PSYC30013 Research Methods for Human Inquiry

PSYC30013 Research Methods for Human Inquiry

Lecture 7: Answers to Weekly Multiple Choice Questions
For each explanation below, the most correct response is shown in RED font. The explanation of
why this is the best option to choose is given in GREEN font.
The order of responses below may not be the same as that which you encountered because their
ordering was done randomly by the testing procedure in LMS.
Q1.
(a)
(b)
(c)
(d)

Which of the following is not a term used when investigating mean differences?
Among groups
Within groups
Between groups
Mixed groups

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This term is not used when specifying designs to investigate group
differences, even though it is often used as synonym for between (which is a term used for
group difference designs).
(b) (Incorrect response) This is a more general term to describe dependent groups (e.g., a design
in which the same people are each measured at k time points).
(c) (Incorrect response) This is a more general term to describe independent groups (i.e., a
design in which each person in the same can belong to only one of the k groups being
investigated.).
(d) (Incorrect response) This term is used when a design for investigating group differences
contains both within-subjects and between-subjects factors.
Q2
(a)
(b)
(c)
(d)

Which of the following is the best explanation of why the analysis of group differences can
be equally understood in terms of analysis of variance and linear regression using dummy
variables?
Because both approaches investigate the extent to which variation on a dependent variable
can be accounted for by variation in group means.
Because the intercept in a linear regression model using dummy variables can have a
meaningful interpretation.
Because the observed test statistic in both techniques is distributed as a theoretical F
probability distribution.
Because the degrees of freedom for the F test will not be the same in both analyses.

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This statement accurately describes the decomposition of total sum of
squares on the dependent variable into that part explained by differences (i.e. variation)
between group means (with the remaining variation being accounted for by variation within
each group). This is directly analogous to linear regression involving the decomposition of
total sum of squares into regression sum of squares and residual sum of squares when
dummy variables are employed. Compared to the other three responses, this one seems the
best (but it not necessarily fully explained).
(b) (Incorrect response) This statement is true, but it has nothing to do with why differences in
group means can be understood in terms of variation the decomposition of total sum of
Week 7

Page 1

PSYC30013 Research Methods for Human Inquiry

(c)
(d)

Q3
(a)
(b)
(c)
(d)

squares in linear regression.

(Incorrect response) This statement is correct as it stands, but it does not explicitly explain
why ANOVA and linear regression can both be used to investigate group mean differences.
(Incorrect response) This statement is at best ambiguous and at least wrong, because it does
not indicate the number of dummy variables or sample size in the respective regression
model and ANOVA model.
Which of the following statements about dummy variables is most correct?
Dummy variables will only ever have values of either 0 or 1.
The value of the k dummy variables is directly related to the k categories from which they are
derived.
The values assigned to the categorical variable are directly calculated from the values of the
dummy variables.
Dummy variables are measures of the categories at a population level.

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This statement is correct, because it describes the possible values any
dummy variable can have by definition (note that a variable with values of 0 and 1 may be
treated as a dummy variable, but it does not necessarily have be specified or regarded as
onee.g., sex can be coded as 0=males and 1=females without it necessarily having to be
referred to as a dummy variable).
(b) (Incorrect response) This statement is incorrect because there are only ever k-1 (rather than
k) dummy variables derived from the k categories.
(c) (Incorrect response) This statement gets the relationship between a categorical variable and
dummy variables around the wrong way: Dummy variables are derived from the categorical
variables, rather than the latter being derived from the former (however, the categories of
the categorical variable can be identified from the scores on the set of dummy variables).
(d) (Incorrect response) This statement is nonsensical and can be readily dismissed outright.
Q4
(a)
(b)
(c)
(d)

What is the best reason for why the values of {1, 2, 3, 4, 5} assigned to a categorical variable
with five categories cannot be directly used in a linear regression?
All three response options are equally good reasons.
Because the regression coefficients will not be meaningful.
Because the sum of squares in the ANOVA Table will be incorrect.
Because the observed R-square value will differ, depending on which category is assigned to
which value.

