Theoretical practice questions
Introduction to research in marketing – Fall 2018-2019
* Exam October 2018 Q&A
** Practice exam 2018 Q&A
What is multicollinearity?
High intercorrelations among independent variables in a multiple regression model. It can lead to
skewed/misleading results.
What kind of data belongs to Non-metric data?
Nominal, Ordinal
What kind of data belongs to metric data?
Interval (scales) and Ratio (measurable and countable)
What are bivariate visualizations?
Relationships between two or more variables
What is the assumption of homoscedasticity and when is it satisfied?
Homoscedasticity means we want to find no differences in the variances across conditions. When
Levene's test finds a significance below .05, it is NOT satisfied. When the significance is above .05, it IS
satisfied. H0 is rejected when it is not satisfied.
What does ANOVA test?
It tests if there are differences in the mean of a metric dependent variable across different levels of one
or more non-metric IVs
What does cluster analysis do?
Combines objects or persons into groups based on a predefined set of characteristics.
What does factor analysis do?
Combine highly correlated variables together.
What does logistic regression do?
Predict the probability that a non-metric variable is "A"
What does Conjoint analysis do?
Instead of rating attributes, customers rate the whole product. The product rating can be decomposed
into values attached to each feature; part worths
What is MDS?
Multidimensional scaling is an exploratory technique to identify dimensions by which objects are
perceived
Eline van de Ven
, Why is it important to use all but 1 of the part worths? (Conjoint) *
Multicollinearity. Part worths have to sum to zero. So only j-1 is independent.
What does the angle between vectors mean (MDS)?*
The smaller the angle, the closer they are, and the more correlated
What does the length of a vector mean (MDS)?*
How much information the attribute adds to the map. A long vector means a lot of information
What does the distance from the origin (0,0) to the point at which the perpendicular crosses the vector
mean?*
How well the object scores on that attribute
A vector =
Indicator of magnitude and direction in which the attribute is increasing in the Euclidian space
What are the axes of the map?
Special set of vectors suggesting the underlying dimensions that best characterize how consumers
differentiate among alternatives.
Does this belong to MDS?
Similar based approach is most appropriate for functional products*
No
When multicollinearity is an assumption to be met instead of a violation, what do we use?*
Factor analysis
What does a scree plot explain?**
How much variation is explained by the factors
exp(x)/1 exp(x)
Does the function get closer to -1 when x decreases?*
No
What is power?
The probability of correctly rejecting the null when it's false
When there is a large effect, do we need a small or large sample to get good/excellent power?
Small
How big is good and excellent power respectively?
0.80 and 0.95
When your sample is too small, which of the following is correct?
a) the effect is probably insignificant
b) the effect is probably significant
c) when the effect is significant, it is probably overstated
Eline van de Ven
Introduction to research in marketing – Fall 2018-2019
* Exam October 2018 Q&A
** Practice exam 2018 Q&A
What is multicollinearity?
High intercorrelations among independent variables in a multiple regression model. It can lead to
skewed/misleading results.
What kind of data belongs to Non-metric data?
Nominal, Ordinal
What kind of data belongs to metric data?
Interval (scales) and Ratio (measurable and countable)
What are bivariate visualizations?
Relationships between two or more variables
What is the assumption of homoscedasticity and when is it satisfied?
Homoscedasticity means we want to find no differences in the variances across conditions. When
Levene's test finds a significance below .05, it is NOT satisfied. When the significance is above .05, it IS
satisfied. H0 is rejected when it is not satisfied.
What does ANOVA test?
It tests if there are differences in the mean of a metric dependent variable across different levels of one
or more non-metric IVs
What does cluster analysis do?
Combines objects or persons into groups based on a predefined set of characteristics.
What does factor analysis do?
Combine highly correlated variables together.
What does logistic regression do?
Predict the probability that a non-metric variable is "A"
What does Conjoint analysis do?
Instead of rating attributes, customers rate the whole product. The product rating can be decomposed
into values attached to each feature; part worths
What is MDS?
Multidimensional scaling is an exploratory technique to identify dimensions by which objects are
perceived
Eline van de Ven
, Why is it important to use all but 1 of the part worths? (Conjoint) *
Multicollinearity. Part worths have to sum to zero. So only j-1 is independent.
What does the angle between vectors mean (MDS)?*
The smaller the angle, the closer they are, and the more correlated
What does the length of a vector mean (MDS)?*
How much information the attribute adds to the map. A long vector means a lot of information
What does the distance from the origin (0,0) to the point at which the perpendicular crosses the vector
mean?*
How well the object scores on that attribute
A vector =
Indicator of magnitude and direction in which the attribute is increasing in the Euclidian space
What are the axes of the map?
Special set of vectors suggesting the underlying dimensions that best characterize how consumers
differentiate among alternatives.
Does this belong to MDS?
Similar based approach is most appropriate for functional products*
No
When multicollinearity is an assumption to be met instead of a violation, what do we use?*
Factor analysis
What does a scree plot explain?**
How much variation is explained by the factors
exp(x)/1 exp(x)
Does the function get closer to -1 when x decreases?*
No
What is power?
The probability of correctly rejecting the null when it's false
When there is a large effect, do we need a small or large sample to get good/excellent power?
Small
How big is good and excellent power respectively?
0.80 and 0.95
When your sample is too small, which of the following is correct?
a) the effect is probably insignificant
b) the effect is probably significant
c) when the effect is significant, it is probably overstated
Eline van de Ven
