Theory of Probabi

lity
351

(11.5) CLASsICAL (OR *A PRIORI" ) DEFINITION OF

PROBABILITY

andom experiment has n

possible oulcomes, which are mutually
erhaustive and equally likely, and m of these are
favourable to an event A,exclsive.
ansbability' of the event is defined as the ratio m/n. In symbols then the
m
PA) =
(11.5.1)
Number of outcomes
Probability of an event = favourable to the event
Total number of mutually
exclusive, exhaustive
and equally likely outcomes
random experiment
of the
It may be noted that
probability', as defined above, is only a ratio of two numbers,
which
in the numerator (m) is the number of favourable cases
and the denominator
in is the total number of possible outcomes certain satisfying conditions. Therefore,
for the calculation of probability the undernoted steps should be followed:
(Step 1): Enumerate all the possible outcomes of the
experiment, such that they
satisfy the 3 criteria of being
'mutually exclusive', 'exhaustive', and
'equally likely' .
Count the number (n) of such outcomes.
(Step 2): Check how many of these cases are favourable to the event
for which
the probability is desired. Let this number be
m.

(Step 3): Divide m by n, and the result gives the probability of the event.
Probability, as defined by (11.5.1), always lies between 0 and 1.
0sps1
minimum value of p, viz. 0, is attained when none of the outcomes is (11.5.2)
The
favourable
to the event, i.e., m 0. The event is then said to be
=
'impossible'. The maximum value
ofp. viz. 1, is attained when all the possible outcomes are favourable to the event, i.e.
m =n. The event is then said to be 'certain'.

Defects of Classical Definition

(i) It is based on the feasibility of subdividing the possible outcomes of the
experiment into 'mutually exclusive', 'exhaustive' and 'equally likely' cases.
Unless this can be done, the formula is inapplicable.
) The phrase "equally likely' appearing in the classical definition, is synonymous
with 'equally probable', which means that we are trying to define probability in
terms of equal probabilities. How do you know whether the probabilities are
cqual, before you can measure them ? The definition is thus circular in nature.
) The definition has only limited applications in coin-tossing, die-throwing and
similar games of chance. Using this definition, we cannot, for example, find
the probability that an Indian aged 25 will die before reaching
the agerates
such probabilities are required to be calculated for fixing the premium
in life insurance). Thus, it may not be practically possible to enumerate all the

Outcomes of a random experiment.

The definition fails, when the number of possible outcomes is infinitely large
 Statistical Methods
352
OD
'Head' in
Example 11.10 What is the prohability of obtaining a si
toss of
an unbiased coin ?

is tossed there are two and .

possible outcomes viz. Head
Solution When a coin
and exhaustive. Moreover, since the
the coin is s
is coin The
two outcomes are mutually exclusive
outcomes are also equally likely. Out
of these two mutually exclusive, exhaustive .and
unbiased, the
eqally
one case is favourable
to the cvent "Head'. Thus,. ng
likely outcomes, only the
definition we have classical
Probability of obtaining 'Head' =

Example 11.11 Two unbiased coins are tossed. What is the probahili
obtaining (a) both heads, (b) one head and one tail, (c) both tails, (d) at of
head? one

Solution The experiment has 4 possible outcomes, viz. HH, HT, TH, T7T (the two la
each case lettersin
denoting the results on the 1st and 2nd coins respectively). These are miu
mutually
exclusive.exhaustiveand equally likely (Examples 11.6c, 11.7c, 11.9c). Thusn =4.Tha
e cases
favourable to the events are as follows:
Event Favourable cases Number of favourable
cases
(a) Both heads HH
(b) One head and tail
one HT TH
(c) Both tails
TT
d) At least one head
HH, HT, TH
Applying the classical definition (11.5.1)
(a) P (both heads) = 1/4
(b) P (one head and
one tail) = 2/4 1/2
(c) P (both tails) = 1/4
d) P(both least one head) = 3/4

Example 11.12 A die is tossed and the number of points

uppermostface is observed. What is the appearing on the
(b) an odd number (c) less than 3, probability of obtaining (a) an even number
(d) a "six" ?
Solution In this
mutually exclusive experiment there are 6
and exhaustive. If possible outcomes, viz. 1, 2, 3, 4. 5, 6. These are
are also the die is
equally likely, Thus 6. Out ofn =
assumed to be unbiased, then the 6
3 cases
(vi., 2,4,6) are
them. outcoc
3 cases
(viz., 2, 3, 5) are favourable to 'even
2 cases favourable to 'odd
number
(viz., 1,2) are
I case (viz., 6) number
favourable to "less than
Thus, is favourable to 'six' 3
P(even number)
3/6 1/2 =
=
Plodd number) 3/6 =

