Section 5.

Mean and Variance (LECTURE NOTES 4) 57

2.5 Mean and Variance

The mean (or expected value) of a discrete random variable X with range R and pmf
f (x) is, provided series converges absolutely,
X
µ = E(X) = xf (x).
x∈R

The variance of discrete random variable X with range R and pmf f (x) is, provided
series converges absolutely,
X
σ 2 = V ar(X) = (x − µ)2 f (x) = E[(X − µ)2 ]
x∈R
X
= x2 f (x) − µ2 = E(X 2 ) − [E(X)]2 = E(X 2 ) − µ2 ,
x∈R

with associated standard deviation, σ.

For uniform random variable X with range R = {1, 2, . . . , k}, and pmf f (x) = k1 ,
r
2
k+1 k − 1 k2 − 1
µ = E(X) = , σ 2 = V ar(X) = , σ = ;
2 12 12
for binomial random variable,
√
µ = E(X) = np, σ 2 = V ar(X) = npq, σ = npq;

for Poisson random variable,

√
µ = E(X) = λ, σ 2 = V ar(X) = λ, σ= λ.

Exercise 2.5 (Mean and Variance)

1. Expected value, variance and standard deviation: seizures. The probability
function for the number of seizures, X, of a typical epileptic person in any
given year is given in the following table.

x 0 2 4 6 8 10
f (x) 0.17 0.21 0.18 0.11 0.16 0.17

(a) Calculating the expected value. The expected value (mean) number of
seizures is given by
X
E(X) = xf (x) = 0(0.17)+2(0.21)+4(0.18)+6(0.11)+8(0.16)+10(0.17) =
x

(i) 4.32 (ii) 4.78 (iii) 5.50 (iv) 5.75.

58 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

P(X = x) P(X = x) P(X = x)

0.20 0.20 0.20

0.15 0.15 0.15

0.10 0.10 0.10

0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10

(a) (b) (c)

Figure 2.11: Expected value: fulcrum point of balance

x <- c(0,2,4,6,8,10) # values of random variable

px <- c(0.17,0.21,0.18,0.11,0.16,0.17) # probabilities
EX <- sum(x*px); EX # expected value

[1] 4.78

(b) Understanding expected value: seizures.

If the expected value is like a fulcrum point which balances the “weight”
of the probability distribution, then the expected value is most likely close
to the point of the fulcrum given in which of the three graphs above?
(i) (a) 9 (ii) (b) 5 (iii) (c) 1.
(c) Variance. The variance in number of seizures is given by

σ 2 = V ar[X] = E (X − µ)2
 
X
= (X − µ)2 f (x)
x
= (0 − 4.78)2 (0.17) + (2 − 4.78)2 (0.21) + · · · + (10 − 4.78)2 (0.17) ≈

(i) 7.32 (ii) 8.78 (iii) 10.50 (iv) 12.07.

VarX <- sum((x-EX)^2*px); VarX # variance

[1] 12.0716

(d) Standard Deviation. The standard deviation in the number of seizures is

p √
σ = V ar(X) ≈ 12.07 ≈

(circle one) (i) 3.47 (ii) 4.11 (iii) 5.07 (iv) 6.25.
SDX <- sqrt(VarX); SDX # standard deviation

[1] 3.474421

In other words, we expect to see about µ ± σ = 4.78 ± 3.47 seizures

according to the probability distribution given here.
Section 5. Mean and Variance (LECTURE NOTES 4) 59

4.78 +- 3.47 4.78 +- 2.47 4.78 +- 1.47

P(Y = y) P(Y = y) P(Y = y)

0.20 0.20 0.20

0.15 0.15 0.15

0.10 0.10 0.10

0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10

(a) seizure distribution (b) another distribution (c) and another distribution

Figure 2.12: Standard deviation: dispersion of distribution

(e) Understanding standard deviation: “dispersion”.

The standard deviation measures the dispersion of a probability distribu-
tion. The most dispersed distribution occurs in
(i) (a) (ii) (b) (iii) (c).

