88 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

Answers to Odd-Numbered “Practice

Using Statistical Software (IBM SPSS)”
Questions: Full Version

Chapter 1
1. There are 27,534 cases in the dataset.
3. a. The value that is recorded for the variable “Marital status of respondent”
[MARSTAT] is “1”.
b. The attribute that is associated with the value “1” is “Married.” The person
is married.
5. a. The value that is recorded for the variable “Total household income - 2012”
[INCMHSD] is “9”.
b. The attribute that is associated with the value “9” is “$50,000 to $59,999.”
The combined income of all of the people in that person’s household in
2012 was $50,000 to $59,999.
7. a. The variable VBR_10 captures whether or not people voted in the last fed-
eral election.
b. The variable has the following attributes: “Yes,” “No,” “Valid skip,” “Don’t
know,” “Refusal,” and “Not stated.” The value “1” is associated with the at-
tribute “Yes,” the value “2” is associated with the attribute “No,” the value
“6” is associated with the attribute “Valid skip,” the value “7” is associated
with the attribute “Don’t know,” the value “8” is associated with the at-
tribute “Refusal,” and the value “9” is associated with the attribute “Not
stated.”
c. The attributes/values that are designated as missing are: “Valid skip”
(6), “Don’t know” (7), “Refusal” (8), and “Not stated” (9).
d. It is a dichotomous variable.
9. a. The variable WHW_120C captures the number of hours that people work
at their jobs each week.
b. The values on this variable represent quantities.
c. The attributes/values that are designated as missing are: “Valid skip”
(999.6), “Don’t know” (999.7), “Refusal” (999.8), and “Not stated” (999.9).
d. It is a ratio-level variable.

noa15214_answers_SPSS_088-138.indd 88 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 89

Chapter 2
1.

a. Overall, 12,613 men answered the survey; 45.8 per cent of survey
respondents are men.
b. Overall, 14,921 women answered the survey; 54.2 per cent of survey
respondents are women.
3.

a. Among the people who gave a valid answer to the question, 55.3 per cent
say that Canadians share the value of human rights to a great extent.
b. Among the people who gave a valid answer to the question, 38.9 per cent
say that Canadians share the value of human rights to a moderate extent.
c. Among the people who gave a valid answer to the question, 94.2 per cent
say that Canadians share the value of human rights to either a great or a
moderate extent.

noa15214_answers_SPSS_088-138.indd 89 01/11/18 09:23 PM

90 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

5.

a. Among the people who gave a valid answer to the question, 62.9 per cent
say that they are very proud to be Canadian.
b. The percentage from question 4(a) is different than the percentage from
question 5(a) because a different denominator is used to calculate it. In
question 4(a) people who had “No opinion” and who are “Not a Canadian
citizen” are included in the denominator used to calculate the percent-
age, whereas in question 5(a) people with these two attributes are excluded
from the denominator used to calculate the percentage.
7. a.

noa15214_answers_SPSS_088-138.indd 90 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 91

b. This bar graph displays the information more effectively than the pie
graph from question 6, because the bar graph makes it easy to see that the
percentage of people who gave each answer gets smaller as you move from
the highest category to the lowest category.
9. a.

b. This graph shows that people born outside Canada are more likely than
people born in Canada to say that Canadians share the value of human
rights to a great extent. In contrast, people born in Canada are more likely
than people born outside Canada to say that Canadians share the value of
human rights to a moderate or a small extent.

Chapter 3
1.

The mode is “Spouse only.” This shows that the largest number of people report
living with only their spouse.

noa15214_answers_SPSS_088-138.indd 91 01/11/18 09:23 PM

92 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

3.

The mode is “$30,000 to $39,999.” This shows that it is most common for people
to report having an annual income of $30,000 to $39,999.
The median is also “$30,000 to $39,999.” This shows that half of people report
an annual income of $30,000 to $39,999 or less, and half of people report an annual
income of $30,000 to $39,999 or more. (Or, half of people report an annual income
of $39,999 or less and half of people report an annual income of $30,000 or more.)

5.

The interquartile range of this variable is from “$15,000 to $19,999” (the attrib-
ute associated with the value “5”) to “$60,000 to $79,999” (the attribute associ-
ated with the value “10”). This shows that the middle 50 per cent of people have
an annual income between $15,000 and $79,999. In other words, the 50 per
cent of people in the middle of the income distribution have annual incomes
between $15,000 and $79,999.

noa15214_answers_SPSS_088-138.indd 92 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 93

7.

a. The twenty-fifth percentile is 5 hours, the fiftieth percentile is 10 hours,

and the seventy-fifth percentile is 20 hours.
b. The interquartile range is from 5 to 20 hours. Excluding outliers, the range
is from 0 to 42 hours.

