Decision Problems, Research Questions and Theory

Introduction

Marketing research begins before a questionnaire is written, a sample is selected, or statistical software is opened. It begins with a decision that matters to an organisation and with a disciplined effort to determine what must be known before that decision can be made responsibly. A manager may say, “Sales are declining,” “Customers are abandoning the digital channel,” or “Should we launch this product?” These statements are important, but they are not yet well-formed research problems. They describe symptoms, concerns, or possible actions. The researcher’s task is to translate them into questions that can be answered with credible evidence.

This translation is difficult because managerial decisions and research investigations have different purposes. Managers must choose an action under conditions of uncertainty, constraints, and competing objectives. Researchers seek systematic evidence about phenomena, relationships, meanings, and patterns. A sound marketing research project connects the two without confusing them. It identifies the decision to be supported, clarifies the information required, formulates research objectives and questions, defines relevant constructs and variables, and grounds the inquiry in theory and prior literature.

The quality of later methodological choices depends heavily on this foundation. A poorly specified problem cannot be rescued by an advanced statistical technique. Conversely, a carefully framed problem can often be addressed with a relatively simple design.

Learning objectives

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

1. Distinguish a managerial decision problem from a marketing research problem.
2. Translate a broad management concern into research objectives, questions, and testable hypotheses.
3. Define constructs, variables, relationships, moderators, mediators, and control variables.
4. Explain how theory and literature inform conceptual frameworks.
5. Identify the information required for a decision and establish decision criteria.
6. Evaluate the quality, feasibility, ethics, and limitations of a proposed research problem.
7. Apply these principles to a contemporary marketing case involving digital customer experience.

1. The architecture of a research problem

A useful research architecture links six elements:

Decision context → managerial problem → research problem → objectives and questions → conceptual framework and hypotheses → evidence for action.

Each element performs a different function.

The decision context

The decision context includes the organisation, market, stakeholders, time horizon, resources, constraints, and consequences of action or inaction. For example, a retailer considering a loyalty programme may face rising acquisition costs, limited technology resources, privacy obligations, and pressure to improve retention within twelve months.

The managerial decision problem

A managerial decision problem asks what the organisation should do. It is action-oriented and normally contains alternatives. Examples include:

• Should the firm reposition the brand toward value-conscious consumers?
• Should the company reduce delivery fees or invest in faster delivery?
• Should a service provider introduce a subscription plan?
• Should the firm enter a new geographic market?

The decision problem should not be confused with a presumed solution. “How can we prove that a mobile application is needed?” is advocacy, not neutral problem definition.

The marketing research problem

A research problem asks what must be understood or measured to support the decision. It is evidence-oriented. The delivery-fee decision, for example, may require knowledge of customer price sensitivity, perceived delivery value, switching intentions, competitor offerings, and the profitability of alternative policies.

A strong research problem is specific enough to guide design but open enough to permit evidence to challenge managerial assumptions. It also identifies the unit of analysis, population, relevant context, and time period.

Symptom versus underlying problem

A symptom is an observable condition; an underlying problem is a mechanism or decision issue that may explain it. A fall in sales may reflect weaker awareness, lower distribution, reduced category demand, poor perceived value, customer migration, stockouts, or measurement error. Treating the symptom as the research problem can lead to an inappropriate study.

A practical diagnostic sequence is:

1. What decision must be made?
2. What evidence would change the decision?
3. What possible explanations are consistent with the observed symptom?
4. Which explanations are controllable, consequential, and investigable?
5. What evidence is already available, and what remains uncertain?

2. From managerial concern to research objectives

Research objectives state what the study will accomplish. They should use observable verbs such as describe, estimate, compare, assess, explain, predict, evaluate, or explore. Objectives should be aligned with the decision and should not promise more than the design can support.

A broad objective might be: “To assess the factors influencing customer retention in the firm’s online channel.” More precise objectives could be:

• To estimate the relationship between perceived service reliability and renewal intention.
• To compare renewal intention among customers receiving different levels of service recovery.
• To assess whether trust changes the relationship between perceived privacy protection and willingness to share data.
• To identify customer interpretations of failed deliveries through qualitative interviews.