P(less than 3) 1/2 =

P(six)
=
2/6 =
1/3
l/6

Example 11.13
obtaining (a) 3 heads,When two
unbiased coins
(b) not more than 3
are ssed. what
tosse is the probability 9
heads?
 Theory of Probability 357

1.5). (2.6). (3, 7). (4, 8), (5,9). (6,. 10). (7, 11). (8. 12)
ane the total number ofarrangements favourable to the event is m = 8(2 x 10 ). By the

classical
detinition of probability (11.5.).
8x2
p = = 82 x10)
n 12! 12xII33
Second method) Of the 12 places, X and Ycould occupy any 2 places in C, = 66 ways.
which are mutuall exclusive, exhausti and equally likely. There will be 3 persons between
Vand Y. ifthe latter occupy places (1. 5)., (2. 6). (3. 7), 4.8). (5. 9), (6. 10), (7, 11). (8. 12)
Lnce the number of favourable cases is 8. So, the required probability = 8/66 = 4/33.

Example11.25 Twelve persons, amongst whom are Xand Y, are seated at random
dt a round table, What is the probability that there are 3 persons between X and Y?
Solution X can occupy any of the 12 seats, and correspondingly Yany of the remaining11
seats. Hence X and Y can occupy seats in 12 x, 11 = 132 ways. There will be 3 persons in
betweenif X and Ybccupy seats (1, 5), (2,6). (3, 7). (4.8). (5. 9). (6, 10). (7, 11), (8, 12). (9. 1).
(10.2). (11.3), (12,4). Since X and Y may interchangetheirseats, the total numberof favourable
cases is 12 x2 =24.

Required probability =
242
132
(Note that the permutation of n persons in a ring or circle is (n - 1)! but around a 'round

table' is n !: because in the former the positions are relative to their neighbours, but in the latter
relative to their seat numbers.

Example 11.26 A lady declares that by tasting a cup of tea with milk. she can
discriminate whether milk or tea infusion was first added to the cup. In order to test
these assertion, 10 cups of tea are prepared-S in one way and 5 in the other, and
presented to the lady for judgement in a random order. Assuming that the lady has no
discrimination power, calculate the probability that she would judge correctly all the
ups, it being known to her that 5 are of each kinda
What is the probability, if the tea cups were presented to the lady in 5 pairs--each
pair consisting of cups of each kind-in a random order?
Solution Since the 5 cups of each kind, prepared with milk (M) or tea (7) infusion first
added. are identical, the total number of permutations of the 10 cups is (seé formula 11.4.5).
10!
5!5 252
Thus, there are 252 different possible ways of presenting the cups to the lady., and these are
mutually exclusive, exhaustive and equally likely. Only one of these agrees with the lady's
assertion, suppose M T T TMM TM T M. So the required probability is 1/252.
In the second case, when the cups are presented in 5 pairs, the total number of permutationss
2x2x2x 2 x2 = 32
becauseeach pair can be permuted in 2 ways-either (M. T) or (T, M). So there are now
possible ways of presenting the cups to the lady, As before, only one of these agrees withtne
is theretore
uy S statement, say (7TM), (M T), (M D, (TM, and (M T). The required probability
Ans. 1/252, 1/32.
1/32
 This follows from the theorem of compound probability. Similarly,

P(A/B) =
P(AB
PB)
(11.6.17)
Thus the Theorem of Compound Probability enables us to find a formula for e
calculation of conditional probability.
Important Formulae and Results
. Addition Theorem of Probability: If events A and B are
"mutually exclusive'
then
P(A + B) = P(A) + P(B)
2. In general, for any two events A and B,
PA + B) = P(A) + P(B) - P(AB)
3. For three mutually exclusive eventsA, B, C
P(A +B + C) = P(A) + P(B) + P(C)
This can be extended to any
number of mutually exclusive events.
4. For any three events A, B, Cc
P(A+B+ C) =
P(A) + P(B) + P(C) - P(AB)
-

P(AC) P(BC) + P(ABC)