2. Variance and standard deviation: rolling a pair of dice. If the dice are fair, the
distribution of X (the sum of two rolls of a pair of dice) is

x 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 5 4 3 2 1
f (x) 36 36 36 36 36 36 36 36 36 36 36

(a) Expected value, g(x) = x,

   
X 1 1
µ = E(X) = xf (x) = 2 + · · · + 12 =
x
36 36

(i) 4 (ii) 5 (iii) 6 (iv) 7.

x <- 2:12 # values of random variable
px <- c(1,2,3,4,5,6,5,4,3,2,1)/36 # probabilities
EX <- sum(x*px); EX # E(X)
[1] 7

(b) If g(x) = x2 ,
   
2
X
2 2 1 2 1
E(X ) = x f (x) = 2 + · · · + 12 =
x
36 36

(i) 35.43 (ii) 47.61 (iii) 54.83 (iv) 65.67.

EX2 <- sum(x^2*px); EX2 # E(X^2)
[1] 54.83333

(c) Variance.
σ 2 = V [X] = E (X − µ)2 = E X 2 − µ2 = 54.83 − 72 ≈
 

(i) 3.32 (ii) 5.83 (iii) 7.50 (iv) 8.07.

60 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

VarX <- EX2 - EX^2; VarX # variance

[1] 5.833333

(d) Standard deviation.

p √
σ= V ar(X) ≈ 5.83 ≈

(i) 2.42 (ii) 3.11 (iii) 4.07 (iv) 5.15.

SDX <- sqrt(VarX); SDX # standard deviation

[1] 2.415229

3. Uniform: die. Fair die has the following uniform pmf.

x 1 2 3 4 5 6
1 1 1 1 1 1
f (x) 6 6 6 6 6 6

(a) Since pmf of uniform is f (x) = k1 , in this case,

k=

(i) 5 (ii) 6 (iii) 7 (iv) 8.

(b) The expected value of die is
X 1 1 1 1 1 1
µ = E(X) = xf (x) = 1 · +2· +3· +4· +5· +6· =
x
6 6 6 6 6 6

(i) 2.5 (ii) 3.0 (iii) 3.5 (iv) 4.0,

x <- 1:6 # values of random variable
px <- rep(1/6,6) # probabilities: 1/6 repeated 6 times
EX <- sum(x*px); EX # E(X)

[1] 3.5

or, using the formula, the expected value of die is

k+1 6+1
µ= = =
2 2
(i) 2.5 (ii) 3.0 (iii) 3.5 (iv) 4.0.
(c) If µ = 11, then
k+1
µ = 11 =
2
so k = (i) 11 (ii) 21 (iii) 22 (iv) 23
that is, the die has 21 sides.
Section 5. Mean and Variance (LECTURE NOTES 4) 61

4. Another die question. Fair six-sided die is labelled in one of three ways: there
are two sides labelled 1, three sides labelled 2 and one side labelled 3. If it costs
$1 to play and you win $1 × result from die, what is the expected value of this
game?

die 1 2 3
x, payoff 1 − 1 2−1 3−1
2 3 1
f (x) 6 6 6

The expected value is

X 2 3 1
µ = E(X) = xf (x) = 0 · +1· +2· =
x
6 6 6
1 3 5
(i) 6
(ii) 6
(iii) 6
(iv) 76 .
x <- 0:2 # values of random variable
px <- c(2,3,1)/6 # probabilities
EX <- sum(x*px); EX # E(X)

[1] 5/6

5. Binomial: Airplane engines. Each engine of four (n = 4) on an airplane fails

11% (p = 0.11, q = 1 − p = 0.89) of the time. Assume this problem obeys the
conditions of a binomial experiment, in other words, X is b(4, 0.11).

x 0 1 2 3 4
f (x) 0.310 0.058 0.005 0.000

(a) Fill in the

 blank:
 the chance no (zero) engines fail is
4
f (0) = 0.110 0.894 = (i) 0.005 (ii) 0.058 (iii) 0.310 (iv) 0.627.
0
dbinom(0,4,0.11) # binomial pmf

