ is unknown.
10.2 Two nursery schools are located in the same town, but one of
them, “The Clearing” is located close to a forest, whereas the
other, “The Brick Wall” is located downtown. The environ-
mental manager suspects that the children at “The Brick Wall”
might get higher content of lead in their blood, due to the
location. In order to find out, five children were randomly
drawn from each nursery school, and their lead contents were
measured (ng/ml). The result was
The Clearing: 0.96 0.43 0.93 0.85 0.48
The Brick Wall: 0.93 0.63 1.21 1.30 0.58
Do these data indicate a difference between the two nursery
schools? Construct a 95% confidence interval to answer the
question.
a) Assume that the variation in lead content between children is
normally distributed with the same variance for the two nursery
schools.
b) As in a) but do not assume equal variances.
10.3 In order to compare two scales, six items were weighed with
each of the two scales. The results were:
item 1 2 3 4 5 6
scale A 1.0 7.7 9.6 21.0 32.3 22.6
scale B 3.1 8.8 12.0 19.5 35.5 32.5

The measurements can be regarded as observations of normal

random variables. Compute a 95% confidence interval for the

16
 difference  B  A
a) when the standard deviations are known to be  A  2 and
B  3
b) when the standard deviations are unknown.
10.4 During 30 days at an upstream location of the Mississippi river
a certain contamination was measured. Since the measurements
were made at different days, they can be considered as indepen-
dent. The sample mean value was x  13.2 and sample standard
deviation s  2.8. At a location downstream, 40 measurements
were made at different days. Here the sample mean was
x  86.1 and the sample standard deviation s  38.7. Compute
a 95% confidence interval that can be used to assess the
difference in contamination between the two locations. (Note
that we do not assume normal distributions.)
10.5 A physicist has made five measurements in order to determine a
certain physical constant. These measurements can be regarded
as observations of a normally distributed random variable with
known variance. He computes a 90% confidence interval to be
(7.02, 7.14). How many measurements would he need in order
to get
a) a confidence interval with half the length?
b) a 99% confidence interval of the same length?
c) a 99% confidence interval with half the length?
Answers
10.1 a) (1132.150, 1132.160) b) (1132.143, 1132.167)
10.2 a) (–0.628, 0.228) b) (–0.632, 0.232) (df = 7.55)
10.3 a) (–0.018, 5.752) b) (–1.131, 6.865)
10.4 (60.87, 84.93) [(60.49, 85.31) df= 39.54]
10.5 a) 20 b) 13 c) 49

17
11.1 Two firms, A and B, offer telephone support. One wanted to
estimate the difference in wait time before a call is answered
between the two firms. The wait time for 420 calls to firm A
was on average 26.0 minutes, and for 376 calls to firm B the
wait time was on average 31.6 minutes. Assume that wait time
is exponentially distributed with expected values  A and  B
respectively. Construct a 95% confidence interval for the
difference  B   A . Can we draw any conclusion (with error
risk 5%) as to which firm has the longer expected wait times?

11.2 The traffic on a certain road can be described such that the
number of cars that pass a certain point on the road during a
time period of t minutes is Po( t ). During a time period of 10
minutes, 400 cars passed. Compute a 95% confidence interval
for  .

11.3 In a psychological experiment 198 economics students were

asked “Should Sweden join the EMU and introduce the euro as
currency?” 107 said “Yes” and 91 “No”. Another 198 students
were asked “Should Sweden join the EMU and abandon the
Swedish krona as currency?” 95 answered “Yes” and 103
“No”.
Compute a 95% confidence interval for the difference in
proportions of students who answer “yes” in the entire student
population between the two questions. Can we draw any
conclusion whether the formulation of the question influences
the answer?

11.4 Assume that the number of fatal traffic accidents during two
consecutive years was 537 and 453, respectively. The minister
in charge described this decline in accidents as a result of his
clever policy, whereas his opponents claimed it was pure
chance.

18
 Compute a 95% confidence interval for the difference in
intensity of fatal accidents between the two years. Can we reject
the hypothesis that the outcome was pure chance?

11.5 The number X of calls to a telephone exchange during the

busiest hour is Po(λ). During eight days the following
observations on X were made:
115, 82, 108, 106, 118, 87, 99, 92.
Compute a 95% confidence interval for λ.

Answers
11.1 a) (1.55, 9.65), firm B has on average longer wait times.

11.2 (36.1, 43.9)

11.3 (–0.038, 0.159), no conclusion can be drawn.
11.4 Poisson distributions. Difference in intensities (22.3, 145.7),
not chance! However, the decision to perform the test seems to
have been suggested by the data (not prior to observing it,) and
if so, the test is invalidated.

11.5 (93.9, 107.8)

19
12.1 Joe plays on a slot machine which gives a win with probability
p. The claim is that p = 0.2, but Joe suspects that p < 0.2.
Define the hypotheses H0: p = 0.2 and H1: p < 0.2. Joe wants to
test H0 against H1, so he plays on the machine until he wins the
first time. This happens at his 11:th attempt. Can he – with risk
level at most 0.1 – reject H0 in favour of H1?

12.2 Eighteen pairs of iron rods have been buried in the soil, where
in each pair one rod has been given anti corrosion treatment,
whereas the other has not. The task is to figure out if the
treatment has any real effect. The rods will be examined after
some prescribed time and x, = the number of pairs where the
treated rods have corroded more than the untreated, will be
noted. It is decided that the null hypothesis H0: “no effect” will
be rejected in favour of H1: “the treatment has some anti
corrosion effect” if x  5.
a) Decide the significance level of the test
b) What is the conclusion if x = 1? x = 5? x = 15?

