Chapter 3: Measurement in Marketing Research
TRUE/FALSE
1) For many research projects, measurement error can be substantially greater than sampling
error, particularly when measurement involves subjective consumer information such as
attitudes, preferences, and intentions.
Answer: True
Measurement error can be a substantial component of total error and is particularly
problematic when measuring subjective areas.
2) The set of marketing phenomena under study is called the abstract system.
Answer: False
The set of marketing phenomena under study is called the empirical system. The abstract
system consists of all quantities used to represent the marketing phenomena.
3) Measurement can be defined as assigning numbers to objects, characteristics, or events in
the empirical system according to a set of known, specific rules in the abstract system.
Answer: True
By definition this statement is true. The objective of the measurement process is to develop a
correspondence between the empirical system and the abstract system.
4) Nominal scales are used for the lowest form of measurement, namely, classification and
identification.
Answer: True
A nominal scale is one in which numbers (if any) serve only as labels to identify or categorize
information.
5) A nominal scale defines the ordered relationship among objects or events.
Answer: False
It is an ordinal scale that defines the ordered relationship among objects or events.
6) An ordinal scale measures whether an object or event has more or less of a characteristic
than some other object or event, but does not measure how much more or less.
Answer: True
Ordinal scales provide order information of objects or events, but not how much they differ.
7) The mean is an appropriate descriptive statistic for ordinal data.
Answer: False
Ordinal data does not possess the property of equal differences necessary for applying the
mean. The mode and median are appropriate measures of central tendency for ordinal data.
8) Because the zero on an interval scales is an arbitrary assignment, comparisons of absolute
magnitudes or ratios is meaningless.
Answer: True
The arbitrary assignment of the zero point on a interval scale places restrictions on the
comparisons that can be made. Intervals can be compared; absolute magnitudes cannot.
9) In marketing, it is common for attitudinal, opinion, and predisposition judgments to be
treated as interval data.
Answer: True
Although technically it is ordinal data, it is common for them to treated as interval data.
10) The zero point on an interval scale corresponds to the absence of the characteristic being
measured in the empirical system.
Answer: False
The zero point on a ratio scale corresponds to the absence of the characteristic being
measured in the empirical system. Zero on an interval scale does not have any particular
meaning.
11) An operational definition defines a construct with other constructs.
Answer: False
An operational definition specifies how a construct is measured. It a constitutive definition
that defines a construct with other constructs.
12) Measurement error is minimized when a direct correspondence exists between the
number system and the marketing phenomena being measured.
Answer: True
Measurement error is minimized when the numbers assigned to the phenomena accurately
represents the characteristics being measured and nothing else.
13) Total measurement error consists of two components: systematic error and random error.
Answer: True
Systematic error and random error are the two components of measurement error.
14) Reliability refers to the extent to which the measurement process is free from random
errors.
Answer: True
Reliability is concerned with the consistency, precision, and predictability of research
findings, and thus addresses the issue of random error.
15) The methods used to measure validity involve determining the extent of agreement
between attempts to measure the same characteristic through maximally similar methods.
Answer: False
These are methods for testing reliability. Validity must determine the extent of agreement
with something in the empirical system: “Are we measuring what we think we are
measuring?”
16) Content validity involves subjective judgments by experts as to the appropriateness of the
measurement.
Answer: True
Sometimes called face validity, content validity uses experts to determine if a measurement is
a appropriate.
17) When using the test-retest method for measuring reliability, the greater the discrepancy in
scores, the greater the systematic error present in the measurement process.
Answer: False
The greater the discrepancy in scores between retests, the greater will be the random error,
not the systematic error.
18) In conducting international marketing research, evidence suggests that measures are fairly
reliable across countries for “hard” variables, such as age and gender, but have limited
reliability for “soft” variables, like lifestyle, product involvement, and attitudes.
Answer: True
Geodemographic variables tend to be more reliable across countries than attitudinal type of
measures.
19) The hierarchy-of-effects model postulates an advertising-purchase-repurchase sequence
resulting from marketing effort.
