MAT 540 Statistical Concepts For Research

235
Chapter 8

DRAWING INFERENCES FROM LARGE
SAMPLES

8.1 (a) (i)
65
125
. . 5.814
n
S E

= = =
(ii) 98% error margin is
0.01
5.814) 13.547 (2.33)(
n
z

= = .
(b) (i)
38
47
. . 5.543
n
S E

= = =
0.025
5.543) 10.864 (1.96)(
n
z

= = .
(c) (i)
20
6
. . 8.165
n
S E

= = =
0.05
8.165) 13.431 (1.645)(
n
z

= = .

8.2 (a) Point estimate 103 x =
95% error margin is
0.025
15
65
) 3.647 (1.96)(
S
n
z = = .
(b) Point estimate 260 x =
97.5% error margin is
0.0125
45
45
) 15.026 (2.24)(
S
n
z = = .
8.3 Point estimate is x . Estimated standard error is
n
s
, where
2
2
( )
1
i
x x
s
n
.
(a)
2
185 116
6.17, 4, . . 0.365
30 29 30
s
x s S E = = = = = =
(b)
2
145 103
5.8, 4.29, . . 0.414
25 24 25
s
x s S E = = = = = =
(c)
2
297 194
6.6, 4.41, . . 0.313
45 44 45
s
x s S E = = = = = =

236 CHAPTER 8. DRAWING INFERENCES FROM LARGE SAMPLES
8.4 (a) The point estimate for is 1.715 x = .
(b) Estimated standard error is
5.475
0.866
40
s
n
= = .
(c) 98% error margin is
0.01
0.866) 2.014 (2.326)(
S
n
z = =

8.5 Estimated mean weakly amount contested is $95.74 x = .
Estimated standard error is
24.63
$3.483
50
s
n
= = .
90% error margin is
0.05
1.645(3.483) $5.730
s
z
n
= = .

8.6 Estimated mean time is 3.8 minutes x = .
Estimated standard error is
1.2
0.1897
40
s
n
= = .
90% error margin is 1.645 1.645(0.1897) 0.31 minutes
s
n
= = .

8.7 (a) Given that the 95% error margin is 1.96 4.2
s
n
= , we know that the estimated
standard error is
4.2
2.143
1.96
s
n
= = .
(b) 90% error margin is 1.645 1.645(2.143) 3.525
s
n
= = .

8.8 The formula is
2
2
z
n
d
(
=
(

, where d is the specified error margin. Since the
sample size must be an integer, we always round up to the nearest integer.
(a)
2
(2.326)(135)
24, 650.57
2
(
=
(

, so the required sample size is 24, 651 n = .
(b)
2
(1.645)(18)
35.070
5
(
=
(

, so the required sample size is 36 n = .

8.9 Recall that the 95% error margin is 1.96
n
.
(a) In order that the 95% error margin be equal to
1
8
, we must have
1
8
1.96
n
= ,
so that solving for n yields
2
(1.96 8) 245.9 n = = . So, the required sample size
is 246 n = .
237

(b) In order that the 95% error margin be equal to 0.15 , we must have
1.96 0.15
n
= ,
so that solving for n yields
2
1.96
170.7
0.15
n
| |
= =
|
\
. So, the required sample size is
171 n = .

8.10 We have
2
0.005
1.4, 3.2, and 2.58 d z z
= = = = . Hence,
2
2.58(3.2)
34.78
1.4
(
=
(

.
So, the required sample size is 35 n = .

8.11 We have
2
0.01
5.00, 25, and 2.326 d z z
= = = = . Hence,
2
2.326(25)
135.26
5.00
(
=
(

.

8.12 We have
2
0.05
0.5, 3.5, and 1.645 d z z
= = = = . Hence,
2
1.645(3.5)
132.6
0.5
(
=
(

.

8.13 We have
2
0.05
0.5, 2.5, and 1.645 d z z
= = = = . Hence,
2
1.645(2.5)
67.6
0.5
(
=
(

.

8.14 Since the sample size is 108, the 90% error margin 1.645 2.9
108
= , so that
solving for yields 18.321 = . Now, to determine the required sample size n for
the case where 1.8, d = 18.321 =
2
0.025
and 1.96 z z
= = , we calculate
2
1.96
397.98
1.8
(
=
(

. So, the required sample size is 398. n =

8.15 Observe that 1 0.98 = implies 0.02 = , so that
2
0.01
2.326 z z
= = . A 98%
confidence interval for is calculated as follows:
0.01
1.303
2.583 2.326 2.583 0.5533
30
s
x z
n
= = or (2.0297, 3.136) .
8.16 Observe that 1 0.99 = implies 0.01 = , so that
2
0.005
2.58 z z
= = . A 99%
0.005
0.19
0.21 2.58 0.21 0.070
49
s
x z
n
= = or (0.14, 0.28) cm.

8.17 95% of 365, namely 347, confidence intervals are expected to cover the true means.
Before the data are obtained each day, the probability of covering the population
mean is 0.95. By the long run frequency interpretation of probability,
approximately 95% will cover, assuming each day of the year is a representative
sample.

8.18 (a) Correct
(b) We will never know whether the interval (2.96, 3.24) covers the true mean .
This is a single realization of the random interval 1.96
100
S
X . In repeated
sampling, about 95% of such intervals will cover the true mean .

8.19 For large n, a 95% confidence interval for is given by
0.025
S
X z
n
. Using

0.025
1.96, 35, z n = = and the summary statistics 30.2 grams, 3.8 grams, x s = = the
95% confidence interval for is given by
3.8
30.2 1.96 30.2 1.26
35
= or (28.94, 31.46) grams.

0.01
S
X z
n
. Using

0.01
2.33, 35, z n = = and the summary statistics 658 mg, 47 mg, x s = = the 98%
confidence interval for (true mean sodium content) is given by
47
658 2.33 658 18.5
35
= or (639.5, 676.5) mg.

0.005
S
X z
n
. Using

0.005
2.58, 120, z n = = and the summary statistics 18.3 days, 5.2 days, x s = = the
99% confidence interval for (true mean survival time) is given by
5.2
18.3 2.58 18.3 1.2
120
= or (17.1, 19.5) days.