[1] 0.6274224

(b) The expected number of failures is

X
µ = E(X) = xf (x) = 0(0.627)+1(0.310)+2(0.058)+3(0.005)+4(0.000) ≈
x

(i) 0.44 (ii) 0.51 (iii) 0.62 (iv) 0.73.

or, using the formula, the expected number of failures is

µ = np = 4(0.11) =
(i) 0.44 (ii) 0.51 (iii) 0.62 (iv) 0.73.
62 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

(c) The variance in number of failures is

σ 2 = V [X] = E (X − µ)2
 
X
= (X − µ)2 f (x)
x
= (0 − 0.44)2 (0.627) + (1 − 0.44)2 (0.0.310) + · · · + (4 − 0.44)2 (0.000) ≈

(i) 0.15 (ii) 0.39 (iii) 0.51 (iv) 0.63.

or, using the formula, the variance in number of failures is

σ 2 = npq = 4(0.11)(1 − 0.11) ≈

(i) 0.15 (ii) 0.39 (iii) 0.51 (iv) 0.63.

(d) The √
standard deviation in number of failures is
σ = 0.39 ≈ (i) 0.45 (ii) 0.56 (iii) 0.63 (iv) 0.83.

6. Bernoulli: mean and variance formulas. Bernoulli pmf is given by:

x 0 1
f (x) 1−p p

(a) The expected value is

X
µ = E(X) = xf (x) = 0(1 − p) + 1(p) =
x

(i) p (ii) 1 − p (iii) (1 − p) (iv) p(1 − p).

(b) The variance is
X
σ 2 = V ar(X) = (x − µ)2 f (x) = (0 − p)2 (1 − p) + (1 − p)2 (p) =
x

(i) p (ii) 1 − p (iii) (1 − p) (iv) p(1 − p).

7. Poisson: accidents. An average of λ = 3 accidents per year occurs along the I–95
stretch of highway between Michigan City, Indiana, and St. Joseph, Michigan.

(a) Expectation. The expected number of accidents is

µ = E(X) = λ =
(i) 1 (ii) 2 (iii) 3 (iv) 4.
(b) Expected cost. If it costs $500,000 per accident, the expected yearly cost is
E(C) = E(500000X) = 500000E(X) = 500000(3) =
(i) $500, 000 (ii) $1, 000, 000 (iii) $1, 500, 000 (iv) $2, 000, 000.
Section 6. Functions of a Random Variable (LECTURE NOTES 4) 63

(c) Variance. The variance in the number of accidents per year is

σ 2 = V ar(X) = λ =
(i) 1 (ii) 2 (iii) 3 (iv) 4.
(d) Standard
√ deviation.
√ Standard deviation in number of accidents per year
σ= λ= 3≈
(i) 1.01 (ii) 1.34 (iii) 1.73 (iv) 1.96.

2.6 Functions of a Random Variable

Expected value of a function u of random variable X, E[u(X)], is
X
E[u(X)] = u(x)f (X).
x

Some properties are

X
E(a) = a f (x) = a,
x
E[au(X)] = aE[u(X)],
E[a1 u1 (X) + a2 u2 (X) + · · · + ak uk (X)] = a1 E[u1 (X)] + a2 E[u2 (X)] + · · · + ak E[uk (X)]

where a, a1 , a2 , . . . ak are constants (numbers, not random variables). Furthermore,

V ar[aX] = a2 V ar(X), V ar[a] = 0.

Exercise 2.6 (Functions of a Random Variable)

1. Functions of random value: seizures. The pmf for the number of seizures, X,
of a typical epileptic person in any given year is given in the following table.

x 0 2 4 6 8 10
f (x) 0.17 0.21 0.18 0.11 0.16 0.17

(a) Recall, the expected value number of seizures is

X
E(X) = xf (x) = 0(0.17)+2(0.21)+4(0.18)+6(0.11)+8(0.16)+10(0.17) =
x

(i) 4.32 (ii) 4.78 (iii) 5.50 (iv) 5.75.

x <- c(0,2,4,6,8,10) # values of random variable
px <- c(0.17,0.21,0.18,0.11,0.16,0.17) # probabilities
EX <- sum(x*px); EX # E(X)

