Deck 17: Scaling Stimuli

Question 1

Which of the following statements are characteristics of stimulus scaling? (circle all that apply)

A) a good stimulus scaling achieves a ratio-level scale of stimuli
B) any individual differences in a stimulus scale are due to measurement error
C) the distances between stimuli on a stimulus scale are equal
D) the relationship between all items on a stimulus scale are transitive

Answer 1

B,D

Question 2

Researchers constructed a stimulus scale by asking participants to organize 15 movies in terms of overall acting performance, with the top movie reflecting the best overall performances, and the bottom the worst overall performances. Which of the following statements would NOT be true about this scale? (circle all that apply)

A) participants' ratings of movie performances are transitive
B) the researchers constructed their scale by having participants select the better movie performances between 105 pairwise comparisons [i.e., 15(14)/2]
C) the top movie on the scale represents excellent acting performances
D) participant differences in movie performance ratings are thought be measurement error

Answer 2

B,C

Question 3

Which of the following characteristics are shared by all types of stimulus scaling techniques?

A) all stimulus scaling techniques generate comparative judgments between people
B) the distances between stimuli on a scale are equal
C) when judging stimuli, participants simultaneously consider all stimuli on the same comparison dimension
D) A and C
E) B and C
F) all of the above

Answer 3

B

Question 4

Which of the following possibilities threatens the transitivity of a stimulus scale? (circle all that apply)

A) not all participants are equally familiar with all presented stimuli
B) some stimuli may be harder to distinguish (between the two of them) than others
C) respondents' differences in ranking or scaling reflect actual individual differences
D) stimuli were apparently judged on different aspects of the same comparison dimension

Question 5

Which of the following statements is true about multidimensional scaling?

A) stimuli are compared on any specified characteristics (e.g., beauty, similarity, smelliness)
B) the "dimensions" that are obtained with multidimensional scaling are inferred by the researcher
C) unlike unidimensional scaling techniques, in MDS, individual differences are not considered part of measurement error
D) A and B
E) A and C
F) none of the above

Question 6

Researchers wanted to investigate how people selected the cars that they wanted to buy, and they thought they would use stimulus scaling to do so. All things being equal, which of the following techniques would be appropriate for this kind of investigation?

A) pair-comparison scaling
B) multidimensional scaling
C) rank-order scaling
D) A and C
E) B and C
F) all of the above

Question 7

Researchers conducted a study of landscaping preferences using pair-comparison scaling that provided participants 28 pairwise comparisons between types of trees. What assumptions are the researchers making by choosing to use pair comparison scaling?

A) that participants will judge the trees on the same dimension
B) that, in the absence of measurement error, people would organize their preferences in exactly the same way
C) that differences between stimuli are meaningful
D) that participants rate stimuli as if they are differentiated in equal intervals

Question 8

What would be appropriate conclusions to draw from this choice matrix? (choose all that apply)

A) Tesla is an excellent car company
B) Ford is a terrible car company
C) 60% of respondents thought Chevy was better than Toyota
D) 40% of respondents thought Ford was preferable over Tesla

Question 9

What are some of the advantages of rank order scaling over pair comparison scaling?

A) rank ordering 10 stimuli is less cognitively taxing for participants than making pairwise comparisons for 10 stimuli
B) when making judgments, participants consider all stimuli at once, reducing the amount of variation in dimensions that the stimuli are compared
C) rank ordering responses are thought to be more reliable than numerous pairwise comparisons
D) A and B
E) B and C
F) all of the above
G) none of the above

Question 10

You are interested in whether there are differences in aesthetic value for 10 types of pens. Which of the following techniques would be able to potentially reveal if pens differed in their aesthetic quality?

