Professional Documents
Culture Documents
Solutions Manual To Accompany Elementary Statistics Using Excel 4th Edition 0321564960
Solutions Manual To Accompany Elementary Statistics Using Excel 4th Edition 0321564960
Chapter 8
Hypothesis Testing
1. Given the large sample size and the fact that 20% is so much less than 50%, it is apparent that
any confidence interval for the proportion of bosses that are good communicators would fall
entirely below 50%. Assuming the magazine has properly interpreted the survey, the results
appear to support the claim that “less than 50% of bosses are good communicators” but not
necessarily that “less than 50% of the people believe that bosses are good communicators” –
those are two different statements that should not be confused.
Given that the responders constitute a voluntary response sample, and not a random sample,
it is likely that they are not representative of the population and consist largely of people with
strong feelings on and/or a personal interest in the topic. The results should not be used to
support the stated claim.
2. Since the P-value gives the probability of obtaining the observed result or more extreme results
by chance alone, the smallest p-value of 0.001 gives the strongest evidence for the alternative
hypothesis and would be the preferred result.
3. No. Since the claim that the mean is equal to a specific value must be the null hypothesis, the
only possible conclusions are to reject that claim or to fail to reject that claim. Hypothesis
testing cannot be used to support a claim that a parameter is equal to a particular value.
4. No. Sample data that is not consistent with a claim cannot be used to support that claim. In
particular, no sample proportion less than 0.5 can ever be used to support a claim that the
population proportion is greater than 0.5.
5. If the claim were not true, and p ≤ 0.5, then getting 90 heads in a sample of 100 tosses would
be an unusual event. There is sufficient evidence to support the claim.
6. If the claim were not true, and p ≤ 0.35, then getting 0.955(2480) = 2368 households with
telephones in a sample of 2480 households would be an unusual event. There is sufficient
evidence to support the claim.
7. If the claim were not true, and µ ≥ 75, then getting a mean pulse rate of 74.4 in a sample of
students would not be an unusual event. There is not sufficient evidence to support the claim.
8. If the claim were not true, and σ ≥ 15, then getting a standard deviation of 14.8 in a sample of
40 movie patrons would be not an unusual event. There is not sufficient evidence to support
the claim.
9. original claim: µ > $60,000
Ho: µ = $60,000
H1: µ > $60,000
10. original claim: p = 0.20
Ho: p = 0.20
H1: p ≠ 0.20
0.1056
<-----------
-1.25 0 Z
30. P-value
= P(z>2.50)
= 1 – 0.9938
= 0.0062
Since 0.0062 < 0.05, reject Ho.
Excel: 1 – NORMSDIST(2.50) = 0.0062
0.9938
<--------------------------------
0 2.50 Z
-1.75 0 1.75 Z
32. P-value
= 2·P(z<-0.55)
= 2·(0.2912)
= 0.5824
Since 0.5824 > 0.05, fail to reject Ho.
Excel: 2* NORMSDIST(-0.55) = 0.5823
0.2912
<-----------
-0.55 0 0.55 Z
33. P-value
= 2·P(z<-2.75)
= 2·(0030)
= 0.0060
Since 0.0060 < 0.05, reject Ho.
Excel: 2* NORMSDIST(-2.75) = 0.0060 0.0030
<-----------
0.707 ^
p
-2.75 0 2.75 Z
34. P-value
= 2·P(z>0.35) <--------------------------------|
0.6368
= 2·[1 – 0.6368]
= 2·[0.3632]
= 0.7264
Since 0.7264 > 0.05, fail to reject Ho.
Excel: 2*(1 – NORMSDIST(0.35)) = 0.7263
0.75 ^
p
-0.35 0 0.35 Z
= 1 – 0.9893
= 0.0107
Since 0.0107 < 0.05, reject Ho.
Excel: 1 – NORMSDIST(2.30) = 0.0107
0.25 ^
p
0 2.30 Z
36. P-value
= P(z<-2.95)
= 0.0016
Since 0.0016 < 0.05, reject Ho.
Excel: NORMSDIST(-2.95) = 0.0016
0.0016
<-----------
0.777 ^
p
-2.95 0 Z
37. There is not sufficient evidence to conclude that the percentage of blue M&M’s is greater than
5%.
38. There is sufficient evidence to conclude that the percentage of on-time U.S. airline flights is
less than 75%.
39. There is not sufficient evidence to reject the claim that the percentage of Americans who know
their credit score is equal to 20%.
40. There is sufficient evidence to reject the claim that the percentage of Americans who believe in
heaven is equal to 90%.
41. original claim: p = 0.41
Ho: p = 0.41
type I error: rejecting Ho when Ho is actually true
rejecting the claim that the percentage of non-smokers exposed to secondhand smoke is
41% when that percentage actually is 41%
type II error: failing to reject Ho when H1 is actually true
failing to reject the claim that the percentage of non-smokers exposed to secondhand
smoke is 41% what that percentage is actually different from 41%
42. original claim: p = 0.20
Ho: p = 0.20
type I error: rejecting Ho when Ho is actually true
rejecting the claim that the percentage of Americans who believe that life exists only on
earth is 20% when that percentage actually is 20%
type II error: failing to reject Ho when H1 is actually true
failing to reject the claim that the percentage of Americans who believe that life exists only
on earth is 20% when that percentage is actually different from 20%
43. original claim: p > 0.70
Ho: p = 0.70
type I error: rejecting Ho when Ho is actually true
rejecting the claim that the percentage of college students who use alcohol is 70% when
that percentage actually is 70%
type II error: failing to reject Ho when H1 is actually true
failing to reject the claim that the percentage of college students who use alcohol is 70%
when that percentage is actually greater than 70%
for c as follows:
0.50 c ^
p
c = p + zc· pq/n 0 1.645 Z
= 0.50 + (1.645)· (0.50)(0.50)/64
= 0.50 + (1.645)·(0.0625)
= 0.6028
Excel: NORMINV(0.95,0.50,SQRT(0.5*0.5/64)) = 0.6028
for c as follows:
0.50 c ^
p
c = p + zc· pq/n 0 1.645 Z
= 0.50 + (1.645)· (0.50)(0.50)/n
= 0.50 + 0.8225/ n
The power calculations are given below
and illustrated by the figure at the right. |-------------------------------->
0.8000
c = p + zc· pq/n
= 0.55 + (-0.84)· (0.55)(0.45)/n
= 0.55 – 0.4179/ n
The final value for n is obtained by 0.2000
equating the two expressions for c. <-----------
0.50 + 0.8225/ n = 0.55 – 0.4179/ n c 0.55 ^
p
1.2404/ n = 0.05 -0.84 0 Z
1.2404/0.05 = n
n = 24.808
n = 615.43, rounded up to 616
NOTE: To reinforce the concept that all z scores are standardized rescalings obtained by
subtracting the mean and dividing by the standard deviation, the manual uses the “usual” z formula
written to apply to p̂ ’s
z p̂ = ( p̂ – µ p̂ )/σ p̂ .
When the normal approximation to the binomial applies, the p̂ ’s are normally distributed with
µ p̂ = p and σ p̂ = pq/n .
And so the formula for the z statistic may also be written as z p̂ = ( p̂ -p)/ pq/n .
In addition, the manual continues to use the more accurate z0.01 = 2.326 taken from the “large”
row of the t table rather than the 2.33 obtained by reading the z table backwards.
1. There were 1261 + 491 + 384 = 2136 total responses. The sample proportion of yes responses
is p̂ = x/n = 491/2136 = 0.230. The symbol p̂ is used to represent a sample proportion.
2. No. The sample is not necessarily representative of the general population for two reasons.
Since the respondents were Internet users who happened to make contact with the survey, the
sample is a convenience sample. Since the responders chose whether or not to participate, the
sample is a voluntary response sample.