9.

noa15214_answers_SPSS_088-138.indd 93 01/11/18 09:23 PM

94 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

The new, recoded variable shows that 9.0 per cent of people have no annual
personal income. About one in five people (20.4 per cent) have an annual
income between $1 and $19,999. A slightly higher percentage of people—25.1
per cent—have an annual income between $20,000 and $39,999. Another 19.2
per cent of people have an annual income from $40,000 to $59,999 and the
remaining 26.2 per cent have higher annual incomes. When the variable is
recoded this way, it shows that the most common annual personal income is
$20,000 to $39,999; this is also the median annual personal income.

Chapter 4
1.

The mean is 37.80. This shows that, on average, people work at their jobs for
37.80 paid hours per week (when people who do not work for pay are excluded).
The standard deviation is 14.38. Since the standard deviation is relatively small
compared to the mean, it shows that the distribution of the “Number of paid
hours worked per week” variable isn’t widely spread out.

noa15214_answers_SPSS_088-138.indd 94 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 95

3.

a. The median is 40.00. This shows that half of people work 40 paid hours
or more at their jobs per week, and half of people work 40 paid hours or
fewer at their jobs per week. The median is slightly higher than the mean,
but not by much.
b. The kurtosis is 0.71. This shows that the distribution of the “Number of
paid hours worked per week” variable is moderately leptokurtic. In other
words, it’s slightly more peaked than a normal distribution.
c. The skew is −0.11. This shows that the distribution of the “Number of paid
hours worked per week” variable is very slightly left-skewed compared to a
normal distribution. In other words, it has a tail that extends very slightly
to the left, because some people work a very low number of paid hours at
their jobs each week. But, since the skew is between −0.5 and +0.5, the
distribution of this variable is considered approximately normal in terms
of its skew.

5.

The mean is 6.48. This shows that, on average, people have 6.48 close friends.
The standard deviation is 9.55. Since the standard deviation is substantially
larger than the mean, it shows that the “Number of close friends” variable has
a wide spread.

noa15214_answers_SPSS_088-138.indd 95 01/11/18 09:23 PM

96 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

7.

a. The mean is 6.48. The median and the mode are both 5.00. The mean
shows that, on average, people have 6.48 close friends. The median shows
that half of people have 5 close friends or more, and half of people have 5
close friends or fewer. The mode shows that it is most common for people
to have 5 close friends.
b. The standard deviation is 9.55 and the range is 200. Both of these statistics
show that the “Number of close friends” variable is widely dispersed.
c. The kurtosis is 141.21. This shows that the distribution of the “Number of
close friends” variable is highly leptokurtic; in other words, the distribu-
tion is very peaked compared to a normal distribution.
d. The skew is 9.26. It shows that the distribution of the “Number of close
friends” variable is highly right-skewed compared to a normal distribu-
tion. In other words, the distribution has a tail that extends far to the right
because some people report very high numbers of close friends.

Chapter 5
1. a.

noa15214_answers_SPSS_088-138.indd 96 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 97

b. <No output>
c.

3. a.

b.

noa15214_answers_SPSS_088-138.indd 97 01/11/18 09:23 PM

98 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

c.

d. The frequency distributions in (a) and (c) have the same total number of
cases, which is the same as the number of cases in the dataset, whereas the
frequency distribution in (b) has 29 million cases that represent the Can-
adian population. The frequency distributions in (b) and (c) have the same
percentages of men and women, whereas the frequency distribution in (a)
has different percentages of men and women.

Chapter 6
1.

a. The mean is 6.54. In the sample, on average, people have 6.54 close friends.
b. The 95 per cent confidence interval for the mean is 6.43 to 6.65. In the popu-
lation, the average number of close friends is likely between 6.43 and 6.65.

noa15214_answers_SPSS_088-138.indd 98 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 99

3.

The round dots for men and women show the mean number of close friends
for men and women in the sample, as reported in the statistics from question
2: 6.90 and 6.18, respectively. The whiskers below and above each dot show the
distance between the lower bound and the upper bound of the 95 per cent con-
fidence interval for the mean for men and women, as reported in the statistics
from question 2. So, for men, the whiskers extend from 6.74 to 7.07, and for
women the whiskers extend from 6.05 to 6.32.