-

5. Probability of the complementary event:

P(A) = 1 - P(A)
6. Boole's inequality:

P(A +B) s P(A) + P(B)

The
sign of equality holds, when A and B are
7.
Multiplication Theorem of Probability: For anymutually exclusive.
two events A and
B,.
P(AB) =
P(A). P(B/A)
8. If events A and B P(B),P(A/B)
are
'independent', then
P(AB) =
PA). P(B)
Conversely, if this relation holds,
then the events A and B
"independent". are said to be
9. For any three
events A, B, C

10. For three P(ABC)= P(A).P(B/A).P(C/AB)

independent events A, B, C,
P(ABC) = P(A).P(B).P(C)
This can be extended
to any number
of
11. f evetns A and
B are independent events.
independent, then
P(A) =
PA/B) P(A/B)=

This is the P(B) =

P(B/A) P(B/ A)
=
conditional and unconditional probability
12 are
equal
PA + B) 1
- P(A B)
=

if A and B are
=
1-PA ).PB).
independent.
 Theory of Probability
367

P(B) = PAB) + P(AB)

13
PA).PB/A) + P( A ).P(B/
14. Conditional Probabiliny:
P(AB)
P(B/A) =
PA) provided P(A) 0.
P(AB)
P(A/B) = P(B) provIded P(B) z 0.

Example 11.35 Given PlA) = ; . P(B) = , PAB)=

(a) find the values ofthe following probabilities:

PA) PA + B), P(A/B), P(AB), P( A B), P( + B)
(h) State whether the events A and B are (i) mutualy exclusive, (ii) exhaustive,
(ii) equaly likely. (iv) independent.
Solution

(a) P(A) =
1-P(A)) = 1
--52
P(A +B) =
P(A) + P(B) -P(AB)

2 12

P(A/B) P(AB)
P(B)
3

P(A B) =
P(B) P(AB)
-
=
4
PA B) = 1-P(A +B) = l-2-5
PA +B) = P(A) + P(B) - P( A B)

(b) i) No; because P(AB) # 0. Here, P(AB) =

() No; because P(A + B) * 1. Here, P(A + B)=

1) No; because P(A) and P(B) are not equal. Here P(A) = and P(B) =

3
(v No; because P(AB)# PlA). P(B). Here P(AB) =, but P(A). PB)= 6

Example 11.36 Given that P(A) . =

3
P(B) =
5
and PA + B) =
Jind PiB
Gnd
P(B/A ). Are A and B [W.B.H.S. '78]
independet:
 (11.7) DRAWING WITHOUT REPLACEMENT
AhOx contains A Wnite and B
black balls. If (a +
probabilitytthat among them hi balls are drawn at
exactly are white and b black
a random, tn
are ack is [see Fig. 11.3(i)]
ACa"C,
a
A+B Ca+b (11.7.1)
There are**°Ca+b
Possible ways of forming groups of
of (a + b) balls out
b) balls out of
of a total of
(A+B) balls, and these groupS are
mutually exclusive, exhaustive and equallytota
+ a

likey
A
a+b A
B a+b+c
B
C

i) Two categories (i1) Three categories

Fig. 11.3 Drawing wi th Replacement (Urn Models)

However, a group of a white balls can be obtained in C, ways because there are A
white balls in the box. Similarly, a group of b black balls can be obtained in BC, ways.
Since any of the AC, groups of a white balls can be combined with any of the C
groups of b black balls, the number of groups of (a + b) balls favourable to the event
is CCh Hence, by the classical definition, the required probability is given by
(11.7.1).
Note: (i) The phrase, 'at random signifies that all possible drawings are "equally likely'.
(ii) If instead of drawing the (a + b) balls all at a time, the balls are drawn one by
one, (a + b) times in succession (a ball drawn once not having been returned to
the box). then the same probability (11.7.1) is obtained. Hence. such drawing is
also call drawing without replacement', as distinct from 'drawing with
Teplacement (Section 11.8).]
 Statistical Methods
374

If the box contains balls of 3 different categories A of them white. B blact

red, and (a +b+ c) balls are drawn, then the probability that among them exacand
are white, b black and c red, is [see Fig. 11.3(ii)]

