Research Methods and Consumer Insight

Introduction

Sensory branding and experiential marketing depend on a deceptively difficult question: what does an experience actually do to a consumer? A brand may claim that a signature scent increases approach behavior, that tactile packaging communicates premium quality, or that an immersive retail environment strengthens loyalty. Such claims cannot be evaluated adequately through preference statements alone. Consumers may not notice the relevant stimulus, may be unable to articulate its influence, or may report what they believe is socially desirable rather than what shaped their behavior.

Research in this field therefore requires methodological pluralism. Qualitative inquiry can reveal meanings and cultural associations; quantitative research can estimate prevalence and test relationships; observation can capture behavior in context; ethnography can explain practices; biometric and neuromarketing methods can measure physiological or attentional responses; experiments can test causal effects; and sensory testing can evaluate the detection, discrimination, and hedonic consequences of stimuli.

The central methodological principle is alignment: the research question, construct, method, sample, analysis, and managerial decision must fit together. No method is inherently superior. A sophisticated study often combines methods in a staged design, using one method to discover relevant phenomena, another to measure their distribution, and a third to test whether an intervention causes a meaningful outcome.

Learning objectives

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

1. Distinguish major qualitative, quantitative, observational, ethnographic, biometric, neuromarketing, experimental, and sensory-testing methods.
2. Select methods according to research questions, constructs, context, and managerial decisions.
3. Evaluate reliability, validity, causality, sampling quality, and interpretation.
4. Design a mixed-method study of a sensory or experiential brand intervention.
5. Recognize ethical, cultural, technological, and practical limitations.
6. Translate consumer insight into defensible brand-management action.

1. From managerial problem to researchable question

A managerial problem is usually broad: “Our store experience is not distinctive.” A researchable question specifies the phenomenon, population, context, comparison, and outcome. For example: “Among current and prospective customers, does a warm ambient scent, compared with an unscented environment, increase perceived brand warmth and dwell time in a controlled retail simulation?”

A useful sequence is:

Decision problem → construct definition → research question → design → evidence → interpretation → action.

A construct is an abstract concept that cannot be observed directly, such as sensory congruence, perceived authenticity, immersion, arousal, brand attachment, or purchase intention. An operational definition states how the construct will be represented empirically. “Brand warmth,” for instance, might be measured through a validated multi-item scale, spontaneous verbal descriptions, facial expression coding, or approach behavior. These measures are related but not interchangeable.

Researchers should distinguish at least four levels of outcome:

• Perception: whether the stimulus is noticed or detected.
• Experience: how it is interpreted and felt.
• Response: attention, emotion, memory, evaluation, or behavior.
• Business consequence: conversion, retention, advocacy, revenue, or customer lifetime value.

A scent may be detected without being liked; liked without improving memory; and associated with positive memory without increasing sales. Research must therefore avoid treating a single proxy as the entire customer journey.

2. Research paradigms and design logic

A positivist or post-positivist approach emphasizes measurement, comparison, prediction, and testing. It is valuable when the organization needs estimates or causal evidence. An interpretivist approach examines how consumers construct meaning through language, culture, identity, and social practice. It is essential when the meaning of an experience is not yet understood. A pragmatic approach selects methods according to the decision and accepts that different forms of evidence answer different questions.

Exploratory, descriptive, and explanatory designs

Exploratory research investigates poorly understood experiences. It commonly uses interviews, focus groups, mobile diaries, social listening, and ethnography.

Descriptive research estimates what is occurring and for whom. Surveys, structured observation, panel studies, and sensory tests are common.

Explanatory research examines why or whether an effect occurs. Experiments, quasi-experiments, longitudinal designs, and causal modeling are relevant.

A cross-sectional survey can identify an association between perceived multisensory coherence and brand preference, but it cannot by itself establish that coherence caused preference. An experiment can provide stronger causal evidence, but may simplify the real-world environment. A longitudinal field study can examine persistence, adaptation, and delayed effects, although it is more expensive and vulnerable to attrition.

3. Qualitative methods: meaning, language, and lived experience

Interviews

Semi-structured interviews use a topic guide while allowing participants to explain unexpected meanings. They are appropriate for exploring sensory memories, identity, cultural symbolism, discomfort, accessibility, and the vocabulary consumers use spontaneously.

