Measurement, Constructs, and Questionnaire Design

Introduction

Marketing decisions depend on information about concepts that are often impossible to observe directly. A manager may want to know whether customers are satisfied, whether a brand is perceived as trustworthy, whether a promotion increases purchase intention, or whether service quality has declined. None of these concepts is directly visible in the way that sales revenue or the number of transactions is visible. Marketing research therefore requires measurement: the systematic assignment of numbers or symbols to characteristics of objects, people, events, or responses according to clearly defined rules.

Measurement is not simply asking questions and recording answers. It involves translating abstract ideas into observable indicators, selecting an appropriate scale, designing understandable questions, and evaluating whether the resulting data are consistent and meaningful. Poor measurement can produce precise-looking statistics that do not represent the phenomenon being studied. Good measurement creates evidence that can support segmentation, product development, communication planning, customer experience management, and strategic decision-making.

This chapter explains how researchers operationalize constructs, distinguish variables and indicators, choose measurement scales, design and test questionnaires, and assess reliability and validity. It also considers measurement error, ethics, digital data collection, and the use of artificial intelligence in questionnaire development.

Learning Objectives

By the end of this chapter, you should be able to:

1. Distinguish constructs, variables, indicators, and operational definitions.
2. Convert abstract marketing concepts into measurable variables.
3. Apply nominal, ordinal, interval, and ratio measurement scales appropriately.
4. Use Likert, semantic differential, ranking, constant-sum, and behavioral measures.
5. Explain reliability and validity and select suitable assessment methods.
6. Identify major sources of measurement error and methods for reducing them.
7. Design, pilot, administer, and refine a professional marketing questionnaire.
8. Evaluate ethical, cultural, technological, and managerial implications of measurement choices.

1. From Marketing Concepts to Measurable Variables

Constructs and variables

A construct is an abstract concept developed to describe a phenomenon. Examples include brand loyalty, perceived value, involvement, trust, satisfaction, perceived risk, and service quality. Constructs are theoretical ideas; they cannot usually be observed directly.

A variable is a characteristic that can take different values across respondents, objects, or occasions. For example, satisfaction may be measured as a score from 1 to 7, while purchase frequency may be recorded as the number of purchases in the previous month.

An indicator is an observable measure used to represent a construct. Customer trust might be indicated by agreement with statements such as “This brand keeps its promises” and “I feel confident purchasing from this brand.” A construct may be represented by one indicator, but important and multidimensional constructs are usually measured with several indicators.

An operational definition specifies exactly how a construct will be measured in a particular study. For example:

In this study, customer loyalty is operationalized as the average response to four seven-point agreement statements concerning repurchase intention, recommendation intention, resistance to switching, and preference for the focal brand.

The operational definition makes the research observable, reproducible, and open to evaluation.

Conceptualization and operationalization

Conceptualization clarifies what a construct means and what it includes or excludes. Operationalization specifies how that construct will be observed and measured. These stages should not be confused. A researcher who begins by writing questions without defining the construct may inadvertently measure a different concept.

For example, “customer loyalty” may include behavioral loyalty, attitudinal commitment, and resistance to switching. If a study measures only the number of purchases, it may capture repeat behavior but not loyalty in situations where customers repurchase because of convenience or lack of alternatives.

A useful operationalization process is:

1. Define the construct theoretically.
2. Review relevant literature and established measurement instruments.
3. Identify dimensions of the construct.
4. Generate several candidate indicators for each dimension.
5. Consult subject-matter experts or experienced practitioners.
6. Select a response format and scale.
7. Pretest the items with members of the target population.
8. Assess reliability and validity after data collection.

Unidimensional and multidimensional constructs

A unidimensional construct represents one underlying attribute, such as perceived convenience. A multidimensional construct contains distinct but related dimensions. Service quality, for example, may involve reliability, responsiveness, assurance, empathy, and tangibles.

A multidimensional construct should not automatically be measured by one broad question. If management needs to know which aspect of service requires improvement, dimension-specific indicators are necessary. Researchers must also decide whether the dimensions form a broad overall construct or should be analyzed separately.