[1] 4.78
64 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

(b) If the medical costs for each seizure, X, is $200; in other words, function
u(x) = 200x, the probability distribution for u(X) is:
x 0 2 4 6 8 10
u(x) = 200x 200(0) = 0 200(2) = 400 800 1200 1600 2000
p(u(x)) 0.17 0.21 0.18 0.11 0.16 0.17
The expected value (mean) cost of seizures is then given by
X
E[u(X)] = E[200X] = (200x)f (x) = [0](0.17)+[400](0.21)+· · ·+[2000](0.17) =
x

(i) 432 (ii) 578 (iii) 750 (iv) 956.

u <- 200*x # U(X), cost of seizures
EU <- sum(u*px); EU # E(U)

[1] 956

Alternatively, since c = 200 is a constant,

E[u(X)] = E[200X] = 200E[X] = 200(4.78) =

(i) 432 (ii) 578 (iii) 750 (iv) 956.
EU <- 200*EX; EU # E(U)

[1] 956

(c) If the medical costs for each seizure, X, is given by function u(x) = x2 ,
x 0 2 4 6 8 10
u(x) = x2 02 = 0 4 16 36 64 100
p(u(x)) 0.17 0.21 0.18 0.11 0.16 0.17
The expected value (mean) cost of seizures in this case is given by
X
E[u(X)] = E[X 2 ] = x2 f (x) = [0](0.17) + [4](0.21) + · · · + [100](0.17) =
x

(i) 34.92 (ii) 57.83 (iii) 75.01 (iv) 94.56.

EX2 <- sum(x^2*px); EX2 # E(X^2)

[1] 34.92

(d) If u(x) = 200x2 + x − 5,

E[u(X)] = E[200X 2 + X − 5]
= E(200X 2 ) + E(X) − E(5)
= 200E(X 2 ) + E(X) − E(5)
= 200(34.92) + 4.78 − 5 =
(i) 4320.67 (ii) 5780.11 (iii) 6983.78 (iv) 8480.99.
Section 6. Functions of a Random Variable (LECTURE NOTES 4) 65

u <- 200*x^2 + x - 5 # U(X), cost of seizures

EU <- sum(u*px); EU # E(U)

[1] 6983.78

2. More functions of random variable: flipping until a head comes up. A (weighted)
coin has a probability of p = 0.7 of coming up heads (and so a probability of
1 − p = 0.3 of coming up tails). This coin is flipped until a head comes up or
until a total of 4 flips are made. Let X be the number of flips. Recall,

x 1 2 3 4
f (x) 0.700 0.210 0.063 0.027

(a) If u(X) = x,
X
µ = E(X) = xf (x) = 1(0.700) + 2(0.210) = 3(0.063) + 4(0.027) =
x

(i) 1.117 (ii) 1.217 (iii) 1.317 (iv) 1.417.

x <- 1:4 # values of random variable
px <- c(0.700,0.210,0.063,0.027) # probabilities
EX <- sum(x*px); EX # E(X)

[1] 1.417

(b) If u(X) = x1 ,
  X
1 1 1 1 1 1
E = f (x) = (0.7) + (0.21) + (0.063) + (0.027) =
X x
x 1 2 3 4

(i) 0.41755 (ii) 0.83275 (iii) 1.53955 (iv) 2.56775.

u <- 1/x # U(X)
EU1 <- sum(u*px); EU1 # E(U)

[1] 0.83275

200 1
(c) If u(X) = x
+ 200x
+ 5,
 
200 1
E[u(X)] = E + +5
X 200X
  
1 1
= E 200 + +5
200 X
   
1 1
= 200 + E + E[5]
200 X
 
1
= 200 + (0.83275) + 5 =
200
(i) 43.20 (ii) 57.80 (iii) 109.35 (iv) 171.55.
66 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

u <- 200/x + 1/(200*x) + 5 # U(X)