5.

noa15214_answers_SPSS_088-138.indd 99 01/11/18 09:23 PM

100 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

The round dots for each age group show the mean number of close friends
for people in each age group in the sample, as reported in the statistics from
question 4. For example, the lowest average number of close friends is among
people aged 45 to 54 and among people aged 55 to 64 (5.99 for both groups),
and the highest average number of close friends is among people aged 15 to 24
(7.60). The whiskers below and above each dot show the distance between the
lower bound and the upper bound of the 95 per cent confidence interval for the
mean for each age group, as reported in the statistics from question 4. So, for
people aged 15 to 24, the whiskers extend from 7.38 to 7.81, whereas for people
aged 75 and over, the whiskers extend from 5.72 to 6.66.

7.

Overall, 35.5 per cent of people volunteered in the past 12 months. The mean of
the recoded variable is 0.355. When 35.5 per cent is converted into a proportion
(by dividing it by 100), the result is 0.355, which corresponds to the mean of
the recoded variable.

noa15214_answers_SPSS_088-138.indd 100 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 101

9.

a. The mean shows that the proportion of men in the sample who volun-
teered in the past 12 months is 0.3248, or 32.48 per cent. Similarly, the pro-
portion of women in the sample who volunteered in the past 12 months is
0.3853, or 38.53 per cent.
b. The 95 per cent confidence interval for the mean shows that the propor-
tion of men in the population who volunteered in the past 12 months is
likely to be between 0.3169 and 0.3326 (or 31.69 and 33.26 per cent). Sim-
ilarly, the 95 per cent confidence interval for the mean shows that the
proportion of women in the population who volunteered in the past 12
months is likely to be between 0.3772 and 0.3934 (or 37.72 and 39.34 per
cent). Since these 95 confidence intervals do not overlap, in the population
it is likely that a higher proportion of women than men volunteered in the
past 12 months.

noa15214_answers_SPSS_088-138.indd 101 01/11/18 09:23 PM

102 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

Chapter 7
1.

a. In the sample, the difference between the mean number of close friends
for men and for women is 0.72, or less than 1 close friend. This isn’t a very
large difference.
b. Cohen’s d is 0.08. Since Cohen’s d is less than 0.1, the effect size is very
small. (In other words, the relationship between people’s sex/gender and
their number of close friends is very weak.)
3. a. The output from all three procedures display the mean, the number of
cases, and the standard deviation for each group. The output from the
Explore procedure and the Independent Samples T-Test procedure both
display the standard error of the mean for each group.
b. The output from the Explore procedure displays additional statistics about
the distribution of the variable within each group, and includes the median,
the variance, the minimum, the maximum, the range, the interquartile
range, the skew, and the kurtosis.
   The output from the Means procedure is the only one to show the mean,
the standard deviation, and the number of cases for the sample overall (not
divided by group).
   The output from the Independent Samples T-test procedure shows
the t-statistic, the degrees of freedom, and the significance test associated
with those results (for both versions of the t-test). It also shows the differ-
ence between the means, the standard error of the difference, and the 95
per cent confidence interval for the difference between means.
c. Yes, the answers to the two questions correspond. In question 2(c) in this
chapter, the t-test results show that there is likely a difference between the
two group means in the population. In question 2(b) in Chapter 6, the 95
per cent confidence intervals for the mean do not overlap, suggesting that
there is likely a difference between the two group means in the population.

noa15214_answers_SPSS_088-138.indd 102 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 103

5.

a. A non-directional research hypothesis for this relationship is this: “In the

population, there is a relationship between people’s general orientation
toward trusting people and their number of close friends.” (An alternative
non-directional hypothesis is this: “In the population, those who think
that most people can be trusted have a different number of close friends,
on average, than those who think you cannot be too careful in dealing
with people.”)
b. The null hypothesis associated with this research hypothesis is this: “In the
population, there is no relationship between people’s general orientation
toward trusting people and their number of close friends.” (An alterna-
tive null hypothesis is this: “In the population, those who think that most
people can be trusted have the same number of close friends, on average,
as those who think you cannot be too careful in dealing with people.”)
c. The t-statistic of 15.86 has a p-value that is less than 0.05, so the null hy-
pothesis is rejected. In the population, there is likely a relationship between
people’s general orientation toward trusting people and their number of
close friends.
7.

noa15214_answers_SPSS_088-138.indd 103 01/11/18 09:23 PM

104 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

a. A non-directional research hypothesis for this relationship is this: “In

the population, there is a relationship between having a post-secondary
education (or not) and the number of close friends that people have.” (An
alternative non-directional hypothesis is this: “In the population, people
with a post-secondary education have a different number of close friends,
on average, than people who do not have a post-secondary education.”)
b. The null hypothesis associated with this research hypothesis is this: “In
the population, there is no relationship between having a post-secondary
education (or not) and the number of close friends that people have.”
(An alternative null hypothesis is this: “In the population, people with a
post-secondary education have the same number of close friends, on aver-
age, as people who do not have a post-secondary education.”)
c. The t-statistic of 5.43 has a p-value that is less than 0.05 so the null hy-
pothesis is rejected. In the population, there is likely a relationship be-
tween having a post-secondary education (or not) and the number of close
friends that people have.