239

8.22 (a) For large n, a 95% confidence interval for is given by
0.025
S
X z
n
. Using

0.025
1.96, 39, z n = = and the summary statistics 4.7, 3.2, x s = = the 95%
confidence interval for is given by
3.2
4.7 1.96 4.7 1.004
39
= or (3.696, 5.704) songs.

(b) This is uncertain. We only know that in the long run, 95% of all confidence
intervals would contain the value of the true parameter .

(c) 95% of all confidence intervals would contain by the interpretation provided
in the text.

0.025
S
X z
n
. Using

0.025
0.7
3.8 1.96 3.8 0.196
49
= or (3.604, 3.996).
(b) This is uncertain. We only know that in the long run, 95% of all confidence
intervals would contain the value of the true parameter .

(c) 95% of all confidence intervals would contain by the interpretation provided
in the text.

0.005
S
X z
n
. Using

0.005
confidence interval for (true mean time) is given by
51.852
307.77 2.58 307.77 10.643
158
= or (297.127, 318.413) units.

0.05
S
X z
n
. Using

0.05
1.645, 140, z n = = and the summary statistics 8.6 miles, 4.3 miles, x s = = the
90% confidence interval for (true mean commuting distance) is given by
4.3
8.6 1.645 8.6 0.60
140
= or (8.00, 9.20) miles.

0.05
S
X z
n
. Using

0.05
1.645, 40, z n = = and the summary statistics 3.8 minutes, 1.2 minutes, x s = =
the 90% confidence interval for (true mean processing time) is given by
1.2
3.8 1.645 3.8 0.31
40
= or (3.49, 4.11) minutes.

0.025
S
X z
n
. Using

0.025
1.96, 50, z n = = and the summary statistics 95.74 dollars, x =
24.63 dollars, s = the 95% confidence interval for (true mean amount contested)
is given by
24.63
95.74 1.96 95.74 6.83
50
= or (88.91, 102.57) dollars.

0.025
S
X z
n
. Using

0.025
1.96, 55, z n = = and the summary statistics 14.9, x = 2.8, s = the 95%
2.8
14.9 1.96 14.9 0.74
55
= or (14.16, 15.64).
0.025
S
X z
n
. Using

0.025
0.0207
0.0017 1.96 0.0017 0.003
154
= or ( 0.0047, 0.0013) .
0.05
S
X z
n
. Using

0.05
1.645, 30 z n = = and the summary statistics 8.12, 1.78 x s = = the 90%
1.78
8.12 1.645 8.12 0.535
30
= or (7.585, 8.655).
0.005
S
X z
n
. Using

0.005
2.58, 40, z n = = and the summary statistics 1.715 centimeters, x =
5.475 centimeters, s = the 99% confidence interval for is given by
5.475
1.715 2.58 1.715 2.233
40
= or (-0.518, 3.948) centimeters.
241

0.025
S
X z
n
. Using

0.025
0.06
0.11 1.96 0.11 0.015
65
= or (0.095, 0.125).

8.33 (a) Uncertain since the true mean growth (a population parameter) is not known.
Refer to the discussion in the text for a detailed explanation.

(b) The very definition of a confidence interval ensures this statement is true see
property 2 in the text.


(b) The very definition of a confidence interval ensures this statement is true see
property 2 in the text.

0.025
S
X z
n
. Using

0.025
confidence interval for (true mean amount of PCBs in the soil) is given by
0.05
3.56 1.96 3.56 0.155
40
= or (3.405, 3.715) PCBs.
(b) The sample mean is 3.56 x = and does indeed lie in this interval. In fact, it is
the midpoint of the interval. (This is always true for the sample mean.)
(c) Uncertain since the true mean growth (a population parameter) is not known.
(d) No. This is an incorrect interpretation of a confidence interval. Refer to the
text for a correct interpretation.

(b) Yes, the sample mean x does indeed lie in this interval. In fact, it is the
midpoint of the interval. (This is always true for the sample mean.)
(c) Cannot tell. The sample mean X is a random variable. As such, its realization
for a future sample may be within this particular numerical interval, or it may
be outside of it. It is not possible to know apriori for certain.
(d) Cannot tell. The concept of a confidence interval has nothing to do with the
locations of the individual sample observations.

8.37 The alternative hypothesis H
1
is the assertion that is to be established; its opposite is
the null hypothesis H
0
.
(a) Let denote the population mean time, in days, to pay a claim. The
hypotheses are:
0 1
: 14, : 14 H H = <
(b) Let denote the population mean amount spent, in dollars. The hypotheses
are:
0 1
: 4.50, : 4.50 H H = >
(c) Let denote the population mean hospital bill, in dollars. The hypotheses
are:
0 1
: 5000, : 5000 H H = <
(d) Let denote the population mean time between purchases, in days. The
hypotheses are:
0 1
: 60, : 60 H H =

8.38 (a) The decision to reject H
0
is correct if 14 < , while it is an incorrect decision if
14 = . A Type I error is incurred when such an incorrect decision is made.
(b) The decision to reject H
0
is correct if 4.50 > , while it is an incorrect decision
if 4.50 = . A Type I error is incurred when such an incorrect decision is
made.
(c) The decision to reject H
0
is correct if 5000 < , while it is an incorrect
decision if 5000 = . A Type I error is incurred when such an incorrect
decision is made.
(d) The decision to reject H
0
is correct if 60 , while it is an incorrect decision
if 60 = . A Type I error is incurred when such an incorrect decision is made.

8.39 Retaining H
0
is a correct decision if H
0
is true, while it is an incorrect decision if H
1

is true. A Type II error is incurred when such an incorrect decision is made.
(a) Correct decision if 14 = and incorrect decision if 14 < (Type II error).
(b) Correct decision if 4.50 = and incorrect decision if 4.50 > (Type II error).
(c) Correct decision if 5000 = and incorrect decision if 5000 < (Type II
error).
(d) Correct decision if 60 = and incorrect decision if 60 (Type II error).