A measurement specification should identify:

2. Levels of Measurement

The level of measurement determines what comparisons and statistical operations are defensible. The four classical levels are nominal, ordinal, interval, and ratio.

Nominal measurement

A nominal scale classifies observations into mutually exclusive categories with no inherent order. Examples include gender identity, payment method, region, product category, and whether a respondent has used a service.

Numbers assigned to nominal categories are labels, not quantities. If “1” represents cash and “2” represents card payment, it is meaningless to calculate an average payment method. Appropriate summaries include frequencies, percentages, the mode, and cross-tabulations.

Categories should be exhaustive and mutually exclusive. “Age 18–24,” “25–34,” and “35–44” are mutually exclusive, but they are not exhaustive if respondents below 18 or above 44 are eligible. Adding “under 18,” “45 or older,” and “prefer not to say” may solve the problem.

Ordinal measurement

An ordinal scale ranks categories, but the distances between adjacent categories are not known to be equal. Examples include class rank, education level, agreement categories, and preference rankings.

If one customer ranks Brand A above Brand B, the researcher knows the order of preference but not how much stronger the preference is. Medians, percentiles, rank comparisons, and nonparametric procedures are appropriate. Researchers should be cautious when treating ordinal responses as if the distances were exactly equal.

Interval measurement

An interval scale has ordered values with equal intervals, but zero does not represent a complete absence of the measured attribute. Temperature measured in Celsius is a familiar example. In marketing research, attitude scales are frequently treated as approximately interval when they contain several balanced response categories and are combined into a composite score. This is a practical convention, not proof that the underlying psychological distance is perfectly equal.

With interval data, differences are meaningful. A score of 6 is two units higher than a score of 4, but a score of 6 cannot necessarily be interpreted as indicating twice as much of the construct as a score of 3.

Ratio measurement

A ratio scale has equal intervals and a meaningful zero. Examples include units sold, purchase amount, number of visits, waiting time, and age in years. Ratio comparisons are meaningful: 20 purchases represent twice as many as 10 purchases, assuming the same measurement rules.

The distinction between levels matters because the scale determines permissible interpretation. A researcher should not claim that a customer with a satisfaction score of 6 is twice as satisfied as a customer with a score of 3 merely because the response options are numbered.

3. Common Marketing Measurement Formats

Likert scales

A Likert item asks respondents to indicate the extent to which they agree or disagree with a statement. A typical seven-point format ranges from “strongly disagree” to “strongly agree.” A scale may also have five points. More response categories can provide greater discrimination, but only if respondents can meaningfully distinguish them.

A good Likert statement expresses one idea, uses familiar language, and has a clear reference period where relevant. “The delivery arrived when promised” is preferable to “The delivery was reliable and convenient,” which contains two ideas.

Balanced response options should include both favorable and unfavorable positions. A midpoint such as “neither agree nor disagree” may be appropriate when genuine neutrality is possible. A “not applicable” or “don’t know” option should be separate from the midpoint; lack of knowledge is not the same as neutrality.

Negatively worded items are sometimes used to reduce acquiescence, the tendency to agree with statements. However, excessive reverse wording can confuse respondents and introduce wording effects. If a reverse item is used, it must be clearly understood and correctly recoded before analysis.

For a multi-item scale, a simple composite score is often the arithmetic mean:

Here, S is the respondent’s composite score, x_i is the response to item i, and k is the number of valid items. This calculation assumes that the items represent the same construct, have been coded in the same direction, and that the researcher has established a defensible rule for missing responses. The mean remains on the original response scale, which aids interpretation.

Semantic differential scales

A semantic differential asks respondents to rate an object between pairs of bipolar adjectives, such as:

Untrustworthy 1 2 3 4 5 6 7 Trustworthy

Complicated 1 2 3 4 5 6 7 Simple

The two adjectives should represent genuinely opposite meanings. The scale is useful for measuring brand image, product personality, and perceived positioning. Adjectives should be familiar, relevant to the object, and not so ambiguous that respondents interpret them differently.

