F71SM STATISTICAL METHODS RJG

TUTORIAL ON §4 SPECIAL DISTRIBUTIONS

1. Let X ~ b(n, p) and Y ~ b(n, 1 − p). Note that P(X ≥ k) = P(Y ≤ n − k).
Hence, using NCST, find the probabilities that in a series of 20 independent, identical Bernoulli
trials with P(success) = 0.6, we obtain (a) 12 or more successes, and (b) 12 or fewer successes.

2. Let X ~ geo(p). Show that, for x = 0, 1, 2, …, and k = 0, 1, 2, … ,

P(X ≥ x + k | X ≥ k) = P(X ≥ x) , and comment in the context of waiting times.

3. Let X ~ P(λ). Verify the following recursion formula

λ
P ( X = x + 1) = P ( X = x ) , x = 0, 1, 2, …
x +1
Use this result to find the mode of the distribution (the mode in this context is the value x for
which P(X = x) ≥ P(X = x + 1) and P(X = x) ≥ P(X = x − 1)).
X −λ
4. Let X ~ P(λ) and let Z be the standardised version, where Z = . Show that the mgf of Z
λ
(
satisfies log M Z ( t ) = − λ t + λ et / λ
)
− 1 and hence show that MZ(t) → exp(t2/2) as λ → ∞.
X −λ
What does this result imply about the limiting distribution of Z = ?
λ
What does it imply about the asymptotic distribution of X ~ P(λ)?

5. Let X ~ b(n, p).

(i) Let λ = np and express f(x) = P(X = x) in terms of λ and n.
λx
Show that f(x) → e − λ as n → ∞ (thus justifying using the Poisson(np) distribution as an
x!
approximation to the binomial(n, p) distribution for large n (and small p)).
n
 x
Note: 1 +  → e x as n → ∞
 n
X − np
(ii) Let Z be the standardised version of X, so Z = . Show that the mgf of Z satisfies
npq
np   t  
log M Z ( t ) = − t + n log  p exp   + q  and hence show that MZ(t) → exp(t /2) as
2

q 
  npq  
n → ∞.
X − np
What does this result imply about the limiting distribution of Z = ?
npq
What does it imply about the asymptotic distribution of X ~ b(n, p)?

6. Let X ~ exp(λ). Show that P(X > x + k | X > k) = P(X > x) and comment in the context of waiting
times. (This illustrates an important property called the memoryless property − and cf Qu2).

1
7. Let X ~ Gamma(α, λ).
1 Γ (r +α )
(a) Show by integration that E  X r  = r , r ≥ 0 and hence specify E[X] and V[X].
λ Γ (α )
−α
 t
(b) Show that the moment generating function of X is given by M ( t ) = 1 −  for t < λ, and hence
 λ
confirm the values of E[X] and V[X] found in part (a).
(c) In the case α = n/2 , λ = 1/2, where n is a positive integer, X has a chi-square distribution with n

degrees of freedom, written X ~ χ n2 .

(i) State the mean and variance of X ~ χ n2 .

(ii) Suppose X ~ N(0,1) and let Y = X 2 . By expressing the distribution function of Y in terms of that
of X, and differentiating, show that Y ~ χ12 .

(d) In the case that 2α is a positive integer, show that X ~ G(α, λ) ⇒ 2λX ~ χ 22α .

8. Skewness
E ( X − µ )  = E  X 3  − 3µ E  X 2  + 2 µ 3 .
3
Firstly recall, or show, that
 
(i) Let X ~ b(n, p). Show that E[(X − µ)3] = np(1 − p)(1 − 2p) and hence show that the coefficient of
q− p
skewness of the b(n, p) distribution is given by γ 1 = .
npq
(ii) Show that the coefficient of skewness of the P(λ) distribution is given by γ 1 = 1/ λ .

(iii) Show that the coefficient of skewness of the G(α, λ) distribution is given by γ 1 = 2 / α .

Hence show that

(a) the coefficient of skewness of the exp(λ) distribution is γ 1 = 2 , and

(b) the coefficient of skewness of the χ n2 distribution is γ 1 = 2 2 / n , and comment on the

symmetry of χ n2 as n increases.

9. Let X ~ N(µ, σ2).

(a) Show, using the mgf of Z = (X − µ)/σ , that the third, fourth, and fifth central moments of X are given
by E[(X − µ)3] = 0, E[(X − µ)4] = 3σ 4 , and E[(X − µ)5] = 0.
(b) Find an expression for E[(X − µ)6].

There follows a series of 11 further questions involving applications of the distributions introduced in this
chapter. The new facts in the notes at the end of Question 19 on linear combinations of r.vs. will be considered
in §5.

2
10. A plant, whose constitution is known to be either A or B, is tested by raising n offspring from its seeds. If
the constitution is A then each offspring has probability 1/4 of being white and 3/4 of being red; if B then
each offspring has an equal probability of 1/2 of being white or red.
(i) Using a binomial model, write down an expression for the probability pA(r, n) that exactly r of n
offspring of an A plant are white, and similarly pB(r, n) for the offspring of a B plant.
If three offspring are raised from an A plant and three are raised from a B plant, find the probability that
the plants produce the same number of white offspring.
(ii) When a plant is tested, the number of white offspring, r, is recorded, and it is assigned to class A if
pA(r, n) > pB(r, n) and to class B if pA(r, n) < pB(r, n). Show that, if 20 offspring are raised from a plant
it will be classified as A if it produces 7 or fewer white offspring and B otherwise.
Hence find the probability that an A plant will be misclassified as a B plant.

11. At a fairground stall the probability that Annie hits the target with a single rifle shot is 0.4. How many
shots must she have to ensure a probability of at least 0.98 that she hits the target at least once?

12. Edward is a learner driver whose successive attempts to pass the driving test can be regarded as
independent trials, each with probability 0.6 of success (he is not an effective learner …).
(a) Find the probability that Edward needs more than 3 attempts to pass the test.
k
(b) Edward’s dad is a generous sort. He offers Edward a present (in units of £1000) of 0.9 , where k
(= 1, 2, 3, …) is the number of attempts he requires to pass the test. Find the expected value of the
present.
(c) Find the probability that Edward needs more than 4 further attempts to pass, given that he has had 2
unsuccessful attempts already.

13. The records of an airline company reveal that 4% of passengers with reservations on flights from London
to New York fail to turn up for their flight. To increase income the airline accepts 310 reservations for such
a flight to be operated using an aircraft which has only 300 seats.
Find the probability that the airline will be unable to allocate a seat to each passenger with a reservation
who turns up for the flight and so will have to “bump” at least one passenger off.

14. The cells in a sample of blood are examined through a microscope and counted. The probability that any
cell is a white cell is p. Given that the second white cell to be observed was the fifth cell to be counted, find
the probability distribution of the number of cells that had to be counted to observe the first white cell,
and comment on the result.

15. Two children (easily amused children) play a lengthy and tedious game as follows: each child throws a pair
of fair six−sided dice 180 times, and at each throw the player scores a point if he/she throws a “double
one” or a “double six”. Calculate (to 3 dp) the probability that the game is drawn, that is, both children
score exactly the same number of points.
3
4
5