8.40 (a) : R X c
(b) : R X c
(c) : R X c
(d) : 60 R X c

8.41 (b) (i)
0 1
: 0.15, : 0.15 H H = <
(ii)
0.15
0.085/ 125
X
Z

=
(iii) : 1.96 R Z
243

(c) (i)
0 1
: 80, : 80 H H =
(ii)
80
8.6/ 38
X
Z

=
(iii) : 2.58 R Z
(d) (i)
0 1
: 0, : 0 H H =
(ii)
0
1.23/ 40
X
Z

=
(iii) : 1.88 R Z

8.42 (a) Using 30.54 x = , the observed value of the test statistic is
30.54 30 0.54
2.00
0.2697 2/ 55
z

= = = .
Since the observed 2.00 z = lies in the rejection region : 1.645 R Z , the null
hypothesis is rejected with 0.05 = .

(b) Using 0.136 x = , the observed value of the test statistic is
0.136 0.15 0.014
1.84
0.0076 0.085/ 125
z

= = = .
Since the observed 1.84 z = does not lie in the rejection region : 1.96 R Z ,
the null hypothesis is not rejected with 0.025 = .
(c) Using 77.35 x = , the observed value of the test statistic is
77.35 80 2.65
1.90
1.395 8.6/ 38
z

= = = .
Since the observed 1.90 z = does not lie in the rejection region : 2.58 R Z ,
the null hypothesis is not rejected with 0.01 = .
(d) Using 0.59 x = , the observed value of the test statistic is
0.59 0 0.59
3.03
0.1945 1.23/ 40
z

= = = .
Since the observed 3.03 z = lies in the rejection region : 1.88 R Z , the null
hypothesis is rejected with 0.06 = .

8.43 Since the claim is that 40 > , we formulate the hypotheses:
0 1
: 40, : 40 H H = >
With 70 n = and 5.6 = , the test statistic (in standardized form) is
40
5.6 / 70
X
Z

= .
(a) 41.31 X is equivalent to
41.31 40
1.96
5.6 / 70
Z

= , or more simply 1.96 Z .
Since [ 1.96] 0.025 P Z = , the level of significance is 0.025 = .
(b) Since [ 1.645] 0.05 P Z = , the rejection region : 1.645 R Z has the level of
significance 0.05 = . Now, since
40
5.6 / 70
X
Z

= , we observe that 1.645 Z
is equivalent to
5.6
40 (1.645) 41.10
70
X + = . Therefore, 41.10 c = .
8.44 To test
0 1
: 2.0 versus : 2.0 H H = > , we use the test statistic
2.0
/
X
Z
S n
= .
With 0.02 = , the rejection region is : 2.054 R Z . The observed value of the
test statistic is
2.538 2.0
2.26
1.303/ 30
z

= = ,
which is in R. Therefore, we reject H
0
at 0.02 = . Furthermore, the p-value is
[ 2.26] 0.0119 P Z = .

8.45 (a) The test statistic remains the same, but this time the rejection region is two-
sided. With 0.02 = , we have
2
0.01
2.33 z z
= = and the two-sided rejection region

is : 2.33 R Z . Note that the observed value 2.26 z = is not in this rejection
region. Hence, H
0
is not rejected at 0.02 = . Furthermore, the p-value is
2 [ 2.26] 2(0.0119) 0.0238 P Z = = .

(b) Since the test statistic was not in the rejection region, it is possible that we
made a Type II error, meaning that the mean number of items returned to the store
really wasnt 2, but we accepted that it was.

8.46 (a) To test
0 1
: 15 versus : 15 H H = , we use the test statistic
15
/ 39
X
Z
S
= .
Since H
1
is two-sided, the rejection region should have the form
2
: R Z z
.
With 0.01 = , the rejection region is : 2.58 R Z . The observed value of
the test statistic is
16.69 15
1.387
7.61/ 39
z

= = ,
which is not in R. Therefore, we do not reject H
0
at 0.01 = .

(b) The p-value is ( ) 2 [ 1.387] 2 0.0828 0.1654 P Z > = = .

8.47 (a) Reject H
0
since the p-value is 0.005, which is less than 0.01.
(b) We could have made a Type I error if H
0
happened to be true.
(c) Prior to sampling, the probability of making a Type I error is 0.01.
(d) If we took more and more samples and ran the same test of hypothesis, about
1% of the time we would make a Type I error (i.e., reject a true H
0
).
(e) The p-value is 0.005, as reported on the Minitab output.
245

8.48 (a) Reject H
0
since the p-value is 0.018, which is less than 0.02.
(b) We could have made a Type I error if H
0
happened to be true.
(c) Prior to sampling, the probability of making a Type I error is 0.02.
(d) If we took more and more samples and ran the same test of hypothesis, about
2% of the time we would make a Type I error (i.e., reject a true H
0
).
(e) The p-value is 0.018, as reported on the Minitab output.

8.49 (a) Since the claim is that 3.5 > , we formulate the hypotheses:
0 1
: 3.5, : 3.5 H H = >
Now, to run the test with sample size 40 n = , we use the test statistic
3.5
/ 40
X
Z
S
= .
Since H
1
is right-sided, the rejection region should have the form : R Z z
. With
0.10 = , the rejection region is : 1.28 R Z . The observed value of the test
statistic is
3.8 3.5
1.58
1.2/ 40
z

= = ,
0
[ 1.58] 0.0571 P Z = .

(b) We could have made a Type I error, meaning that the average time required to
process the rebate was really 3.5 minutes, even though we rejected this claim.

8.50 Let denote the population mean amount contested (in dollars). We wish to run
the following hypothesis test:
0 1
: 105, : 105 H H =
105
/ 50
X
Z
S
= .
Since H
1
2
: R Z z
. With
0.01 = , the rejection region is : 2.58 or 2.58 R Z Z . The observed value of
the test statistic is
95.74 105
2.658
24.63/ 50
z

= = ,
0
2 [ 2.658] 2(0.00393) 0.0079 P Z = = .

8.51 (a) Let be mean battery life. We test the hypotheses:
0 1
: 183, : 183 H H = >
(b) To run the test with sample size 64 n = , we use the test statistic
183
/ 64
X
Z
S
= .
Since H
1
is one-sided, the rejection region should have the form : R Z z
. With
statistic is
190.5 183
1.875
32 / 64
z

= = ,
0
[ 1.875] 0.0304 P Z = .

(c) We could have made a Type I error, meaning that we rejected the fact that the
mean battery life was 183 days when in fact it was.

