Introduction to SPSS
Activities Part 2
- Answers
1. The label for the variable Courseinfo is
missing. Insert a label like “How useful is
Course Information”.
9.
i. Median is 7. As the variable ranges from 5 to
15 this quite low and suggests that most people
did not find Blackboard easy to use.
ii. No, it is probably not normal. It is not close
to the normal curve.
iii. The distribution is skewed to the bottom
end. This is also indicated by the low values of
the median and the mean.
4.
i. No, Pathway is a categorical or nominal
variable.
ii. Crim
iii. 102, 102
5.
i. No, the average gender makes no sense. The
mode tells us that the most common value is
male.
ii. 47 females, 55 males
iii. Most students on these course are female so
the sample may not be representative.
10.
i. There is a line for each value of the variable.
As it ranges from 10 to 40 there can be up to 31
values.
iii. None. There is no one with a score of 35
iii. The mean is 24.32 and is the average score
the people got on this variable.
iv. This distribution looks close to normal (it
follows the shape of the curve. But this needs
confirming with statistical tests.
6.
i. the value that divides the ordered distribution
of cases into two equal halves lies in the value
‘once a month’ and this has a value of 3.
ii. the most common response is 3, this means
‘once a month’
iii. It looks a bit normal – i.e. close to the curve
shown. But this needs to be checked by testing
for normality.
17. Criminology.
19. The mean of TotuseBb, Overall usefulness
of Blackboard is 24.32. The median is 24 and
thus there are more respondents below the
mean than above it.
7.
i. 18
ii. 2, 42 respondents thought Blackboard
somewhat difficult to use and this was the most
common response.
20. Yes, it looks normally distributed.
21 Use the Kolmogorov-Smirnov test as there
are more than 50 respondents.
22. No. We cannot reject the null hypothesis
that the distribution is normally distributed as
the probability is .200 which is above the
normally used alpha of 0.05. In other words the
evidence suggests that the distribution is
normal, and therefore we can use parametric
statistics on it.
8.
i. 2 tiscali, which is the most common ISP
ii. 2.61 and it makes no sense because this is a
categorical or nominal variable.
1
Activities Part 3 - Answers
3. 22 males scored 2. Expected value is larger
(17.0 vs. 15). No, a greater percentage of
females scored 4 (25.5% vs. 20.0% of men).
4. Women tend to have lower scores i.e. they
tend to find it easier to use. The Chi square
value is 1.248 and this is not significant
(p=0.742) which is above the normal alpha of
0.05. But N.B. 25% of the cells had expected
values of less than 5 so the Chi-square
calculation is unreliable.
5. There is no difference. Chi-square = 7.440 but
p=0.114 so this is not significant. Although
one cell has an expected value of less than 5
this is just below and because the p is well
above 0.05 there really is no evidence that
there is a gender difference.
6. There is very little consistent difference
between males and females. Chi-square =
2.631 and this is not significant p=0.452.
There are no cells with expected values of less
than 5 so the Chi-square value is reliable.
7. Most of the cells have very low expected
values so there is not much you can say about
this relationship. More males use the facility.
8. No table exists for this as no-one used online
quizzes.
10. Year 1 finds it least easy. Years 2 and 3 find
it easier to about the same degree. Chi-square
= 110.542 and is significant (p=0.000) and
although one third the cells have expected
values of less than 5 the result is so significant
that we can probably rely on it.
11. Year 1 finds it least easy. Years 2 and 3 find
it easier to navigate to about the same degree.
Chi-square = 109.330 and is significant
(p=0.000) N.B. even though most cells have
low expected values, inspecting the row
percentages (% within Year of study) shows a
very clear picture.
2
12. No-one in year 1 used online discussion
forums and most who did were in year 3. With
such small numbers it is hard to tell if there is
any difference between year 2 and year 3. The
statistics are not significant.
13. It looks like first years find Blackboard
harder to use and less useful than the other
years. Perhaps they are not as used to it as
second and third years and perhaps their
studies do not rely so much on using
Blackboard.
14. The Chi-square value is 85.711 and its
significance is 0.000. But the cell values are
very small and 60% have expected counts of
less than 5 so the statistic is very unreliable.
15. The counts (and column percentages suggest
year 1 students use Blackboard far less
frequently than second and third years. This
may explain the result in question 13, where
first years find Blackboard harder to use and
less useful. Or maybe they use Blackboard
less because they find it less useful?
16. This crosstabulation does not make any
sense. The variable Overall usefulness of
Blackboard has values from 10 to 40, so there
are far too many rows and most cells are
empty. This tells us nothing.
17. The Chi-square in not significant (Chi-square
=7.056, 4 df, p=0.133) and anyhow is
unreliable. However, looking at the table you
can see that, surprisingly, those with offcampus Internet are less likely to use
Blackboard frequently.
18. This 2x3 table produces a Chi-square value
of 6.881 with 2 df and p=0.032 which is
significant. The table suggests that more
frequent users of Blackboard are less likely to
have off-campus Internet access.
Activities Part 4 - Answers
3. There is not a strong relationship but as the
ease of use score increases so does the
usefulness score. I.e. as students found it easier
to use they also found it more useful. Or
maybe vice versa. There is no way to
determine cause and effect from scatterplots
and correlations.
5. It makes it slightly clearer that as ease of use
increases so does usefulness.
6. The line shows the best (linear) fit between the
variables. I.e. where the points would all lie if
there were no variability in the relationship of
ease of use to usefulness.