3. The value of the sample proportion is p̂ = x/n = 123/280 = 0.439. Since 0.979 > 0.05, the
observed results could easily occur by chance whenever the null hypothesis is true. There is
not sufficient evidence to reject the hypothesis that the proportion of correct responses is p =
0.50 (i.e., the proportion expected by chance alone) in favor of the claim that p > 0.50.
4. a. The value p is the unknown true proportion of correct guesses in the population; the claim in
Exercise 3 was that p > 0.50. The P-value is the probability of getting results (in the
direction of the alternative hypothesis) as extreme as or more extreme than the observed
results whenever the null hypothesis is true. The P-value for Exercise 3 is 0.979.
b. If the P-value is small (such as less than or equal to 0.05), there is sufficient evidence to
reject the null hypothesis. If the P-value is large (such as greater than 0.05), there is not
sufficient evidence to reject the null hypothesis.
To facilitate understanding of the process, problems 5-6 are worked by hand. The Excel output
obtained following the procedure in the text is given in the box at the right.
5. a. p̂ = x/n = 530/1000 = 0.530
z p̂ = ( p̂ – µ p̂ )/σ p̂
= ( p̂ -p)/ pq/n
= (0.530 – 0.50)/ (0.50)(0.50)/1000
= 0.030/0.0158 = 1.90
b. z = ±zα/2 = ±z0.005 = ±2.575
c. P-value = 2·P(z>1.90) = 2·(1 – 0.9713) = 2·(0.0287) = 0.0574
d. Do not reject Ho; there is not sufficient evidence to reject the claim that the percentage of
all college applications that are submitted online is 50%.
e. No. A hypothesis test will either “reject” or “fail to reject” a claim that a population
parameter is equal to a specified value.
= 0.041/0.01586 = 2.56
P-value = 2·P(z>2.56) = 0.0104
conclusion:
Do not reject Ho; there is not sufficient evidence
to reject the claim that p = 0.75. There is
not sufficient evidence to reject the claim that the proportion of adults who say it is morally
wrong to not report all income is 75%.
10. original claim: p = 0.43
p̂ = x/n = 308/611 = 0.504
Ho: p = 0.43
H1: p ≠ 0.43
α = 0.01
C.V. z = ±zα/2 = ±z0.005 = ±2.575
calculations: 0.005 0.005
z p̂ = ( p̂ – µ p̂ )/σ p̂ ^
0.75 p
= (0.504 – 0.43)/ (0.43)(0.57) / 611 -2.575 0 2.575 z
= 0.074/0.02003 = 3.70
P-value = 2·P(z>3.70) = 0.0002
conclusion:
Reject Ho; there is sufficient evidence to reject
the claim that p = 0.43 and conclude that
p ≠0.43 (in fact, that p > 0.43). There is sufficient evidence to reject the claim that the
proportion of adults who believe they voted for the winning candidate is 43%. Either the
voters are deliberately not telling the truth, or they have faulty memories about how they
actually voted.
11. original claim: p > 1/3
p̂ = x/n = 327/839 = 0.390
Ho: p = 1/3
H1: p > 1/3
α = 0.01
C.V. z = zα = z0.01 = 2.326 0.01
calculations: 0.333 ^
p
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0 2.326 z
conclusion:
Reject Ho; there is sufficient evidence to conclude that p >1/3. There is sufficient evidence
to support the claim that the proportion of tennis challenges that are successful is greater
than 1/3. It appears that the referees are erring on more than 1/3 of the challenged calls –
which is not an enviable record, even if the proportion of challenged calls is quite small.
12. original claim: p < 0.10
p̂ = x/n = 27/300 = 0.090
Ho: p = 0.10
H1: p < 0.10
α = 0.05
C.V. z = -zα = -z0.05 = -1.645
calculations:
0.05
z p̂ = ( p̂ – µ p̂ )/σ p̂
0.10 ^p
= (0.090 – 0.10)/ (0.10)(0.90) / 300 -1.645 0 z
= -0.010/0.01732 = -0.58
P-value = P(z<-0.58) = 0.2810
conclusion:
Do not reject Ho; there is not sufficient evidence
to conclude that p < 0.10. There is not
sufficient evidence to support the claim that the proportion of test results that are incorrect
is less than 10%. No; the test appears to have too high of an error rate to be considered
reliable for most purposes.
13. original claim: p > 0.06
p̂ = x/n = 72/724 = 0.099
Ho: p = 0.06
H1: p > 0.06
α = 0.05
C.V. z = zα = z0.05 = 1.645
calculations:
0.05
z p̂ = ( p̂ – µ p̂ )/σ p̂
0.06 ^
p
= (0.099 – 0.06)/ (0.06)(0.94) / 724 0 1.645 z
= 0.039/0.008826 = 4.47
P-value = P(z>4.47) = 0.0001
conclusion:
Reject Ho; there is not sufficient evidence to
conclude that p > 0.06. There is sufficient
evidence to support the claim that the
proportion of Tamiflu recipients who experience nausea is greater than the 6% rate
experienced by those who took the placebo. Yes; nausea does appear to be a legitimate
concern for recipients of Tamiflu.
conclusion:
Reject Ho; there is sufficient evidence to conclude that p > 0.50. There is sufficient
evidence to support the claim that women with more than 12 years of education have a
proportion of correct predictions greater than 0.50. Yes; these women do appear to have an
ability to correctly predict the gender of their babies.
17. original claim: p < 0.20
p̂ = x/n = 1299/(1299 + 5686) = 1299/6985 = 0.186
Ho: p = 0.20
H1: p < 0.20
α = 0.01
C.V. z = -zα = -z0.01 = -2.326
calculations:
0.01
z p̂ = ( p̂ – µ p̂ )/σ p̂
0.20 ^
p
= (0.186 – 0.20)/ (0.20)(0.80) / 6985 -2.326 0 z
= -0.014/0.004786 = -2.93
P-value = P(z<-2.93) = 0.0017
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p < 0.20. There is sufficient
evidence to support the claim that less than 20% of Michigan gas pumps are inaccurate.
No; from the perspective of the consumer, the rate does not appear to be low enough.
While the point estimate of 0.186 indicates the rate is lower than 20%, it should probably
be about 1/10 of that.
18. original claim: p > 0.50
p̂ = x/n = 172/(172 + 39) = 172/211 = 0.815
Ho: p = 0.50
H1: p > 0.50
α = 0.01
C.V. z = zα = z0.01 = 2.326
calculations: 0.01
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0.50 ^
p
0 2.326 z
= (0.815 – 0.50)/ (0.50)(0.50) / 211
= 0.315/0.3442 = 9.16
P-value = P(z>9.16) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence conclude
that p > 0.50. There is sufficient evidence
to support the claim that with this method the probability of a baby being a boy is greater
than 0.50. Yes; the YSORT method of gender selection appears to work.
Ho: p = 0.20
H1: p < 0.20
α = 0.01
C.V. z = -zα = -z0.01 = -2.326
calculations:
z p̂ = ( p̂ – µ p̂ )/σ p̂
= (0.15 – 0.20)/ (0.20)(0.80) / 5000 0.01
= -0.05/0.005657 = -8.84 0.20 ^
p
-2.326 0 z
P-value = P(z<-8.84) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p < 0.20. There is sufficient
evidence to support the advertiser’s claim that
the proportion of households tuned to 60
Minutes is less than 20%.
22. original claim: p > 0.25
use x = (0.30)(30) = 9
p̂ = x/n = 9/30 = 0.300
Ho: p = 0.25
H1: p > 0.25
α = 0.05 [assumed]
C.V. z = zα = z0.05 = 1.645 0.05
calculations: ^
0.25 p
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0 1.645 z
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that p = 0.50. There is
not sufficient evidence to reject the claim that the proportion of executives who say the
most common interview mistake is failure to know the company is 50%. The important
lesson is that a person going for a job interview should prepare by learning about the
company at which he is applying.