8.52 (a) Yes. H
0
would be retained at 0.01 = because the observed value of the test
statistic failed to be in the rejection region with 0.02 = , which was inclusive
of the rejection region with 0.01 = .
(b) Cannot tell. The rejection region with 0.02 = is part of the rejection region
with 0.05 = . Knowing that the observed test statistic was not in R with
0.02 = , we cannot say whether or not it was in the larger R with 0.05 = .
(c) No. A p-value smaller than 0.02 = means that H
0
would be rejected at
0.02 = .

8.53 Let denote the population mean hold time (in minutes). Since the claim is that
3.0 > , we formulate the following hypotheses:
0 1
: 3.0, : 3.0 H H = >
3.0
/ 75
X
Z
S
= .
Since H
1
is right-sided, the rejection region should have the form : R Z z
. With
statistic is
3.4 3.0
1.44
2.4/ 75
z

= = ,
which is not in R. Therefore, we do not reject H
0
at 0.05 = . Furthermore, the p-
value is [ 1.44] 0.0749 P Z = . Since this is not small, support for the claim is
weak.

8.54 Let denote the true mean content (in percent-weight) of cashews in the mixed-nut
cans. Since the inspector wants to substantiate the belief that 25 < , we formulate
the following hypotheses:
0 1
: 25, : 25 H H = <
25
/ 35
X
Z
S
= .
Since H
1
is left-sided, the rejection region should have the form : R Z c . The
observed value of the test statistic is
247

23.5 25
2.86
3.1/ 35
z

= = ,
Note that the p-value is [ 2.86] 0.0021 P Z = . So, the null hypothesis would be
rejected with as small as 0.0021. This extremely small p-value is strong
evidence in support of the inspectors belief.

8.55 Let denote the true mean BOD. Since the intent is to establish that is different
from 3000, we formulate the following hypotheses:
0 1
: 3000, : 3000 H H =
3000
/ 43
X
Z
S
=
Since H
1
2
: R Z z
. With
0.05 = ,
0.025
1.96 z = , so that the rejection region is : 1.96 R Z . The observed
value of the test statistic is
3246 3000
2.13
757 / 43
z

= = ,
0
at 0.05 = . Moreover, the p-value is
[ 2.13] 0.0166 P Z > = . This means that there is strong evidence that the BOD is
significantly off the target.

8.56 Let denote the true mean amount of SS (in pounds per day). Since the claim is
that 6000 < , we formulate the following hypotheses:
0 1
: 6000, : 6000 H H = <
6000
/ 43
X
Z
S
= .
Since H
1
is left-sided, the rejection region should have the form : R Z c . The
observed value of the test statistic is
5710 6000
1.11
1720/ 43
z

= = ,
Note that the p-value is [ 1.11] 0.1335 P Z = . So, the null hypothesis would be
rejected with as small as 0.1135. Because this is not small, support for the claim
is weak.

8.57 (a) (i) p = proportion of adult population for which reading is a favorite leisure
time activity.
(ii) Point estimate
295
986
0.299 p = =
(iii) Estimated S.E. =
(0.299)(0.701)
986
0.0146
pq
n
= =
So, the 95% error margin = 1.96(Estimated S.E.) 1.96(0.0146) 0.0286 = = .


(b) (i) p = proportion of pet owners who revealed that they buy their pets holiday
presents.
(ii) Point estimate
293
440
0.666 p = =
(iii) Estimated S.E. =
(0.666)(0.334)
440
0.022
pq
n
= =
So, the 95% error margin =1.96(Estimated S.E.) 1.96(0.022) 0.043 = = .

8.58 The 98% error margin is given by 2.33(Estimated S.E.). The results are:
(a) 2.33(0.0146) = 0.0340
(b) 2.33(0.022) = 0.051

8.59 (a) We have 0.3, 0.7, 0.03, 0.10 p q d = = = = . So,
2
0.05
1.645 z z
= = . We
calculate
2
1.645
(0.3)(0.7) 631.40
0.03
n
(
= =
(

. So, the required sample size is 632.
(b) Since p is unknown, the conservative bound on n yields
2
1 1.645
751.67
4 0.03
(
=
(

,
so that the required sample size in this instance is 752.

8.60 We use 0.3 p = (to be conservative). Then, 0.7. q = Also, we are given that
0.03, 0.02 d = = . So,
2
0.01
2.33 z z
= = . We calculate
2
2.33
(0.30)(0.70) 1266.74
0.03
n
(
= =
(

. So, the required sample size is 1132.

8.61 (a) Denote
M
p = population proportion of ERS calls involving serious
mechanical problems. Since there were 849 calls involving serious
mechanical problems out of 2927 n = ERS calls, the point estimate of
M
p is
849
2927
0.290
M
p = = . Also, the 95% error margin is
(0.29)(0.71)
2927
1.96 0.016 = .
(b) Denote
S
p =population proportion of ERS calls involving starting problems.
Since there were 1499 calls involving starting problems out of 2927 n = ERS
calls, the point estimate of
S
p is
1499
2927
0.512
S
p = = , and so
1 0.512 0.488
S
q = = . The 98% confidence interval for
S
p is then given by
(0.512)(0.488)
2927
2.33 0.512 2.33 0.512 0.0215
S S
S
p q
n
p = =
or (0.491, 0.534).

8.62 Let p = the population proportion of all possible policies that would have at least
one claim in a year.
Point estimate
2073
12,299
0.169 p = =
249

So, the 95% error margin =
(0.169)(0.831)
12,299
1.96(Estimated S.E.) 1.96 0.007 = = .
Thus, the 95% confidence interval for p is given by
0.025
(0.169)(0.831)
12,299
0.169 1.96 0.169 0.007
pq
n
p z = = or (0.16, 0.18).

8.63 Let p = population proportion of all drive-up window purchases made with a major
credit card.

(a) The sample proportion is
27
0.287
94
p = = .
(b) Estimated standard error is
(0.287)(0.713)
94

0.047
pq
n
= =
(c) The 98% confidence interval for p is given by

0.01

0.287 2.33(0.047) 0.287 0.110
pq
n
p z = =
or (0.177, 0.397). That is, (17.7%, 39.7%).

8.64 Let p = population proportion of all brands of cola that are Brand P.
(a) The sample proportion is
249
0.6225
400
p = = , and the estimated standard error is

(0.6225)(0.3775)
400

0.024
pq
n
= = . So, the 95% confidence interval for p is given by

0.025

0.6225 1.96(0.024) 0.6225 0.047
pq
n
p z = =
or (0.575, 0.670). That is, (57.5%, 67.0%).