Objectives commonly fall into four categories:

These categories can coexist. An investigation may first explore why customers leave, then describe the prevalence of those reasons, and finally test whether a service intervention reduces churn.

3. Research questions and hypotheses

A research question is a focused, answerable question derived from the objectives. It should identify the population, context, concepts, and relationship or phenomenon of interest where appropriate.

Weak question: “Do customers like our website?”

Stronger questions:

• How do customers evaluate the website’s ease of navigation and information quality?
• Is perceived ease of use associated with purchase intention among first-time visitors?
• How do customers explain abandonment during mobile checkout?

A question may be descriptive, comparative, relational, explanatory, or exploratory. Not every question should be converted into a hypothesis. Hypotheses are most appropriate when theory and prior evidence justify a directional or non-directional proposition that can be assessed with data.

A hypothesis is a provisional, logically derived statement about an expected relationship or difference. For example:

H₁: β₁ > 0

This notation expresses the claim that the population coefficient for a specified predictor is positive. It is meaningful only when the variables, population, model, and direction have been defined. A corresponding null hypothesis might be H₀: β₁ = 0, although the substantive interpretation of a null result requires attention to statistical power, measurement quality, and the size of an effect—not merely whether a p-value crosses a threshold.

Examples of well-formed hypotheses include:

• H1: Perceived delivery reliability is positively associated with repurchase intention.
• H2: The effect of a free-return policy on purchase intention is stronger for customers with high perceived purchase risk than for customers with low perceived purchase risk.
• H3: Service satisfaction mediates the relationship between complaint resolution quality and customer loyalty.

A hypothesis must be testable in principle. Terms such as “excellent,” “appropriate,” or “successful” require operational definition. A hypothesis should also be plausible, non-tautological, and open to disconfirmation.

Questions, hypotheses, and objectives compared

4. Theory, literature, and conceptual frameworks

What theory contributes

Theory is a structured set of concepts and propositions that explains how or why phenomena are related. In marketing research, theory can guide construct selection, explain mechanisms, identify boundary conditions, and support predictions. It prevents the study from becoming an unconnected collection of variables.

Relevant theoretical traditions differ by problem. The Technology Acceptance Model may inform perceived usefulness, perceived ease of use, and technology adoption. Expectancy–value reasoning may help explain how beliefs and evaluations shape intentions. Relationship marketing theory may support analysis of trust, commitment, satisfaction, and loyalty. Signalling theory may explain how brand certifications or reviews reduce perceived uncertainty. Institutional, behavioural, service-dominant, and consumer-culture perspectives may be more suitable for other questions.

Theory should not be selected by fashion. The appropriate theory is the one that offers a defensible explanation of the phenomenon and fits the unit, context, and level of analysis.

Literature review as analytical work

A literature review is not a list of article summaries. It should establish:

1. What is already known.
2. How key constructs have been defined and measured.
3. Where findings agree or conflict.
4. Which populations and contexts have been neglected.
5. What theoretical mechanism justifies the proposed relationships.
6. Why the proposed study is consequential.

A graduate-level review compares definitions, measurement choices, research designs, and boundary conditions. It should distinguish original theoretical contributions from later empirical applications and should avoid presenting correlation as proof of causation.

Construct and variable

A construct is an abstract concept inferred from observations, such as trust, perceived value, involvement, brand authenticity, or switching intention. A variable is a measurable representation that can take different values. Constructs are conceptual; variables are operational.

A construct may be measured with several survey items, observed behaviour, experimental manipulation, archival records, or qualitative evidence. The choice depends on the construct’s meaning and the research purpose.

A moderator is not simply another predictor. If risk moderates the effect of a return policy, the effect of the policy differs across risk levels. A mediator explains a pathway: a policy may improve satisfaction, which then increases repurchase intention. These distinctions affect both the conceptual model and the analysis.