Chapter 8
1.

a. In the sample, the average number of close friends among people with less
than a high school education and among people with only a high school
education is the same: 6.19. Among people with a post-secondary diploma,
the average number of close friends is 0.25 higher. People with a university
degree have one more close friend (0.99), on average, than people with
only a high school education or less than a high school education.

noa15214_answers_SPSS_088-138.indd 104 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 105

b. For people with the three lowest levels of education (less than high school,
high school only, and post-secondary diploma), the 95 per cent confidence
intervals for the mean number of close friends all overlap. As a result, we
cannot be confident that, in the population, the average number of close
friends among people with each of these three levels of education is differ-
ent. However, among people with a university degree, the lower bound of
the 95 per cent confidence interval for the mean is higher than the upper
bound of the 95 per cent confidence interval for the mean for the other
three levels of education. Thus, it is likely that, in the population, people
with a university degree have more close friends, on average, than people
with lower levels of education.

3.

a. The post-hoc tests shows that the average number of close friends
among people with a university degree is significantly different than
the average number of close friends among people with lower levels
of education. The p-values of all of the significance tests that include
the university degree group are less than 0.05, whereas the others are
greater than 0.05.
b. Yes, the answers to the two questions correspond. In (a) the post-hoc tests
show that, in the population, the average number of close friends among
people with a university degree is likely different than the average number
of close friends among people with lower levels of education. In question

noa15214_answers_SPSS_088-138.indd 105 01/11/18 09:23 PM

106 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

1(b), the 95 per cent confidence intervals for the mean suggest that, in the
population, people with a university degree have more close friends, on
average, than people with lower levels of education.

5.

a. In the overall sample, people spend an average of 38.04 hours per week
doing paid work. In the overall population, the average number of hours
that people spend doing paid work each week is likely between 37.84 and
38.24.
b. In the sample, people with less than a high school education spend an
average of 33.12 hours per week doing paid work. People with only a high
school education spend an average of 36.82 hours per week doing paid
work—3.70 hours more than the average of people with less than a high
school education. People with a post-secondary diploma spend an aver-
age of 38.93 hours per week doing paid work, compared to 39.60 hours
among people with a university degree; there is only a small difference
(0.67 hours) between these two averages in the sample.
c. For the four educational groups, none of the 95 per cent confidence inter-
vals for the mean number of hours spent doing paid work overlap. Thus,
it is likely that, in the population, people with less than a high school edu-
cation spend the lowest number of hours doing paid work each week, on
average. People with only a high school education spend slightly more
hours doing paid work per week, on average, followed by people with
a post-secondary diploma. People with a university degree are likely to

noa15214_answers_SPSS_088-138.indd 106 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 107

spend the highest number of hours doing paid work each week, on aver-
age. Thus, in the population, people with higher levels of education are
likely to spend more time doing paid work each week, on average, than
people with lower levels of education.

7.

a. The post-hoc tests show that the average number of hours spent doing paid
work each week among people in each educational group is significantly
different than the average number of hours spent doing paid work each
week in every other educational group. The p-values of all of the signifi-
cance tests are less than 0.05.
b. Yes, the answers to the two questions correspond. In (a), the post-hoc
tests show that, in the population, people in each educational group likely
spend a different number of hours doing paid work each week, on aver-
age, than people in every other educational group. In question 5(c), the
95 per cent confidence intervals for the mean suggest that, in the popula-
tion, people with higher levels of education are likely to spend more time
doing paid work each week, on average, than people with lower levels of
education.

noa15214_answers_SPSS_088-138.indd 107 01/11/18 09:23 PM

108 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

Chapter 9
1.

In the sample, 79.9 per cent of women donated money or goods in the past 12
months, compared to only 70.4 per cent of men, a difference of 9.5 percentage points.
3.

a. A non-directional research hypothesis for this relationship is this: “In the

population, people’s sex/gender is related to whether or not they donated
money or goods in the past 12 months.”

noa15214_answers_SPSS_088-138.indd 108 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 109

b. The null hypothesis associated with this research hypothesis is this: “There
is no relationship in the population between people’s sex/gender and
whether or not they donated money or goods in the past 12 months.”
c. The chi-square statistic of 335.44 has a p-value that is less than 0.05 so the
null hypothesis is rejected. In the population, there is likely a relationship
between people’s sex/gender and whether or not they donated money or
goods in the past 12 months.