Good interview questions are open, non-leading, and experience-based. “Tell me about the last time a brand environment made you feel comfortable” is generally better than “Did the relaxing music make you comfortable?” Probing should clarify sequence, context, emotion, interpretation, and behavior. Researchers should distinguish what participants experienced from what they infer after reflection.

Focus groups

Focus groups reveal shared language, disagreement, norms, and symbolic negotiation. They are useful for comparing reactions to packaging, soundscapes, or retail concepts. However, dominant participants, conformity, and public self-presentation may suppress private or socially sensitive views. Groups should be composed with attention to power differences, familiarity, language, and the possibility that participants influence one another.

Projective and elicitation techniques

Collage, sensory sorting, metaphor elicitation, image selection, object association, and personification may help participants express tacit meanings. These techniques are not psychological “truth machines.” Their value lies in generating interpretive material that must be analyzed transparently and, where necessary, corroborated with other evidence.

Qualitative analysis

A defensible analysis involves familiarization, coding, category development, comparison, interpretation, and reporting. Codes may be deductive, based on theory, or inductive, emerging from the data. Researchers should maintain an audit trail showing how raw material became themes. Reflexivity is important: the researcher’s assumptions, sensory preferences, identity, and relationship with the sponsor may shape data collection and interpretation.

Intercoder agreement can support consistency, but it does not guarantee validity. A highly consistent coding scheme can still represent a poor conceptualization. Quality is strengthened through negative-case analysis, member reflection where appropriate, triangulation, peer debriefing, and thick contextual description.

4. Quantitative methods and measurement

Surveys and scales

Surveys are useful for estimating attitudes, awareness, perceived congruence, sensory sensitivity, emotional response, and behavioral intentions across a defined population. Items should be clear, singular, balanced where appropriate, and suitable for the respondent’s language and culture. A multi-item construct is generally preferable to one ambiguous question when the concept is broad.

Reliability concerns consistency. Internal consistency is often summarized using coefficient alpha, although alpha depends on assumptions such as tau-equivalence and can be inflated by many redundant items. A common expression is:

Here, k is the number of items, σᵢ² is the variance of item i, and σₜ² is the variance of the summed scale. Alpha is not a measure of unidimensionality, validity, or causal importance. It should be interpreted alongside item content, factor structure, and test–retest evidence. For ordinal or congeneric measures, alternatives such as omega may be more appropriate.

Validity concerns whether the evidence supports the intended interpretation. Content validity asks whether the measure adequately covers the construct. Construct validity includes convergent and discriminant evidence. Criterion-related validity examines relationships with relevant outcomes. Ecological validity concerns whether findings generalize to realistic contexts.

Sampling

Probability sampling gives known, non-zero selection probabilities and supports statistical generalization when implemented well. Simple random, stratified, cluster, and systematic sampling are examples. Non-probability methods—including convenience, purposive, quota, and respondent-driven approaches—may be appropriate for exploratory work or hard-to-reach communities, but they do not automatically represent a population.

A sample can be large yet biased. Coverage error occurs when parts of the target population are missing from the sampling frame. Nonresponse bias occurs when participation is related to the topic or outcome. Self-selection is especially likely in sensory research because people with strong preferences or sensitivities may be more willing to participate.

For a proportion estimated with a simple random sample, an approximate margin of error is:

Here, p is the estimated population proportion, n is the sample size, and z is the standard-normal critical value for the selected confidence level. The approximation assumes a probability sample, independence, and adequate sample size; finite-population, clustering, weighting, and design effects may alter it. It does not correct for noncoverage or nonresponse bias. Sample-size planning should also consider the smallest effect that is managerially meaningful, statistical power, attrition, and the analysis model.

Quantitative analysis

Descriptive statistics should precede modeling. Researchers should inspect distributions, missingness, outliers, scale direction, subgroup patterns, and measurement invariance. Correlation indicates association, not causation. Regression can adjust for measured covariates, but statistical adjustment does not eliminate unmeasured confounding or reverse causality.

When comparing groups, researchers should report effect sizes and uncertainty, not only p-values. A statistically significant result may be commercially trivial, while a practically important effect may be estimated imprecisely in a small sample. Multiple comparisons, researcher degrees of freedom, selective reporting, and post hoc subgroup analysis increase the risk of false discoveries. Preregistration, transparent exclusion rules, replication, and reporting of null results improve credibility.