Conceptual frameworks

A conceptual framework is a reasoned representation of the proposed constructs and relationships. It may be expressed in prose, a table, or a diagram. A framework should identify the outcome, central explanatory variables, mechanisms, boundary conditions, and relevant controls. It should not include every variable that could possibly matter.

For example, a framework for digital retention might propose:

Perceived service reliability → satisfaction → repurchase intention, with privacy concern moderating the relationship between personalisation and repurchase intention, and customer tenure included as a control.

The arrows represent proposed explanatory relationships, not proof of causation. Causal interpretation requires appropriate temporal ordering, control of competing explanations, and a design capable of supporting causal claims.

5. Operationalisation and measurement logic

Operationalisation converts an abstract construct into observable indicators or conditions. A useful operational definition specifies:

• the conceptual meaning;
• the population and context;
• the unit of analysis;
• the indicator or manipulation;
• the scale, coding, and time frame;
• the source of evidence.

For example, “customer loyalty” could mean stated intention to remain, actual renewal, share of wallet, recommendation behaviour, or a composite. These are related but not identical outcomes. A study should not claim to measure behavioural loyalty if it measures only intention.

Measurement quality has at least two dimensions. Reliability concerns consistency, while validity concerns whether the measure represents the intended construct. A measure may be reliable but invalid—for example, a consistent measure of general satisfaction used as if it measured trust. Researchers should consider content validity, convergent validity, discriminant validity, criterion-related validity, and the possibility of common-method bias when predictor and outcome are collected from the same respondent at the same time.

Operationalisation also requires attention to level of analysis. Individual attitudes cannot automatically be treated as organisational capabilities, and country-level indicators cannot automatically explain individual decisions. Ecological fallacy occurs when group-level relationships are assumed to apply to individuals; atomistic fallacy is the reverse error.

6. Information requirements and decision criteria

A decision-oriented study begins by specifying what information is needed and how it will be used. A useful information requirement includes four parts:

1. Decision alternative: what action is under consideration?
2. Uncertainty: what is not known?
3. Evidence indicator: what observable information would reduce that uncertainty?
4. Decision rule: how would the evidence influence action?

Consider a proposed loyalty programme. Management may need estimates of enrolment, incremental purchase frequency, redemption costs, customer profitability, privacy acceptance, and effects on non-members. If the programme is profitable only when incremental contribution exceeds programme cost, the decision criterion must distinguish incremental behaviour from purchases that would have occurred anyway.

A decision criterion can be qualitative or quantitative. A quantitative rule might be: launch if estimated incremental contribution exceeds implementation and operating costs under the organisation’s acceptable risk threshold. The research may not determine the decision automatically, but it should produce information that maps to the criterion.

Decision criteria should address:

• financial performance;
• customer and stakeholder effects;
• strategic fit;
• operational feasibility;
• legal and ethical acceptability;
• uncertainty and downside risk;
• timing and reversibility.

A statistically significant relationship may have little managerial value if its effect is too small, too uncertain, or too expensive to influence. Conversely, a practically important effect may be estimated imprecisely because the sample is limited. Statistical evidence and managerial significance must therefore be considered together.

7. A disciplined translation process

The following process is suitable for research briefs, proposals, and client discussions.

Step 1: Clarify the decision owner and decision

Identify who will act, what choices are available, when the decision must be made, and what consequences follow from each option.

Step 2: Audit existing evidence

Review sales records, web analytics, customer-service logs, prior research, competitor information, and relevant secondary data. This prevents the study from answering a question already resolved by available evidence.

Step 3: Diagnose competing explanations

Use interviews with decision makers and frontline employees, exploratory customer conversations, process mapping, and literature review. Do not accept the first explanation as fact.

Step 4: Specify the research problem

State the knowledge gap in terms of population, context, constructs, outcomes, and uncertainty. Include the boundaries of the study.

Step 5: Develop objectives and questions

Use a small number of aligned objectives. Each objective should generate one or more answerable questions, and each question should require identifiable evidence.

Step 6: Build the framework

Select theory, define constructs, propose relationships, identify mediators and moderators, and state assumptions. Remove variables that do not serve the decision or theoretical explanation.