5.

a. The relationship between sex/gender and making a donation does not

change that much when visible minority status is taken into account.
Among people in the sample who are visible minorities, 64.1 per cent of
men donated money or goods in the past 12 months, compared to 75.0 per
cent of women, for a difference of 10.9 percentage points. Among people

noa15214_answers_SPSS_088-138.indd 109 01/11/18 09:23 PM

110 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

in the sample who are not visible minorities, 71.7 per cent of men donated
goods or services in the past 12 months, compared to 80.9 per cent of
women, for a difference of 9.2 percentage points.
b. In the zero-order relationship shown in question 1, there is a 9.5 percent-
age point difference between men and women. The two partial relation-
ships are fairly similar to the zero-order relationship: 10.9 and 9.2. So, this
is a weak example of specification: the relationship between sex/gender
and donating money or goods in the past 12 months is slightly stronger
among people who are visible minorities than among people who are not
visible minorities. (Alternatively, you could argue that this is an example
of replication, since the difference between the partial relationships and
the zero-order relationship is relatively small.)

7.

The gamma of 0.611 shows that the error in predicting how likely people are to
vote in the next federal election can be reduced by 61.1 per cent if we know how
interested they are in politics.

9.

noa15214_answers_SPSS_088-138.indd 110 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 111

Cramér’s V is 0.111. Since Cramér’s V is close to 0.1, the effect size is small. (In
other words, the relationship between people’s sex/gender and whether or not
they donated money or goods in the past 12 months is weak.)

Chapter 10
1.

a. The Pearson’s correlation coefficient for the relationship between these

two variables is 0.26. Since the Pearson’s correlation coefficient is less than
0.3, the relationship between the number of relatives people feel close to
and their number of close friends is weak.
b. The direction of the relationship is positive. In the context of these two
variables, this indicates that people who feel close to more relatives tend
to have more close friends. Conversely, people who feel close to fewer rela-
tives tend to have fewer close friends.

3.

a. The Spearman’s correlation coefficient for the relationship between these

two variables is 0.36. Since the Spearman’s correlation coefficient is between
0.3 and 0.5, the rank-order relationship between the number of relatives
people feel close to and their number of close friends is weak to moderate.

noa15214_answers_SPSS_088-138.indd 111 01/11/18 09:23 PM

112 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

b. The direction of the rank-order relationship is positive. In the context of

these two variables, this indicates that people who ranked higher in terms
of the number of relatives they feel close to tend to be ranked higher in
terms of their number of close friends. Conversely, people who ranked
lower in terms of the number of relatives they feel close to tend to be
ranked lower in terms of their number of close friends.

5. a.

b. The scatterplot shows that many people have low values on both variables;
that is, they have relatively few relatives that they feel close to and relatively
few close friends. But it’s hard to discern the overall pattern of the relation-
ship between the two variables since there are many overlapping cases.

7.

noa15214_answers_SPSS_088-138.indd 112 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 113

a. When sex/gender is taken into account, the relationship between the

number of relatives people feel close to and the number of close friends
they have becomes weaker among men and stronger among women.
Among men, the correlation between the number of relatives people feel
close to and their number of close friends is 0.23. Among women, the cor-
relation between the same two variables is 0.30.
b. The zero-order correlation in question 1 is 0.26. The partial correlation
among men (0.23) is smaller than the zero-order correlation, and the par-
tial correlation among women (0.30) is larger than the zero-order correla-
tion. So, this is an example of specification. The relationship between the
number of relatives people feel close to and the number of close friends
they have is weaker among men and stronger among women.

Chapter 11
1. a–b. <No output>
c.

noa15214_answers_SPSS_088-138.indd 113 01/11/18 09:23 PM

114 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

3.

The 95 per cent confidence interval for the slope coefficient shows that, in the
population, for each additional group that people participated in during the
past 12 months, the regression line capturing the relationship with the number
of close friends is predicted to rise between 0.78 and 0.89. In other words, the

noa15214_answers_SPSS_088-138.indd 114 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 115

general pattern of the relationship between the variables suggests that partici-
pating in one additional group during the past year is associated with an in-
crease in the number of close friends that is between 0.78 and 0.89.
The 95 per cent confidence interval for the constant coefficient shows that,
in the population, the regression line capturing the relationship between the
number of groups people participated in during the past 12 months and the
number of close friends is predicted to cross the vertical axis between 5.11
and 5.39. In other words, the general pattern of the relationship between the
variables suggests that participating in no groups in the past year is associated
with having a number of close friends that is between 5.11 and 5.39.