A) 10 semantic differential scales (i.e., one for each pen)
B) 10 Likert scale items (one for each pen; 1 = ugly, 10 = beautiful)
C) 45 comparisons of "how similar are these two stimuli"
D) 45 comparisons of "which one do you prefer"
E) ranking 10 pens from most to least aesthetically pleasing

Question 11

Unlike scaling individuals, when scaling stimuli:

A) differences in responses across individuals are deemed a result of error
B) differences observed between stimuli are considered meaningful
C) differences in responses across individuals are meaningful
D) differences observed between stimuli are deemed a result of error

Question 12

The method of pair comparison:

A) always yields data that are on an interval scale
B) entails presenting all possible pairs of stimuli to a sample
C) key results involve differences in judgments between participants
D) produces the order of stimuli according to a group of participants

Question 13

An intransitive set of judgments could result from:

A) an unrepresentative sample
B) unequal intervals
C) participants' lack of familiarity with the stimuli
D) unclear or complex dimensions that participants use to judge stimuli
E) none of the above

Question 14

The method of comparisons by rank order:

A) provides results that compare rated stimuli on a ratio scale
B) provides results that compare rated stimuli on an interval scale
C) is cognitively more demanding of participants than the method of pair comparison, if the number of rated stimuli is held constant
D) requires participants to evaluate all stimuli at once
E) results in transitive data

Question 15

Multidimensional scaling methods:

A) are generally analogous to factor analysis
B) identify underlying dimensions based on binary judgments
C) identify underlying dimensions based on raked preferences
D) identify underlying dimensions based on ratings of similarity of paired stimuli
E) explore dimensions of individual differences

Question 16

Brian, Sam, and Jenny asked three respondents to rank their preferences for different types of seating. The table below is the result of that ranking. Given this information, draw an (approximate) example of Coombs' unfolding technique on the provided linear dimension below.
$\begin{array}{clll}\hline \text { Ranked Preference } & \text { Katrina } & \text { Roger } & \text { Alice } \\\hline 1 & \text { Couch } & \text { Lawn Chair } & \text { Beanbag } \\2 & \text { Reclinirg Chair } & \text { Floor } & \text { Lawn Chair } \\3 & \text { Beanbag } & \text { Beanbag } & \text { Couch } \\4 & \text { Lawn Chair } & \text { Couch } & \text { Reclining Chair } \\5 & \text { Floor } & \text { Reclining Chair } & \text { Floor } \\\hline\end{array}$

Question 17

Explain two reasons for why you might obtain an intransitive stimulus scale. In your response, (a) describe what it means for a stimulus scale to be intransitive, (b) explain two reasons for why intransitivity occurs, and (c) provide an example for each of those reasons that illustrates how or why this would be problematic for a stimulus scale.

Question 18

In terms of stimulus scaling, explain what it means for a scale to be of interval quality. What are the key features of an interval scale (as it applies to stimulus scaling), and what are the key features that cannot be achieved with an interval scale? Be sure to sufficiently explain each feature and why it can or cannot be achieved with an interval scale.

Question 19

Define individual difference scaling and stimulus scaling. What are the key features and underlying assumptions of each type of scaling? What is considered "true" difference and what is considered "error?"

Question 20

You are interested in the ways that people decide between five types of take-out food (i.e., pizza, Chinese food, Indian food, Thai food, fast food [e.g., Burger King]) on any given night. So, on Tuesday, you conduct a unidimensional stimulus scaling study. On Thursday, with a different group of participants, you conduct a multidimensional scaling study using the same five types of take-out food. Explain the procedures you used in your studies on Tuesday and Thursday. Specifically, in your response, describe (a) the key features of the different methodologies you used, (b) the main assumptions for each type of scaling, and (c) the specific ways you would measure participants' comparisons between take-out food. Furthermore, provide example results for each type of scaling that demonstrate the key features of the methodology, and be sure to explain how and why your examples represent unidimensional and multidimensional scaling techniques (for example, draw a two-dimensional graph for the multidimensional scaling, label the dimensions, and indicate where the five types of food would fall along those dimensions).

Question 21

How does stimulus scaling differ from the scaling of individuals?

Question 22

Describe the steps involved in Thurstone's method of pair comparison and the method of rank order. Is one method superior to the other?

Question 23

When might a researcher want to employ multidimensional scaling techniques rather than conducting a factor analysis?