Step 7: Decide whether hypotheses are warranted

Use hypotheses for theory-based tests. Use exploratory questions where the phenomenon is poorly understood or where imposing categories could suppress important meanings.

Step 8: Define information and action criteria

Specify what will be measured, how uncertainty will be reduced, and how results will affect alternatives. Discuss thresholds, trade-offs, and risk with decision makers before data collection.

Step 9: Test feasibility and ethics

Assess access to respondents, data quality, time, budget, statistical power where relevant, privacy, consent, and potential harms. Revise the problem if the proposed evidence cannot realistically be obtained.

8. Applied case: declining retention in a subscription service

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

Background

StreamNow, a fictional video-subscription service, has experienced a rise in monthly cancellations. Executives initially propose reducing the subscription price. The analytics team observes that cancellations are concentrated among customers who have contacted support and among users who report difficulty finding relevant content.

Managerial problem

Should StreamNow reduce price, improve content discovery, redesign service recovery, or pursue a combination of actions to improve retention?

Initial research problem

The organisation does not know whether cancellations are driven primarily by price-value perceptions, poor discovery, service failures, or their interaction. It also does not know which intervention would produce sufficient incremental retention to justify its cost.

Objectives and questions

The study sets four objectives:

1. Estimate the association between perceived value, discovery usability, service recovery satisfaction, and renewal intention.
2. Identify customer explanations for cancellation and non-cancellation after a service failure.
3. Test whether service recovery satisfaction mediates the relationship between complaint resolution quality and renewal intention.
4. Estimate whether a redesigned discovery interface produces incremental viewing and renewal relative to the existing interface.

Corresponding questions include:

• Which factors best distinguish renewing and cancelling customers?
• How do customers explain the trade-off between price and perceived content value?
• Does effective recovery restore trust and satisfaction?
• Does the redesigned interface improve retention after accounting for customer tenure and prior usage?

Conceptual framework

The proposed framework places perceived value, discovery usability, and complaint resolution quality as explanatory variables. Satisfaction and trust are proposed mediators, while prior engagement and customer tenure are controls. Price sensitivity is proposed as a moderator of the relationship between perceived value and renewal intention.

Approach

A sequential mixed-method design is appropriate. First, qualitative interviews with recent cancellers and renewers explore meanings and unanticipated explanations. Second, a survey measures constructs using carefully evaluated items. Third, behavioural data are linked, where consent and governance permit, to examine actual renewal. Finally, a controlled interface experiment compares the redesigned and existing discovery experiences.

The experiment is especially valuable for the interface decision because random assignment can reduce systematic differences between groups. A survey association alone would not establish that redesigning the interface caused retention to improve.

Decision criteria

Management agrees in advance that an intervention must satisfy four conditions: evidence of meaningful improvement in renewal or engagement, acceptable implementation cost, no unacceptable increase in privacy risk, and operational feasibility within the next product cycle. A price reduction is not evaluated only by its effect on cancellations; it must also account for lost revenue from customers who would have renewed without a discount.

Outcome and learning points

Suppose the research finds that discovery usability is strongly related to renewal, but the relationship is concentrated among low-engagement customers. Interviews reveal that customers interpret poor discovery as evidence that the service lacks relevant content, even when relevant content exists. The interface experiment shows improved content starts and a modest but credible increase in renewal, while a blanket price reduction produces little incremental retention but substantial revenue loss.

The case illustrates five principles. First, a visible symptom can have several causes. Second, theory helps distinguish mechanisms such as value perception, satisfaction, and trust. Third, behavioural outcomes are preferable to intentions when available. Fourth, the appropriate design depends on the claim: exploration, description, association, and causation require different evidence. Fifth, research becomes decision-useful only when findings are connected to explicit criteria.

9. Critical issues, ethics, and contemporary practice

The danger of premature closure

Managers may commission research to validate a preferred action. Researchers should preserve neutrality, disclose alternative explanations, and report results that challenge expectations. Question wording, sampling decisions, selective reporting, and post hoc hypotheses can all bias conclusions.