Answer: False
The hierarchy-of-effects model postulates a cognitive–affective–behavioral sequence
resulting from a marketing effort.
20) The cognitive component of attitude concerns a person’s feelings about the object or
event.
Answer: False
The cognitive component of attitude concerns a person’s knowledge or beliefs about an
object or event; it is the affective component that concerns the person’s feelings.
21) The construct of attitude, because it exists in the minds of individuals and is not directly
observable, is difficult to measure and typically involves the lower scales of measurement,
i.e. nominal and ordinal scales.
Answer: True
Attitude is difficult to measure and quantify, with properties that do not meet the requirement
of equal differences of an interval or ratio scale; therefore nominal and ordinal scales are
typically used. (However, ordinal attitude data is often treated as if it were interval within the
field.)
22) A dummy variable is an ordinally scaled variable on that is coded “0” or “1.”
Answer: False
A dummy variable is coded with a “0” or “1” but is a nominally scaled variable.
23) A graphic rating scale requires the respondents to indicate their position on a continuum
labeled with visual images, ranging from one extreme of the attitude in question to the other
extreme.
Answer: True
While the format varies, the graphic rating scales requires respondents to indicate on a
graphical scale their relative position.
24) A limitation of rank-order scales is the lack of cross-scale comparability.
Answer: True
This lack of cross-scale comparability is a substantial stumbling block to extracting realworld meaning from ranked questions.
25) The primary limitation of the semantic differential scale is the difficulty in composing
two adjectives or phrases to anchor the scale that are truly synonyms.
Answer: False
The difficulty is finding true bipolar adjectives or phrases, i.e., antonyms.
MULTIPLE CHOICE
1) In the measurement process, the _______________ consists of all quantities used to
represent marketing phenomena.
a. empirical system
b. abstract system
c. data analysis system
d. population
Answer: B
The empirical system is the marketing phenomena under study, and the abstract system is
used to represent marketing phenomena.
2) Which of the following arranges the measurement scales in order of complexity (simplest
to most complex)?
a. nominal, interval, ordinal, ratio
b. interval, ratio, nominal, ordinal
c. ratio, interval, ordinal, nominal
d. nominal, ordinal, interval, ratio
e. interval, nominal, ordinal, ratio
Answer: D
Increasing number system characteristics apply to the different measurement scales: nominal
(labels only), ordinal (labels plus order), interval (labels, order, and equal differences), ratio
(labels, order, equal differences, and equal ratios).
3) The numbers used to identify basketball players are representative of which measurement
scale?
a. nominal
b. ordinal
c. interval
d. ratio
Answer: A
Nominal scales uses numbers only as labels to identify or categorize information, just as the
numbers assigned to players in basketball.
4) The main tools for handling nominal data go under the name of
a. parametric analysis
b. categorical data analysis
c. monotonic transformation
d. linear regression
Answer: B
Because nominal scales or classification scales, categorical data analysis is used to examine
the data.
5) The most common statistical test used for nominally scaled variables is
a. regression analysis
b. ordinal regression
c. chi-square
d. coefficient of variation
Answer: C
The most common tests for nominal scales is the chi-square test for contingency tables.
6) If respondents are asked to rank five different brands of cold medicine from best to worst,
it would result in a(n)
a. nominal scale
b. ordinal scale
c. interval scale
d. ratio scale
Answer: B
It would be ordinal data since the brands are ranked from best to worst, but distance between
each ranking is not determined.
7) On an ordinal scale, which of the following series of numbers represent a permissible
substitution for the series 1, 2, 3?
a. 1, 3, 30
b. 5, 10, 15
c. 3, 2, 1 f. both a and b
d. all of the above
e. none of the above
Answer: False
Ordinal scales can employ any set of numbers, the only restriction being that the numbers
must correspond to the ordinal relationships. One could use {1, 3, 5}, {1, 3, 30}, or {14, 73,
1128} equivalently to {1, 2, 3}.