24. original claim: p < 0.27
use x = (0.183)(785) = 143.66, rounded to 144
p̂ = x/n = 144/785 = 0.183
Ho: p = 0.27
H1: p < 0.27
α = 0.01
C.V. z = -zα = -z0.01 = -2.326 0.01
calculations: ^
0.27 p
z p̂ = ( p̂ – µ p̂ )/σ p̂ -2.326 0 z
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0.75 ^
p
-1.96 0 1.96 z
= (0.73 – 0.75)/ (0.75)(0.25) / 3011
= -0.02/0.007891 = -2.53
P-value = 2·P(z<-2.53) = 0.0114
conclusion:
Reject Ho; there is sufficient evidence to reject
the claim that p = 0.75 and conclude that
p ≠ 0.75 (in fact, that p < 0.75). There is sufficient evidence to reject the claim that the
proportion of adults who use the Internet is ¾. While the difference between 0.73 and 0.75
may be of little practical significance, in the interest of accuracy the reporter should not
write that ¾ of all adults use the Internet.
26. NOTE: The value for x is not given. In truth, any 293 ≤ x ≤ 307 rounds to the given
p̂ = x/1501 = 20%. For want of more precise information, use p̂ = 0.20 in the hand
calculations and x = 300 for the Excel input.
original claim: p < 0.25
p̂ = x/n = x/1501 = 0.20
Ho: p = 0.25
H1: p < 0.25
α = 0.01
C.V. z = -zα = -z0.01 = -2.326
0.01
calculations:
0.25 ^p
z p̂ = ( p̂ – µ p̂ )/σ p̂ -2.326 0 z
= (0.20 – 0.25)/ (0.25)(0.75) /1501
= -0.05/0.01118 = -4.47
P-value = P(z<-4.47) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p < 0.25. There is sufficient evidence to support the claim that the
proportion of persons who believe there is not solid evidence for global warming is less
than 25%. One possible consequence of too many people incorrectly believing that there is
not solid evidence of global warming is that no personal or political action will be taken to
deal with the problem in a timely manner.
27. NOTE: The problem as stated is not possible. No whole number x yields x/59 = 43%, as
25/59 = 42.4% and 26/59 = 44.1%. Use the closest possible value, x = 25.
original claim: p = 0.50
p̂ = x/n = 25/59 = 0.424
Ho: p = 0.50
H1: p ≠ 0.50
α = 0.05
C.V. z = ±zα/2 = ±z0.025 = ±1.96
calculations:
0.025 0.025
z p̂ = ( p̂ – µ p̂ )/σ p̂
0.50 ^
p
= (0.424 – 0.50)/ (0.50)(0.50) / 59 -1.96 0 1.96 z
= -0.076/0.06509 = -1.17
P-value = 2·P(z<-1.17) = 2·(0.1210) = 0.2420
conclusion:
Do not reject Ho; there is not sufficient evidence
to reject the claim that p = 0.50. There is not
sufficient evidence to reject the claim that women with 12 years of education or less
have no ability to predict the gender of their babies. Conclude that the guesses of these
women do not differ significantly from the results of random guesses.
28. NOTE: The value for x is not given. In truth, any 335 ≤ x ≤ 343 rounds to the given
p̂ = x/870 = 39%. For want of more precise information, use p̂ = 0.39 in the hand
calculations and x = 339 for the Excel input.
original claim: p < 0.791
p̂ = x/n = x/870 = 0.39
Ho: p = 0.791
H1: p < 0.791
α = 0.01
C.V. z = -zα = -z0.01 = -2.326
calculations: 0.01
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0.791 ^
p
-2.326 0 z
= (0.39 – 0.791)/ (0.791)(0.209) / 870
= -0.401/0.01378 = -29.09
P-value = P(z<-29.09) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p < 0.791. There is sufficient
evidence to support the claim that the selection process results in under representation of
Americans of Mexican ancestry. The jury selection process appears to be unfair – although
the under representation could be caused by persons opting out and not by a bias in the
selection process.
29. NOTE: The value for x is not given. In truth, any 15,720 ≤ x ≤ 16,336 rounds to the given
p̂ = x/61,647 = 26%. For want of more precise information, use p̂ = 0.26 in the hand
calculations and x = 16028 for the Excel input.
original claim: p > 0.25
p̂ = x/n = x/61,647 = 0.26
Ho: p = 0.25
H1: p > 0.25
α = 0.05
C.V. z = zα = z0.05 = 1.645
0.05
calculations:
^
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0.25 p
0 1.645 z
= (0.26 – 0.25)/ (0.25)(0.75) / 61,647
= 0.01/0.001744 = 5.73
P-value = P(z>5.73) = 1 – 0.9999 = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p > 0.25. There is sufficient evidence to support the claim that the
proportion of employees who say that bosses scream at employees is greater than 25%. If
the survey was done with a voluntary response sample, then the sample is not necessarily
representative of the population and the above conclusion may not be valid.
30. NOTE: The value for x is not given. In truth, any 13,555 ≤ x ≤ 13,768 rounds to the given
p̂ = x/21,346 = 64%. For want of more precise information, use p̂ = 0.64 in the hand
calculations and x = 13661 for the Excel input.
original claim: p > 0.50
p̂ = x/n = x/21,346 = 0.64
Ho: p = 0.50
H1: p > 0.50
α = 0.01
C.V. z = zα = z0.01 = 2.326 0.01
calculations:
0.50 ^
p
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0 2.326 z
= (0.64 – 0.50)/ (0.50)(0.50) / 21,346
= 0.14/0.003422 = 40.91
P-value = P(z>40.91) = 0.0001
conclusion:
Reject Ho; there is not sufficient evidence to
conclude that p > 0.50. There is sufficient evidence to support the claim that most people
believe that the Loch Ness monster exists. If the survey was done with a voluntary
response sample, then the sample is not necessarily representative of the population and the
above conclusion may not be valid.
31. NOTE: The value for x is not given. In truth, any 426 ≤ x ≤ 432 rounds to the given
p̂ = x/703 = 61%. For want of more precise information, use p̂ = 0.61 in the hand
calculations and x = 429 for the Excel input.
original claim: p > 0.50
p̂ = x/n = x/703 = 0.61
Ho: p = 0.50
H1: p > 0.50
α = 0.05
C.V. z = zα = z0.05 = 1.645
calculations: 0.05
z p̂ = ( p̂ – µ p̂ )/σ p̂ 0.50 ^
p
0 1.645 z
= (0.61 – 0.50)/ (0.50)(0.50) / 703
= 0.11/0.01886 = 5.83
P-value = P(z>5.83) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to
conclude that p > 0.50. There is sufficient
evidence to support the claim that most (more than 50%) workers get their jobs through
networking. The results suggests that when it comes to getting a job, who you know may
be just as important as what you know.
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that p = 0.50. There is
not sufficient evidence to reject the claim that the students were selected from a population
in which the percentage of males is equal to 50%.
35. There are 54 total bears, 35 of which are males.
original claim: p = 0.50
p̂ = x/n = 35/54 = 0.648
Ho: p = 0.50
H1: p ≠ 0.50
α = 0.05
C.V. z = ±zα/2 = ±z0.025 = ±1.96 0.025 0.025
calculations: 0.50 ^
p
z p̂ = ( p̂ – µ p̂ )/σ p̂ -1.96 0 1.96 z
= -0.207/0.08409 = -2.46
P-value = 2·P(z<-2.46) = 0.0138
conclusion:
Do not reject Ho; there is not sufficient evidence
to reject the claim that p = 0.55. There is
not sufficient evidence to reject the claim that the movies were selected from a population
in which the percentage of R-rated movies is 55%.