EU <- sum(u*px); EU # E(U)
EU <- (200 + 1/200)*EU1 + 5; EU # E(U) again

[1] 171.5542

3. Another function. Assume random variable X has uniform pmf

1
f (x) = , x = 1, 2, 3, 4.
4
(a) If u(x) = x,
1 1 1 1 10
µ = E(X) = 1 · +2· +3· +4· = =
4 4 4 4 4
(i) 1.0 (ii) 1.5 (iii) 2.0 (iv) 2.5
x <- 1:4 # values of random variable
px <- rep(1/4,4) # probabilities: 1/4 repeated 4 times
EX <- sum(x*px); EX # E(X)

[1] 2.5

(b) If u(x) = x2 ,
1 1 1 1 30
E[X 2 ] = 12 · + 2 2 · + 32 · + 4 2 · = =
4 4 4 4 4
(i) 6.75 (ii) 7.00 (iii) 7.25 (iv) 7.50
EX2 <- sum(x^2*px); EX2 # E(X^2)

[1] 7.5

and also
V ar(X) = E(X 2 ) − µ2 = 7.5 − 2.52 =
(i) 1.25 (ii) 1.50 (iii) 1.75 (iv) 2.50
VarX <- EX2 - EX^2; VarX # variance

[1] 1.25

(c) If u(x) = 2x,

E[2X] = 2E(X) = 2(2.5) =
(i) 5 (ii) 6 (iii) 7 (iv) 8
u <- 2*x # U(X)
EU <- sum(u*px); EU # E(U)

[1] 5

and also
V ar[2X] = 22 V ar(X) = 4(1.25) =
(i) 5 (ii) 6 (iii) 7 (iv) 8
VarU <- 2^2*VarX; VarU # Var(U)
Section 6. Functions of a Random Variable (LECTURE NOTES 4) 67

[1] 5

(d) If u(x) = 2x + 3x2 ,

E[2X + 3X 2 ] = 2E(X) + 3E(X 2 ) = 2(2.5) + 3(7.5) =
(i) 23.5 (ii) 24.5 (iii) 25.5 (iv) 27.5.
u <- 2*x + 3*x^2 # another U(X)
EU <- 2*EX + 3*EX2; EU # E(U)
[1] 27.5

2
4. Random variable X has mean µX = µ, and variance σX = σ 2 . If u(x) = 3x + 4,
then
µU = E[3X + 4] = 3E(X) + 4 =
(i) µ + 4 (ii) 2µ + 4 (iii) 3µ + 4 (iv) 4µ + 4

and also
σU2 = V ar[3X + 4] = 32 V ar(X) + 0 =
(i) 8σ 2 + 4 (ii) 8σ 2 (iii) 9σ 2 (iv) 10σ 2
5. Consider random variable X where E[X + 2] = 4 and E[X 2 + 4X] = 3, then
E[X + 2] = E(x) + 2 = 4,
so µ = E(X) = (i) 1 (ii) 2 (iii) 3 (iv) 4

and also since

E[X 2 + 4X] = E(X 2 ) + 4E(X) = E(X 2 ) + 4(2) = 3
then E(X 2 ) = (i) −5 (ii) −3 (iii) −1 (iv) 1

so V ar(X) = E(X 2 ) − µ2 = (i) −9 (ii) −7 (iii) −6 (iv) −5

6. For Poisson random variable X where λ = 3, determine E[X 2 − 3X].

Since
µ = E[X] = λ =
(i) 1 (ii) 2 (iii) 3 (iv) 4

and also since V ar(X) = E(X 2 ) − µ2 = λ, then

E(X 2 ) = λ + µ2 = λ + λ2 = 3 + 32 =
(i) 9 (ii) 10 (iii) 12 (iv) 13

so
E[X 2 − 3X] = E(X 2 ) − 3E(X) = 12 − 3(3) =
(i) 3 (ii) 4 (iii) 5 (iv) 6
68 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

2.7 The Moment-Generating Function

The moment generating function (mgf ) of random variable X (taken about the origin)
with pmf f (x) and range R is defined by, assuming the expectation exists,

X tX
M (t) = E etX = e f (x).
x∈R

Furthermore, if random variable X and its mgf M (t) exists for all t in an open interval
containing 0, then
• M (t) uniquely determines the distribution of X,
• M 0 (0) = E(X), M 00 (0) = E(X 2 ).
Also, if Y = aX + b,

MY (t) = E[eY t ] = E e(aX+b)t = E e(at)X ebt = ebt E e(at)X ,

     

but E[e(at)X ] = MX (at), so

MY (t) = ebt MX (at).
The probability-generating function (pgf ) is
∞
X
X
P (t) = E[t ] = tx f (x),
x=0

where P 0 (1) = E(X).

