EXERCISES IN STATISTICS

Series A, No. 5
1. Find the moment generating function of x f (x) = 1, where 0 < x < 1,
1
.
and thereby confirm that E(x) = 12 and V (x) = 12
Answer: The moment generating function is
Z
xt
M (x, t) = E(e ) =

=
0

ext dx

0
t

xt 1

e
t

1
e
.
t
t

But

t
t2
t3
t0
+ + + + ,
e =
0! 1! 2! 3!
so

1
t
t2
t3
1
M (x, t) =
+ 1 + + + +
t
2! 3! 4!
t
2
3
t
t
t
+ .
=1+ + +
2
6
24
By the process of differentiating M (x, t) with respect to t and the setting t = 0,
we get

M (x, t)
1
1 2t 3t2
E(x) =
+
+
+

,
=
=

t
2
3!
4!
2
t=0
t=0

6t
2 M (x, t)
1
2
2
E(x ) =
+
+
=
= .

2
t
3!
4!
3
t=0
t=0
t

Combining these results gives

2
1
1 1
V (x) = E(x2 ) E(x) = =
.
3 4
12
2. Find the moment generating function of x f (x) = aeax ; x 0.
Answer: The moment generating function is
Z
Z
xt
xt ax
e ae
dx =
M (x, t) = E(e ) =

x(at)

ae
at

aex(at) dx

1
a
=
.
at
1 t/a

SERIES A No.5 : ANSWERS

3. Prove that x f (x) = xex ; x 0 has a moment generating function of
1/(1 t)2 . Hint: Use the change of variable technique to integrate with
respect to w = x(1 t) instead of x.
Answer: The moment generating function is
Z
xt

M (x, t) = E(e ) =

xt

e xe

dx =

xex(1t) dx.

Define w = x(1 t). Then

x=

w
1t

and

dx
1
=
.
dw
1t

The change of variable technique indicates that

Z

Z
g(x)dx =

g{x(w)}

dx
,
dw

where g(x) = xex(1t) . Thus we find that

Z

w w 1
e
dw
1t
1t
0
Z
1
1
=
wew dw =
.
2
(1 t) 0
(1 t)2

M (x, t) =

Here the value of the final integral is unity, since the expression wew , which
is to be found under the integral sign, has the same form as the p.d.f. of x.
To demonstrate directly that the value is unity, we can use the technique
of integrating by parts. The formula is
Z

dv
u dx = uv
dx

Z
v

du
dx.
dv

Within the expression wew , we take w = u and ew = dv/dw. Then we get

Z

we

dw =

wew 0

Z
=
0

Z
+

e
0

dw

ew dw = ew 0 = 1.

SERIES A No.5 : ANSWERS

4. Using the theorem that the moment generating function of a sum of independent variables is the product of their individual moment generating
functions, find the m.g.f. of x1 ex1 and x2 ex2 when x1 , x2 0 are
independent. Can you identify the p.d.f. of f (x1 + x2 ) from this m.g.f.?
Answer: If x1 and x2 are independent, then their joint p.d.f. can be written
as f (x1 , x2 ) = f (x1 )f (x2 ); and it follows that
Z

e(x1 +x2 )t f (x1 , x2 )dx1 dx2 )

Zx2 x1
Z
x1
=
e f (x1 )dx1
ex2 f (x2 )dx2 = M (x1 , t)M (x2 , t),

M (x1 + x2 , t) =

x1

x2

or simply that M (x1 + x2 , t) = M (x1 , t)M (x2 , t). If x1 , x2 are independent

with x1 ex1 and x2 ex2 , then
M (x1 , t) = M (x2 , t) =

1
1t

and M (x1 + x2 , t) =

1
.
(1 t)2

But, according to the answer to question (3), this implies that

f (x1 + x2 ) = (x1 + x2 )e(x1 +x2 ) .

5. Find the moment generating function of the point binomial

f (x) = px (1 p)1x
where x = 0, 1. What is the relationship between this and the m.g.f. of
the binomial distribution ?
Answer: If f (x) = px (1 p)1x with x = 0, 1, then
M (x, t) =

ext f (x) = e0 p0 (1 p) + et p(1 p)0

x=0,1

= (1 p) + pet = q + pet .
Pn
But the binomial outcome z =
i=1 xi is the sum of n inependent pointbinomial outcomes; so it follows that the binomial m.g.f. is
M (z, t) = (q + pet )n .