= 0.019/0.00949 = 2.00
P-value = 2·P(z>2.00) = 0.0456
conclusion:
Reject Ho; there is sufficient evidence to reject
the claim that p = 0.10 and conclude that
p ≠ 0.10 (in fact, that p > 0.10). There is sufficient evidence to reject the claim that the
proportion of zeros equals 0.10.
a. As seen above, the traditional method leads to rejection of the claim that p=0.10 because the
calculated z=2.00 is greater than the critical value of 1.96.
b. As seen above, the P-value method leads to rejection of the claim that p=0.10 because the
calculated P-value=0.0456 is less than the level of significance of 0.05.
c. α = 0.05 and p̂ = x/n = 119/1000 = 0.1190
ˆˆ
p̂ ± zα/2 pq/n
0.1190 ± 1.96 (0.1190)(0.8810)/1000
0.1190 ± 0.0201
0.0989 < p < 0.139
Since 0.10 is inside the confidence interval, p=0.10 is a reasonable claim that should not be
rejected.
d. The traditional method and the P-value method are mathematically equivalent and will
always agree. As seen by this example, the confidence interval method does not always lead
to the same conclusion as the other two methods.
= -0.10/0.0424 = -2.36
P-value = 2·P(z<-2.36) = 2·(0.0091) = 0.0182
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that p = 0.10. There is
not sufficient evidence to reject the claim that the proportion of M&M’s that are blue is
equal to 0.10.
40. The test of hypothesis is given below.
Ho: p = 0.40
H1: p < 0.40
α = 0.05 [assumed]
C.V. z = -zα = -z0.05 = -1.645
The c corresponding to z = -1.645 is found
by solving zc = (c-p)/ pq/n for c as follows:
0.05
c = p + zc· pq/n
c 0.40 ^
p
= 0.40 + (-1.645) (.40)(.60)/50 -1.645 0 z
= 0.40 – (1.645)(0.0693)
= 0.2860
a. The power calculations are given below and illustrated by the figure at the right.
The z corresponding to c= 0.2860 is
found as follows:
zc = (c-p)/ pq/n
= (0.2860-0.25)/ (.25)(.75)/50
= 0.0360/0.0612
= 0.59
power = P(rejecting Ho|Ho is false)
= P( p̂ <0.2860|p=0.25) ^
0.25 0.2860 p
= P(z<0.59) 0 0.59 z
= 0.7224
b. β = P(type II error)
= P( not rejecting Ho|Ho is false)
= 1 – P(rejecting Ho|Ho is false)
= 1 – 0.7224
= 0.2776
c. No test of hypothesis can lead the researcher to the right conclusion 100% of the time. If Ho
is true, there is a 95% chance Ho will (correctly) not be rejected; if Ho is false and p=0.25 is
true, there is a 72% chance that Ho will (correctly) be rejected. These are reasonable
probabilities of success, and the test is reasonably effective.
1. In order to use the methods of this section to test the claim that µ = 1.5,
a. the sample must be a simple random sample.
b. the population standard deviation σ must be known.
c. the population distribution must be approximately normal (since n ≤ 30).
2. We consider the normality requirement to be satisfied if there are no outliers and the histogram
of the sample data is approximately bell-shaped. More formally, a normal quantile plot could
be used to determine whether the sample data are approximately normally distributed.
3. A one-tailed test at the 0.01 level of significance rejects the null hypothesis if the sample
statistic falls into the extreme 1% of the sampling distribution in the appropriate tail. The
corresponding (two-sided) confidence interval test that places 1% each tail would be a 98%
confidence interval.
4. The mean weight loss of 2.1 pounds is statistically significant because it led to the conclusion
that the true mean weight change is less than zero – i.e., negative, indicating a loss of weight.
The mean weight loss of 2.1 pounds (after one year on the program) is not practically
significant because most people would not consider such a small weight loss worth the effort.
5. original claim: µ = 5 cm
Ho: µ = 5 cm
H1: µ ≠ 5 cm
α = 0.05 [assumed]
C.V. z = ±zα/2 = ±z0.025 = ±1.96
calculations:
z x = (x - µ x )/σ x
0.025 0.025
= 1.34 [TI-83/84] _
P-value = 0.1797 [TI-83/84] 5 x
-1.96 0 1.96 z
conclusion:
Do not reject Ho; there is not sufficient
evidence to reject the claim that µ = 5. There is not
sufficient evidence to reject the claim that women have a mean wrist breadth equal to 5 cm.
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ > 65. There is sufficient evidence
to support the claim that the sample is from a population with a mean greater than 65 mph.
21. The mean length of the n = 50 screws is x = 0.74682 inches.
original claim: µ = 0.75 inches
Ho: µ = 0.75 inches
H1: µ ≠ 0.75 inches
α = 0.05
C.V. z = ±zα/2 = ±z0.025 = ±1.96
calculations:
z x = (x - µ x )/σ x 0.025 0.025
_
= (0.74682 – 0.75)/(0.012/ 50 ) 0.75 x
= -0.00318/0.001697 = -1.87 -1.96 0 1.96 z
P-value = 2·P(z<-1.87) = 0.061
conclusion:
Do not reject Ho; there is not sufficient evidence
to reject the claim that µ = 0.75. There is
not sufficient evidence to reject the claim that the screws have a mean length of ¾ in. as
indicated on the label. Yes; the lengths appear to be consistent with the label.
22. The mean length of the n = 40 readings is x = 123.6625 volts.
original claim: µ = 120 volts
Ho: µ = 120 volts
H1: µ ≠ 120 volts
α = 0.01
C.V. z = ±zα/2 = ±z0.005 = ±2.575
calculations:
z x = (x - µ x )/σ x
0.005 0.005
= (123.6625 – 120)/(0.24/ 40 ) _
120 x
= 3.6625/0.03795 = 96.52 -2.575 0 2.575 z
P-value = 2·P(z>96.52) = 0.0001
conclusion:
Reject Ho; there is sufficient evidence to reject
the claim that µ = 120 and conclude that
µ ≠ 120 (in fact, that µ > 120). There is
sufficient evidence to reject the claim that the mean home voltage amount is 120 volts.
23. a. In general, the power of a test is the probability of correctly rejecting the hull hypothesis
when some specified alternative is actually correct. In this context, a power of 0.2296
indicates that when the true population mean is actually 170 lbs, the test has a 22.96%
probability of correctly rejecting the null hypothesis that the mean is 166.3 lbs in order to
conclude that the mean is greater than 166.3 lbs.
b. In general, β = P(type II error)
= P( not rejecting Ho|Ho is false)
= 1 – P(rejecting Ho|Ho is false)
In this context, β = 1 – P(rejecting Ho:µ = 166.3| µ = 170))
= 1 – 0.2296
= 0.7704
1. Yes; since n ≤ 30, the sample should be from a population that is approximately normally
distributed. We consider the normality requirement to be satisfied for such data if there are no
outliers and the histogram of the sample data is approximately bell-shaped. More formally, a
normal quantile plot could be used to determine whether the sample data are approximately
normally distributed.
2. In statistics, df denotes the degrees of freedom. In general, the degrees of freedom give the
number of pieces of information that are free to vary without changing the mathematical
constraints of the problem. When using a t test with n = 20 sample values to test a claim about
the mean of a population, df = 19.
3. A t test is a hypothesis test that uses the Student t distribution, typically to perform a test about
µ when the true value of σ is not know. The letter t is used because that was the notation
chosen by William Gosset (1786-1937), the developer of the distribution. He wrote under the
pseudonym Student, and t is most prominent letter in “student” (considering that “s” was
already being used to denote the standard deviation).
4. This section and the preceding section involve making inferences about an unknown
population mean. This section, which does not require knowing the value of the population
standard deviation, is the more realistic. Since the standard deviation measures the spread
around the mean, it is not realistic to have a situation in which the standard deviation is known
and the mean is not.
5. Use t. When σ is unknown and the x’s are approximately normally distributed, use t.
6. Neither the z nor the t applies. When σ is unknown and the x’s are not normally distributed,
sample sizes n≤30 cannot be used with the techniques in this chapter.
7. Neither the z nor the t applies. When σ is unknown and the x’s are not normally distributed,
sample sizes n≤30 cannot be used with the techniques in this chapter.
8. Use t. When σ is unknown and the x’s are not normally distributed, use t may be used
whenever n>30.