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) Yes, all three responses here contain reasons for why a categorical
variable cannot be used as a single IV in a regression analysis. The key underlying reason is
that (i) the results will depend on which number from 1 to 5 is assigned to which category, (ii)
the numbers themselves are arbitrary (they could just as well be 1, 5, 11, 100, 1001), and (iii)
the unit difference between each adjacent pair of values from 1 to 5 does not equate to a
unit difference between adjacent categories in how much is being measured by the number
1, 2, etc.
(b) (Incorrect response) This is a good reason, but the other two are equally good (and therefore
not the best response). There will only be one regression coefficient (because the there is
only one categorical variable being used as an IV in the analysis) and the value of the
Week 7

Page 2

PSYC30013 Research Methods for Human Inquiry

(c)

(d)

Q5

(a)
(b)
(c)
(d)

coefficient will be different, depending on which number from 1 to 5 is assigned to which

category (the only way the sum of squares will be unaffected by this aspect is when the
frequency count is the same for each category).
(Incorrect response) This is a good reason, but the other two are equally good (and therefore
not the best response). The decomposition of total sum of squares on the DV will be
different, depending on which number from 1 to 5 is assigned to which category (the only
way the sum of squares will be unaffected by this aspect is when the frequency count is the
same for each category).
(Incorrect response) This is a good reason, but the other two are equally good (and therefore
not the best response). Because the decomposition of total sum of squares on the DV will be
different, depending on which number from 1 to 5 is assigned to which category, the Rsquare value will subsequently be different depending on which number is assigned to which
category (the only way that R-squared will be unaffected by this aspect is when the frequency
count is the same for each category).
The sample means of three groups are as follows: Grp1 = 10, Grp2 = -4, Grp3 = 6. If the third
group is defined as the reference category in a linear regression model using two dummy
variables, which will be the value of the partial regression coefficient for the dummy
variable assigned to Grp2?
10
+10
+2
2

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) The dummy variable assigned to Grp2 will have a value of either 0 or 1. A
value of 0 on this dummy variable corresponds to the mean on the DV for the reference
category (defined in Q5 to be Grp3s mean equal to +6). The regression coefficient given by
the dummy variable assigned to Grp 2 therefore must equal the difference between the
respective means for Grp2 and Grp3; i.e., 4 (+6) = 10.
(b) (Incorrect response) This option is not true because a regression coefficient value of +10
would correspond to using the mean for Grp2 as the reference category rather than the
mean for Grp3i.e., calculating the regression coefficient as +6 (4) = 10, which is not
correct. The regression coefficient for the dummy variable corresponding to Grp2 is instead
correctly given by the difference between the mean of Grp2 (i.e., 4) and the mean of the
reference category Grp3 (i.e., +6), which translates into 4 - (+6) = 10.
(c) (Incorrect response) This option is not true because a regression coefficient of +2 implies
adding the means for Grp2 and Grp 3, which is the reference category, respectively rather
than calculating their difference. The regression coefficient for the dummy variable
corresponding to Grp2 is correctly given by the difference between the mean of Grp2 (i.e.,
4) and the mean of the reference category Grp3 (i.e., +6), which translates into 4 (+6) =
10.
(d) (Incorrect response) This option is not true because it is the one calculation that is most
difficult to come up with a way in which it was calculated. A regression coefficient of 2
seems to imply subtracting the mean for Grp3 from the absolute mean of Grp2 (which is
nonsensical). The regression coefficient for the dummy variable corresponding to Grp2 is
correctly given by the difference between the mean of Grp2 (i.e., 4) and the mean of the
reference category Grp3 (i.e., +6), which translates into 4 (+6) = 10.
Week 7

Page 3

PSYC30013 Research Methods for Human Inquiry

Q6
(a)
(b)
(c)
(d)

If there are 7 distinct groups in one categorical variable and 3 distinct groups in a second
categorical variable, how many more unique differences can be examined for the former
categorical variable compared to the latter?
4
2
Either 3 or 7
6

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This is correct because there will be 6 dummy variables for the variable
with 7 distinct groups, and 2 dummy variables for the variable with 3 distinct groups, which
therefore means the larger categorical variable will have 4 more dummy variables than the
smaller one.
(b) (Incorrect response) This is incorrect because this value is the number of dummy variables for
the variable with 3 distinct groups.
(c) (Incorrect response) This is incorrect because these are just the number of categories in the
respective categorical variables, and it makes no sense to have two possible values given the
structure of the question.
(d) (Incorrect response) This is incorrect because this value is the number of dummy variables for
the variable with 7 distinct groups.
Q7
(a)
(b)
(c)
(d)

Which of the following is the most correct statement for what can be inferred from the
rejection of an omnibus F test in ANOVA?
At least one unidentified group mean is different from all remaining group means.
There is a difference between the largest and smallest group means.
There is a difference among all the group means.
Either the largest or smallest group mean differs from all other group means.