5. Observation and ethnography

Observation records behavior in context rather than relying entirely on recall. Structured observation may code approach, touch, dwell time, path, queue abandonment, interaction with displays, and social behavior. It requires an explicit coding frame, observer training, and checks for inter-observer consistency. Cameras and digital traces create additional privacy responsibilities.

Ethnography is deeper and more contextual. It may involve participant observation, home visits, store shadowing, cultural analysis, diaries, and artifact interpretation over an extended period. Ethnography can reveal that a supposedly individual sensory preference is shaped by family routines, religious practice, workplace norms, disability, climate, or local retail conventions.

The observer effect is a persistent limitation: people may alter behavior when watched. Covert observation may reduce reactivity but raises serious ethical concerns and should not be used merely for convenience. In digital ethnography, publicly accessible content is not automatically ethically unrestricted; expectations of privacy, platform terms, vulnerability, and traceability must be considered.

6. Biometric and neuromarketing methods

Biometric methods measure physiological or behavioral signals associated with attention, arousal, workload, or affect. Examples include eye tracking, pupil diameter, electrodermal activity, heart-rate measures, facial action coding, respiration, and sometimes electroencephalography (EEG) or functional neuroimaging.

These measures are informative but rarely psychologically unique. Increased electrodermal activity can reflect excitement, anxiety, effort, temperature, or movement. A pupil response can be influenced by luminance as well as cognitive or emotional processes. Eye fixation demonstrates visual attention, not necessarily liking, comprehension, or purchase. Facial movements may be absent, culturally moderated, voluntarily controlled, or difficult to interpret in masked populations.

Neuromarketing should therefore be treated as measurement of responses, not access to unconscious consumer truth. Strong studies use stimulus controls, baseline periods, counterbalancing, preregistered preprocessing, artifact rejection, appropriate sampling rates, and theoretically justified outcome measures. Commercial claims should not exceed the evidence.

A simple signal-to-noise perspective is useful. If an observed measure Y consists of a systematic response S plus measurement error E, then:

This is a conceptual measurement model, not a claim that all error is random. Systematic confounds—such as lighting, room temperature, device placement, or participant movement—can bias results and are not solved by averaging alone.

7. Experimental and quasi-experimental research

Experiments manipulate an independent variable, observe an outcome, and use random assignment to make treatment groups comparable in expectation. A basic sensory-brand experiment might randomly assign participants to a congruent scent, incongruent scent, or no-scent condition while holding visual identity, price, and task constant.

The causal estimand should be stated clearly. For a treatment T and outcome Y, the average treatment effect is conceptually:

Here, Y(1) is the potential outcome under treatment, Y(0) is the potential outcome under control, and E denotes the population or sample average under the stated design. The fundamental problem is that the same individual cannot be observed simultaneously under both conditions. Randomization provides a design-based solution under assumptions including consistency, adequate treatment implementation, and appropriate handling of interference and missing outcomes.

Internal validity can be threatened by unequal attrition, demand characteristics, contamination between conditions, manipulation failure, order effects, novelty, and experimenter expectancy. External validity can be limited when a laboratory scent strip or virtual store differs from an actual store over repeated visits.

Within-subject designs increase precision by allowing each participant to serve as a comparison, but they create carryover, fatigue, learning, and sensitization risks. Between-subject designs reduce these risks but usually require more participants. Counterbalancing, washout periods, manipulation checks, and realistic exposure durations are essential.

Quasi-experiments are useful when randomization is impractical. Difference-in-differences designs compare changes over time in a treated unit with changes in a comparison unit. Their credibility depends on assumptions such as parallel pre-treatment trends. Interrupted time series, matched comparisons, natural experiments, and regression discontinuity designs can be valuable, but each requires explicit design assumptions rather than causal language by default.

8. Sensory testing

Sensory testing differs from ordinary consumer research because it focuses on human responses to products or stimuli under controlled presentation. Three broad purposes are common:

• Analytical tests: detect or discriminate differences, such as triangle tests or paired comparisons.
• Descriptive analysis: characterize attributes and intensities using trained panels.
• Affective or consumer tests: measure liking, acceptance, preference, purchase interest, or fit with a brand.

Panel selection must match the purpose. Trained assessors may provide reliable attribute descriptions but may not represent target consumers. Consumer panels provide market relevance but greater individual variability. Protocols should standardize serving order, coding, temperature, lighting, cleansing intervals, exposure duration, and instructions. Randomized or balanced presentation reduces order and position effects.