NOTE: Exercises 9–12 may be worked as follows.
table: find the correct df row in Table A-3, and see what values surround the given t.
Excel: use TDIST(t-value, df, # of tails) – but be advised that the TDIST function requires a
positive value for t, and probabilities associated with negative values of t must be determined by
appealing to the symmetry of the t distribution.
For the remainder of this manual, exact P-values for the t distribution USING THE
UNROUNDED VALUES of t will be determined as described above and as in exercises #9-#12.
Critical values for df values not in Table A-3 will be determined using TINV(p,df) – where p
denotes the two-tailed α necessary to produce the desired tail probability value(s). And as in the
previous chapter, a subscript (df) may be used to identify which t distribution to use in the tables.
9. P-value = P(t24 > 0.430)
table for area in one tail: [0.430 < 1.318] P- value > 0.10
TDIST(0.430,24,1) = 0.3355
10. P-value = 2·P(t14 > 1.495)
table for area in two tails: [1.761 > 1.495 > 1.345] 0.10 < P- value < 0.20
TDIST(1.495,14,2) = 0.1571
11. P-value = 2·P(t8 < -1.905)
table for area in two tails: [-2.306 < -1.905 < -1.860] 0.05 < P- value < 0.10
TDIST(1.905,8,2) = 0.0932
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ > 210. There is sufficient
evidence to support that claim that the sample is from a population with a mean greater
than 210 seconds. The results suggest that the advice in the manual is not good advice.
14. original claim: µ < 5.4 cells/microliter
Ho: µ = 5.4 cells/microliter
H1: µ < 5.4 cells/microliter
α = 0.01 and df = 49
C.V. t = -tα = -t0.01 = -TINV(0.02,49) = -2.405
calculations:
t x = (x - µ)/s x 0.01
= -2.226 [TI-83/84+] _
5.4 x
P-value = 0.0153 [TI-83/84+] -2.405 0 t 49
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that µ < 5.4. There is not
sufficient evidence to support the claim that the sample is from a population with a mean
red blood cell count less than 5.4 cells/microliter If 5.4 is the upper limit for the range of
normal individuals, and if the mean of the sample group is not significantly below the
upper limit for an individual, then that group may have unusually high red cell counts. The
µ ± 2σ guideline for normal values suggests that the population mean is approximately
5.4 – 2(0.54) = 4.32, and the sample mean of 5.23 is considerably higher than that.
15. original claim: µ < 21.1 mg
Ho: µ = 21.1 mg
H1: µ < 21.1 mg
α = 0.05 and df = 24
C.V. t = -tα = -t0.05 = -1.711
calculations:
t x = (x - µ)/s x 0.05
_
= (13.2 – 21.1)/(3.7/ 25 ) 21.1 x
-1.711 0 t 24
= -7.9/0.7400 = -10.676
P-value = P(t24 < -10.676) = TDIST(10.676,24,1) = 6.752E-11 = 0.00000000007
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ < 21.1. There is sufficient
evidence to support the claim that filtered 100 mm cigarettes have a mean tar amount less
than 21.1 mg. The results suggest that filters are effective in reducing the amount of tar.
= -1.9/3.3500 = -0.567
P-value = 2·P(t99 < -0.567) = TDIST(0.567,99,2) = 0.5719
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that µ = 49.5. There is
not sufficient evidence to reject the claim that the sample is from a population with a mean
equal to 49.5 cents. The results suggest that the cents portion of credit card charges could
be uniformly distributed from 0 to 99, although there are many other possibilities that
would produce a similar sample mean.
= 0.3/0.2444
= 1.227
P-value = P(t80 > 1.227) = TDIST(1.227,80,1) = 0.1117
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that µ > 4.5. There is not
sufficient evidence to support the claim that the mean time for all college students to earn
their bachelor’s degrees is greater than 4.5 years.
20. original claim: µ = 120 volts
Ho: µ = 120 volts
H1: µ ≠ 120 volts
α = 0.05 and df = 39
C.V. t = ±tα/2 = ±t0.025 = ±2.023
calculations:
t x = (x - µ)/s x 0.025 0.025
_
= (123.59 – 120)/(0.31/ 40 ) 120 x
-2.023 0 2.023 t 39
= 3.59/0.04902
= 73.242
P-value = 2·P(t39 > 73.242) = TDIST(73.242,39,2) = 2.2E-43 ≈ 0
conclusion:
Reject Ho; there is sufficient evidence to reject the claim that µ = 120 and conclude that
µ ≠ 120 (in fact, that µ > 120). There is sufficient evidence to reject the claim that the
mean voltage amount is 120 volts.
21. original claim: µ = 49.5 cents
Ho: µ = 49.5 cents
H1: µ < 49.5 cents
α = 0.01 and df = 99
C.V. t = -tα = -t0.01 = -TINV(0.02,99) = -2.365
calculations:
0.01
t x = (x - µ)/s x _
49.5 x
= (23.8 – 49.5)/(32.0/ 100 ) -2.365 0 t 99
= -25.7/3.2000
= -8.031
P-value = 2·P(t99 < -8.031) = TDIST(8.031,99,2) = 2.1E-12 ≈ 0
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ < 49.5. There is sufficient
evidence to conclude that the cents portion of all checks has a mean that is less than 49.5
cents. The results suggest that the cents portions of checks are not uniformly distributed
from 0 to 99 cents.
= -0.22/0.1909 = -1.152
P-value = 2·P(t17 < -1.152) = TDIST(1.152,31,2) = 0.2580
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that µ = 8.00. There is
not sufficient evidence to reject the claim that all cars have a mean greenhouse gas
emission of 8.00 tons.
24. original claim: µ > 63.6 inches
Ho: µ = 63.6 inches
H1: µ > 63.6 inches
α = 0.01 and df = 8
C.V. t = tα = t0.01 = 2.896
calculations: 0.01
t x = (x - µ)/s x _
63.6 x
0 2.896 t8
= (70.0 – 63.6)/(1.5/ 9 )
= 6.4/0.5000 = 12.800
P-value = P(t8 > 12.800) = TDIST(12.800,8,1) = 6.5E-7 = 0.0000007
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ > 63.6. There is sufficient
evidence to support the claim that supermodels have a mean height that is greater than the
63.6 inches of the general population of women. Yes; assuming that the heights of
supermodels are approximately normally distributed around their mean, the test and the
conclusion are valid.
= -296.33/111.3429
= -2.661
P-value = P(t5 < -2.661) = 0.0224
conclusion:
Do not reject Ho; there is not sufficient evidence
to conclude that µ < 1000. There is not
sufficient evidence to support the claim that the population mean is less than 1000 hic. No;
since one of the sample values is 1210, there is proof that not all of the child booster seats
meet the specified requirement.
26. preliminary values: n = 10, Σx = 533, Σx2 = 30,615, x = 53.3, s2 = 245.12, s = 15.66
original claim: µ > 48.0 words
Ho: µ = 48.0 words
H1: µ > 48.0 words
α = 0.05 and df = 9
C.V. t = tα = t0.05 = 1.833
calculations:
0.05
t x = (x - µ)/s x _
48.0 x
= (53.3 – 48.0)/(15.66/ 10 ) 0 1.833 t 9
= 5.3/4.9510
= 1.070
P-value = P(t9 > 1.070) = 0.1561
conclusion:
Do not reject Ho; there is not sufficient evidence
to conclude that µ > 48.0. There is not sufficient evidence to support the claim that the
mean number of words per page is greater than 48.0. There is not enough evidence to
support the claim that there are more than 70,000 words in the dictionary.