5.

a. The constant coefficient shows that those who met no new people in the
past month are predicted to have 5.90 close friends. The slope coefficient
shows that each additional new person met in the past month is associ-
ated with having an additional 0.16 close friends; in other words, meeting

noa15214_answers_SPSS_088-138.indd 115 01/11/18 09:23 PM

116 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

approximately six new people in the past month is associated with having
one additional close friend.
b. The t-statistic of 27.22 has a p-value that is less than 0.05, so there is likely
a relationship in the population between the number of new people met in
the past month and the number of close friends that people have.

7.

The 95 per cent confidence interval for the slope coefficient shows that, in the
population, for each additional new person met in the past month, the regres-
sion line capturing the relationship with the number of close friends is pre-
dicted to rise between 0.15 and 0.17. In other words, the general pattern of
the relationship between the variables suggests that meeting an additional new
person in the past month is associated with an increase in the number of close
friends that is between 0.15 and 0.17.
The 95 per cent confidence interval for the constant coefficient shows that,
in the population, the regression line capturing the relationship between the
number of new people met in the past month and the number of close friends
is predicted to cross the vertical axis between 5.79 and 6.02. In other words, the
general pattern of the relationship between the variables suggests that meeting
no new people in the past month is associated with having a number of close
friends that is between 5.79 and 6.02.

9.

noa15214_answers_SPSS_088-138.indd 116 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 117

a. The “Model Summary” and the “ANOVA” tables are identical to the regres-
sion produced in question 5. As well, the independent variable rows of
the “Coefficients” table are identical (the rows for SCP_110 and SCP_110_
CENTRED). The only part of the output that is different is the constant
row of the “Coefficients” table.
b. The constant coefficient shows that those who met four new people in the
past month are predicted to have 6.55 close friends. Meeting one additional
new person in the past month is associated with a 0.16 increase in people’s
number of close friends; similarly, meeting one less new person in the past
month is associated with a 0.16 decrease in people’s number of close friends.

Chapter 12
1.

noa15214_answers_SPSS_088-138.indd 117 01/11/18 09:23 PM

118 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

a. The constant coefficient shows that those who participated in no groups

in the past 12 months, and who met no new people in the past month, are
predicted to have 4.89 close friends.
b. The unstandardized slope coefficient of the “Number of groups” variable
shows that each additional group that people participated in during the
past 12 months is associated with a 0.72 increase in their number of close
friends, controlling for the number of new people they met in the past
month.
The unstandardized slope coefficient of the “Number of new people
met” variable shows that each additional new person met in the past
month is associated with a 0.14 increase in the number of close friends
that people have, controlling for the number of groups they participated
in during the past 12 months.
c. The standardized slope coefficients show that the number of groups people
participated in during the past 12 months has a stronger relationship with
the dependent variable (“Number of close friends”) than the number of
new people they met in the past month.

3.

noa15214_answers_SPSS_088-138.indd 118 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 119

5.

noa15214_answers_SPSS_088-138.indd 119 01/11/18 09:23 PM

120 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

a. The constant coefficient shows that men who did not participate in any
groups in the past 12 months, and who did not meet any new people in the
past month, are predicted to have 5.21 close friends.
b. The unstandardized slope coefficient of the “Number of groups” variable
shows that each additional group that people participated in during the
past 12 months is associated with a 0.72 increase in their number of close
friends, controlling for sex/gender and the number of new people they met
in the past month.
   The unstandardized slope coefficient of the “Number of new people
met” variable shows that each additional new person met in the past
month is associated with a 0.13 increase in the number of close friends that
people have, controlling for sex/gender and the number of groups they
participated in during the past 12 months.
   The unstandardized slope coefficient of the “Women” dummy vari-
able shows that women are predicted to have 0.63 fewer close friends than
men, controlling for the number of new people they met in the past month
and the number of groups they participated in during the past 12 months.
c. The standardized slope coefficients show that the number of groups people
participated in during the past 12 months has a stronger relationship with
the dependent variable (“Number of close friends”) than sex/gender or the
number of new people met in the past month.