Sensory data are context-dependent. A stimulus that is preferred in isolation may be unsuitable in a brand environment because of incongruence, fatigue, cultural meaning, or interaction with other senses. Repeated exposure may generate adaptation or habituation. Screening for allergies, asthma, migraine, sensory processing differences, and other relevant risks is an ethical necessity, not merely a procedural detail.

9. Mixed methods and triangulation

Mixed methods integrate qualitative and quantitative evidence in a single research program. A sequential exploratory design may begin with ethnography and interviews, develop constructs and stimuli, then use a survey and experiment. A sequential explanatory design may begin with a quantitative pattern and use interviews to explain an unexpected subgroup difference. A convergent design collects qualitative and quantitative evidence in parallel and compares interpretations.

Triangulation does not mean forcing all methods to produce identical results. Divergence may reveal context dependence, measurement mismatch, or a meaningful distinction between stated and observed behavior. A useful integration matrix maps each research question to its construct, method, evidence, limitation, and managerial implication.

For example, a retail scent project might find that interviews describe “comfort,” observation shows longer dwell time, a biometric measure indicates increased arousal, and a randomized test shows no increase in purchase. The responsible conclusion is not that the scent “works.” It may enhance engagement without improving conversion, or the purchase task may be underpowered, unrealistic, or mediated by price and assortment.

10. Interpretation, causal reasoning, and managerial translation

Interpretation should separate four claims:

1. What was measured?
2. What pattern was observed?
3. What explanation is plausible?
4. What decision is justified?

A finding that an ambient sound is associated with premium ratings does not prove that sound caused the ratings. A causal claim requires temporal precedence, covariation, and credible exclusion of alternative explanations. Even then, causal effects may be heterogeneous: a stimulus can help some consumers, have no effect on others, and harm consumers with sensory sensitivities.

Managers should evaluate effect magnitude, operational feasibility, cost, brand fit, equity implications, and downside risk. A small average effect may justify implementation if it is inexpensive and consistent with strategy; a larger effect may not be actionable if it requires intrusive data collection or produces unacceptable discomfort.

11. Applied case: developing a multisensory hotel identity

The following is an illustrative case, not a report of an actual company.

Background

A regional hotel group sought a distinctive sensory identity across its lobby, mobile booking experience, and guest rooms. Management proposed a woody ambient scent, low-tempo instrumental sound, textured paper, and warm lighting. The concern was that the concept reflected executives’ preferences rather than guest insight.

Problem

The group needed to determine whether the proposed cues communicated calm, local authenticity, and premium quality; whether the cues worked together; and whether they improved guest outcomes without causing discomfort or reducing accessibility.

Analysis and approach

The research team used four stages. First, ethnographic visits to competing hotels and in-room diary tasks explored arrival rituals, travel fatigue, local identity, and sensory irritants. Interviews included frequent travelers, families, older guests, international visitors, and participants who reported scent sensitivity.

Second, a qualitative stimulus workshop compared scent, sound, paper, and lighting concepts. Participants described meanings, associations, and perceived congruence rather than merely rating attractiveness. The team learned that “woody” could signify authenticity to some participants but cleaning products or smoke to others.

Third, a controlled experiment used a simulated lobby with three conditions: visual identity only, visual identity plus congruent sensory cues, and visual identity plus deliberately incongruent cues. Participants were randomly assigned. Outcomes included perceived brand authenticity, comfort, memory after a delay, booking intention, and observed time spent exploring information. A manipulation check tested whether participants detected the cues and whether they interpreted them as intended.

Fourth, a four-week pilot in two hotels used a matched comparison property. Observation measured queue behavior and information interaction; a short post-stay survey measured experience; operational data tracked complaints, upgrades, and repeat booking indicators. The pilot avoided collecting identifiable biometric data from guests.

Outcome

The experiment suggested that the congruent package improved perceived coherence and authenticity relative to visual identity alone, but only among participants who detected and liked the scent. The incongruent package reduced comfort. Ethnography showed that some guests preferred scent-free rooms and that local authenticity was conveyed more reliably through materials, staff interaction, and regionally relevant content than through scent alone.

The pilot produced no clear overall booking increase, but it generated fewer positive effects than management expected: lobby exploration increased modestly, while room-level scent complaints appeared among a small but important group. The group therefore retained tactile materials, local sound elements at low volume, and warm lighting in public areas, but introduced scent-free zones, transparent guest controls, ventilation standards, and opt-out room policies.