27. preliminary: n = 5, Σx = 32061, Σx2 = 220,431,831, x = 6412.2, s2 = 3,712,581.7, s = 1926.80
original claim: µ = $5000
Ho: µ = $5000
H1: µ ≠ $5000
α = 0.05 and df = 4
C.V. t = ±tα/2 = ±t0.025 = ±2.776
calculations:
t x = (x - µ)/s x 0.025 0.025
_
= (6412.2 – 5000)/(1926.80/ 5 ) 5000 x
-2.776 0 2.776 t4
= 1412.2/861.6927
= 1.639
= -3.732
P-value = P(t9 < -3.732) = 0.0023
conclusion:
Reject Ho; there is sufficient evidence to
conclude that µ < 20.16. There is sufficient
evidence to support the claim that the recent winners are from a population with BMI less
than 20.16, which was the BMI for winners in the 1920’s and 1930’s. Yes; recent winners
appear to be significantly different from those in the earlier years.
29. summary statistics from Excel: n = 50, x = 0.74682, s = 0.012322
original claim: µ = 0.75 inches
Ho: µ = 0.75 inches
H1: µ ≠ 0.75 inches
α = 0.05 and df = 49
C.V. t = ±tα/2 = ±t0.025 = ±TINV(0.05,49) = ±2.010
calculations:
0.025 0.025
t x = (x - µ)/s x _
0.75 x
= (0.74682 – 0.75)/(0.012322/ 50 ) -2.010 0 2.010 t 49
= -0.00318/0.001743
= -1.825
P-value = 2·P(t49 < -1.825) = 0.0741
conclusion:
Do not reject Ho; there is not sufficient evidence
to reject the claim that µ = 0.75. There is not sufficient evidence to reject the claim that the
screws have a mean length of ¾ in. as indicated on the label. Yes; the lengths appear to be
consistent with the label.
37. NOTE: Throughout this exercise the manual uses the s = 26.32 given in Example 1. Going
back to the original Data Set 1 in Appendix B to find a more precise value will give slightly
different answers (beginning at the third decimal place) in parts (a) and (b).
The test of hypothesis is given below
and illustrated by the figure at the right.
Ho: µ = 166.3 lbs
H1: µ > 166.3 lbs
α = 0.05 and df = 39
C.V. t = tα = t0.05 = 1.685
The c corresponding to t = 1.685 is
found by solving tc = (c-µ)/(s/ n )
0.05
for c as follows. _
166.3 c x
c = µ + tc·(s/ n ) 0 1.685 t
39
= 166.3 + 1.685(26.32/ 40 )
= 166.3 + 7.01
= 173.31
a. The power calculations are given below
and illustrated by the figure at the right.
The t corresponding to c= 173.31 is
tc = (c-µ)/(s/ n )
= (173.31 – 180)/(26.32/ 40 )
= -6.688/4.1616 = -1.607
power = P(rejecting Ho|Ho is false)
= P( x >173.31|µ=180) _
173.31 180 x
= P(t39>-1.607) -1.607 0 t 39
= TDIST(1.607,39,1)
= 0.9419
b. β = P(type II error) = P( not rejecting Ho|Ho is false)
= 1 – P(rejecting Ho|Ho is false)
= 1 – 0.9419
= 0.0481
The test is very effective in recognizing that the mean is greater than 166.3 lbs when the
mean is actually equal to 180 lbs, for it will do so 94.19% of the time and fail to do so only
4.81% of the time.
1. The normality requirement for an hypothesis test about a standard deviation is stricter than the
normality requirement for an hypothesis test about a mean. Deviations from normality that
were tolerated when testing a claim about a mean may be serious enough to invalidate the
results when testing a claim about a standard deviation.
2. A one-tailed test at the 0.01 level of significance rejects the null hypothesis if the sample
statistic falls into the extreme 1% of the sampling distribution in the appropriate tail. The
corresponding (two-sided) confidence interval test that places 1% each tail would be a 98%
confidence interval. When testing claims about a standard deviation, the confidence interval
method gives the same results as tests using the traditional method or the P-value method.
3. No. Unlike tests and confidence intervals involving the mean, which do not require normality
when n>30, test and confidence intervals involving standard deviations require approximate
normality for all sample sizes.
4. Yes. The claim that the variance is equal to 36 mm2 and the claim that the standard deviation
is equal to 6 mm are equivalent claims, and their corresponding tests are equivalent.
5. a. test statistic: χ2 = (n-1)s2/σ2 = (24)(645)2/(696)2 = 20.612
b. critical values for α = 0.05 and df = 24: χ2 = χ 1-2 α / 2 = χ 0.975
2
= 12.401
χ2 = χ α2 /2 = χ 0.025
2
= 39.364
c. P-value limits: 15.659 < 20.612
P-value > 0.20
P-value exact: 2*(1 – CHIDIST(20.612,24)) = 0.6770
d. conclusion: Do not reject Ho; there is not sufficient to conclude that σ ≠ 696
= (39)(11.3)2/(10)2 39 54.572 χ2
39
= 49.799
P-value = CHIDIST(49.799,39) = 0.1152
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that σ > 10. There is not
sufficient evidence to support the claim that the pulse rates of men have a standard
deviation greater than 10 beats/minute.
11. original claim: σ ≠ 3.2 mg
Ho: σ = 3.2 mg
H1: σ ≠ 3.2 mg
α = 0.05 and df = 24
C.V. χ2 = χ 1-α/2
2
= χ 0.975
2
= 12.401
χ2 = χ α/2
2
= χ 0.025
2
= 39.364 0.025 0.025
calculations: 39.364 χ2
12.401 24
χ2 = (n-1)s2/σ2 24
= (24)(3.7)2/(3.2)2 = 32.086
P-value = 2*CHIDIST(32.086,24) = 0.2498
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that σ ≠ 3.2. There is not
sufficient evidence to support the claim that the tar content of such cigarettes has a
standard deviation different from 3.2 mg.
12. original claim: σ = 6.0 lbs
Ho: σ = 6.0 lbs
H1: σ ≠ 6.0 lbs
α = 0.01 and df = 39
CV χ2 = χ 1-α/2
2
= χ 0.995
2
= CHIINV(0.995,39) = 19.996
0.005 0.005
χ2 = χ α/2
2
= χ 0.005
2
= CHIINV(0.005,39) = 65.476
calculations: 19.996 39 65.476 χ2 39
χ2 = (n-1)s2/σ2
= (39)(4.9)2/(6.0)2 = 26.011
P-value = 2*(1 – CHIDIST(26.011,39)) = 0.1101
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that σ =6.0. There is
not sufficient evidence to reject the claim that the weight losses from this diet have a
standard deviation of 6.0 lbs.
13. There are n=9 heights.
original claim: σ < 2.5 inches
Ho: σ = 2.5 inches
H1: σ < 2.5 inches
α = 0.05 and df = 8
C.V. χ2 = χ 1-α
2
= χ 0.95
2
= 2.733
0.05
calculations:
2 2 2
χ = (n-1)s /σ 2.733 8 χ2 8
2 2
= (8)(1.5) /(2.5) = 2.880
P-value = 1 – CHIDIST(2.880,8) = 0.0583
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that σ < 2.5. There is not
sufficient evidence to support the claim that the heights of supermodels have a standard
deviation that is less than 2.5 inches, the value for the general female population. The
conclusion indicates that while supermodels may be more homogeneous than the general
population, that cannot be concluded from this sample of size n=9.
14. original claim: σ < 14.1
Ho: σ = 14.1
H1: σ < 14.1
α = 0.01 and df = 26
C.V. χ2 = χ 1-α
2
= χ 0.99
2
= 12.198
calculations: 0.01
χ2 = (n-1)s2/σ2
= (26)(9.3)2/(14.1)2 = 11.311 12.198 26 χ226
= (39)(12.5)2/(10)2 = 60.9375
P-value = 2·CHIDIST(60.9375,39) = 0.0277
conclusion:
Reject Ho; there is sufficient evidence to reject the claim that σ = 10 conclude that σ ≠ 10
(in fact, that σ > 10). There is sufficient evidence to reject the claim that the pulse rates of
women have a standard deviation equal to 10 beats/minute.