7.

noa15214_answers_SPSS_088-138.indd 120 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 121

a. The constant coefficient shows that people who are currently in a relation-
ship (married or common-law) are predicted to have 6.40 close friends.
(People who are currently in a relationship have a “0” value on the “Single”
dummy variable and the “Previous relationship” dummy variable.)
b. The unstandardized slope coefficient of the “Previous relationship”
dummy variable shows that people who were previously in a long-term
relationship (widowed, separated, or divorced) are predicted to have 0.73
fewer close friends than people who are currently in a relationship (mar-
ried or common-law).
   The unstandardized slope coefficient of the “Single” dummy variable
shows that people who are single (never married) are predicted to have
0.82 more close friends than people who are currently in a relationship
(married or common-law).

9.

noa15214_answers_SPSS_088-138.indd 121 01/11/18 09:23 PM

122 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

Chapter 13
1. a.

b.

noa15214_answers_SPSS_088-138.indd 122 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 123

3. a-d. <No output>

e.

noa15214_answers_SPSS_088-138.indd 123 01/11/18 09:23 PM

124 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

5.

noa15214_answers_SPSS_088-138.indd 124 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 125

noa15214_answers_SPSS_088-138.indd 125 01/11/18 09:23 PM

126 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

a. The constant coefficient of the second model shows that 45-year-old men
who are not visible minorities, who have only a high school education,
and who are currently in a relationship are predicted to spend 42.35 hours
doing paid work per week.
b. The unstandardized slope coefficient of the “Less than high school”
dummy variable shows that people with less than a high school education
are predicted to spend 4.24 fewer hours doing paid work per week than
people with only a high school education, controlling for age, visible min-
ority status, sex/gender, and relationship status.
   The unstandardized slope coefficient of the “Post-secondary dip-
loma” dummy variable shows that people with a post-secondary diploma
are predicted to spend 1.49 more hours doing paid work per week than
people with only a high school education, controlling for age, visible min-
ority status, sex/gender, and relationship status.
   The unstandardized slope coefficient of the “University degree”
dummy variable shows that people with a university degree are predicted
to spend 2.29 more hours doing paid work per week than people with only
a high school education, controlling for age, visible minority status, sex/
gender, and relationship status.
c. The unstandardized slope coefficient of the “Previous relationship”
dummy variable shows that people who were previously in a long-term
relationship (widowed, separated, or divorced) are predicted to spend 0.67
more hours doing paid work per week than people who are currently in a
relationship (married or common-law), controlling for age, visible minor-
ity status, sex/gender, and highest level of education. But, since the p-value
associated with this slope coefficient is greater than 0.05, we cannot be
confident that, in the population, there is any difference between people
who are currently in a relationship and people who were previously in a
long-term relationship in terms of the number of hours they spend doing
paid work each week.

noa15214_answers_SPSS_088-138.indd 126 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 127

   The unstandardized slope coefficient of the “Single” dummy variable

shows that people who are single (never married) are predicted to spend
5.22 fewer hours doing paid work per week than people who are currently
in a relationship (married or common-law), controlling for age, visible
minority status, sex/gender, and highest level of education.
d. The standardized slope coefficients show that among the independent
variables used in regression, sex/gender has the strongest relationship with
the number of hours spent doing paid work each week.
e. The R 2 shows that, overall, 10.9 per cent of the variation in the number of
hours spent doing paid work each week can be explained by age, visible
minority status, sex/gender, highest level of education, and marital status.
Since the R 2 and the adjusted R 2 are the same, it suggests that all of the
independent variables in this model are good predictors of the dependent
variable.
f. The tolerances and variance inflation factors indicate that there are no col-
linearity problems among the independent variables in this regression: all
of the tolerances are above 0.1 and all of the variance inflation factors are
below 10. The three variables with the lowest tolerances (and thus high-
est variance inflation factors) are the “University degree” dummy vari-
able (0.64), the “Post-secondary diploma” dummy variable (0.65), and the
“Single” dummy variable (0.66).

9. a.

noa15214_answers_SPSS_088-138.indd 127 01/11/18 09:23 PM

128 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

These results show that there is a relationship between the regression residuals
and the dependent variable. Ideally, there should be no relationship between
the regression residuals and the dependent variable. (The only remaining vari-
ation should be random.)
The model systematically over-predicts the number of paid hours
worked for people working fewer than 40 hours per week, and systematically
under-predicts the number of paid hours worked for people working more
than 40 hours per week. This suggests that the model still needs improvement.
Independent variables that help to explain why people work very low and very
high numbers of hours each week should be added.

b.

These results show that there is no relationship between the regression resid-
uals and people’s highest level of education. The median of each box plot is near
0, the ideal value of a residual. The box plots also show that there is slightly
more variation in the residuals—a wider range and interquartile range—for
people who have less than a high school education than for people with higher
levels of education.