8) An appropriate measure of central tendency for ordinal scales is the
a. arithmetic mean
b. mode
c. harmonic mean
d. standard deviation
Answer: B
For ordinal scales, the mode and median are appropriate measures of central tendency.
9) One of the most common methods of assessing opinions and attitudes involves an “agreedisagree” scale, with 5 to 7 points, to statements such as “This appliance would make an
attractive addition my kitchen.” This method of assessing a person’s opinion results in a(n)
_______________ variable.
a. nominally scaled
b. ordinally scaled
c. interval scaled
d. ratio scaled
Answer: B
Although data from these types of scales are often treated as if they are interval, they are
technically ordinal because equal distance between points on the scale cannot be confirmed.
10) A crucial advantage of _______________ scales is that they allow the use of linear
regression techniques.
a. nominal
b. ordinal
c. interval
d. graphic rating
Answer: C
The nature of interval scales allows for linear regression, which is the dominant analysis
method in all of social science.
11) Suppose three brands are rated on an interval scale regarding degree of preference. Brand
X receives a 6 (most preferred), Brand Y a 3, and Brand Z a 2. What conclusions can be
drawn from this data?
a. Brand X is three times as preferred as Brand Z.
b. Brand Y is more preferred than Brand Z.
c. The difference between Brand X and Brand Y is three times greater than the difference
between Brand Y and Brand Z.
d. all of the above
e. both b and c
Answer: E
Interval scales preserve the ordinal relationship (Brand Y is more preferred than Brand Z) and
equal differences (the difference between Brand X and Brand Y is three times greater than the
difference between Brand Y and Brand Z). However, equal ratios (Brand X is three times as
preferred as Brand Z) would require a ratio scale.
12) With a(n) _______________ scale, the absolute or natural zero point refers to the
assignment of the number zero to the absence of the characteristic being measured.
a. nominal
b. ordinal
c. interval
d. ratio
Answer: D
The ratio scale has an absolute zero, which is an absence of the item being measured.
13) Marketing phenomena such as sales, market share, costs, ages, and number of customers
possess the properties of a(n)
a. nominal scale
b. ordinal scale
c. interval scale
d. ratio scale
Answer: D
Because each of these phenomena possess the properties of both equal differences and a
natural or absolute zero, it is a ratio scale.
14) Which type of scale is being used if a respondent’s age is measured by the question “How
many years have elapsed since your 18th birthday?”
a. nominal scale
b. ordinal scale
c. interval scale
d. ratio scale
Answer: C
It is an interval scale, because the property of equal differences applies, but the property of
equal ratios does not (someone 10 years past their 18th birthday is not twice as old as
someone 5 years past their 18th birthday).
15) In marketing, a(n) _______________ is an abstraction we build in our minds based on
perceptions of a phenomenon.
a. graphical interface
b. construct
c. monotonic transformation
d. perceptual map
Answer: B
Constructs, or concepts, are abstractions we build in our minds. In marketing, these serve as
the building blocks for marketing models.
16) An operational definition
a. defines a construct with other constructs
b. defines a construct by assigning numbers to different quantities of a characteristic
c. specifies how a construct is measured
d. directs the development of a constitutive definition
Answer: C
An operational definition defines or gives meaning to a variable by spelling out what the
investigator must do to measure it.
17) Common situational factors that are potential sources of measurement error include
a. factors such as a person’s mood, fatigue and willingness to participate
b. variations in the environment such as time of day, temperature, presence of family
members
c. variations in how the questions are administered, such as by phone or through e-mail
d. the degree of ambiguity and difficulty of the questions and the ability of respondents to
answer
Answer: B
Situational factors, such as the time of the day the survey is completed and point-of-purchase
displays, can cause measurement errors.
18) Common data analysis factors that are potential sources of measurement error include
a. errors made in the coding and tabulation process
b. variations in the environment such as time of day, temperature, presence of family
members
c. variations in how the questions are administered, such as by phone or through e-mail
d. the degree of ambiguity and difficulty of the questions and the ability of respondents to
answer
Answer: A
Data analysis factors, such as coding and tabulating responses, can cause measurement errors.