Data quality and construct drift

Digital platforms generate abundant behavioural data, but abundance does not guarantee relevance. Clicks may reflect confusion rather than engagement; time on page may reflect distraction; a purchase may reflect a promotion rather than loyalty. Construct definitions should precede data selection, not be altered opportunistically to match available fields.

AI and automated analytics

AI can assist with literature discovery, text classification, interview summarisation, segmentation, and prediction. However, automated outputs require human evaluation for validity, bias, privacy, reproducibility, and disclosure. Generated summaries may omit minority viewpoints; predictive models may reproduce historical discrimination; and commercially sensitive data may be exposed through inappropriate tools. Researchers remain responsible for definitions, assumptions, evidence quality, and interpretations.

Privacy and informed consent

Research involving customer data should follow data minimisation, purpose limitation, access control, retention limits, and applicable privacy law. Linking survey responses to transaction records requires a clear legal and ethical basis, transparent communication, and appropriate de-identification. Consent is not a substitute for good governance, and anonymisation claims should be assessed realistically because datasets can sometimes be re-identified.

Replicability and transparency

A credible project documents its decision context, research questions, measures, sampling logic, exclusions, analytic choices, limitations, and deviations from the original plan. Preregistration may be useful for confirmatory studies. Transparency is especially important when multiple outcomes or models are examined.

Best-practice checklist

Before approving a research brief, ask:

• Is the decision and decision owner explicit?
• Has the symptom been distinguished from plausible underlying causes?
• Are the research problem, objectives, questions, and hypotheses aligned?
• Are constructs defined and operationalisable?
• Does theory explain the proposed relationships?
• Are causal claims matched to an appropriate design?
• Are information requirements connected to decision criteria?
• Are practical significance, uncertainty, cost, and risk considered?
• Are privacy, consent, fairness, and data governance addressed?
• Could a reasonable finding change the decision?

Chapter Summary

Marketing research is a bridge between organisational decisions and systematic evidence. A managerial problem asks what the organisation should do; a research problem asks what must be known to support that decision. Effective problem definition separates symptoms from underlying causes and translates the decision into aligned objectives, research questions, and—when justified—testable hypotheses.

Theory and literature provide more than background. They define constructs, explain mechanisms, identify moderators and mediators, guide measurement, and establish why a relationship should be expected. Conceptual frameworks make these relationships explicit, while operationalisation connects abstract concepts to observable evidence. Information requirements and decision criteria ensure that a study produces findings that are relevant, feasible, ethically obtained, and useful for action.

Graduate researchers must also recognise the limits of evidence. Associations do not automatically establish causation, statistical significance does not guarantee managerial importance, and abundant digital data do not guarantee valid measurement. Ethical governance, transparency, critical evaluation, and appropriate use of AI are integral to research quality rather than optional additions.

Key Terms

Managerial decision problem: An action-oriented question about what an organisation should do.

Marketing research problem: An evidence-oriented statement of what must be understood or measured.

Research objective: A specific statement of what a study intends to accomplish.

Research question: A focused, answerable question derived from the research problem.

Hypothesis: A theory-based, testable proposition about a relationship or difference.

Theory: An organised set of concepts and propositions explaining phenomena.

Literature review: A critical synthesis of existing theoretical and empirical knowledge.

Construct: An abstract concept inferred from observable indicators.

Variable: A measurable representation that can take different values.

Independent variable: A presumed predictor or explanatory factor.

Dependent variable: The outcome a study seeks to explain.

Mediator: A variable describing the mechanism through which an influence operates.

Moderator: A variable that changes the strength or direction of a relationship.

Control variable: A factor included to improve interpretation of a focal relationship.

Confounder: An alternative influence associated with both a predictor and an outcome.

Operationalisation: The process of translating a construct into measurable indicators or conditions.

Conceptual framework: A reasoned representation of constructs and their proposed relationships.

Decision criterion: A rule or standard used to evaluate alternatives in light of evidence.

Construct validity: The extent to which a measure represents the intended construct.

Unit of analysis: The entity about which conclusions are drawn.