Stapel scales

A Stapel scale uses one adjective and a numerical range, commonly from −5 to +5, with no opposing adjective. It can be useful when finding a natural bipolar pair is difficult. However, respondents may find it less intuitive than a semantic differential.

Ranking measures

In a ranking question, respondents place alternatives in order, for example, ranking five smartphone features from most to least important. Ranking provides comparative information but does not show the size of the differences between alternatives. Asking respondents to rank too many items increases cognitive burden.

A paired-comparison format asks respondents to choose between two alternatives at a time. It can provide detailed preference information but becomes lengthy as the number of alternatives increases. The number of unique pairs among n alternatives is:

This formula applies when every pair is compared once. For example, six alternatives require 15 pairwise comparisons, which may be burdensome in a questionnaire.

Constant-sum scales

A constant-sum question asks respondents to allocate a fixed total, such as 100 points, across attributes. For example, a respondent may allocate 100 points among price, design, durability, and warranty according to purchase importance. The measure captures relative importance, but respondents may struggle when attributes are numerous or abstract. The total should be explicit, and the questionnaire should validate that the allocation equals the required sum.

Behavioral measures

Behavioral measures record actions or events rather than attitudes. Examples include purchase frequency, basket value, website visits, complaint counts, redemption rates, and time to repurchase. They may come from self-reports, transaction systems, loyalty programs, observation, or digital analytics.

Behavioral measures are often valuable because they relate to actual conduct, but they are not automatically superior. A transaction may not reveal motivation, satisfaction, or future intention. Self-reported behavior is vulnerable to memory error, while digital behavior can be distorted by shared accounts, device changes, tracking limitations, or automated traffic. Combining behavioral and attitudinal measures often produces a more complete explanation.

4. Reliability and Validity

Reliability

Reliability concerns the consistency or dependability of a measurement. A reliable instrument produces similar results when the underlying attribute has not changed, although reliability is always evaluated under particular conditions.

Several forms are important:

• Test–retest reliability examines stability by administering the same measure to the same respondents on two occasions. It is most appropriate when the construct should remain stable over the interval.
• Internal consistency examines whether items intended to measure the same construct are related to one another.
• Inter-rater reliability examines agreement among observers or coders, especially in content analysis and observational research.
• Parallel-forms reliability compares equivalent versions of an instrument.

Cronbach’s alpha is a commonly reported internal-consistency coefficient:

In this expression, k is the number of items, σ_i² is the variance of item i, and σ_t² is the variance of the total score formed by summing the items. Alpha assumes that items are scored consistently and are intended to measure a common construct. A high value does not prove unidimensionality or validity. Alpha can also increase simply because many similar items are included. Researchers should examine item content, corrected item–total relationships, and dimensionality rather than applying a universal cutoff mechanically.

Reliability is necessary but not sufficient. A bathroom scale that is consistently two kilograms too high may be reliable but not valid for measuring true body weight.

Validity

Validity concerns whether evidence supports the interpretation and use of scores for their intended purpose. Validity is not a permanent property of a questionnaire; it depends on the construct, population, context, and proposed interpretation.

Important forms of evidence include:

• Face validity: Whether items appear relevant and appropriate to informed readers. It is useful for communication but is not strong scientific evidence by itself.
• Content validity: Whether the items adequately cover the construct’s relevant domain. Experts and literature reviews are important.
• Construct validity: Whether the measure behaves as theory predicts. It includes convergent validity, in which related measures correlate, and discriminant validity, in which distinct constructs are not excessively correlated.
• Criterion-related validity: Whether scores relate to an external criterion, such as actual renewal, market share, or complaint behavior. Predictive validity concerns future outcomes; concurrent validity concerns outcomes measured at approximately the same time.
• Nomological validity: Whether the construct fits expected relationships within a broader theoretical network.

A questionnaire can be reliable without being valid, but a measure with very poor reliability cannot generally provide strong validity evidence because its scores contain substantial random variation.