19) One component of total measurement error is random error, which
a. causes a constant bias in the measurements
b. involves a consistent error in the measuring instrument
c. involves a bias caused by the way sample is selected
d. involves influences that bias measurements, but not in a systematic way
Answer: D
Random error involves errors and biases in measurement that occur randomly rather than
systematically.
20) If the measurement process involves high systematic error and low random error, the
measurements are
a. valid but not reliable
b. valid and reliable
c. reliable but not valid
d. neither valid nor reliable
Answer: C
Validity means low systematic error; reliability means low random error. Therefore, high
systematic error and low random error means reliable but not valid. The measure gets
consistent results, but it is not measuring what it was designed to measure.
21) All of the following are methods of estimating validity except
a. test-retest validity
b. construct validity
c. content validity
d. concurrent validity
e. predictive validity
Answer: A
The four methods of estimation are construct, content, concurrent, and predictive. Test-retest
is a means of estimating reliability.
22) _______________ involves understanding the theoretical rationale underlying the
obtained measurement.
a. Convergent validity
b. Discriminant validity
c. Concurrent validity
d. Construct validity
e. Test-retest validity
Answer: D
Construct validity addresses the issue “Does the construct measure what we think it
measures?”
23) _______________ involves correlating two different measurements of the same
marketing phenomenon administered at the same point in time.
a. Split half reliability
b. Concurrent validity
c. Pragmatic judgment
d. Convergent validity
Answer: B
Concurrent validity evaluates a measure by comparing it to another measure of a
phenomenon, with both measures taken at the same time. It is excellent for validating a new
measurement instrument with another that has already been established.
24) _______________ involves the ability to use a measured marketing phenomenon at one
point in time to predict another marketing phenomenon at a future point.
a. Test-retest validity
b. Split half reliability
c. Predictive validity
d. Concurrent validity
Answer: C
Predictive validity involves measures at two different points of time, whereas concurrent
validity involves two measures at the same point in time.
25) The primary methods of measuring reliability include all of the following except
a. test-retest
b. construct
c. alternative-forms
d. split-half
Answer: B
Construct validity is a method for measuring validity, not reliability.
26) _______________ reliability involves repeated measurement of the same person or group
using the same scaling device under conditions judged to be similar.
a. Alternative-forms
b. Test-retest
c. Split-half
d. Concurrent
e. Repeated scale
Answer: B
Test-retest reliability uses repeated measures and compares the scores to determine similarity
in responses. The greater the discrepancy in scores, the greater the random error.
27) _______________ reliability involves giving the subject two forms of an instrument that
are judged to be equivalent but not identical.
a. Alternative-forms
b. Test-retest
c. Split-half
d. Concurrent
e. Repeated scale
Answer: A
By giving a subject two forms of the same instrument, scores can be compared to determine
the level of reliability.
28) The affective component of the hierarchy-of-effects model can be described by which of
the following terms?
a. feelings
b. beliefs
c. preference
d. a and b only
e. a and c only
Answer: D
The affective component of attitude is a person’s feelings, likes, dislikes, and preferences.
Beliefs and knowledge fall under the cognitive component.
29) _______________ refers to the various operational definitions developed to measure the
construct of attitude and accompanying beliefs, intentions, and other subjective mental states.
a. Attitude scaling
b. Predictive validity
c. Heterogeneity
d. Hierarchy-of-effects model
Answer: A
Attitude scaling refers to the various operational definitions that focus on measuring
respondents’ beliefs and their feelings.
30) If a researcher uses storytelling, word-association tests, and sentence completion, the
researcher is using the _______________ approach to direct communication.
a. self-reports
b. responses to stimuli
c. performance of objective tasks
d. overt behavior
Answer: B
With the responses to stimuli method, respondents are shown unstructured, partially
structured, or pictorial stimuli and are asked to express their reaction.