Learning points

The case demonstrates that sensory congruence is not equivalent to universal liking. A mixed-method design identified meanings, tested causal effects, examined behavior, and surfaced inclusion risks. It also showed why a positive laboratory result should not be translated directly into full-scale rollout. The appropriate outcome was not a single “winning scent,” but a differentiated and governable experience system.

12. Ethics, governance, and contemporary developments

Sensory and biometric research can affect bodily comfort, privacy, autonomy, and dignity. Informed consent should explain procedures, foreseeable discomfort, recording, storage, withdrawal rights, and data use. Researchers should minimize deception, debrief when deception is necessary, and obtain ethics review for vulnerable populations or sensitive physiological data.

Data governance should follow purpose limitation, data minimization, access control, retention limits, and secure deletion. Facial images, eye movements, voice recordings, and physiological signals may become identifiable when combined with other data. “Anonymous” should not be claimed unless re-identification risk has been carefully assessed.

Accessibility is part of research validity. Samples and protocols should include people with disabilities and sensory differences where relevant. A brand experience that performs well for an average participant may exclude people with migraine, autism, hearing loss, visual impairment, allergies, or culturally specific restrictions.

AI can assist with transcription, semantic clustering, stimulus generation, survey personalization, and anomaly detection. It can also reproduce cultural stereotypes, hallucinate themes, obscure analytic decisions, and intensify privacy risk. Human researchers remain responsible for construct definition, consent, validation, contextual interpretation, and final decisions. AI-generated coding should be audited against human-coded examples, and confidential data should not be transferred to systems without appropriate contractual and technical safeguards.

Best-practice checklist

• Begin with a decision and a precise research question.
• Define constructs before selecting measures.
• Use qualitative work to understand meanings and generate hypotheses.
• Use probability sampling when population generalization is required.
• Randomize experimental conditions and verify manipulation success.
• Distinguish detection, liking, arousal, memory, intention, and behavior.
• Report uncertainty, effect sizes, missingness, exclusions, and limitations.
• Test subgroup and accessibility implications where theoretically justified.
• Treat biometric signals as context-dependent indicators, not direct emotions.
• Integrate methods transparently rather than averaging incompatible evidence.
• Protect privacy and avoid collecting data that the decision does not require.
• Replicate important findings in realistic settings before scaling.

Chapter Summary

Research in sensory branding and experiential marketing is strongest when methods are matched to questions. Qualitative interviews, focus groups, elicitation, and ethnography reveal meanings, practices, and cultural context. Surveys and structured observation estimate patterns, while sensory testing distinguishes detection, discrimination, description, and consumer acceptance. Biometric and neuromarketing measures can enrich understanding of attention and physiological response, but their signals are not uniquely interpretable. Experiments and credible quasi-experiments provide the strongest basis for causal claims, provided that randomization, manipulation, measurement, and assumptions are sound.

Reliability, validity, sampling quality, ecological realism, and ethical governance must be evaluated together. Mixed methods can connect lived experience, measured response, observed behavior, and business outcomes while preserving the distinctions among them. Ultimately, consumer insight is not a colorful collection of findings; it is a defensible interpretation that improves a decision while acknowledging uncertainty, heterogeneity, and limitations.

Key Terms

Construct: An abstract concept represented through observable indicators.

Operationalization: The process of defining how a construct will be measured or manipulated.

Qualitative research: Inquiry focused on meanings, experiences, language, and context.

Quantitative research: Inquiry using numerical measurement and statistical analysis.

Ethnography: Contextual study of people’s practices, cultures, and lived environments.

Observation: Systematic recording of behavior in a natural or designed setting.

Biometric measure: Physiological or behavioral measurement associated with response.

Neuromarketing: Application of neuroscience-related methods to marketing questions.

Sensory testing: Controlled evaluation of human responses to products or sensory stimuli.

Reliability: Consistency of a measurement or procedure.

Validity: The degree to which evidence supports an intended interpretation.

Sampling error: Random discrepancy between a sample estimate and a population value.

Nonresponse bias: Bias arising when participation relates to the outcome or topic.

Internal validity: Credibility of a causal conclusion within a study.

External validity: Generalizability across people, settings, stimuli, or times.

Manipulation check: Measure confirming whether an experimental treatment was experienced as intended.

Triangulation: Use of multiple methods, sources, or perspectives to strengthen interpretation.