CCC
AB+C Ca tb+c (11.7.2)
Example 11.47 A bag contains 8 white and 6 black balls. If5 balls aredraw
random, what is the probability that 3 are white and 2 black? twn at
Solution 5 balls can be drawn out of 14 in Cs
ways, and these cases are mutually exclusive,
exhaustive and equally likely. However, a group of 3 18WHITE
white balls can be drawn out of 8 in *C; ways, and 2 6 BLACK (3+2)
black balls out of 6 in °C, ways. So the number of
favourable cases is *C3.°C2. By the classical definition

$CC2 Fig. 1.4 Drawing Balls

P14 Cs
But C 8x1x6 56: 6C=
C=X2
6X =15; (see p. 56)
1x 2x3

14 x 13x 12 x11x10
14C = 14x 13x 11
1x 2x3x 4x5

P = 56 x15 60
14x 13x 11 143
(Second method) We may assume that the balls have been drawn
replacement.
one by one without
Of the 5 balls drawn, the first one
may be any of the 14 in the bag, the 2nd
remaining 13, and so on. So, the total number of ways in which 5 balls may be one of the
the order in which they may be drawn, considering
appear, is 14x 13 x 12 x I1x 10.
Now, let us consider the number of ways in which 3 white
and 2 black balls be obtained
in a particular order; for
example, the first 3 are white and the last 2 are black.may
This number is
(8 x7x6) x (6 x 5). Hence the
probability of drawing only white balls in the first 3
only black balls in the last 2 draws is draws and
(8 x 7x 6) x (6 x 5) 6
14x 13 x 12 x 11x 10 143
But, obviously this is the
probability of having 3 white and 2 black balls in any other
In our problem, the order is
immaterial, and hence the required oruer
and 2 black balls is, by the Theorem of
Total
probability of having 5 W
Probability,.
143 143 t4 (k times) xk
143
where k is the
number of arrangements (i.e., permutations) in which 3 white and 2 black
may appear. Since this is given by
5
6
k 32 =
10, the required probability is 14310=
x 10 60 =

143
 Theory of Probabili ty 375

Example 11.48 Five men in

company of 20 are graduates. If 3 men are picked
a

t of the 20 at randoi., what is the probability that they are all graduates? What is
the probability of at least one graduate? C.U., B.A. (Econ) *73]
Solution i) Applying (11.7.1). let
A Numberof graduates in the company
a =Number of graduates in the sample
B Number of non-graduates in the company = 20 -5 15
Number of non-graduates in the sample
=
0 =

5 G 3 5G 3
15 N-G (3+0) 15 N-G (0+3)

(i) All graduates (ii) None graduate

Fig. 11.5 Drawing from Two Categories
The probability that all are graduates is

P 20 C3
5x 4x3
But. C Ix 2x3
= 10

C 1; 20C= 20*19x18 = 1440,

1x2x3
10x1_
i) In order to find the probability of at least one
P11140
114
graduate, it will be easier to find the
probability of the complementary event, viz. that 'none is a graduate' (i.e. all 3 are
so that A 5, a 0, B 15, b 3. non-graduates),
Applying (11.7.1), the complementary probability
= = = =

is
CC 1x45591
Hence, by
C
(11.6.5), the required probability
I140
is
228

P2 1-9 137
228 228
Alternative method)
(i1) The event at least
.
graduate' can be split up
one into three mutually exclusive events:
exactly 1graduate and 2 non-graduates.
2. exactly 2 graduates and I non-graduatec.
3 exactly 3 graduates and 0 non-graduate.
The probabilities of these cases are, rspectively
525 C x "C I50
C I140
C I140

C,x C 10
20 C 140
By the Theoren of Total Probability (11.6.2), the required probability p, is given by he
Sun of these
probabilities. Hence,
 376 Stat istical Met hods

52 150 10 137
Ans 137
4
140 140 1140 228 228

Example 11.49 A bag red and 5 white balls. Two successive drau.
contains 8
that he first drawins7
halls a r made without eplacement. Find the probuhility will
Rive white halls and the second 3 red balls. ICA. May 7%
Solution Let Adenote the event first drawing gives 3white balls, and B denote the.
event
3 red balls, We have to find the probability of the event "A and e"
'second drawing gives ie
A as well as ". wh:ch in symbols is P ( AB).

PAB) = P(A).

