CE_08A_Solutions: Hypothesis Testing

For questions 1 and 2: To determine the reliability of experts used in interpreting the
results of polygraph examinations in criminal investigations, 280 cases were studied. The
results were:

True Status
Innocent Guilty Total
Innocent 131 15
Examiner’s Guilty 9 125
Decision Total

If the hypotheses were H 0 : suspect is innocent vs H 1 : suspect is guilty, then

1. What is the probability of making a type II error?

Probability of committing Type II error is equal to P(Non-Rejecting Ho / Ho is false). So we have to

calculate P(Non-Rejecting Ho / Ho is false)

=P (Examiner decides that suspect is innocent / Suspect is actually guilty)

15
P( A∩Y ) 280 15
P( A / Y )= = =
P( Y ) 140 140
= 280

2. What is the probability of making a type I error?

Probability of committing Type I error is equal to P(Rejecting Ho / Ho is true). So we have to

calculate P(Rejecting Ho / Ho is true)

=P (Examiner decides that suspect is guilty / Suspect is actually innocent)

9
P ( B∩ X ) 280 9
P ( B / X )= = =
P( X ) 140 140
= 280

3. In a statistical test of hypothesis, what happens to the rejection region when alpha, the level of
significance, is reduced?

If the level of significance decreases, then obviously the size of the rejection region decreases too.

For the below questions (4, 5 & 6), use hypothesis testing by critical value approach to make the statistical
decision and managerial conclusion.
4. A drug company produces a certain type of antibiotics in 100 mg capsules. A researcher wants to test
if the machines are filling the capsules with the right amount of medicine, not more or less. She
takes a sample of 64 and finds the mean is 96 mg. Past data shows that the population standard
deviation is 12 mg. ( α=0 . 05 )

ANSWER

1. Check for CLT:

As n = 64 > 30, hence CLT is applicable and thus sampling distribution of mean becomes
normally distributed.

2. State the null hypothesis, H0 and the alternative hypothesis, H1

H 0 ;μ=100mg ¿}¿ ¿¿
Two-tailed test

3. It is a two-tailed test

4. Determine which table to use:

Population SD is known, use z-table

5. Determine the critical values that divide the rejection and non-rejection regions

For α = 0.05 and two-tail test, zcv = ±1.96

6. Compute test statistic

X̄−μ 96−100 −4
z= = = =−2 . 67
σ 12 1. 5
√n √64
7. Make the statistical decision

z=-2.67 < -1.96, the test statistic falls into the rejection region, reject the null hypothesis.

8. Make a managerial conclusion in the context of the real-world problem:

As H0 is rejected, hence with 95% level of confidence it can be concluded that the machines
are NOT filling the capsules with the right amount of medicine. The company should invest
more money to repair the machines or to renew them.

9. Determine the type of error:

As H0 is rejected, hence type I error might have been committed. P(Type I) = 5%.
In this context the error means:
It is concluded that the machines are NOT filling the capsules with the right amount of
medicine but actually they might be filling the capsules with the right amount of medicine.
5. The mean life of a large shipment of light bulbs at a light bulb factory is 360 hours. The quality
control manager hopes that with the arrival of high quality raw materials, the mean life of the final
product might have increased significantly. The population standard deviation is known to be 90
hours. A random sample of 25 bulbs indicates a sample mean life of 380 hours. ( α=0 . 01 )

1. Check for CLT:

As n = 25 < 30, hence CLT is not applicable and hence population is assumed to be normally
distributed and thus sampling distribution of mean becomes normally distributed.

2. State the null hypothesis, H0 and the alternative hypothesis, H1

H o :μ≤360hr ¿} ¿ ¿¿
Upper tail test

3. It is an upper tail test

4. Determine which table to use:

Population SD is known, use z-table

5. Determine the critical values that divide the rejection and non-rejection regions

For α = 0.01, upper tail test, zcv = 2.33

6. Compute test statistic

X̄−μ 380−360 20
z= = = =1. 11
σ 90 18
√n √ 25
7. Make the statistical decision

z = 1.11 < 2.33, the test statistic falls into the non-rejection region, do not reject the null
hypothesis H0.

8. Make a managerial conclusion in the context of the real-world problem:

As H0 is not rejected, hence with 99% level of confidence it can be concluded that arrival of
high quality raw materials might NOT have increased the mean life of the final product
significantly.

9. Determine the type of error:

As H0 is not rejected, hence type II error might have been committed.

In this context the error means:
It is concluded that arrival of high quality raw materials might NOT have increased the mean
life of the final product significantly but actually arrival of high quality raw materials might
have increased the mean life of the final product significantly.

6. A tire manufacturer claims that its tires will last atleast an average of 50,000 km before they need to
be replaced. A consumer group wishes to challenge this claim and have a sample of 49 tires to
 measure their lifetime. They found the mean value of this sample is 48,000 km. Population standard
deviation is known as 7000 km. ( α=0 .10 )

1. Check for CLT:

As n = 49 > 30, hence CLT is applicable and thus sampling distribution of mean becomes
normally distributed.

2. State the null hypothesis, H0 and the alternative hypothesis, H1

H o :μ≥50,000km ¿} ¿ ¿¿
Lower tail test

3. It is a lower tail test

4. Determine which table to use:

Population SD is known, use z-table

5. Determine the critical values that divide the rejection and non-rejection regions

 = 0.10 lower tail test: zcv = -1.28

6. Compute test statistic

X̄−μ 48 , 000−50 , 000 −2000
z= = = =−2
σ 7000 1000
√n √ 49
7. Make the statistical decision

As z-statistic falls in the rejection region (z = -2 < -1.28), hence H0 is rejected.

8. Make a managerial conclusion in the context of the real-world problem:

As H0 is rejected, hence with 90% level of confidence it can be concluded that the tires will
NOT last atleast an average of 50,000 km before they need to be replaced.

9. Determine the type of error:

As H0 is rejected, hence type I error might have been committed in this case.P(Type I) = 10%.
In this context the error means:
It is concluded that that the tires will NOT last atleast an average of 50,000 km before they
need to be replaced but actually the tires might last atleast an average of 50,000 km before
they need to be replaced

For the above questions (4, 5 & 6), calculate the p-value and make a statistical decision.
4. p-value = P(having a sample mean which is lower than 96 mg or higher than 104 mg given than the
population mean is 100 mg)

96−100 104−100
p−value=P( X̄ ≤96 or X̄ ≥104 /μ=100)=P z≤
( 12
√ 64
or z≥
12
√ 64 )
¿ P( z≤−2 .67 or z≥2 . 67)=0 .0076
p-value is lower than the significance level (0.0076 < 0.05), therefore the null hypothesis is rejected.

5. p-value = P(having a sample mean which is higher than 380 hours given than the population mean is
360 hours)

380−360
p−value=P( X̄ ≥380/ μ=360)=P z≥
( 90
√ 25 )
=P( z≥1. 11)=0 . 1335

p-value = 0.1335 and this is higher than the significance level of 0.01. Therefore, we do not reject the
null hypothesis.

6. p-value = P(having a sample mean which is lower than 48,000km given than the population mean is
50,000 km)

48000−50000
P−value=P( X̄ ≤48000 /μ=50000)=P z≤
( 7000
√ 49 )
=P( z≤−2)=0. 0228

p-value is lower than the significance level (0.0228 < 0.1), therefore the null hypothesis is rejected.