DISCRETE f(x)
 M (t) µ σ2
n
Binomial px q n−x (pet + q)n np npq
x
x
λ(et −1)
Poisson e−λ λx! e λ λ
pet
Geometric q x−1 p 1−qet
1/p q/p2
 
x−1 
pet
r
Negative Binomial pr q x−r 1−qet
r/p rq/p2
r−1

Exercise 2.7 (The Moment-Generating Function)

1. Deriving mgf from pmf. If random variable X has range R = {1, 2, 3} with pmf
f (1) = 21 , f (2) = 13 and f (3) = 16 , then

X tX
M (t) = E etX = e f (x) =
x∈R

t t t
(i) 13 et + 21 e2t + 16 e3t (ii) 12 et + 13 e2t + 16 e3t (iii) e 2 + 2e 3 + 3e 3
Section 6. The Moment-Generating Function (LECTURE NOTES 4) 69

2. Deriving pmf from mgf. What is the pmf of X if

1 1 1
M (t) = et + e2t + e3t ?
2 3 6
(a) pmf A
x 1 2 3
1 1 1
f (x) 3 2 6

(b) pmf B
x 1 2 3
1 1 1
f (x) 2 3 6

(c) pmf C
1 1 1
x 2 3 6
f (x) 1 2 3
3. Expected value using mgf. What is the expected value of
1 1 1
M (t) = et + e2t + e3t ?
2 3 6
On the one hand, since M (t) is equivalent to

x 1 2 3
1 1 1
f (x) 2 3 6

X 1 1 1
E(X) = xf (x) = 1 · +2· +3· =
x
2 3 6
3 4 5 6
(i) 3
(ii) 3
(iii) 3
(iv) 3

On the other hand, since

1 1 1
M (t) = et + e2t + e3t ,
2 3 6
the first derivative of M (t) with respect to t is
 
0 d 1 t 1 2t 1 3t 1 1 1
M (t) = e + e + e = et + 2 · e2t + 3 · e3t
dt 2 3 6 2 3 6
and so evaluating this derivative at t = 0,
1 1 1
M 0 (0) = e0 + 2 · e2(0) + 3 · e3(0) = E(X) =
2 3 6
3 4 5 6
(i) 3
(ii) 3
(iii) 3
(iv) 3

which (i) the same (ii) different from before.

70 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

4. Variance using mgf. What is the variance of

1 1 1
M (t) = et + e2t + e3t ?
2 3 6
On the one hand, since M (t) is equivalent to

x 1 2 3
1 1 1
f (x) 2 3 6

 2  2  2
X
2 5 1 5 1 5 1
V ar(X) = (x − µ) f (x) = 1 − · + 2− · + 3− · =
x
3 2 3 3 3 6
3 4 5 6
(i) 9
(ii) 9
(iii) 9
(iv) 9

On the other hand, since

1 1 1
M (t) = et + e2t + e3t ,
2 3 6
the second derivative of M (t) with respect to t is
 
00 d 1 t 1 2t 1 3t 1 1 1
M (t) = e +2· e + 3 · e = et + 4 · e2t + 9 · e3t
dt 2 3 6 2 3 6

and so evaluating this derivative at t = 0,

1 1 1
M 00 (0) = e0 + 22 · e2(0) + 32 · e3(0) = E(X 2 ) =
2 3 6
3 4 6 10
(i) 3
(ii) 3
(iii) 3
(iv) 3

so  2
102 2 5
V ar(X) = E(X ) − µ = − =
3 3
3 4 5 6
(i) 9
(ii) 9
(iii) 9
(iv) 9

which (i) the same (ii) different from before.