Using (11.7.1)
x 10 5
P(A) =
1C3 286 143

in order to find the conditional probability. we assume that event A has

actually
happencd. ie. white balls have been taken out, so that there remain 8 red and 2 white balls ir
the hag after the first drawing. The probability of getting 3 red balls now is

56 x1
- 1C 120 15

7
Hence. the required probability is P(AB) = 143 429

Example 11.50 Four cards are drawn at random from a full pack. What is the
probability that they belong to different suits?
Solution In the pack of 52 cards, there are 13 cards of each suit-13 spades. 13 hearts. 13
diamonds and 13 clubs. If the 4 drawn cards are to belong to four different suits, we should
have I card from each suit. Hence the required probability is (extension of formula 1.7.2).

| 13 SPADE 4
13 HEART
13 DIAMOND
(1+1+1+1)
13 CLUBB

Fig. 11.6 Drawing from Four Categories

2197
C 20825
(see Exanmple I1.28)
 Theory of Probability
377

11.8) REPEATED TRIALS-DRAWING WITH REPLACEMENT

a a.Certain experiment, the probability of occurrence of an event is p and
the probability of its non-occurrence is I - p =q, suppose. In n repeatedconsequently
trials of the
XDeriment, if p remains a constant in each trial, then the probability that the event
occurs i úme is

"C,p (11.8.1)
where p + q =1.
The performance of anexperiment is usually called a "trial": the occurrence of he
event is called a "'success and its non-occurrence a "failure". Thus, the probability
rsuccesses (and mutuallyn - r failures) in n independent trials is given by(I.8.1)

where p is the probability of success in each trial.

Example 11.51 Find the probability that there will be exactly successes inn
independent trials (n 2 r), where p is the probability of success in a single trial.
Solution Let us, at the beginning, consider the event in which the firstr trials result in
success in each trial, and the remaining n-r urials result in failures only. Now. the probability
ofsuccessis p in each trial, so that the probability of failure is 1 -p = q. say. Since the n trials
are independent, the probability of r consecutive success, followed by n -rconsecutive failures
is. by the theorem of compound probability, given by

PXpX.. pXqXqX... Xq
r times (n r ) times p'q"-
The probability of obtaining r successes (and hence n-r failures) in any other specified
order is similarly p' q'-". However, we are not interested in any particular arrangement of the
r successes and the n- r failures, but overall number of r successes in n trials, whatever be the
arangement in which they appear. But, r successes and n - r failures in n trails may arise in

n
="C, mutually exclusive ways, in each of which the probability is the same, viz.
r(n r)
P.Hence. by the theorem oftotal probability, the probability that there will be r successes.
irespective of the order in which they appear, is
pq-+p'q'-+ .. ("C, times) = "C,p' q=".
Note: This problem may be identified with drawing n balls from a box in which the
proportions of white and black balls are p and q. If balls are drawn one by one, n times in
Succession. the ball drawn each time being returned to the box before the next drawing, then
theprobability of obtaining r white balls is given by (11.8.1). Such experiments are, therefore
called "drawing with replacement". [Aso see Note (i) Section I 1.7, page 31].

Example 11.52 A coin is tossed 10 times. Find the probability of getting

i)exactly 6 heads, and (ii) 9 heads and I tail.
utton Let us describe the appearance of "head' in one toss of the coin as "suceess". Then.
assuming that the coin is unbiased,

P probability of success in each trial =

q =1-p= l-
 Statistical Methods
aftected by
378 Is n o t

any
tossing
.),
(iI8.1 ) with
a
tail in apply
head ot we
of a T h e r e f o r e ,

Also, since the probability of occ

c c u r r e n c e

independen1.

the trials are

Is
the resuits of heads
any other tossing. the u probability of6
Therefore,
the
n 10.p . Her, r
=6.
2
210 05
is a
i0 512 the femaining
obviously
s u c c e s s e s
(and
(ii) We have of 9
nthe
to find the
e requied probabi
probability

Tallure). Putting r = 9,
10
10 512

a graduate
student will be
Example 11.53 Theprobabiliry that an entering
college
students, (i) none, fi) one
the probabilin that out of 5 entering
(iii) at least
e
one, will be a
graduate. be called a "success
Solution Let the event 'an student
student will be a graduate
n entering college
centering wI or
henp= prohahil..
of success each trial =0.4. Assuming that the success
or
lanure OiE
Probability in
be considered as naepenueni
t docs not affect the result of anv other student, the trials may
Hence. putting n = 5, p =0.4. q = 1 - p =0.6 in (11.6.1)
(1)
Probability that none is a graduate
probability of Osuccess =
"Co (0.4)° (0.6)-
= (1) (1) (0.6) = 0.07776
ii) Probability that one is a graduate
probability of one success = "C, (0.4) (0.6)
5 (0.4) (0.6 = 0.2592
(11) The complementary event is 'none is a graduate'. Hence, applying (11.6.5).
Probability that at least one is graduate
= 1- (probability tat none is a graduate)
1 07776, from (i) above.
92224.