(b) You cannot determine this. We only know that 95% of all such confidence
intervals would contain the true value of p.

(c) 95% of all confidence intervals would contain p by the interpretation provided
in the text.

8.65 (a) Let p be the proportion of students that hold a part time job. The hypotheses
are:
0 1
: 0.26, : 0.26 H p H p = >
(b) Let p be the proportion of subscribers that have complaints against the cable
company. The hypotheses are:
0 1
: 0.13, : 0.13 H p H p = <
(c) Let p be the proportion of subscribers that have complaints against the cable
company. The hypotheses are:
0 1
: 0.13, : 0.13 H p H p = >
(b) Let p be the proportion of boards that would break under the standard load.
The hypotheses are:
0 1
: 0.05, : 0.05 H p H p = <

8.66 (a) Let p be the proportion of smokers who would quit on the first attempt. The
hypotheses are:
0 1
: 0.40, : 0.40 H p H p = <
(b) Let p be the proportion of drivers that use the lane legally. The hypotheses
are:
0 1
: 0.25, : 0.25 H p H p = >
(c) Let p be the proportion of patients that would wait over half an hour. The
hypotheses are:
0 1
: 0.20, : 0.20 H p H p = <

8.67 (b) (i)
0 1
: 0.75, : 0.75 H p H p = >
(ii)
(0.75)(0.25)
228
0.75 0.75
0.0287
p p
Z

= =
(iii) Since
0.02
0.02, 2.05 z = = and H
1
is right-sided, the rejection region is
given by : 2.05 R Z .

(c) (i)
0 1
: 0.60, : 0.60 H p H p =
(ii)
(0.60)(0.40)
77
0.60 0.60
0.0558
p p
Z

= =
(iii) Since
2 0.01
0.02, 0.01, 2.33 z
= = = and H
1
is two-sided, the rejection
region is given by
2
: 2.33 R Z z
= .

(d) (i)
0 1
: 0.56, : 0.56 H p H p = <
(ii)
(0.56)(0.44)
86
0.56 0.56
0.0535
p p
Z

= =
(iii) Since
0.10
0.10, 1.28 z = = and H
1
is left-sided, the rejection region is
given by : 1.28 R Z .

8.68 (a) The value of the observed test statistic is
0.233 0.32
2.04
0.0426
z

= = . Since
2.04 is in the rejection region : 1.645 R Z , H
0
is rejected at 0.05 = .
(b) The value of the observed test statistic is
0.818 0.75
2.369
0.0287
z

= = . Since
0
(c) The value of the observed test statistic is
0.709 0.60
1.95
0.0558
z

= = . Since
1.95 is not in the rejection region : 2.33 R Z , H
0
is not rejected at 0.02 = .
(d) The value of the observed test statistic is
0.387 0.56
3.23
0.0535
z

= = . Since
0

8.69 Let p = proportion of all freshmen dorms that have a poster of a rock group
hanging.
(a) Test the hypotheses
0 1
: 0.30, : 0.30 H p H p = >
251

(b) Since H
1
is one-sided, the rejection region : R Z z
, and since 0.05 = , this

becomes : 1.645 R Z .
(c) We would reject H
0
in favor of H
1
. So, a type I error could be made.
(d) Test statistic:
(0.70)(0.30)
60
0.30 0.30
0.059
p p
Z

= =
If 25 x = , then
25
0.417
60
p = = . Then, the test statistic value is given by
0.417 0.30
1.983
0.059
Z

= = , which is inside R. Hence, we reject H
0
at 0.05 = .
(e) Since we are rejecting H
0
, we could have made a Type I error, meaning that
the true proportion of dorms with posters of a rock group hanging really is
0.30.

8.70 Let p = sample proportion.
(a) Test statistic:
(0.30)(0.70)
0.30
n
p
Z

=
Since
0.05
1.645 z = and H
1
is right-sided, the rejection region is : 1.645 R Z .
(b) The observed sample proportion is
19
48
p = , and the test statistic has the value

19
48
(0.30)(0.70)
48
0.30
1.45 Z

= = , which is not in R. Hence,H
0
is not rejected at 0.05 = .
Furthermore, the p-value is [ 1.45] 0.0735 0.07 P Z = . Therefore, must
be taken larger than 0.07 in order to reject H
0
.

8.71 (a) Since the intent is to establish that 0.5 p < , we formulate the following
hypotheses:
0 1
: 0.5, : 0.5 H p H p = <
(b) The sample proportion of support is
228
500
0.456 p = = . Since 0.05 = ,
0.05
1.645 z = and H
1
is left-sided, the rejection region is : 1.645 R Z . The
test statistic is calculated to be
228
500
(0.5)(0.5)
500
0.5
1.97 Z

= = , which is in R. Hence, H
0

is rejected at 0.05 = and the claim is supported. Furthermore, the p-value is
[ 1.97] 0.0244 P Z = , so that H
0
would still be rejected with as small as
0.0244.

8.72 Let p denote the population proportion of ERS calls involving flat tires or lockouts.
Since the intent is to establish that 0.19 p < , we formulate the following
hypotheses:
0 1
: 0.19, : 0.19 H p H p = <
The sample proportion of support is
498
2927
0.170 p = = . Since 0.05 = ,
0.05
1.645 z =
and H
1
is left-sided, the rejection region is : 1.645 R Z . The test statistic is
calculated to be
498
2927
(0.19)(0.81)
2927
0.19 0.170 0.19
2.76
0.00725
Z

= = = , which is in R. Hence, H
0
is
rejected at 0.05 = and the claim is supported.

8.73 Let p denote the population proportion of brands of cola that are Brand P. Since the
intent is to establish that 0.55 p > , we formulate the following hypotheses:
0 1
: 0.55, : 0.55 H p H p = >
Since H
1
is right-sided, the rejection region is of the form : R Z z
. Using the
information from Exercise 8.64, we have 0.6225 p = and . . 0.024 S E = . The test
statistic is calculated to be
0.6225 0.55
3.021
0.024
Z

= = . The associated p-value to
this observed z is certainly < 0.01, so that there is very strong evidence in support of
our claim.