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This answer is correct because the omnibus F test assesses a null
hypothesis that all group means are equal. If this hypothesis is rejected, then the vague
alternative hypothesis can only be inferred as implying at least one group mean is different
(without identifying which particular group that might be).
(b) (Incorrect response) although this will almost certainly be the case if the omnibus null
hypothesis is reject (assuming homogeneity of variance assumptions hold), it cannot be
inferred directly from a statistically significant omnibus F test.
(c) (Incorrect response) This is incorrect because a statistically significant omnibus F test implies
only that at least one mean of different from all remaining means; it does not imply
necessarily that all group means are different among each other.
(d) (Incorrect response) This is incorrect because a statistically significant omnibus F test implies
only that at least one mean of different from all remaining means; it does not imply that
either the smallest or largest group mean is necessarily different.
Q8
(a)
(b)

If a between-subjects factor contains five levels, how many non-redundant planned

comparisons can be specified for that factor?
4
More than 5

Week 7

Page 4

PSYC30013 Research Methods for Human Inquiry

(c)
(d)

5
5 * 4 / 2 = 10

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This is correct, as there can only be k-1 fundamental differences on a
single dependent variable among k objects.
(b) (Incorrect response) This is incorrect because, while there can be a very large number of ways
that differences may occur among the five groups, there are only 4 ways that such
differences can fundamentally described that contain no redundant information.
(c) (Incorrect response) This is incorrect because it is k-1 fundamental differences on a single
dependent variable among k objects, not k differences.
(d) (Incorrect response) This is incorrect because (i) there are only k-1 fundamental, nonredundant differences on a single dependent variable among k object, and (ii) this is the
number of possible pairwise differences (which will contain some redundancy).
Q9
(a)
(b)
(c)
(d)

If a null hypothesis for group mean differences among the five means in Q8 proposed that
the first two groups differed from the last three groups, what would be the contrast
weights?
{+3 +3 -2 -2 -2}
{+2 +2 -3 -3 -3}
{+1 +1 -1.5 -1.5 -1.5}
{+1 +1 -1 -1 -1}

Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This is correct because the sum of the positive weights (i.e., +6) equals the
sum of the negative weights (-6) in absolute terms, and the values reflect the relative size of
the two sets (one containing 2 groups, and the other containing 3 groups).
(b) (Incorrect response) This is incorrect because the sum of the positive weights (i.e., +4) does
not equals the sum of the negative weights (-9) in absolute terms, and therefore the total
sum of weights is non-zero (by definition, contrast weights sum to zero).
(c) (Incorrect response) This is incorrect because the sum of the positive weights (i.e., +2) does
not equals the sum of the negative weights (-4.5) in absolute terms, and therefore the total
sum of weights is non-zero (by definition, contrast weights sum to zero).
(d) (Incorrect response) this is incorrect because the sum of the positive weights (i.e., +2) does
not equals the sum of the negative weights (-3) in absolute terms, and therefore the total
sum of weights is non-zero (by definition, contrast weights sum to zero).
Q10 Which of the following will be most likely to result in meaningful differences between
groups being found (all other things be equal)?
(a)
Larger differences between group means with smaller group variances.
(b) Larger differences between group means with larger group variances.
(c)
Smaller differences between group means with smaller group variances.
(d) Smaller differences between group means with larger group variances.
Why are the first option correct and the remaining three options incorrect?
(a) (Correct response) This is true because larger differences between group means will result in
a larger between-groups sum of squares, and smaller group variances will result in a smaller
within-groups sum of squares.
Week 7

Page 5

PSYC30013 Research Methods for Human Inquiry

(b)

(c)

(d)

(Incorrect response) This is not true because, although larger differences in-group means will
result in a larger between-groups sum of squares, larger group variances will result in a larger
within-groups sum of squares. Therefore the observed F statistic will be relatively smaller
(for the same number of groups and same total sample size).
(Incorrect response) This is not true because, although smaller group variances will result in a
smaller within-groups sum of squares, smaller differences between group means will result in
a smaller between-groups sum of squares. Therefore the observed F statistic will be
relatively smaller (for the same number of groups and same total sample size).
(Incorrect response) This is not true because smaller differences in group means will result in
a smaller between-groups sum of squares and larger group variances will result in a larger
within-groups sum of square. Therefore the observed F statistic will be relatively smaller (for
the same number of groups and same total sample size).

Week 7

Page 6