7. The patterns look very similar. There may be a
slight tendency for the males to have more
extreme values on the usefulness axis.
8. All the Year 1 students have low values on the
ease of use axis, all less than 10, whereas all
second and third years score 9 or above on ease
of use. We saw this from the crosstabulation
tables last week.
10. The Pearson Correlation (r) is 0.485 and this
is statistically significant, p<.000. The value is
not very high (r varies from 0 to 1) and the
relationship between Overall ease of use of
Blackboard and Overall mark at end of year is
not strong although it is statistically significant.
11. The Pearson Correlation (r) is 0.887 and this
is statistically significant, p<.000. The value is
quite high and the relationship between Overall
usefulness of Blackboard and Overall mark at
end of year is quite a strong one and it is
statistically significant.
12. 0.354 (.595 squared) or about 35%.
14. No, the distribution does not show a gathering
towards the centre (in fact there is a dip at the
centre of the range) and is strongly skewed to
the lower end of the scale. Therefore we should
not use a parametric statistic.
3
15. Yes. The histogram follows the normal curve
fairly closely. We may use a parametric
statistic.
16. Yes. The histogram follows the normal curve
fairly closely. We may use a parametric
statistic.
17. The histogram looks close to a normal
distribution but the variable is ordinal.
Frequency of use is put into order but the
difference between more than once a week and
once a week is not the same as the difference
between less than once a month and once only.
In this case we should use a non-parametric
statistic.
19. The Spearman’s rho for Overall mark at end of
year vs. how often used Blackboard is 0.337.
This is a low figure so there is only a poor
relationship between the variables, although it is
significant (p <.001) so we couldn’t have got it
by chance. However, only 0.114 (i.e. 11.4%) of
the variation in one can be explained by the
other, which shows how poor the correlation is.
Other things explain the other 89% of the
variation.
20. The Spearman’s rho for Overall mark at end of
year vs. overall usefulness of Blackboard is
0.906. This is a high figure so there is a very
strong relationship between the variables, and it
is significant (p =.000) so we couldn’t have got
it by chance. About 82% of the variation in one
can be explained by the other, which shows how
strong the correlation is.
21. There is a strong correlation but we cannot
infer causality from this. It might be that using
Blackboard helped these students get a good
mark, but it might just be that hard-working
students both got good marks and found
Blackboard useful.
Activities Part 5 - Answers
3. Looking at the histogram the distribution is
bimodal and is very far from the normal curve,
so it is probably not normally distributed.
4. Looking at the histogram the distribution is
close to the normal curve, so it is probably
normally distributed.
5. The mean for ‘Overall ease of use of
Blackboard’ is 8.57 and the mean for ‘Overall
usefulness of Blackboard’ is 24.32. There are
102 cases (N=102) so the KolmogorovSmirnov test should be used. For the variable
‘Overall ease of use of Blackboard’ the
Kolmogorov-Smirnov is .240 and has a
significance of .000 so we can conclude that
this variable is NOT normally distributed. For
the variable ‘Overall usefulness of Blackboard’
the Kolmogorov-Smirnov is .059 and has a
significance of .200 so we can conclude that
this variable IS normally distributed.
6. The Standard Deviation of ‘Overall usefulness
of Blackboard’ for Females is 5.811 and for
Males is 7.876. (Standard deviation is the
square root of the Variance.) So they are
different but we need to test to see if this
difference is statistically significant. Levene’s
test has a significance of .013. This is below
the commonly used alpha of .05 so we reject
the null hypothesis (no difference) and
conclude that the variances ARE statistically
significantly different. So we use the ‘Equal
variances not assumed’ results for the t-test.
7. The mean values of ‘Overall usefulness of
Blackboard’ for Females is 24.57 and for
Males is 24.11. The values are different and
Females find Blackboard slightly more useful
than Males but we need to see if this difference
is statistically significant. The t value is .342
and the significance is .733. So we accept the
null hypothesis and there is NO evidence that
the mean value of Overall usefulness of
4
Blackboard’ for Females is difference from that
for Males.
8. Females have the higher mean rank (52.44 to
Males 50.70) but the Asymp. Significance (2tailed) is .765, this is well above the standard
alpha of 0.05 so there is no evidence that there
is any gender difference in how easy students
find Blackboard to use.
9. Those who use Blackboard more than once a
week find it the most useful (mean score of
30.13) and those who use it less than once a
month find it least useful overall (mean score of
20.91). The difference between all the groups is
statistically significant. F=10.251, 4df. with a
significance of .000 which is well below the
alpha of .05.
10. Aside from the 6 people who had used
Blackboard only once, there is a tendency for
those who use it more often to find it more
useful. But those who use it once a month or
less find it much less useful than those who use
it weekly or more often.
11. Almost all the statistically significant pairwise
differences (p<.05, indicated by asterisks) are
between the two most frequent users categories
and the monthly or less often users. This
confirms the picture that there are two
groupings of how useful Blackboard is found,
the frequent users (once a week or more often)
who found it significantly more useful overall
than the infrequent users (once a month and less
than once a month).
12. Year 3 has the highest mean rank (92.84) and
year 1 the lowest mean rank (34.06).
13. There is a statistically significant difference
between the three year cohorts (2 =71.838, 2df,
p<0.000), with the third year finding
Blackboard the easiest to use overall and year 1
finding it the hardest