5. Measurement Error

An observed score can be represented conceptually as:

Here, X is the observed score, T is the underlying true score under the measurement model, and E is measurement error. This is a conceptual model rather than a claim that a respondent possesses one perfectly known, unchanging true value.

Random error varies unpredictably across observations and tends to reduce consistency. Systematic error produces a consistent bias, such as a question that leads respondents toward a favorable answer. Both threaten the usefulness of research findings.

Common sources include:

• respondent misunderstanding or lack of knowledge;
• faulty memory and telescoping of events into the wrong time period;
• social desirability and fear of judgment;
• acquiescence, extreme-response, and midpoint-response styles;
• interviewer influence and inconsistent administration;
• ambiguous, double-barrelled, leading, or overly technical wording;
• order, context, priming, and fatigue effects;
• nonresponse and coverage error;
• translation and cultural interpretation differences;
• technical failures, duplicate submissions, and poor mobile design.

Researchers can reduce error by using short and specific questions, defining unfamiliar terms, providing appropriate time frames, separating sensitive topics, randomizing suitable item blocks, using neutral introductions, training interviewers, ensuring privacy, and conducting cognitive pretesting.

6. Designing an Effective Questionnaire

Questionnaire design should begin with the information requirements, not with a favorite question format. Every question should serve a research objective or a necessary screening, classification, or quality-control function.

Step 1: Specify the information required

Translate the management problem into research questions and variables. If the objective is to improve a mobile banking app, required information may include task success, perceived ease of use, security concerns, satisfaction, support experiences, and intended continued use.

Step 2: Define the target respondent and eligibility

A screening section should establish whether the person belongs to the population and has sufficient experience to answer. For example: “Have you personally used the app during the past 90 days?” Screening should be limited and respectful. Disqualified respondents should receive a clear, courteous termination message.

Step 3: Select existing or new measures

Established scales can improve comparability and provide prior evidence of reliability and validity. They must not be copied mechanically: wording, response format, language, and context may require adaptation. Permission and attribution should be obtained when required.

Step 4: Write clear questions

Effective questions:

• ask one thing at a time;
• use plain, concrete language;
• avoid unnecessary jargon;
• specify the object, time frame, and context;
• avoid assumptions that all respondents have knowledge;
• include all reasonable response categories;
• distinguish “not applicable” from “don’t know.”

Weak question: “How satisfied are you with the app’s speed and security?”

Improved questions: “How satisfied are you with the speed of the app?” and “How satisfied are you with the security of the app?”

Weak question: “Do you frequently purchase premium products?” The word “frequently” is subjective.

Improved question: “How many times have you purchased a premium product in the past three months?” with suitable numeric or categorical options.

Step 5: Choose response formats

Closed-ended questions simplify coding and comparison. Open-ended questions allow unexpected answers and are valuable in exploratory work, but they require coding and may be difficult for respondents. A mixed questionnaire often begins with structured measures and includes a small number of open questions for explanation.

For behavioral frequency, use meaningful categories or a numeric response with a defined period. For expenditure, specify currency and whether the amount refers to one transaction or a period. For sensitive financial or demographic questions, provide ranges and a “prefer not to say” option where appropriate.

Step 6: Establish logical order

A common sequence is:

1. introduction, consent, and screening;
2. easy questions about the relevant experience;
3. core evaluations and behavioral measures;
4. explanations or open-ended questions;
5. sensitive questions;
6. demographic and classification questions;
7. closing message.

This sequence is not universal. If earlier questions could prime later answers, researchers may randomize blocks or use alternative forms. Filter instructions should be implemented electronically where possible and tested carefully.

Step 7: Provide a professional introduction

The introduction should identify the research purpose in neutral terms, expected duration, voluntary nature, confidentiality or anonymity arrangements, contact information where appropriate, and any incentive conditions. It should not overstate confidentiality if identifiable data are retained.

Step 8: Format for accessibility and mobile use

Use readable typography, sufficient spacing, clear progress information, limited scrolling, and response options that work on small screens. Avoid matrix questions with many rows, particularly on mobile devices. Accessibility requires attention to screen readers, color contrast, keyboard navigation, language, and alternative text.