Example 11.54 A machine produces on the average 2 per cent defectives. f 4

articles are chosen randomly, find the probability that there are at least 2 defective

articles.
Sobution Denoting the occurrence of a detectuve article as "success", we find that

defective article 20%

probability of a
=
00 02
p
= l -0.2 = 98
y
occurrence of a delective
article does not affect the
of another artic
Since the trials are independent.
or not, hence the
u c c e s s e s in 4 trials The
Theretore, applying (11.8.ih.
we have
being defective of al least 2
2 successes
s
trals. he complementary event onsists
the probability
to find cases,
exclusive
viz.
Viz. (i)
(i) exactly
exact.
oc
success ill 4 trials, and
(ii) exacth I succes
mutually cases dre respetne
Two ol these

The probab1iities =
(iiit approx.
4 trials. C.02 (98)*"
C(.02) ( 9 8 ) : (4 (02 98 )75
(appox
 neory, bul

theorem' is derived as a rule (see 11.17.2) which follows from

'multiplication
in axiomatic theory.
the definition of conditional probability
of "conditional probability" and *"independent events" are
(vi) The concepts
whereas in the
introduced in the classical theory by logical arguments,
axiomatic theory these are defined by mathematical statements.

(11.19 RANDOM VARIABLE

Let S be a sample space of some given random experiment. It has been observed tnat
the outcomes (i.e. sample points of S) are not always numbers. We may howevet
an
assign a
sample point according to some definite rule. Sucn
real number to each
assignment given us a "function defined on the sample space S". This Tune
called a random variable (or
stochastic variable).
 Theory of Probability
407

RandomVariable X may be defined as a function which assigns a real number X(e)

h sample point e of a given sample space S.

lustratnon 1.
rando expeniment of tossing 2 coins, let the sample space be S (HH, =
In the
the randor variable denoting the 'number heads', HT*".
TH.77).If is
of then wehave
assigned
a number to each sample point as follows:
XHH) = 2. X(HT) =I, X(TH) = 1. X(T) = 0

Note: Many other rand variables could also be proposed on this sample
h esQuare of the number of tails, the difference of the space.
numbers of heads and
tails.
llustration 2.

Let S= |1. 2. 3, 4, 5, 6} be a sample space of the random experiment

die The sample points are 1, 2. 3, 4, 5, 6. We introduce the randomof throwing a
number of points obtained when the die is thrown. Then
variable X =

X1)
1. X(2) 2. X(3) 3. X(4) 4.
= = =
X5) = 5, X(6) 6 =
=

Suppose, we introduce another random variable Y on the same

Number of points less 3¥. Then sample space, Y =

Y1) (1-3) =4 =
Y(4) = (4 -3)2 = 1
Y2) = (2-3) = 1 Y5) = (5 -3) = 4
Y3) = (3-3 3)2:= 0 Y6) = (6 -3) = 9
These are shown in the table below:
Sample point 5
Random variable Y 4 4 9
A random variable which assumes a
finite number or a
of values is called a Discrete random countably infinite number
variable. If the random variable
assumes an
uncountably infinite number of values, it is called a Continuous
random variable.
llustration3.
The random variable
Xin Illustration 1 takes
takes 6values 1,2, 6, and Ytakes 4 values only
3 values 0, 1, 2. In
Illustration 2.X
...

0, 1,4,9. These are all discrete random

variables with a finite number of values.
Discrete Probability Distribution
Let X be a discrete random variable which can assume
the values X1, X2, Xj,
in an
increasing order of magnitude) with probabilities pi.
(arranged ...

2= 1). The specification P2» P3 respectively

.

of the set
P,i= 1,2, 3, ...) defines the
of values x, together with their
probabilities
Let us write flx)
discrete probability distribution of X.
to denote the probability that X takes a specified value x.
x ) = P(X =x)
he (11.19.1)
function flx) is called the
probability mass
probubility function, of the diserete random variable X.furnction (p.m.f.), or simply
It satisfies two conditions
i) fAx)20,
Ie
(ii) Efx) 1 =
(11.19.2)
Summation in (ii) being taken over all values
fx) = p when x = x(i = 1, 2,...)
of x =
A1, , Ag .

