Statistical Tests for Means

STAT 400
July 27, 2016

Example 1 The overall standard deviation of the diameters of the ball bearings is = 0.005mm.
The overall mean diameter of the ball bearings must be 4.300 mm. A sample of 81 ball bearings
had a sample mean diameter of 4.299 mm. Is there a reason to believe that the actual overall mean
diameter of the ball bearings is not 4.300 mm?
a. What hypothesis are we interested in testing?
H0 : 0 = 4.3 vs. H1 : 0 6= 4.3
b. What is the value of the appropriate test statistic?
In this case, we know , so we can use the standard normal distribution or Z-test,
0
X
4.299 4.3
= 1.80
(1)
Z=
=

0.005

81

c. What is the rejection region for = 0.10?

This is a two-tail test with Z/2 = 1.6449, so we reject H0 if, |Z| > Z0.05 = 1.6449.
d. Perform an appropriate test at a 5% level of significance ( = 0.05). What do you conclude?
We compare our test statistic to the rejection region and in this case |Z| > Z0.025 = 1.96,
so we fail reject H0 because 1.80 > 1.96. We could also compute the p-value as,
p value = P (Z 1.80) + P (Z 1.80) = 0.0359 + 0.0359 = 0.0718
Clearly, 0.0718 > 0.05, so we fail to reject the null.
1

(2)

STAT 400

Tests for

Example 2 A trucking firm believes that its mean weekly loss due to damaged shipments is at
most $1800. Half a year (26 weeks) of operation shows a sample mean weekly loss of $1921.54 with
a sample standard deviation of $249.39.

a. What hypothesis are we interested in testing?

H0 : 0 1800 vs. H1 : 0 > 1800.

b. What is the value of the appropriate test statistic?

In this case, we do not know , so we must use students t-distribution or t-test,

0
X
1921.54 1800
T =
= 2.485
=
s
249.39

(3)

26

c. What is the rejection region for = 0.10?

This is a right-tail test with t,25 = 1.3163, so we reject H0 if, t > t0.10,25 = 1.3163

d. Perform an appropriate test at a 10% level of significance ( = 0.10). What do you conclude?

We compare our test statistic to the rejection region and in this case t > t0.10,25 = 1.3163,
so we reject H0 because 2.485 > 1.3163. Alternatively, we could have computed the
p-value as 0.01 with a computer or calculator.

Stepanov

STAT 400

Tests for

Example 3 Metaltech Industries manufactures carbide drill tips used in drilling oil wells. The
life of a carbide drill tip is measured by how many feet can be drilled before the tip wears out.
Metaltech claims that under typical drilling conditions, the life of a carbide tip follows a normal
distribution with mean of at least 32 feet. Suppose some customers disagree with Metaltechs
claims and argue that Metaltech is overstating the mean (i.e. the mean is actually less than 32).
Metaltech agrees to examine a random sample of 25 carbide tips to test its claim against the
customers claim. If the Metaltechs claim is rejected, Metaltech has agreed to give customers a
price rebate on past purchases. Suppose Metaltech decided to use a 5% level of significance and
the observed sample mean is 30.5 feet with the sample variance 16 feet sq.

a. What hypothesis are we interested in testing?

H0 : 0 32 vs. H1 : 0 < 32.

b. What is the value of the appropriate test statistic?

In this case, we do not know , so we must use students t-distribution or t-test,

0
X
30.5 32
T =
=
= 1.875
s
4

25

(4)

c. What is the rejection region for = 0.05?

This is a left-tail test with t,24 = 1.7109, so we reject H0 if t < t0.05,24 = 1.7109.

d. Perform an appropriate test at a 5% level of significance ( = 0.05). What do you conclude?

We compare our test statistic to the rejection region and in this case t < t0.05,24 =
1.7109, so we reject H0 . Alternatively, we could have computed the p-value with R
and compared it to 0.05.

Stepanov

STAT 400

Tests for

Definition:
H0 true

H0 NOT true

Do Not Reject H0

Correct

Type II Error

Reject H0

Type I Error

Correct

= P (Type I Error) = P (Reject H0 |H0 true)

= P (Type II Error) = P (Do Not Reject H0 |H0 is not true)
Power = 1 = P (Reject H0 |H0 is not true)

Example 4 A car manufacturer claims that, when driven at a speed of 50 miles per hour on a
highway, the mileage of a certain model follows a normal distribution with mean 0 = 30 miles
per gallon and standard deviation = 4 miles per gallon. A consumer advocate thinks that the
manufacturer is overestimating average mileage. The advocate decides to test the null hypothesis
H0 : = 30 against the alternative hypothesis H1 : < 30.

a. Suppose the consumer advocate tests a sample of n = 25 cars. What is the significance level
associated with the rejection region Reject H0 if x < 28.6?


< 28.6| = 30) = P Z <

= P (X

28.630

4
25

= P (Z < 1.75) = 0.0401.

b. Suppose the consumer advocate tests a sample of n = 25 cars. Find the rejection region in
with the significance level = 0.05.
terms of X

Reject H0 if,
Z=
Z=

0
X

30
X
4
25

< z
< z

< 30 1.645 4 = 28.684

X
25
4

(5)
Stepanov

STAT 400

Tests for

c. Suppose the consumer advocate tests a sample of n = 25 cars and uses a 5% level of significance.
Find the power of the test if the true mean is 1 = 29.5.
< 28.684|1 = 29.5)
P (Reject H0 |H0 is not true) = P (X
28.684 29.5
=P Z<
4/5

= P (Z < 1.02) = 0.1539

(6)

d. Repeat part (c) for the case when the true value of the mean is 1 = 28.5.
< 28.684|1 = 29.5)
P (Reject H0 |H0 is not true) = P (X
28.684 28.5
=P Z<
4/5

= P (Z < 0.23) = 0.591

(7)

e. Repeat parts (b) - (d) using a 10% level of significance.

Reject H0 if,
Z=

30
X
4
25

< z0.10

< 30 1.2816 4 = 28.9744

X
25

(8)

Power if the true mean is 1 = 29.5 is,

< 28.9744|1 = 29.5) = P Z < 28.9744 29.5
P (X
4/5

= P (Z < 0.657) = 0.2556

Power if the true mean is 1 = 28.5 is,

< 28.9744|1 = 28.5) = P Z < 28.9744 28.5
P (X
4/5

= P (Z < 0.593) = 0.7234

Stepanov

STAT 400

Tests for

f. Repeat parts (b) - (d) using a larger sample size of n = 49.

Reject H0 if,
Z=

30
X
4
49

< z0.05

< 30 1.6449 4 = 29.06

X
49

(9)

Power if the true mean is 1 = 29.5 is,

< 29.06|1 = 29.5) = P Z < 29.06 29.5
P (X
4/7

= P (Z < 0.77) = 0.2206

Power if the true mean is 1 = 28.5 is,

< 29.06|1 = 28.5) = P Z < 29.06 28.5
P (X
4/7

= P (Z < 0.98) = 0.8365

g. What is the minimum sample size required if we want to have the power of at least 0.80 at
1 = 29.5 for the test with a 5% level of significance?

The test statistic is Z =

X30
4
n

< 0.8289.

We want,

30
X

4
n

<

30
X

4
n

<




1.645 1




1.645 1


= 29.5 0.80

= 29.5 = P

29.5
X

4
n

29.5 30

4
n

<




1.645 1 =


29.5 30
29.5 30
= P Z
< 1.645
4
4

Note that P (Z < 0.842) 0.80, so 1.645

29.530
4
n

29.5
(10)

0.842. Solving for n yields,

n 396 after rounding.

Stepanov