You are on page 1of 2

Probability and Statistics for Data Analytics

Coursework 8
We will discuss some parts of this coursework during the next tutorial.

1. For X1 , . . . , Xn1 ∼ Bernoulli(p1 ) and Y1 , . . . , Yn2 ∼ Bernoulli(p2 ) i.i.d. Assume that the
X’s and Y ’s are independent. Suppose we would like to test
H0 : p1 = p2 versus H1 : p1 6= p2
at significance level α. Let us write p = p1 = p2 under H0 . Show that the following rejection
rule
|X − Y |
r   > zα/2 ,
p(1 − p) n11 + 1
n2

has a Type I error that is approximately α when n1 , n2 → ∞.

2. A survey on climate change was conducted by QMUL scientists, with 1000 men and
950 women as participants. They were asked about their awareness regarding the effects
of climate change. During the study, 390 women and 450 men said they are aware of the
effects.
a) Test the hypothesis at the 5% level that the same proportion of men and women are
aware of climate change effects.
b) Find the 95% Wald confidence interval for the difference in population proportions
who are aware of climate change effects.

3. Suppose we have data X1 , . . . , Xn ∼ Bernoulli(p) i.i.d. We know that by The Central


Limit Theorem,
 
X −p
P −zα/2 ≤ q ≤ zα/2  ≈ 1 − α.
p(1−p)
n

Find a way to rewrite


X −p
−zα/2 ≤ q ≤ zα/2
p(1−p)
n

into the form


L(X) ≤ p ≤ U (X),

1
so that we can use [L(X), U (X)] as an approximate 100(1 − α)% confidence interval for p.
Here L(X) and U (X) do not depend on p and they are functions of X. Try to do this
without using substitutions like in the Wald interval.

Hint: Rewrite
X −p
−zα/2 ≤ q ≤ zα/2
p(1−p)
n

as
|X − p|
q ≤ zα/2 .
p(1−p)
n

Square both sides

(X − p)2 2
p(1−p)
≤ zα/2
n

and solve for p.

4. In January 2002 the Euro was introduced and soon after stories started to circulate that
some of the Euro coins would not be fair coins, because the “national side” of some coins
would be too heavy or too light. Suppose a French 1 Euro coin was tossed 250 times: 140
heads and 110 tails. Using the new CI derived in Question 3, construct a 95% confidence
interval for the probability of getting heads with this coin.

You might also like