= 0 otherwise
(11.19.3)
 408 Statistical Methods

Example 11.74 Find the probability distributions of the random variahle

llustrations I and 2. in

Solution
1)
Assuming that the coin is unbiased. the sample space is tinite and
equiprobable
PX =
0) =P([TT})=
PX =
1)= P(|HT, TH}) = 4 2
POX =
2) =
P(|HH)) =
The probability distribution of X is
0 Total

fx)
4 2
(11) In llustration 2, assuming that the die is unbiased, the probability distribution of X is
2 3 4 5 6 Total

flx) 6 6 6

Again. P(Y 0) =P({3}))=

P(Y =1)= P({2, 4]) =
P(Y 4) =P{1,5})= 3
P(Y = 9)= P({6}) =

The probabiity distribution of Y is therefore

9 Total

6 6
Note the distinction between capital letters X, Y which
denote the random variables. and
small letters x, y which denote any arbitrary values of these
random variables.

Example 11.75 Find the probability distribution of the number of tails when a
coin is thrown repeatedly until the first head appears.
Solution Here S = {H, TH, TIH, TTTH . .
Sample point H TH TTH TTTH
X 0 2 3
Assuming that the coin is unbiased,

PX =
0) =P({H}) =

PX = )= P[I#) = |
 Theory of Probability
409

PX = 2)=
P|TTH)) =|G
PX= 3) = P({T1TH})= etc.

distribution of X is
The probability
0 .. Total

ontinuous Probability Distribution

If X is a continuous random variable, the number of possible values which it can
ume is uncountably infinite, and hence the probability function cannot be defined
the same manner as
for a discrete random varaible. In this case we introduce a
in
such that it satisfies two conditions.
function fx)
) f)20 ( ) f x ) dr = 1. (11.19.4)
The probability that X lies between two specified values c and d, is defined as

PcSXSd =
f ) dr (11.19.5)
It may be shown that this definition of probability satisfies all the Axioms of
probability (page 409). The function fx) is called the probability density function
random Any function
(p.d.f.), or simply densityfunction,
of the continuous variable X.
satisfying both the conditions (11.19.4) may be accepted as a density function.

Example 11.76 (a) Is the following a probubility density function?

2x, 0<xSI
Ax)= 4-2x, 1<xs2
0, elsewhere
(b) If the random variable X has the probability density function

-2SS2
fr)=
0 elsewhere
obtain P[(2xr + 3) > 5]. (Here P denotes probability)
C.U., B.Sc. (Econ) '81 (New)|

Solution
must satisty the two conditions (i) and (i) at (11.194).
a) Ifthe function fr) is to be a p.d.f. it satisfies (i). It is not negative at any
value of
x.
The function flr) defined here clearly

flxdr =fdr +d + fmds +.fmds

+0
-
0+21)ds+4-2ds detined
being zero, since f(x)itselfis
(the first and the last integrals
those intervals)
to be zero in
two
 410
Statistical Methods

l+1 =2
Since condition (ii) is not satisfied, the given function f(a) Is nol a
probability dles
(b)
function. density
The event (21 +3) > 5 2r>2 > >1. Therefore
P(2x+3) > 5) = Plu> I}|

S ) da , since flx) = 0 for x > 2.

Ans. No.

(11.20 CUMULATIVE DISTRIBUTION FUNCTION (C.D.F.)

The probability that a random variable X takes a value less than or equal to a
specified
value of x, and is usually This function is
x a
is function by denoted
Cumulative distribution function (c.d.f.), or simply distribution function, of the randon
Fo). called
variable X.
Fu) = P(XSx)
Since Flr) represents probability, it must lie between 0 and 1.
(11.20.1
0S Ft) s i (11.20.2)
For the discrete probability distribution of X, which assumes a finite number of
values , X2 . .