Review Questions

Short Answer Questions

1. What is the difference between a managerial decision problem and a marketing research problem?
2. Why is it important to distinguish a symptom from an underlying problem?
3. Define a construct and explain how it differs from a variable.
4. What is the difference between a mediator and a moderator?
5. Why should decision criteria be specified before data collection?

Long Answer Questions

1. Explain the process of translating a broad managerial concern into objectives, research questions, and hypotheses. Use an original marketing example.
2. Discuss how theory and literature improve the quality of a conceptual framework.
3. Explain the role of operationalisation in producing valid marketing research evidence.
4. Compare exploratory, descriptive, explanatory, and predictive objectives, including the kinds of decisions each can support.
5. Discuss why statistical significance, practical significance, and decision relevance should be considered together.

Analytical / Application Questions

1. A retailer observes declining sales after introducing a mobile application. Develop a managerial decision problem, research problem, three objectives, four research questions, and two hypotheses. Justify your choices.
2. Construct a conceptual framework explaining repeat purchase in an online grocery service. Include at least one independent variable, mediator, moderator, and control variable, and explain the proposed relationships.
3. A company wants to use customer-service transcripts and purchase histories to predict churn. Identify the information requirements, possible decision criteria, measurement and validity risks, and ethical governance issues that should be addressed before implementation.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a marketing research problem?

A. What action should management take immediately?

B. What information is needed to support a managerial decision?

C. Which department should receive the research budget?

D. How can a preferred managerial solution be justified?

Correct Answer: B. What information is needed to support a managerial decision?

Question 2

A decline in sales is best classified initially as what?

A. A confirmed causal explanation

B. A research hypothesis

C. An observable symptom

D. A decision criterion

Correct Answer: C. An observable symptom

Question 3

Which objective is primarily descriptive?

A. To explore how customers interpret brand authenticity

B. To estimate the percentage of customers who intend to renew

C. To test whether service recovery causes retention

D. To forecast demand under alternative prices

Correct Answer: B. To estimate the percentage of customers who intend to renew

Question 4

Which statement correctly describes a moderator?

A. It is the outcome variable in every model

B. It explains the mechanism through which an effect operates

C. It changes the strength or direction of a relationship

D. It is always an unmeasured source of bias

Correct Answer: C. It changes the strength or direction of a relationship

Question 5

Which proposition is most clearly a testable hypothesis?

A. Customers should receive excellent service

B. Branding is important in modern markets

C. Perceived delivery reliability is positively associated with repurchase intention

D. The company needs a better digital strategy

Correct Answer: C. Perceived delivery reliability is positively associated with repurchase intention

Question 6

What is the primary purpose of a conceptual framework?

A. To list every variable that might affect a market

B. To represent theoretically justified constructs and relationships

C. To replace the need for operational definitions

D. To guarantee that proposed relationships are causal

Correct Answer: B. To represent theoretically justified constructs and relationships

Question 7

A researcher measures stated renewal intention but claims to have measured actual loyalty. Which issue is most directly raised?

A. Construct validity

B. Random assignment

C. Decision ownership

D. Sample frame size

Correct Answer: A. Construct validity

Question 8

Which question is most relevant when establishing a decision criterion?

A. What font should be used in the questionnaire?

B. What evidence would lead management to choose one alternative over another?

C. Which theory is most popular in marketing journals?

D. How many variables can be included in a framework?

Correct Answer: B. What evidence would lead management to choose one alternative over another?

Question 9

Why would a controlled interface experiment be preferable to a simple survey association when evaluating a redesign?

A. It guarantees a large sample

B. It eliminates the need for theory

C. Random assignment can strengthen causal inference

D. It measures only attitudes rather than behaviour

Correct Answer: C. Random assignment can strengthen causal inference

Question 10

Which practice best reflects responsible use of AI in marketing research?

A. Accepting generated summaries without verification

B. Using customer data in any tool if prediction improves

C. Treating automated classifications as valid constructs automatically

D. Reviewing outputs for validity, bias, privacy, and reproducibility

Correct Answer: D. Reviewing outputs for validity, bias, privacy, and reproducibility