Causal estimand: Precisely defined causal quantity, such as an average treatment effect.

Researcher reflexivity: Examination of how the researcher influences the research process.

Ecological validity: Relevance of findings to realistic contexts and behavior.

Review Questions

Short Answer Questions

1. Why is a managerial problem different from a researchable question?
2. Distinguish reliability from validity.
3. What can ethnography reveal that a standardized survey may miss?
4. Why does a biometric signal rarely identify one specific emotion?
5. What is the purpose of a manipulation check in an experiment?

Long Answer Questions

1. Compare interviews, focus groups, surveys, and ethnography for researching sensory brand meaning.
2. Explain how sampling, nonresponse, and ecological validity affect interpretation of consumer research.
3. Discuss the conditions required for a credible causal claim in a sensory-brand experiment.
4. Evaluate the benefits and limitations of biometric and neuromarketing methods.
5. Design a mixed-method research program for testing a new multisensory retail identity.

Analytical / Application Questions

1. A scent increases dwell time but also increases complaints among a minority of visitors. How should the brand interpret and act on this finding?
2. A survey finds that perceived sensory congruence predicts purchase intention, but a randomized field test finds no sales increase. Develop at least three plausible explanations and propose follow-up research.
3. Construct a research plan that tests whether a soundscape communicates local authenticity across two countries while addressing measurement equivalence, cultural interpretation, and ethical concerns.

Multiple Choice Questions with Answers

Question 1

Which method is most appropriate for discovering how consumers culturally interpret a new brand scent before scale measurement?

A. Ethnographic and qualitative interviewing

B. A large sales forecast

C. A post hoc significance test

D. Random assignment without participant discussion

Correct Answer: A. Ethnographic and qualitative interviewing

Question 2

What does random assignment primarily strengthen in an experiment?

A. Population coverage

B. Internal causal validity

C. Cultural translation

D. Content validity of a scale

Correct Answer: B. Internal causal validity

Question 3

A high value of coefficient alpha does not by itself establish which conclusion?

A. That items have some internal consistency

B. That the scale is necessarily valid and unidimensional

C. That item responses can be summarized in analysis

D. That item variance contributes to the scale estimate

Correct Answer: B. That the scale is necessarily valid and unidimensional

Question 4

Why should eye fixation not automatically be interpreted as brand liking?

A. Fixation is unrelated to visual attention

B. Fixation can indicate attention without indicating positive evaluation

C. Eye tracking measures only purchase behavior

D. Fixation cannot be recorded in experiments

Correct Answer: B. Fixation can indicate attention without indicating positive evaluation

Question 5

Which statement best describes a non-probability convenience sample?

A. Every population member has a known selection probability

B. It always produces unbiased population estimates

C. Participants are selected through accessibility or availability

D. It eliminates nonresponse bias

Correct Answer: C. Participants are selected through accessibility or availability

Question 6

In sensory testing, what is the main purpose of randomizing presentation order?

A. To increase the fragrance intensity

B. To reduce order and position effects

C. To guarantee representative sampling

D. To replace informed consent

Correct Answer: B. To reduce order and position effects

Question 7

Which finding most directly supports a causal claim that a scent changed purchase intention?

A. People who like scents report higher purchase intention

B. Scent preference correlates with brand attachment

C. Randomly assigned participants exposed to the scent show higher intention than controls

D. Managers believe the scent fits the brand

Correct Answer: C. Randomly assigned participants exposed to the scent show higher intention than controls

Question 8

What is a central ethical concern in biometric consumer research?

A. Physiological data are always impossible to store

B. Biometric signals may create privacy and re-identification risks

C. Biometric data cannot vary across individuals

D. Consent is unnecessary when data are numerical

Correct Answer: B. Biometric signals may create privacy and re-identification risks

Question 9

If a treatment increases positive responses only among consumers who detect and like the stimulus, this illustrates which issue?

A. Treatment-effect heterogeneity

B. Perfect external validity

C. Guaranteed measurement invariance

D. Sampling frame completeness

Correct Answer: A. Treatment-effect heterogeneity

Question 10

What is the most defensible interpretation when qualitative, biometric, and sales findings diverge?

A. Discard all findings except sales data

B. Assume the biometric result is unconscious truth

C. Investigate context, construct differences, and methodological limitations

D. Average all results into one score

Correct Answer: C. Investigate context, construct differences, and methodological limitations