12.3 We have 10 independent data from a N(μ, 0.2)-distribution:

3.9 4.1 4.4 4.0 3.8 4.0 3.9 4.3 4.2 4.4
a) Test, with a risk level 0.05, the hypothesis H0: H0: “μ = 4”
against the alternative H1: “μ > 4”.
b) As a), but assume that the standard deviation (0.2) is unknown.

12.4 A researcher has made a new alloy and has theoretically

calculated its melting temperature to 1050 centigrade. In order
to test this result, she has measured the melting temperature for
ten samples, and come up with the following results
(centigrade:)

1054.8 1052.9 1051.0 1049.8 1051.6

1047.9 1051.8 1048.5 1050.2 1050.7
The variation in these numbers depends on imperfections of the

20
 thermometer employed. It is known that the errors are
independent between measurements, unbiased and have a
standard deviation of 2.3 centigrade
a) Test the hypothesis that the melting temperature is 1050
centigrade against the alternative that it is different from that.
Use error level 5%.
b) The same as in a), but assume that the standard deviation is
unknown.
12.5 The traffic police sets up a road check to test drivers for
intoxication. The test is as a null hypothesis test: H0: “the driver
is sober” is tested against the alternative H1: “the driver is
intoxicated” with an error risk of 1%. What will happen in the
long run? Is it true that
a) at most 1% of those acquitted are in fact intoxicated
b) at most 1% of those who are intoxicated are acquitted?
c) at most 1% of those who are caught are in fact sober.
Answers
12.1 The test is “reject if the number of attempts needed is large”.
p-value = P(lose first 10 games) = 0.810 = 0.1074 > 0.1 . No
conclusion.
12.2 a) 0.048 b) accept H1, accept H1, no conclusion.

12.3 Confidence interval μ > 3.996, so no conclusion (but it was

close!) b) μ > 3.975, no conclusion
12.4 a) confidence interval (1049.5, 1052.3), so no conclusion
b) confidence interval (1049.5, 1052.4), so no conclusion
12.5 They are all false. A true statement is “at most 1% of those
who are sober are caught”.

21
13.1 In exercise 12.2, compute the power of the test against the
alternative that p = 0.25, where p = probability that a treated
rod will corrode more than an untreated.
13.2 In exercise 12.3, compute the power function for the test. What
is the power if the true value of μ is μ = 3.8? If μ = 4.3?
13.3 in exercise 12.4, compute the power function of the test. What
is the power if the true value is 1051 centigrade? If it is 1053
centigrade?
13.4 Eight persons measure their heights (cm) morning and evening.
Result:
individual A B C D E F G H
morning 172 168 180 181 160 163 165 177
evening 172 167 177 179 159 161 166 175

The difference between morning and evening height is assumed

to be from N(  ,  ). Test the hypothesis H 0 :   0 against
H1 :   0 with error risk 5%.
Answers
13.1 0.7175
4 
13.2 h(  )  1    10  1.64485  ,
 0.2 
h(3.8)  7.67  10  0, h(4.3)  0.9990
7

 1050   
13.3 h(  )  1    10  1.9600 
 2.3 
 1050   
  10  1.9600 
 2.3 
h(1051)  0.280 , h(1053)  0.9848

13.4 Confidence interval morning – evening (0.178, 2.32).

Morning height is more than evening height on average.

22
14.1 SIFO made opinion polls for the political parties in May and
June 2016. In May, 498 individuals out of 1937 said they would
vote for “Moderaterna” (M). In June, 531 individuals out of
1937 said they would vote for M. Test with error risk 5% if
there has been a shift in the support for M between the two
months
a) employing a confidence interval for the difference in
proportions in favour of M.
b) employing a  2 -test.
14.2 In a genetic experiment the offsprings were studied when two
types of guinea pigs were interbred. Out of 87 offsprings 43
were red, 10 black and 34 white.
According to the genetic model, the probabilities for these
colours are 9/16, 3/16 and 4/16. Should that hypothesis be
rejected at 5% risk level?
14.3 During 81 days the number of cars that passed a certain point
on a road in a ten minute period were counted. The result was:
#cars 0 1 2 3 4 5 6
#days 14 12 25 16 10 3 1

Test, with 5% risk level, if the number of cars passing can be

regarded as coming from a Poisson distribution.
14.4 Out of each of three populations of humans, P1, P2 and P3, a
random sample was selected. One wanted to test if the gender
distribution was the same in the three populations. The result
was
males females
P1 46 54
P2 78 72
P3 143 107
Test, at a 5% risk level, if the gender distribution is the same in
the three populations.

23
14.5 From 500 accidents on a country road data have been collected
according to this table:
injuries safety belt no safety belt
none or light 101 143
severe 58 198
Test, with error risk 5% if safety belts have an impact on the
severity of injuries.
Answers
14.1 a) June – May (–0.0108, 0.04485). No conclusion
b) p = 0.2300, no conclusion
14.2 p = 0.00653 (df = 2). Reject
14.3 p = 0.1130 (df = 3). No conclusion
14.4 p = 0.1519 (df = 2). No conclusion
14.5 p  6.89  106. Safety belts have an impact

24