7. Pretesting, Piloting, and Refinement

A questionnaire should be tested before full deployment. Expert review checks content coverage, technical terminology, response categories, and alignment with objectives. Cognitive interviewing explores how respondents understand, retrieve, judge, and respond to each question. Techniques include think-aloud interviews and probing questions such as “What did that term mean to you?”

A pilot study administers the near-final questionnaire to a small sample resembling the target population. The researcher examines completion time, missing responses, item distributions, dropout points, technical problems, coding, and preliminary reliability. Pilot results should not be treated as definitive evidence from the main sample, especially when the pilot is small or nonprobability-based.

Questions should be revised when respondents consistently misunderstand them, use an unavailable response option, leave them blank, or report that the requested information cannot be recalled. Changes should be documented. If substantial changes are made after a pilot, another test may be necessary.

8. Applied Case: Measuring Customer Experience for FreshCart

Background

FreshCart, an illustrative online grocery retailer, experienced declining repeat orders despite stable delivery coverage. Management believed the problem might involve delivery reliability, product availability, app usability, or customer support, but its existing survey asked only one question: “How satisfied are you with FreshCart?”

Problem

The single item provided an overall score but did not identify the source of dissatisfaction. It also asked respondents to evaluate orders without specifying a time period, increasing recall variation. Management needed a measurement system that could diagnose service problems and relate perceptions to repeat behavior.

Analysis and approach

The research team defined overall experience satisfaction as the customer’s evaluative response to the most recent order. They specified four dimensions: ordering process, product availability, delivery performance, and problem resolution. They also added behavioral indicators: whether the customer had placed another order within 30 days, number of orders in the prior 90 days, and intention to reuse the service.

The questionnaire began with consent and a screening question confirming a recent order. It asked about the most recent order to reduce recall burden. Each service dimension used multiple seven-point satisfaction items, plus “not applicable” where necessary. A separate semantic differential assessed the brand as “unreliable–reliable,” “complicated–simple,” and “unresponsive–responsive.” The questionnaire ended with one open question: “What is the one improvement FreshCart should make first?”

The team conducted eight cognitive interviews. Participants interpreted “delivery reliability” inconsistently, so it was replaced with “delivery arrived within the time promised.” A pilot with 80 eligible customers revealed that the original matrix format led to skipped items on mobile phones. The team converted the matrix into individual screens and added a progress indicator.

After the main survey, the researchers examined item distributions, missingness, internal consistency, and relationships with actual repeat-order data. They found that delivery and product availability formed distinct dimensions, while two nearly identical support items added little information. The redundant item was removed. Satisfaction scores were then analyzed alongside behavioral records rather than being treated as a substitute for them.

Outcome

Management discovered that overall satisfaction was acceptable for most customers, but customers reporting substitutions without clear notification were substantially less likely to reorder. The company improved substitution alerts and added a delivery-exception protocol. Follow-up research used the revised questionnaire and transaction data to monitor whether the intervention improved both perceptions and repeat ordering.

Learning points

The case demonstrates that a broad question may be useful as an overall indicator but is insufficient for diagnosis. Construct definition, recent time frames, multiple indicators, mobile-friendly design, cognitive testing, and behavioral validation produced more actionable evidence. It also illustrates that reliability statistics alone would not have identified the managerial issue; content coverage and criterion-related evidence were equally important.

9. Ethics, Governance, and Contemporary Practice

Measurement decisions have ethical consequences. Researchers should collect only information necessary for the stated purpose, obtain informed consent, protect sensitive data, and explain how responses will be used. Sensitive questions should be optional where possible and should not expose respondents to avoidable embarrassment or risk. Incentives must not be coercively large for the target population.

Privacy regulations may require lawful processing, purpose limitation, data minimization, retention controls, access rights, and secure transfer. Researchers should distinguish anonymous data, where identity cannot reasonably be linked to responses, from confidential data, where identity is protected but retained by the research team.