16. original claim: σ = 28.866 cents
Ho: σ = 28.866 cents
H1: σ ≠ 28.866 cents
α = 0.01 and df = 99
CV χ2 = χ 1-α/2
2
= χ 0.995
2
= CHIINV(0.995,99) = 66.510
χ2 = χ α/2
2
= χ 0.005
2
= CHIINV(0.005,99) =138.987 0.005 0.005
χ2 = (n-1)s2/σ2
= (99)(33.5)2/(28.866)2 = 133.337
P-value = 2*CHIDIST(133.337,99) = 0.0244
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that σ = 28.866. There
is not sufficient evidence to reject the claim that the sample is from a population with a
standard deviation equal to 28.866 cents. If the amounts from 0 to 99 cents follow a
uniform distribution, then the requirement of a normal distribution is not satisfied and the
conclusion is not necessarily valid.
17. preliminary values: n = 10, Σx = 187.6, Σx2 = 3532.04, x = 18.76, s2 = 1.4071, s = 1.1862
original claim: σ = 1.34
Ho: σ = 1.34
H1: σ ≠ 1.34
α = 0.01 and df = 9
C.V. χ2 = χ 1-α/2
2
= χ 0.995
2
= 1.735
χ2 = χ α/2
2
= χ 0.005
2
= 23.589
0.005 0.005
calculations:
χ2 = (n-1)s2/σ2 1.735 9 23.589 χ2 9
= (9)(1.1862)2/(1.34)2 = 7.053
P-value = 2·P( χ 92 < 7.053) = 0.7368
conclusion:
Do not reject Ho; there is not sufficient
evidence to reject the claim that σ = 1.34.
There is not sufficient evidence to reject
the claim that recent Miss America winners are from a population whose BMI values have
a standard deviation of 1.34. No; recent winners do not appear to have MBI variation that
is different from that of the 1920’s and 1930’s.
18. preliminary values: n = 16, Σx = 58.80, Σx2 = 222.5710, x = 3.675, s2 = 0.4321, s = 0.6573
original claim: σ = 0.470 kg
Ho: σ = 0.470 kg
H1: σ ≠ 0.470 kg
α = 0.05 [assumed] and df = 15
C.V. χ2 = χ 1-α/2
2
= χ 0.975
2
= 6.262
χ2 = χ α/2
2
= χ 0.025
2
= 27.488 0.025 0.025
calculations:
2.262 15 27.488 χ2
χ2 = (n-1)s2/σ2 15
= (15)(.6573)2/(0.470)2 = 29.339
P-value = 2·P( χ 15 2
> 29.339) = 0.0291
conclusion:
Reject Ho; there is sufficient evidence to
reject the claim that σ = 0.470 and
conclude that σ ≠ 0.470 (in fact, that
σ > 0.470. There is sufficient evidence to reject that claim that male babies born to
mothers on the supplement are from a population whose birth weights have a standard
deviation equal to 0.470 kg. Yes; the supplement appears to increase the variation among
birth weights.
19. preliminary values: n = 12, Σx = -196, Σx2 = 33452, x = -16.3, s2 = 2750.06, s = 52.441
original claim: σ > 32.2 ft
Ho: σ = 32.2 ft
H1: σ > 32.2 ft
α = 0.05 and df = 11
C.V. χ2 = χ α2 = χ 0.05
2
= 19.675
calculations:
χ2 = (n-1)s2/σ2 0.05
2 2
= (11)(52.441) /(32.2) = 29.176 11 19.675 χ2
11
P-value = P( χ 11
2
> 29.176) = 0.0021
conclusion:
Reject Ho; there is sufficient evidence
to conclude that σ > 32.2. There is
sufficient evidence to conclude that the production method has errors with a standard
deviation greater than 32.2 ft. A greater standard deviation means that the new altimeters
will have more spread in the errors (i.e., generally larger errors) than in the past – which
make the new method worse than the old method. Yes; the company should take action.
20. preliminary values: n = 16, Σx = 4154, Σx2 = 1,123,116, x = 259.6, s2 = 2975.58, s = 54.549
original claim: σ < 60 sec
Ho: σ = 60 sec
H1: σ < 60 sec
α = 0.05 and df = 15
C.V. χ2 = χ 1-α
2
= χ 0.95
2
= 7.261
calculations:
χ2 = (n-1)s2/σ2 0.05
2 2
= (15)(54.549) /(60) 7.261 15 χ2 15
= 12.398
P-value = P( χ 15
2
< 12.398 = 0.3513
conclusion:
Do not reject Ho; there is not sufficient
evidence to conclude that σ < 60.
There is not sufficient evidence to
support the claim that the songs are from a population with a standard deviation less than
one minute.
1. For testing a claim that µ > 0, that must be the alternative hypothesis – and the null hypothesis
must be that µ = 0. A P-value of 0.0091 indicates that the probability of obtaining these results
when the null hypothesis is true is only 0.0091. Such a P-value calls for the rejection of the
null hypothesis and support of the original claim that the mean rainfall amount is greater than 0
inches. The memory aid indicates that a small (e.g., less than 0.05) P-value calls for the
rejection of the null hypothesis and that a large (i.e., greater than 0,05) P-value indicates there
is not sufficient evidence to reject the null hypothesis.
2. Yes; since P-value = 0.0041 < 0.05, there is statistical significance. No; since 0.019 inches is
such a trivial amount (and is it even possible to measure heights of males that accurately?),
there is not practical significance.
3. A voluntary response sample is one in which the respondents themselves decide whether or not
to participate. Since such samples tend to include mostly those with a special interest in the
topic and/or strong feelings about the topic, they are not necessarily representative of the
general population and should not be used to make inferences.
4. A procedure is robust against departures from normality if it works well (i.e., is “correct” 1-α
of the time) even when the sample data are from a population that does not follow a normal
distribution. Yes; the t test of a population mean is robust against departures from normality.
No; the χ2 test of a population standard deviation is not robust against departures from
normality.
1. Since the claim that the proportion of males is greater than 0.5 does not contain the equality, it
must be the alternative hypothesis. The null hypothesis is that the proportion of males is equal
to 0.5. In symbolic form, Ho: p = 0.5 and H1: p > 0.5.
2. The t distribution is appropriate for the indicated test. The others are not appropriate for the
following reasons.
normal – used to perform a test about µ when σ is known..
chi-square – used to perform a test about σ.
binomial – used to perform an exact test about p.
uniform – not an appropriate sampling distribution for any sample statistic.
3. The chi-square distribution is appropriate for the indicated test. The others are not appropriate
for the following reasons.
normal – used to perform a test about µ when σ is known..
t – used to perform a test about µ when σ is unknown.
binomial – used to perform an exact test about p.
uniform – not an appropriate sampling distribution for any sample statistic.
4. True; the null hypothesis is either rejected or not rejected, but it cannot be supported.
5. Table A-2: P-value = 2·P(z>1.50) = 2·(1 – 0.9332) = 2·(0668) = 0.1336
Excel: P-value = 2*(1 – NORMSDIST(1.50)) = 0.1336
Review Exercises
= -0.01011/0.01313 = -0.77
P-value = P(z<-0.77) = 0.2206
conclusion:
Do not reject Ho; there is not sufficient evidence
to conclude that p < 0.25. There is not
sufficient evidence to support the claim that less than ¼ of such adults smoke.
= -1.068
P-value = P(t9 < -1.068) = TDIST(1.068,31,1) = 0.1469
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that µ < 3700. There is not
sufficient evidence to support the claim that the mean weight of cars is less than 3700 lbs.
While the mean weight of cars might be a factor in determining long run wear and tear, the
most relevant factor for determining the required strength for highways is the weight of the
heaviest vehicle that will be using the highway.
= -14.7/1.6706 = -8.799
P-value = P(t9 < -8.799) = TDIST(8.799,174,1) = 6.6E-16
conclusion:
Reject Ho; there is sufficient evidence to conclude that µ < 281.8. There is sufficient
evidence to support the claim that the thinner cans have a mean axial load that is less than
281.8 lbs. Yes; even though the thinner cans are not as strong as the thicker cans currently
in use, they apparently can easily withstand the necessary pressure of 158 to 165 lbs.
