You are on page 1of 9

200032 STATISTICS FOR BUSINESS

Question 1
The Bank of Canberra is interested in estimating the difference between the mean
credit card balances at two of its branches. Independent samples of credit card
customers provide the following results:

Barton Branch: Size n = 12,


Mean of 𝑥̅ = $250,
Standard deviation of s = $10;

Deakin Branch: Size n = 16,


Mean of 𝑥̅ = $175,
Standard deviation of s = $9.

Find a 95% confidence interval for the difference between mean credit card
balances.

You can assume that the credit card balances are normally distributed with equal
population variances.
2

1
Question 1

Let Population 1 = Barton Branch and


Population 2 = Deakin Branch.

From the question, we know n1 = 12 n2 = 16


x̅1 = 250 x̅2 = 175
s1 = 10 s2 = 9

Assuming equal variances and normal populations, the confidence interval here for the
difference in means (independent sample) is

1 1
𝑥̅ − 𝑥̅ ± 𝑡 × 𝑠 × +
𝑛 𝑛

( n1  1)  s12  (n2  1)  s22 11  10²  15  9²


sp    9.436
n1  n2  2 26

Question 1

α = 0.05 d.f. = n1 + n2 – 2 = 12 + 16 – 2 = 26 t0.025, 26 = 2.056

1 1
𝑥̅ − 𝑥̅ ± 𝑡 × 𝑠 × +
𝑛 𝑛

1 1
250 − 175 ± 2.056 × 9.436 × +
12 16

75 ± 7.41

(75 – 7.41 = 67.59, 75 + 7.41 = 82.41)

The 95% confidence interval for the difference between mean credit card balances
at the two branches, Barton and Deakin, is between $67.59 and $82.41.
4

2
Question 2

A survey firm conducts door-to-door surveys on a variety of issues. Some


individuals cooperate with the interviewer and complete the
questionnaire, while others do not.

The following sample information shows the response data for men and
women. Out of 250 men, 150 cooperated with the survey; while out of 300
women, 210 cooperated.

Construct a 99% confidence interval for the difference between the true
proportions of each gender who cooperate with the interviewer and
complete the questionnaire.

Question 2

Let Population 1 = Men and


Population 2 = Women.

From the question, we know n1 = 250 n2 = 300


x1 = 150 x2 = 210

We want to estimate the difference in proportions; therefore we need the sample proportions!

𝑝̂ = = = 0.6

𝑝̂ = = = 0.7

3
Question 2

Here α = 0.01. Since our samples are large, we have z0.005 = 2.575.

𝑝 − 𝑝 ±𝑧 +

(0.6 × 0.4) (0.7 × 0.3)


0.6 − 0.7 ± 2.575 × +
250 300

− 0.1 ± 0.105

− 0.205 < 𝑝 − 𝑝 < 0.005

That is, we are 99% confident that the difference in the proportions of men and women who
cooperate with the interviewer and complete the questionnaire is between –20.5% and 0.5%.

Question 3

An air-conditioning supplier uses two installers. He is interested in comparing the time


each installer takes to install an air-conditioning unit. Data concerning installation times
was collected over a period of time, and random samples from the two installers’ times
yielded the following results. (Note: the installation times are in hours).

Installer A 4 2 3 5 2 2 4 2

Installer B 4 6 5 3 2 5 3

Is there evidence, at the 1% level of significance, of a difference in the average time of


installation between the two installers? (You may assume that all installation times are
normally distributed, and Installer A and Installer B have equal population variances).

4
Question 3
Let Population 1 = Installer A Is there evidence of a difference in
and Population 2 = Installer B. the average time of installation
between the two installers?
1. H0: μ1 = μ2 H0: μ1 – μ2 = 0
HA: μ1 ≠ μ2 HA: μ1 – μ2 ≠ 0

2. Since the populations standard deviations are unknown and assumed equal, and we
can assume that all installation times are normally distributed:

𝑥 − 𝑥 − (𝜇 − 𝜇 )
𝑡=
1 1
𝑠
𝑛 + 𝑛

3. Here α = 0.01; and df = n1 + n2 – 2 = 8 + 7 – 2 = 13. Hence

t0.005, 13 df = 3.012.
9

Question 3
4. Reject H0 if t > 3.012 or t < –3.012.

5. From the data, we need to calculate the following: n1 = 8 n2 = 7


𝑥̅ = 3 𝑥̅ = 4
s1 = 1.195 s2 = 1.414

× × × . ² × . ²
𝑠 = = = 1.301

( ) ( )
𝑡=  
= = −1.485
.

6. Since –1.485 is between –3.012 and 3.012, we cannot reject H0.


We cannot conclude that there is evidence, at the 1% level of significance, of a
difference in the average time of installation between the two installers.

10

5
Question 3 (Using Excel)

3. Here α = 0.01.

4. Reject H0 if p-value < 0.01.

5. Enter the data:

11

Question 3

Now in the Data Analysis window, we select


t-Test: Two-Sample Assuming Equal Variances

• We select Installer A as ‘Variable 1’ and


Installer B as ‘Variable 2’, and specify the
appropriate ranges.

• The ‘Hypothesised Mean Difference’ should be set to 0.

• Since our data have Labels in Row 1,


tick ‘Labels”.

• Change Alpha to 0.01.

• Finally, specify the location of the


output.

12

6
Question 3

t-Test: Two-Sample Assuming Equal Variances

Installer A Installer B
Mean 3 4
Variance 1.428571429 2
Observations 8 7
Pooled Variance 1.692307692
Hypothesized Mean Difference 0
df 13
t Stat -1.485281322
P(T<=t) one-tail 0.080655625
t Critical one-tail 2.650308838
P(T<=t) two-tail 0.16131125
t Critical two-tail 3.012275839

6. Since 0.1613 > 0.01, we cannot reject H0.


We cannot conclude that there is evidence, at the 1% level of
significance, of a difference in the average time of installation
between the two installers.
13

Question 4

A company attempts to evaluate the potential for a new bonus plan


by selecting a random sample of five salespersons to use the bonus
plan for a trial period. Weekly sales volumes before and after
implementing the bonus plan are shown below.

Salesperson 1 2 3 4 5
Before 15 12 18 15 16
After 18 14 19 18 18

Using a significance level of α = 0.05, test to see if the bonus plan


will result in an increase in the mean weekly sales.
14

7
Question 4
Since we are taking 2 measurements from the same salesperson, the data is paired!

Let d = Before – After.

Looking at the question posed to determine the HA:

… bonus plan will result in an increase in the mean weekly sales …

Thus we want to see if “AFTER” is greater than “BEFORE”; hence the difference should be
negative.

Finally, it talks about the mean weekly sales; therefore our alternative hypothesis will be

HA: µd < 0.

Again, here d = Before – After.

15

Question 4
1. H 0 : µd = 0
H A: µ d < 0

2. Assume the differences are normally distributed; and since the population
standard deviation is unknown,

𝑑̅ − 𝜇
𝑡= 𝑠
 
𝑛
3. Here α = 0.05; and since the test is one tailed with n – 1 = 5 – 1 = 4 df, we have

t0.05, 4df = 2.132.

16

8
Question 4

4. Reject H0 if t < –2.132.

5. Here d = Before – After. d (–3, –2, –1, –3, –2)

n=5 𝑑̅ = −2.2 𝑠 = 0.8367

.
𝑡= = . = −5.879
   

6. Since –5.879 < –1.645, we can reject H0.

We can conclude that the bonus plan will result in an increase in the mean weekly
sales.
17

You might also like