Translation requires more than word-for-word substitution. Back translation, bilingual review, cognitive testing, and attention to culturally specific response styles help establish measurement equivalence. A scale that functions differently across countries or demographic groups may not support valid comparisons. Researchers should investigate differential item functioning or other evidence of measurement invariance when group comparisons are central.

Online platforms and AI tools can assist with item drafting, readability checks, translation suggestions, randomization programming, and detection of duplicate or nonsensical responses. However, AI-generated questions may contain leading language, cultural assumptions, fabricated citations, or privacy risks. A qualified researcher must verify every item, protect respondent data, document tool use, and retain human responsibility for construct definitions and final decisions. Automated attention checks should be used carefully because they can unfairly exclude careful respondents, people with disabilities, or respondents using translation tools.

Best-Practice Checklist

Before launching a questionnaire, confirm that:

• every question maps to a research objective;
• each construct has a defensible conceptual and operational definition;
• items are single-purpose, clear, neutral, and time-bounded where needed;
• response categories are exhaustive, mutually exclusive, and appropriate to the scale;
• “not applicable,” “don’t know,” and “prefer not to say” are used appropriately;
• established measures have been adapted and documented responsibly;
• the sequence and randomization rules have been tested;
• the instrument works on mobile devices and supports accessibility;
• cognitive interviews and a pilot have been completed;
• reliability and validity evidence will be evaluated;
• consent, privacy, retention, and data-security procedures are clear;
• changes made during pretesting are documented.

Chapter Summary

Measurement translates abstract marketing constructs into observable evidence. Constructs must be defined conceptually before they are operationalized through variables and indicators. Nominal scales classify, ordinal scales rank, interval scales represent equal differences without a meaningful zero, and ratio scales include a meaningful zero and support ratio comparisons.

Likert scales are useful for agreement or evaluation, semantic differentials measure perceived meanings and images, ranking measures comparative preference, constant-sum measures relative allocation, and behavioral measures record actions. No format is universally best; the choice depends on the construct, respondent capability, analytical purpose, and context.

Reliability concerns consistency, while validity concerns the defensibility of score interpretations and uses. A reliable measure may still be invalid. Measurement error arises from respondents, instruments, interviewers, contexts, sampling processes, translation, and technology. Clear wording, suitable response options, privacy, cognitive testing, piloting, and careful administration reduce error.

Professional questionnaire design is an iterative process involving information requirements, operational definitions, item writing, response-format selection, logical ordering, accessibility, pretesting, and refinement. Ethical research protects respondents and uses data only for legitimate, transparent purposes. Contemporary digital and AI tools can improve efficiency, but they do not replace theoretical reasoning, human review, or responsibility for validity.

Key Terms

Construct: An abstract theoretical concept such as satisfaction, trust, or loyalty.

Variable: A characteristic that can take different values across cases.

Indicator: An observable measure used to represent a construct.

Operational definition: A precise description of how a construct will be measured.

Conceptualization: Clarification of a construct’s meaning and boundaries.

Nominal scale: A classification scale with categories but no order.

Ordinal scale: A scale with ordered categories but unknown equal distances.

Interval scale: A scale with equal intervals but no meaningful absolute zero.

Ratio scale: A scale with equal intervals and a meaningful zero.

Likert scale: A response scale measuring agreement or disagreement with statements.

Semantic differential: A scale positioned between bipolar descriptive adjectives.

Ranking measure: A method requiring respondents to order alternatives.

Constant-sum scale: A measure requiring allocation of a fixed total across alternatives.

Reliability: The consistency or dependability of a measurement.

Validity: Evidence supporting the interpretation and use of measurement scores.

Measurement error: The difference between an observed score and the intended underlying value.

Cognitive interview: A pretest exploring how respondents understand and answer questions.

Pilot study: A small-scale trial of a near-final research instrument.

Internal consistency: The degree to which items intended to measure one construct relate to one another.

Measurement invariance: Evidence that a measure functions comparably across groups or contexts.