5. Binomial mgf. With some effort, it can be shown mgf for binomial is
n  
tX
X
tx n
n
M (t) = E[e ] = e px q n−x = pet + q .
x
x=0
Section 6. The Moment-Generating Function (LECTURE NOTES 4) 71

(a) Determine E(X) using M (t).

n
0 d (pet + q) h
n−1 t i
n−1
E(X) = M (0) = = n pet + q pe = n pe0 + q pe0 =
dt t=0 t=0

(i) λ (ii) np (iii) 2np (iv) npq.

(b) Determine E (X 2 ).
n
2 d2 (pet + q)
00
h
t

n−1 t 2

t

n−1 t i
E(X ) = M (0) = = n(n − 1) pe + q pe + n pe + q pe
dt2 t=0 t=0
0

n−1 0 2

0

n−1 0
= n(n − 1) pe + q pe + n pe + q pe =

(i) np(n − 1) (ii) np2 (n − 1)2 + np (iii) np2 (n − 1) + np.

(c) Determine V ar (X).

V ar (X) = E X 2 − E (X)2 = (np2 (n − 1) + np) − (np)2 = np(1 − p) =

(i) n (ii) np (iii) 2np (iv) npq.

6. Identify binomial pmf with mgf. What is the pmf of random variable X with

11
M (t) = 0.3et + 0.7 ?

Since
11
(pet + q)n = 0.3et + 0.7 ,
where p = 0.3, q = 0.7 and n = 11, this is a binomial distribution b(n, p) =

(i) b(11, 0.3) (ii) b(0.3, 11) (iii) b(11, 0.7) (iv) b(0.7, 11).

7. Identify geometric pmf with mgf.

(a) What is the pmf of random variable X with

0.3et
M (t) = ?
1 − 0.7et
Since, from the table above,

pet 0.3et
= ,
1 − qet 1 − 0.7et
this is a geometric distribution where

(i) p = t, q = 0.3 (ii) p = 0.7, q = 0.3 (iii) p = 0.3, q = 0.7

72 Chapter 2. Discrete Random Variables (LECTURE NOTES 4)

(b) Expected value. From table above,

1 1
E(X) = µ = = =
p 0.3
3
(i) 3
(ii) 34 (iii) 63 (iv) 10 3
library(MASS) # call up library(MASS)
fractions(1/0.3) # fractional form of 1/0.3
[1] 10/3

(c) Variance. From table above,

q 0.7
V ar(X) = σ 2 = 2
= =
p 0.32
70
(i) 9
(ii) 69
9
(iii) 68
9
(iv) 67
9
fractions(0.7/0.3^2) # fractional form of 0.7/0.3^2
[1] 70/9

8. Deriving mgf from function of X. If mgf of X is

1 1 1
MX (t) = et + e2t + e3t
2 3 6
then mgf of Y = aX + b = 3X − 2 is
 
bt −2t −2t 1 3t 1 6t 1 9t
MY (t) = e MX (at) = e MX (3t) = e e + e + e =
2 3 6
(i) 12 et + 31 e4t + 16 e5t (ii) 12 et + 13 e2t + 16 e3t (iii) 21 et + 13 e4t + 61 e7t
9. Deriving mgf from function of binomial X. If mgf of X is b(11, 0.3),

n
11
M (t) = pet + q = 0.3et + 0.7 ,
then mgf of Y = 11 − X or Y = aX + b = −X + 11 is

11 
11  11
MY (t) = ebt MX (at) = e11t MX (−t) = e11t 0.3e−t + 0.7 = et 0.3e−t + 0.7 = 0.3 + 0.7et
which is (i) b(11, 0.3) (ii) b(0.3, 11) (iii) b(11, 0.7) (iv) b(0.7, 11)
10. Binomial pgf. With some effort, it can be shown pgf for binomial is
n  
X n
X
P (t) = E[t ] = t x
px q n−x = (q + pt)n .
x
x=0

So to determine E(X) using P (t),

d (q + pt)n

0
= n (q + pt)n−1 p t=1 = n (q + p(1))n−1 p =
 
E(X) = P (1) =
dt t=1

(i) λ (ii) np (iii) 2np (iv) npq. because p + q = p + (1 − p) = 1.