31) The most common tool of attitude measurement is
a. performance of objective tasks
b. self-reports
c. observed overt behavior
d. physiological reactions
e. responses to unstructured or partially structured stimuli
Answer: B
Of the general methods for measuring attitudes, self-reporting is by far the most widely used.
32) The simplest self-reporting scale is a(n) _______________, where the respondent’s
beliefs are classified in two or more categories.
a. nominal scale
b. ordinal scale
c. verbal rating scale
d. unidimensional measure
Answer: A
Nominal scales are the simplest form of self-reporting by classifying aspects of attitude into
categories.
33) Probably the most frequently used scales in all of social science, and certainly in
marketing research, are
a. nominal scales
b. graphic rating scales
c. verbal rating scales
d. semantic differential scales
Answer: C
Verbal rating scales are the most frequently used scales and ask respondents to indicate their
positions by selecting among verbally identified categories.
34) All of the following are legitimate arguments for assigning numbers to the response
categories of a verbal rating scale except
a. a neutral point is needed to allow respondents to demonstrate indifference to an issue
b. the researcher believes the respondents’ judgments can be treated as interval data
c. numerical labels can help guide the respondent as the progression of the categories and
demonstrate the scale is indeed ordinal, not merely a nominal set of labels
d. the numbers correspond to some quantity explicitly represented in the question itself, such
as a frequency
Answer: A
The issue of whether to include a neutral point is pertinent to the decision of whether to use
an odd or even number of categories, not to whether the categories are numbered.
35) A rating scale without a “no opinion” or “don’t know” category is an example of what
kind of scale?
a. balanced
b. unbalanced
c. noncomparative
d. forced
e. nominal
Answer: D
Forced scales force respondents to indicate a position, unless they leave the question blank.
36) _______________ scales provide respondents with an explicit referent or example with
which to compare while expressing their attitudes.
a. Noncomparative
b. Symmetric
c. Comparative
d. Asymmetric
Answer: C
With comparative scales, respondents are given a reference or comparison point.
37) Having respondents place various objects in strict order with regard to the attitude in
question is a
a. verbal rating scale
b. Stapel scale
c. semantic differential scale
d. rank-order scale
Answer: D
The rank-order technique involves having the respondent place various objects in a strict
order based on some attitude criteria.
38) In using a paired-comparison scale, how many evaluations would have to be made if five
objects are evaluated?
a. 5
b. 6
c. 8
d. 10
Answer: D
For n objects, [n(n – 1)/2] comparisons are required. [5(5 – 1)/2] = 5 × 2 = 10
39) The semantic differential utilizes
a. unipolar ten-point scales
b. bipolar adjectives
c. [n(n – 1)/2] comparisons
d. all of the above
Answer: B
The semantic differential typically requires the respondents to evaluate an object on a rating
scale bounded at each end by bipolar adjectives.
40) The _______________ scale is a unipolar scale, typically 10-point, with rating scales
from +5 to -5.
a. rank-order
b. paired-comparison
c. semantic differential
d. Stapel
Answer: D
Stapel scales are a modification of semantic differential scales, usually using endpoints of +5
and -5.
SHORT ANSWER
1) What are the four characteristics of the number system, and how do these correspond to the
four types of measurement scales?
Answer: 1) each number is unique
2) numbers are ordered
3) equal differences (e.g., 3–1 = 4–2)
4) equal ratios (e.g., 6/3 = 4/2)
Nominal scales meet the first characteristic only; ordinal scales meet the first two
characteristics; interval scales meet the first three; and ratio scales meet all four.
2) What are the components of measurement error and how do they relate to reliability and
validity?
Answer: The components of measurement error are systematic error, which causes a constant
bias in measurement, and random error. Validity refers to the extent to which the
measurement process is free from systematic error, reliability to the extent to which the
measurement process is free from random error.
[Instructors note that there is a discrepancy in the text regarding whether validity means
freedom from both systematic and random error or just systematic. See note in Instructor’s
Manual for Chapter 3 under “Validity and Reliability: Definition and Types” for detail.]
3) What is content validity?