X, With probabilities p1. P2 P, respectively, the.

c.d.f. is given as
follows:
Fx)= 0 when x<X1
P +
when XSx<X2
=
P P2 when Sx<xz
PtP2+P3 when X3Sr<X4

=
Pi +P2 +. +P=l when x,Sx (11.20.3)
The probability that X lies in interval,
an
say between x and x7 (both inclusive). 1s
then given by

Plx S XSx) =
P(X S x7) -

P(X Sx)
For the continuous
=
Fa) F(x,) -

(11.20.4)
probability distribution of X, with probability density function
fx) in the interval (a SISb). the c.d.f. is given as
1.20.5)
If follows that the
with respect to x.
p.d.f. flx) can be obtained from the c.d.f. Flx) dy differentia

If the
Ax) =Fx) (11.20.6)
c.d.f. is known, the
probabilities of intervals can be obtained from the on
P(c SXSd) ran
particular when c and d are equalFid)
=
In
-

Fc) 11.20.7)

i.e. Plc XSc) Fic)- =

Fic) = 0
PX = c) = 0 (11.20.8)
 Theory of Probability
411

Thus, the arobability

probability that a continuous random variable takes any
o
specified value
consequence.
As a
zero.
js
Pc SX Sd) = Plc S X< d
=
Plr <
Xsd) =Plr <X<d).
Example 11.77 An unbiased coin
thown three times. If the randm
is
variable
rnotes the umber of heads ohtained. find the cumulative distribution fune tiom
muma

cdfofN.

caltion lt can be easily shown that the probabilities of obtaining 0 head. I head, 2 heads
I331
i heads are respectively R'RR The probability distribution of X is therefore
0
Total

f)
8 8
From this table, we can findthe probabilities of obtaining 0 or less head. I or less
less heads. 3 or less heads, as follows:
head, 2 or

F(0) = PX < 0) = PX = 0) =

F1) = PX<1)

PX=0)+ PX =1) =
+
F2) = P(X< 2)
=
P(X =0) + P(X =
1) + P(X =
2)

F3) = P(X < 3)

=
P(X =
0) + P(X =1) + P(X =
2) + P(X =
3)

Thus we have
0
3
Fx) 7
8 8
In general, if denotes any
X) is
x
arbitrary real number, the probability that the number of heads
x or less can be given as
Fx) = 0, when x< 0
Decause we cannot have less than O head.

Fx)R when 0 X<l

ror
example F(0.7) =
because the probability of having 0.7 or less head is the siane as
that of having () head.

Fi)= when ISr<2

for
example. F(I1.34) = the probability of having 11.34 or less heads the
because
2
is

fme as
that of having 0 or I head. Similariy,
 412 St atistical Methods

when 2 < 3
8 8 8
3
and Fu)=+ when 3 Sr
For example F(8.75) = 1, because the probability of having 8.75 heads or less is the same
ame as
that of having 0. 1. 2. 3 heads
The cdf. of X is thus given as follows:

Fa) = 0, when X<0

when 0 SI<l

when 1 I < 2

when 2 a < 3
8
when 3 x

Example 11.78 The p.df. of a continuous random variable is

y = k(r-)(2-x);
(1xS2)
Determine
i) the value of the constant k;
(ii) the cumulative distribution function;

(ii) the probabiliry that X is less than

3
(iv) the probubility that X is greater than

5
(v) the probability that Xlies between and.
4
Solution y= klx - 1)(2- x) k(3x
-x 2)
= -

i) Using the second relation of (11.19.4)

I = k(3x- -2) dx

k
Thus,l:
6 i.e. k =6
The complete p.d.f. is therefore y =6(3x x - 2)
-

(ii) Fl)=63 -2) dt, by (11.20.5)

 Theory of Probabi l i ty
413

5- 12r+ 9 2

(ii) Probability
that Xis less than

by (11.20.1)

5-12x+9x -2

v) Probability that Xis less than

2

3 9 27
=

5-12x +9x -2x

Probability that Xis greater than

(v) Probability that Xlies between 3

and

xG}'E)
1_5- I
2 2 32 32

11.21) JOINT DISTRIBUTION OF TWO VARIABLES

(DISCRETE)
Let S be a
real
sample space of some given random experiment. We may assign
numbers X(e) and Ye) each sample point e of S
to tw
Section 1.19). We then have two random according to some givenrules
space S.
variables X and Y defined on the sample
The possible
pairs of values (x, y) assumed by the two random variables X and
ogether with the
of X and probabilities for all the pairs of values, gives the Joint Distributicom
Y, or Bivariate Distribution.
Example 11.79 Four unbiased coins are lossed. X and
If Y denote
respectivey
number of heads ' and the 'longest run of heads', construct the joint distrihution
ofX and Y.