Review Questions

5 Short Answer Questions

1. Distinguish a construct from an indicator and give a marketing example of each.
2. Why is a true zero important in distinguishing ratio from interval measurement?
3. What is the difference between reliability and validity?
4. Identify two problems caused by double-barrelled questions.
5. Why should “not applicable” be distinguished from a neutral response?

5 Long Answer Questions

1. Explain the process of operationalizing a multidimensional marketing construct, using service quality or brand loyalty as an example.
2. Compare nominal, ordinal, interval, and ratio scales and discuss the implications of each for marketing analysis.
3. Describe the major forms of reliability and validity evidence relevant to a multi-item questionnaire.
4. Explain how questionnaire wording, question order, response formats, and administration can create measurement error.
5. Develop a complete design plan for a customer-experience questionnaire, including pretesting, piloting, ethics, and data quality controls.

3 Analytical / Application Questions

1. A retailer asks, “How satisfied are you with our prices, product range, and delivery speed?” Diagnose the measurement problems and rewrite the instrument for valid analysis.
2. A brand-awareness study finds that respondents recognize the brand but rarely purchase it. Propose a measurement framework that distinguishes awareness, attitudes, intentions, and behavior.
3. A seven-item loyalty scale has high internal consistency, but its scores are unrelated to renewal behavior and experts say it omits switching resistance. Analyze what this implies about reliability, content validity, and criterion-related validity, and recommend next steps.

Multiple Choice Questions with Answers

Question 1

Which statement best defines an operational definition?

A. A theoretical explanation of why customers behave in a certain way

B. A precise specification of how a construct will be measured in a study

C. A list of all respondents included in a sample

D. A statistical test used to compare two means

Correct Answer: B. A precise specification of how a construct will be measured in a study

Question 2

Which measurement scale classifies payment methods such as cash, card, and mobile wallet without implying an order?

A. Nominal

B. Ordinal

C. Interval

D. Ratio

Correct Answer: A. Nominal

Question 3

A semantic differential scale typically asks respondents to:

A. Allocate exactly 100 points among attributes

B. Rank alternatives from first to last

C. Rate an object between bipolar adjectives

D. Report the number of purchases made

Correct Answer: C. Rate an object between bipolar adjectives

Question 4

What does test–retest reliability primarily assess?

A. Whether items cover all dimensions of a construct

B. Whether scores are stable when the construct should remain stable

C. Whether a scale predicts a future criterion

D. Whether categories are mutually exclusive

Correct Answer: B. Whether scores are stable when the construct should remain stable

Question 5

A measure can be highly reliable but invalid when it:

A. Produces consistent scores that do not represent the intended construct

B. Contains several response categories

C. Uses a time frame for behavioral recall

D. Includes both positive and negative items

Correct Answer: A. Produces consistent scores that do not represent the intended construct

Question 6

Which question is double-barrelled?

A. How many orders did you place last month?

B. How satisfied are you with the delivery time and product packaging?

C. Which payment method did you use?

D. Have you used the service in the past 90 days?

Correct Answer: B. How satisfied are you with the delivery time and product packaging?

Question 7

In the conceptual model X = T + E, what does E represent?

A. The sampling frame

B. The observed score

C. Measurement error

D. The target population

Correct Answer: C. Measurement error

Question 8

Which activity is most directly associated with cognitive interviewing?

A. Estimating market share from transaction records

B. Exploring how respondents interpret and answer questionnaire items

C. Calculating the number of pairwise comparisons

D. Assigning respondents to geographic strata

Correct Answer: B. Exploring how respondents interpret and answer questionnaire items

Question 9

If six alternatives are compared in every possible unique pair, how many paired comparisons are required?

A. 6

B. 12

C. 15

D. 36

Correct Answer: C. 15

Question 10

Which practice best reduces recall error in a customer-experience survey?

A. Using vague terms such as “often”

B. Asking respondents to evaluate an unspecified period

C. Focusing questions on a clearly defined recent experience

D. Removing all behavioral questions

Correct Answer: C. Focusing questions on a clearly defined recent experience