8.74 Let p denote the population proportion of persons who rate the services as being
excellent. Since the intent is to establish that 0.46 p , we formulate the following
hypotheses:
0 1
: 0.46, : 0.46 H p H p =
The sample proportion of support is
258
505
0.5109 p = = . Since H
1
is two-sided and
0.10 = ,
0.10
2
0.05
1.645 z z = = , the rejection region is : 1.645 R Z . The test
statistic, based on a sample of size 505 n = , is calculated to
be
(0.46)(0.54)
505
0.46 0.5109 0.46
2.29
0.0222
p
Z

= = = , which is in R. Hence, we reject H
0
at
0.10 = . Furthermore, the associated p-value is 2 [ 2.29] 0.0220 P Z = . This low
p-value signifies strong support of H
1
.

8.75 (a) We have 505 n = and
258
505
0.5109 p = = . So, a 90% confidence interval for p
is given by:
0.05
(0.5109)(0.4881)
505
0.5109 1.645 0.5109 0.0366
pq
n
p z = =
or (0.4743, 0.5475).
(b) The unknown total number of customers who would rate the service as being
excellent is 8200p. The lower endpoint of the 90% confidence interval for the
total is then 8200(0.4743) = 3889, and the upper endpoint is 8200(0.5475) =
4490.

8.76 Let p denote the population proportion of adults supporting a tougher legislation.
Since the intent is to establish that 0.75 p > , we formulate the following
hypotheses:
0 1
: 0.75, : 0.75 H p H p = >
The sample proportion based on a sample of size 980 n = is 0.78 p = . Since H
1
is
right-sided, the rejection region is of the form : R Z c . The test statistic is
253

calculated to be
(0.75)(0.25)
980
0.75 0.78 0.75
2.17
0.0138
p
Z

= = = . The associated p-value to this
observed z is [ 2.17] 0.0150 P Z = . This very small p-value signifies a strong
support of the conjecture.

8.77 We have 42 n = and
30
42
0.7143 p = = . So, a 95% confidence interval for p
is given by:
0.025
(0.7143)(0.2857)
42
0.7143 1.96 0.7143 0.1366
pq
n
p z = =
or (0.58, 0.85). That is, (58%, 85%).

8.78 (a) 1.949 x = is the point estimate of .
(b)
1.558
59
. . of 0.203
n
S E X

= = =
0.01
0.203) 0.473 (2.33)(
n
z

= = .
8.79 Recall that the point estimate of is
i
x
x
n
=

and the estimated standard error is
n
s
, where
2
2
( )
1
i
x x
s
n
.
(a)
3230.84
85.02
38
x = =
(b)
2028.35
37
S.E. 1.201
38
s
n
= = =
0.025
1.201) 2.354 (1.96)(
n
z

= =

8.80 (a) A point estimate of is 47 x = minutes.
(b) An approximate 95.4% error margin is given by 2 (Estimated S.E.) 2
s
n
= ,
which in this case is
5
2 2 1.58
40
s
n
= = minutes.
8.81 Recall that the standard error of X is
n
, where n is the sample size.

(a) Since
1
2 4 n n

= , we need a sample of size 4n. Therefore, we must increase
the sample size by a factor of 4.
(b) Since
1
4 16 n n

= , we need a sample of size 16n. Therefore, we must
increase the sample size by a factor of 16.

8.82 We have
2
0.025
10 and 1.96 d z z
= = = .
(a) Since
2
1.96(40)
61.5
10
(
=
(

, the required sample size is 62 n = .
(b) Since
2
1.96(80)
245.8
10
(
=
(

, the required sample size is 246 n = .

8.83 (a) The population mean is estimated by 126.9 x = .

10.5
Estimated S.E. 1.416
55
s
n
= = = . So, the approximate 95.4% error margin
is given by 2 (Estimated S.E.) 2(1.416) 2.8 = ,
(b) Observe that 1 0.90 = implies 0.10 = , so that
2
0.05
1.645 z z
= = . A 90%
1.645 126.9 1.645(1.416) 126.9 2.329
s
x
n
= = or (124.571,129.228) .

8.84 (a) A 98% confidence interval for is calculated as follows:
0.01
1.607
1.925 2.33 1.925 0.684
30
s
x z
n
| |
= =
|
\
or (1.241, 2.609) .
(b) In the long run, 98% of such confidence intervals would cover .

8.85 (a) Correct
(b) We will never know whether this particular interval covers the true mean .
This is a single realization of the random interval 1.645
S
X
n
. In repeated
sampling, about 90% of such intervals will cover the true mean .

8.86 (a) To test the hypotheses:
0 1
: 3.7 : 3.7 H H = <
we use the test statistic
3.7
/
X
S n
Since H
1
is left-sided, the rejection region
should have the form : R Z z
. Since
0.05
1.645 z = , the rejection region is
: 1.645 R Z . Since the value of the test statistic, namely
3.6 3.7
1.265
0.5 / 40
= , is not in R, we do not reject H

0
at 0.05 = .
(b) Since we do not reject H
0
, we could be making a Type II error.

255

8.87 The alternative hypothesis H
1
is the assertion that is to be established; its opposite is
the null hypothesis H
0
.
(a) Let denote the population mean mileage. The hypotheses are:
0 1
: 50, : 50 H H = <
(b) Let denote the population mean number of pages per transmission. The
hypotheses are:
0 1
: 3.4, : 3.4 H H = >
(c) Let p denote the probability of success with the method. The hypotheses are:

0 1
: .50, : .50 H p H p = >
(d) Let denote the mean fill. The hypotheses are:
0 1
: 16, : 16 H H =
(e) Let denote the mean percent fat content. The hypotheses are:
0 1
: 0.04, : 0.04 H p H p = >

8.88 Let denote the mean number of words per sentence.
(a) We test the hypotheses:
0 1
: 9.1, : 9.1 H H =
(b) The test statistic is
9.1
/
X
Z
s n
= .
(c) Since H
1
is two-sided, the rejection region
2
: R Z z

(d) Using 8.6, 1.2, 36 x s n = = = , we see that the test statistic is

8.6 9.1
2.5
1.2 / 36
Z

= = . For 0.10 = , the rejection region is
0.025
: 1.645 R Z z = . Since the test statistic value is in R, we reject H
0
at
0.10 = .
(e) The associated p-value is ( ) 2 ( 2.5) 2 0.0062 0.0124 P Z = = .
(f) Since we rejected H
0
, we could have made a Type I error in that we rejected the
fact that there are 9.1 words per sentence, on average, when this is in fact the
case.