Answer: Content validity involves a subjective judgment by an expert as to the
appropriateness of the measurement.
4) Which method of validity or reliability estimation involves dividing a multi-item
measurement device into equivalent groups and correlating the item responses?
Answer: Split-half reliability.
5) Identify and briefly describe the three components of attitude.
Answer: The three components are cognitive, affective, and behavioral. The cognitive
consists of knowledge and beliefs, affective of feelings and preferences, and behavioral of
readiness to act.
ESSAY
1) Explain the difference between a constitutive definition of a construct and an operational
definition.
Answer: A constitutive definition defines a construct with other constructs. An example
would be defining brand loyalty in terms of a consumer’s preference, attitude, or behavioral
response to a target brand over time. An operational definition specifies how a construct is to
be measured. Thus, it defines or gives meaning to a variable by spelling out what the
investigator must to do to measure it.
2) Identify and briefly describe the common potential sources of measurement error.
Answer: The potential sources of error are:
1) respondent characteristics—personal factors such as mood, fatigue, willingness to
participate
2) situational factors—variations in the environment in which measurements are reached,
such as time of day and temperature
3) data collecting factors—variations in how the questions are administered, such as phone,
personal contact, e-mail
4) measuring instrument factors—the degrees of ambiguity and difficulty of the questions and
the ability of the respondent to answer them
5) data analysis factors—errors made in the coding and tabulation process
3) Identify and discuss the four methods of estimating validity and the three methods of
estimating reliability.
Answer: The four methods of estimating validity are:
1) construct—understanding the theoretical rationale underlying the obtained measure. (It can
be estimated using convergent or discriminant validity. Convergent validity measures a
construct with independent measurement techniques and looks for a high correlation among
the measures. Discriminant validity involves demonstrating a lack of correlation amount
differing constructs.)
2) content—subjective judgments by experts as to the appropriateness of the measurement
3) concurrent—correlating two different measurements of the same marketing phenomenon
administered at the same point in time
4) predictive—the ability of a measured marketing phenomenon at one point in time to
predict another marketing phenomenon at a future point
The three main methods of estimating reliability are:
1) test-retest—repeated measurement of the same person or group using the same scaling
device under conditions judged to be similar
2) alternative-forms—giving the subject two forms judged equivalent, but not identical
3) split-half—dividing a multi-item measurement device into equivalent groups and
correlating the item responses
4) Discuss methods of attitude scaling, including the three categories of communication
methods and the two categories of observation methods.
Answer: Attitude scaling refers to the various operational definitions developed to measure
the construct of attitude and accompanying beliefs, intentions, and other subjective mental
states.
Communication techniques fall into three main categories:
1) self-reports—respondents are directly asked to report their beliefs or feelings by
responding to a questionnaire
2) responses to stimuli—respondents are shown unstructured, partially structured, or pictorial
stimuli and are asked to express their reaction
3) performance of tasks—respondents are asked to memorize and/or report factual
information about products, and responses are analyzed to uncover the nature of the
respondent’s beliefs and feelings
Observation techniques fall into two categories:
1) overt behavior—individuals are placed in a situation that allows their behavior patterns to
be exhibited and recorded, allowing inferences about underlying beliefs and feelings
2) physiological reactions—researchers measure physiological reactions (such as galvanic
skin response, eye dilation, or neuroimaging of the brain) as respondents are exposed to
products
5) Discuss the intrinsic issues related to the construction of useful, appropriate verbal rating
scales.
Answer: The issues are:
1) the overall number of categories
2) an odd or even number of categories
3) a balanced versus an unbalanced scale
4) the extent of verbal description
5) category numbering
6) a forced versus a nonforced scale
7) a comparative versus a noncomparative scale
8) a symmetric versus an asymmetric scale
9) scale direction
10) choice of endpoints
Test Bank for Modern Marketing Research: Concepts, Methods, and Cases
Fred M. Feinberg, Thomas Kinnear, James R. Taylor
9781133188964, 9781133191025, 9780759391710