7. original claim: µ = 74
Ho: µ = 74
H1: µ ≠ 74
α = 0.05
C.V. z = ±zα/2 = ±z0.025 = ±1.96
calculations:
z x = (x - µ x )/σ x 0.025 0.025
= (74.4 – 74)/(12.5/ 100 ) _
74 x
= 0.4/1.25 -1.96 0 1.96 z
= 0.32
P-value = 2·P(z>0.32) = 2*(1 – NORMSINV(0.32) = 0.7490
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that µ = 74. There
is not sufficient evidence to reject the claim that the sample comes from a population with
a mean equal to 74. Yes; based on these results, the calculator’s random number generator
appears to be working correctly.
8. original claim: µ = 74
Ho: µ = 74
H1: µ ≠ 74
α = 0.05 and df = 99
C.V. t = ±tα/2 = ±t0.025 = ±TINV(0.05,99) = ±1.984
calculations:
t x = (x - µ)/s x
0.025 0.025
= (74.4 – 74)/(11.7/ 100 ) _
74 x
= 0.4/1.17 -1.984 0 1.984 t 99
= 0.342
P-value = 2·P(t99 > 0.342) = TDIST(0.342,99,2) = 0.7332
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that µ = 74. There
is not sufficient evidence to reject the claim that the sample comes from a population with
a mean equal to 74. Yes; based on these results, the calculator’s random number generator
appears to be working correctly.
χ2 = (n-1)s2/σ2
= (99)(11.7)2/(12.5)2
= 86.734
P-value = 2*(1 – CHIDIST(86.734,99)) = 0.3883
conclusion:
Do not reject Ho; there is not sufficient evidence to reject the claim that σ = 12.5. There
is not sufficient evidence to reject the claim that the sample comes from a population with
a standard deviation equal to 12.5.
31
= 28.856
P-value = 1 – CHI(28.856,31) = 0.4233
conclusion:
Do not reject Ho; there is not sufficient evidence to conclude that σ < 520. There is not
sufficient evidence to support the claim that the standard deviation of the weights of cars is
less than 520 lbs.
1. scores in numerical order: 10.54 10.75 10.82 10.93 10.94 10.97 11.06 11.07 11.08
summary statistics: n = 9, Σx = 98.16, Σx2 = 1070.8508
a. x = (Σx)/n = (98.16)/9 = 10.907 seconds Mean 10.90667
b. x = 10.94 seconds Standard Error 0.059208
c. s2 = [n(Σx2) – (Σx)2]/[n(n-1)] Median 10.94
= [9(1070.8508) – (98.16)2]/[9(8)] Mode #N/A
= 2.2716/72 = 0.03155 Standard Deviation 0.177623
s = 0.178 seconds Sample Variance 0.03155
d. s2 = 0.03155, rounded to 0.032 seconds2 Kurtosis 1.057734
e. R = 11.08 – 10.54 = 0.54 seconds Skewness -1.15628
Range 0.54
2. a. Ratio, since differences are meaningful and there is a
Minimum 10.54
meaningful zero.
Maximum 11.08
b. Continuous, since time can be any value on a continuum.
Sum 98.16
c. No; the times were not selected at random from some
Count 9
population, but they were determined by being the
winning times for 9 consecutive Olympics.
d. The sample statistics in Exercise 1 do not consider the chronological order of the data.
e. A time series plot will reveal tendencies over time, while still giving a general idea (by
looking from the vertical axis) of the central tendency and variation of the data.
3. σ unknown, use t with df=8
α = 0.05, tdf,α/2 = t8,0.025 = 2.306
x ± tα/2·s/ n
10.907 ± 2.306(0.178)/ 9
10.907 ± 0.136
10.770 < µ < 11.043 (seconds)
No; this result cannot be used to estimate winning times in the future because there is a pattern
of decreasing times and no fixed population mean.
4. original claim: µ < 11 seconds
Ho: µ = 11 seconds
H1: µ < 11 seconds
α = 0.05 and df = 8
C.V. t = -tα = -t0.05 = -1.860
calculations:
t x = (x - µ)/s x
0.05
= (10.907 - 11)/(0.178/ 9 ) _
11 x
= -0.0933/0.05921 = -1.576 -1.860 0 t 8
5. Some parts of this exercise are better answered from the x f f·x .
frequency distribution representation at the right, where the 0.70 1 0.70
x column gives the (approximate) midpoint value of each bar. 0.72 0 0.00
a. Yes; the histogram is approximately bell-shaped. 0.74 1 0.74
0.76 2 1.52
b. Σf = 100 0.78 5 3.90
c. Using the first two midpoints, 0.72 – 0.70 = 0.02 grams 0.80 8 6.40
d. Using the formula for summarized grouped data, 0.82 8 6.56
x ≈ (Σf·x)/(Σf) = 85.70/100 = 0.857 grams 0.84 13 10.92
e. No. All we can tell, for example, is that there are 30 values 0.86 30 25.80
0.88 15 13.20
in the class centered at 0.86 grams, but we cannot determine 0.90 3 2.70
their individual weights. 0.92 4 3.68
0.94 6 5.64
0.96 1 0.96
0.98 2 1.96
1.00 0 0.00
1.02 1 1.02
100 85.70
6. Using a vertical scale that does not start at zero exaggerates differences between the classes
and gives a distorted impression of the data. The visual impression of the first class (x=1), for
example, is that it contains about twice as many data points as the second class (x=2) – and in
reality those values are 16 and 13.
7. The requested frequency distribution is given by the first two x f f·x
columns of the table at the right. Using the formula for grouped 1 16 16
data, 2 13 26
x = (Σf·x)/(Σf) 3 22 66
4 21 84
= 347/100 5 13 65
= 3.47 6 15 90
100 347
8. For one test of hypothesis at the α = 0.05 level, the probability of making an error by rejecting
a true null hypothesis is given by P(E) = 0.05. For two such independent tests,
P(E1 and E2) = P(E1)·P(E2|E1) = (0.05)(0.05) = 0.0025.
9. For each part, use the formula z = (x – µ)/σ
a. P(x<700) = P(z<-1.18) = 0.1190
Excel: NORMDIST(700,739,33,1) = 0.1186
b. P(x>750) = P(z>0.33) = 1 – 0.6293 = 0.3707 = 37.07%
Excel: 1 – NORMDIST(750,739,33,1) = 0.3694 = 36.94%
c. The x ’s are normally distributed with
µ x = µ = 739 and
σ x = σ/ n = 33/ 50 = 4.6669.
P( x <730) = P(z<-1.93) = 0.0268
Excel: NORMDIST(730,739,33/SQRT(50),1) = 0.0269
d. From Table A-2, the z score with 0.9000 [closest entry is 0.8997]below it is 1.28.
x = µ + zσ = 739 + (1.28)(33) = 781.24 mm
Excel: NORMINV(0.90,739,33) = 781.29
10. The sample heights appear to come from a normal distribution. The histogram shows a
distribution that is approximately bell-shaped, and the normal quantile plot indicates that the
data points approximate a straight line.
Normal Quantile Plot
5
2.0
1.0
Frequency
z score
0
2
-1.0
1
-2.0
0
740 760 780 800 820 840 700 750 800 850 900
Sitting Eye Heights (mm) Sitting Eye Heights (mm)
NOTE: Other representations of the data may give different impressions, and so it is always
a good idea not to base a conclusion about normality on a
single figure. The reasonable representation of the data height (mm) frequency
given at the right, for example, suggests that the data may be 725 – 749 2
750 – 774 6
bimodal – perhaps representing two different subpopulations. 775 – 799 3
While frequency distributions and histograms are sensitive to 800 – 824 7
how the classes are defined, the normal quantile plot is not 825 – 846 2 .
subject to such arbitrary decisions. 20