8.89 Let denote the mean length of time that a 7 oz. tube of toothpaste lasts.
0 1
: 30.5, : 30.5 H H = >
(b) The test statistic is
30.5
/
X
Z
s n
= .
(c) Since H
1
is one-sided, the rejection region : R Z z

(d) Using 32.3, 6.2, 75 x s n = = = , we see that the test statistic is

32.3 30.5
2.514
6.2 / 75
Z

= = . For 0.10 = , the rejection region is
0.10
: 1.28 R Z z = . Since the test statistic value is in R, we reject H
0
at
0.10 = .
(e) The associated p-value is ( 2.514) 0.00597 P Z > = .
0
, we could have made a Type I error in that we rejected the
fact that there are 9.1 words per sentence, on average, when this is in fact the
case.

8.90 (a) Cannot tell. The rejection region for 0.03 = is smaller than that for
0.05 = .
(b) Yes. The observed test statistic was in the rejection region with 0.05 = , and
now the rejection region is even larger.
(c) No. The p-value is the smallest value of at which H
0
can be rejected.

8.91 (a) Reject H
0
since the p-value is 0.014, which is less than 0.03 = .
(b) To test the hypotheses:
0 1
: 84 : 84 H H = <
we use the test statistic
84
/
X
S n
(the observed value of which, from the

printout, is 2.45 ). Since H
1
is left-sided, the rejection region should have
the form : R Z z
. Since
0.01
2.33 z = , the rejection region is : 2.33 R Z .
Since the value of the test statistic, namely 2.45 , is in R, we reject H
0
at
0.01 = .

8.92 A 99% confidence interval for is calculated as follows:
0.01
2.3
3.4 2.33 3.4 0.619
75
s
x z
n
| |
= =
|
\
or (2.781, 4.019) .

8.93 (a) [ ] 0(0.5) 1(0.3) 2(0.2) 0.7 E X = + + =
(b) Since
2 2 2 2
[ ] 0 (0.5) 1 (0.3) 2 (0.2) 1.1 E X = + + = , we observe that

2 2 2
var[ ] [ ] ( [ ]) 1.1 (0.7) 0.61 X E X E X = = = , and so 0.61 0.781 = =
(c) To test
0 1

0.7
/
X
Z
S n
= with 64 n = and we will use the standard deviation from the

new sample because the distribution of sales may have changed. With
statistic is
0.84 0.7
2.80
0.4/ 64
z

= = ,
0
at 0.05 = . Furthermore, the p-value
is [ 2.80] 0.0026 P Z = . So, the evidence in favor of increased mean sales is
very strong.

257

8.94 (a) To be 95% certain that the error does not exceed 2, d = we use

2
0.025
1.96 z z
= = and evaluate
2 2
0.025
(1.96)(25)
600.25
2
z s
n
d
( (
= = =
( (

. So,
the required sample size is 601 n = .
(b) For large n, a 98% confidence interval for is given by
0.01
S
X z
n
. Using

0.01
2.33, 100, z n = = and the summary statistics 155, 22, x s = = the 98%
22
155 2.33 155 5.126
100
= or (149.87, 160.13).

8.95 (a) The point estimate of the proportion unemployed is
175
2000
0.0875 p = = .
(b) The 95% error margin = 1.96(Estimated S.E.), which is given by
(0.0875)(0.9125)
2000
1.96 1.96 0.0124
pq
n
= = .

49
78
0.0875 p = = , so that 0.9125 q =
. We also are given that

0.02 = (so that
2
0.01
2.33 z z
= = ) and. Thus, the 98% confidence interval is

(0.0875)(0.9125)
0.01 2000

0.0875 2.33 0.0875 0.0147
pq
n
p z = = or (0.073, 0.102).
That is, (7.3%, 10.2%).

139
625
0.2224 p = = , so that 0.7776 q =
. We also are given that

0.05 = (so that
2
0.025
1.96 z z
= = ) and. Thus, the 95% confidence interval is

(0.2224)(0.7776)
0.025 625

0.2224 1.96 0.2224 0.0333
pq
n
p z = = or (0.189, 0.256).
That is, (18.9%, 25.6%).

8.98 Let p = proportion of all students that missed at least one class last week.
0 1
: 0.20, : 0.20 H p H p = >
(b) Denoting the sample proportion by p , the test statistic is
(0.2224)(0.7776)
625
0.20 0.20
0.017
p p
Z

= =
(c) Since H
1
is one-sided, the rejection region with 0.05 = is : 1.645 R Z .
(d) Using 0.2224 p = , 625, . . 0.017 n S E = = , the value of the test statistic is
0.2224 0.20
1.318
0.017
Z

= = . Since this is not in R, we do not reject H
0
at
0.05 = .
(f) Since we do not reject H
0
, we could have made a Type II error, meaning that the
claim is valid when we assert it is not.

8.99 Let p = proportion of all flights from Chicago to Austin that arrive late.
0 1
: 0.25, : 0.25 H p H p = >
(b) Denoting the sample proportion by p , the test statistic is

0.25
pq
n
p
Z

=
(c) Since H
1
is one-sided, the rejection region with 0.05 = is : 1.645 R Z .
(d) Using
44
137
0.321 p = = and 137 n = , the value of the test statistic is
(0.321)(0.679)
137
0.321 0.25
1.775 Z

= = . Since this is in R, we reject H
0
at 0.05 = .
0
, we could have made a Type I error, meaning that 25% of
all flights from Chicago to Austin arrive late.

8.100 (a) Let p denote the probability of at least one claim on a policy in the population
of all possible policies that could be written. Since the intent is to establish
that 0.16 p , we formulate the following hypotheses:
0 1
: 0.16, : 0.16 H p H p = The test statistic for a sample of size
12, 299 n = is
(0.16)(0.84)
12,299
0.16 0.16
0.0033
p p
Z

= = . With 0.05 = ,
0.05
2
1.96 z = and H
1

is two-sided, the rejection region is : 1.96 R Z . The observed sample
proportion is
2073
12,299
0.169 p = = , and the test statistic has the
value
0.169 0.16
2.73
0.0033
Z

= = , which is in R. Hence, H
0
is rejected at
0.05 = . So, we conclude that the proportion of policies with at least one
claim is different from 16%.