Chapter 14
1.

noa15214_answers_SPSS_088-138.indd 128 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 129

3. a.

noa15214_answers_SPSS_088-138.indd 129 01/11/18 09:23 PM

130 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

b.

5.

7. a. <No output>
b.

noa15214_answers_SPSS_088-138.indd 130 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 131

Chapter 15
1.

3.

noa15214_answers_SPSS_088-138.indd 131 01/11/18 09:23 PM

132 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

The 95 per cent confidence interval for the odds ratio of the “Visible minority”
dummy variable shows that, in the population, a regression capturing the re-
lationship between visible minority status and experiencing discrimination is
likely to show that people who are visible minorities have between 87 per cent and
114 per cent higher odds of experiencing discrimination. In other words, the
general pattern of the relationship between the variables suggests that being
a visible minority is associated with odds of experiencing discrimination
that are 87 per cent to 114 per cent higher than for people who are not visible
minorities.

5.

noa15214_answers_SPSS_088-138.indd 132 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 133

The odds ratio of the “Age” variable shows that each one-year increase in age is
associated with 2 per cent lower odds of experiencing discrimination, control-
ling for visible minority status and sex/gender.
The odds ratio of the “Visible minority” dummy variable shows that
people who are visible minorities are predicted to have 76 per cent higher odds

noa15214_answers_SPSS_088-138.indd 133 01/11/18 09:23 PM

134 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

of experiencing discrimination than people who are not visible minorities,

controlling for age and sex/gender.
   The odds ratio of the “Women” dummy variable shows that women are
predicted to have 45 per cent higher odds of experiencing discrimination than
men, controlling for age and visible minority status.

7. a.

b.

c. The standardized slope coefficient of the “Age” variable is −0.832.

   The standardized slope coefficient of the “Visible minority” variable
is 0.454.
   The standardized slope coefficient of the “Women” variable is 0.404.
d. The standardized slope coefficients show that age has a stronger relation-
ship with the dependent variable (“Experienced discrimination”) than
visible minority status or sex/gender.

noa15214_answers_SPSS_088-138.indd 134 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 135

9.

noa15214_answers_SPSS_088-138.indd 135 01/11/18 09:23 PM

136 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

noa15214_answers_SPSS_088-138.indd 136 01/11/18 09:23 PM

 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version 137

a. The odds ratio of the “Christian” dummy variable shows that people who
are Christians are predicted to have 17 per cent lower odds of experiencing
discrimination than people who have no religious affiliation, controlling
for age, visible minority status, and sex/gender.
   The odds ratio of the “Other religion” dummy variable shows
that people who are affiliated with another (non-Christian) religion have
1 per cent higher odds of experiencing discrimination than people who
have no religious affiliation, controlling for age, visible minority status,
and sex/gender. But, since the p-value associated with this odds ratio is
greater than 0.05, we cannot be confident that, in the population, there
is any difference between people with no religious affiliation and people
affiliated with another (non-Christian) religion in terms of their odds of
experiencing discrimination.
b. The odds ratio of the “Age” variable does not change substantially once
religious affiliation is controlled for (0.98 in both models).
   The odds ratio of the “Visible minority” dummy variable becomes
smaller once religious affiliation is controlled for (changing from 1.77
to 1.70). When religious affiliation is not taken into account, people
who are visible minorities are predicted to have 77 per cent higher odds
of experiencing discrimination than people who are not visible min-
orities (controlling for age and sex/gender). Once religious affiliation
is accounted for, people who are visible minorities are only predicted
to have 70 per cent higher odds of experiencing discrimination than
people who are not visible minorities (also controlling for age and
sex/gender).
   In contrast, the odds ratio of the “Women” dummy variable be-
comes slightly larger once religious affiliation is controlled for (changing
from 1.47 to 1.49). When religious affiliation is not taken into account,
women are predicted to have 47 per cent higher odds of experiencing dis-
crimination than men (controlling for age and visible minority status).
Once religious affiliation is accounted for, women are predicted to have

noa15214_answers_SPSS_088-138.indd 137 01/11/18 09:23 PM

138 Answers to Odd-Numbered IBM SPSS Practice Questions: Full Version

49 per cent higher odds of experiencing discrimination than men (also

controlling for age and visible minority status).
c. The Nagelkerke R 2 of the second model is 0.074, whereas the Nagelkerke
R 2 of the first model is 0.072. Since the difference between them is small,
it suggests that accounting for religious affiliation does not substantially
improve the fit of the logistic regression model.

noa15214_answers_SPSS_088-138.indd 138 01/11/18 09:23 PM