(b) The p-value is [ 2.73] [ 2.73] 2(0.0032) 0.0064 P Z P Z + = = . So, the
evidence in favor of 0.16 p is strong.

8.101 (a) Let p denote the population proportion of plants that are of the dwarf variety.
Since the intent is to establish that 0.8 p , we formulate the following
hypotheses:
0 1
: 0.8, : 0.8 H p H p = The test statistic for a sample of size
200 n = is
(0.8)(0.2)
200
0.8 0.8
0.0283
p p
Z

= = . With 0.05 = ,
0.05
2
1.96 z = and H
1
is two-
sided, the rejection region is : 1.96 R Z . The observed sample proportion is
136
200
0.68 p = = , and the test statistic has the value
0.68 0.8
4.24
0.0283
Z

= = ,
which is in R. Hence, H
0
is rejected at 0.05 = . Furthermore, the associated
p-value is given by
259

[ 4.24] [ 4.24] 0.0001 P Z P Z + < ,
which is extremely small. As such, a contradiction of the genetic model is
strongly indicated.

(b) A 95% confidence interval for p is given by
(0.68)(0.32)
200
1.96 0.68 1.96 0.68 0.0555
pq
n
p = = or (0.6245, 0.7355).

8.102 (a) Under the alternative
1
10.5, X = is normally distributed with mean 10.5 and
standard deviation = 0.25
n
= . So,
10.5
0.25
X
Z

= is N(0,1). Therefore, the
power at
1
10.5 = is given by
1
10.49 10.5
[ 10.49 when 10.5] [ ]
0.25
[ 0.04] 0.5160.
P X P Z
P Z

= =
= =

(b) For
1
10.8 = ,
10.8
0.25
X
Z

= is N(0,1). Therefore, the power at
1
10.8 = is
given by
1
10.49 10.8
[ 10.49 when 10.8] [ ]
0.25
[ 1.24] 0.8925.
P X P Z
P Z

= =
= =

8.103 (a) A 99% confidence interval will be larger than the 95% confidence interval
given in the printout. In fact, the greater the confidence level, the larger the
confidence interval will be for a given sample size. To verify this, observe
that
0.01
2
0.4840
2.8157 2.58 2.8157 0.2110
35
s
x z
n
= = ,
or (2.605, 3.026).

(b) To test
0 1
2.6
0.484/ 35
X
Z

= . With 0.05 = , since H
1
is right-sided, the rejection region
is : 1.645 R Z . The observed value of the test statistic is
2.8157 2.6
2.637
0.484/ 35
z

= = ,
0
at 0.05 = .

8.104 (a) A 98% confidence interval will be larger than the 95% confidence interval
given in the printout. In fact, the greater the confidence level, the larger the
confidence interval will be for a given sample size. To verify this, observe
that
0.02
2
1.101
77.458 2.33 77.458 0.4056
40
s
x z
n
= = ,
or (77.052, 77.834).

(b) To test
0 1
: 77 versus : 77 H H = > , we use the test statistic
77
1.101/ 40
X
Z

= . With 0.05 = , since H
1
is right-sided, the rejection region
is : 1.96 R Z . The observed value of the test statistic is given to be Z=2.630
0
at 0.05 = .

8.105 Enter the data into column C2 in a Minitab worksheet.

(a) Run a 1-sample t-test since the population standard deviation is unknown. The
output is as follows.

T Confidence Intervals
Variable N Mean StDev SE Mean 97.0 % CI
C2 40 1.7150 0.4748 0.0751 (1.5459, 1.8841)

So, the 97% confidence interval is (1.5459, 1.8841).

(b) Testing the hypotheses
0 1
: 1.9 versus : 1.9 H H = using Minitab yields
the following output:

T-Test of the Mean
Test of mu = 1.9000 vs mu not = 1.9000

Variable N Mean StDev SE Mean T P
C2 40 1.7150 0.4748 0.0751 -2.46 0.018

Since the p-value is 0.018, we reject H
0
at 0.03 = .

8.106 Enter the data into column C3 in a Minitab worksheet. Run a 1-sample t-test
since the population standard deviation is unknown. The output is as follows.

C3 29 1.698 0.565 0.105 ( 1.483, 1.913)


261

8.107 Enter the data into column C4 in a Minitab worksheet. Run a 1-sample t-test

C4 28 110.39 23.18 4.38 (102.93, 117.85)


8.108 (a) Enter the data into column C5 in a Minitab worksheet. Run a 1-sample t test


C5 81 38.62 9.06 1.01 ( 36.39, 40.84)


(b) The histogram is as follows:
60 55 50 45 40 35 30 25 20
20
10
0
C5
F
r
e
q
u
e
n
c
y

The graph appears to be nearly normal, with a slight imbalance to the left.

8.109 (a) Enter the data into column C6 in a Minitab worksheet. Run a 1-sample t test


C6 151 4.072 9.065 0.738 ( 2.614, 5.530)


(b) The histogram is as follows. Note that the histogram has a long tail to the
right (much more extreme than in Exercise 8.108).
0 10 20 30 40 50 60 70 80
0
50
100
BPH
F
r
e
q
u
e
n
c
y

8.110 (a) Enter the data into column C1 of a Minitab worksheet. The output is as
follows:


Weight 61 184.3 124.3 15.9 (152.5, 216.1)

So, the 95% confidence interval for the mean weight of all bears is given
by (152.5, 216.1) pounds.

(b) A histogram for the weight data for all bears is as follows:

600 500 400 300 200 100 0
15
10
5
0
Weight (all bears)
F
r
e
q
u
e
n
c
y

It is skewed a bit with a tail to the right.

MAT 540 Statistical Concepts For Research

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MAT 540 Statistical Concepts For Research

Uploaded by

Copyright:

Available Formats

235

= = and the two-sided rejection region

, and since 0.05 = , this

, where n is the sample size.

= , is not in R, we do not reject H

(the observed value of which, from the

= with 64 n = and we will use the standard deviation from the

. We also are given that

= = ) and. Thus, the 98% confidence interval is

. We also are given that

= = ) and. Thus, the 95% confidence interval is

You might also like