Foundations of Marketing Research

Introduction

Marketing research is the systematic process of defining a marketing problem or opportunity, collecting and analyzing relevant evidence, and communicating findings for decision making. It connects questions about markets, customers, competitors, brands, channels, and societal contexts with disciplined evidence. In graduate marketing management, its significance lies not merely in producing data, but in improving the quality, transparency, and consequences of managerial judgment.

Organizations operate under uncertainty. A firm may not know why sales have declined, which customer segment is most attractive, whether a new product solves a meaningful problem, or how a campaign affects brand perceptions. Research cannot eliminate uncertainty, and it cannot substitute for managerial responsibility. It can, however, reduce avoidable uncertainty by making assumptions explicit, using appropriate evidence, and estimating the limits of what is known.

Marketing research is therefore both a technical and a strategic activity. It includes problem formulation, research design, measurement, sampling, data collection, analysis, interpretation, communication, and the ethical governance of information. Poor research may produce precise-looking but misleading results; good research produces decision-relevant evidence whose limitations are understood.

Learning Objectives

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

1. Define marketing research and explain its purpose in managerial and societal contexts.
2. Distinguish marketing research from marketing intelligence, marketing analytics, and management decision making.
3. Describe the evolution and contemporary ecosystem of the research industry.
4. Evaluate the value, limitations, and sources of error in research evidence.
5. Apply a structured framework to determine whether research is warranted and how it should support a decision.
6. Identify major ethical, privacy, governance, and technology-related issues in contemporary research.

1. The Nature and Purpose of Marketing Research

The American Marketing Association describes marketing research broadly as the function linking consumers, customers, and the public to marketers through information used to identify and define opportunities and problems; generate, refine, and evaluate actions; monitor performance; and improve understanding of marketing as a process. This definition emphasizes several features.

First, research is purposeful. It begins with a decision context or knowledge need. A survey conducted merely because data are available is not necessarily useful research. The central question is what someone may do differently as a result of credible findings.

Second, research is systematic. Systematic does not mean that every study must be quantitative. A well-designed ethnography, interview study, experiment, or social listening project can be systematic when its questions, evidence, procedures, and interpretation are explicit and defensible.

Third, research is empirical but theory-informed. Evidence may consist of observed behavior, reported attitudes, digital traces, organizational records, or cultural artifacts. Theory helps determine what to observe and how to interpret it; evidence tests, extends, or challenges explanations.

Fourth, research is concerned with inference. A researcher often observes a sample, a test market, or a period of activity and draws conclusions about a broader market, future behavior, or a causal relationship. The strength of those conclusions depends on design, measurement, sampling, analysis, and context.

Fifth, research supports but does not make decisions. A decision includes objectives, constraints, values, risk tolerance, timing, and accountability. Research informs these considerations but does not decide them automatically.

1.1 The information–decision relationship

A useful distinction is between a research question and a management question. A management question asks what action should be taken: Should the firm enter the market? Should it change the price? Should it reposition the brand? A research question asks what must be learned to evaluate those actions: How large is the addressable market? How does price affect purchase intention among target users? Which associations distinguish the brand from competitors?

A decision-oriented research brief normally identifies:

• the decision maker and stakeholders;
• the decision to be supported;
• the alternatives under consideration;
• the information already available;
• the consequences of making the decision incorrectly;
• timing, budget, and access constraints; and
• the standards by which evidence will be judged.

The distinction prevents a common failure: answering an interesting question that does not resolve the actual managerial uncertainty.

1.2 Descriptive, diagnostic, predictive, and prescriptive purposes

Marketing research can serve several purposes, often in sequence.

These categories are related but not interchangeable. A descriptive survey can show that customer satisfaction is lower in one region; it does not by itself establish why. A predictive model can identify customers likely to churn; it does not automatically demonstrate that a particular intervention will prevent churn. Prescriptive recommendations require assumptions about actions, costs, benefits, and constraints.

2. Core Terminology

A construct is an abstract concept such as trust, involvement, perceived value, or loyalty. A variable is a characteristic that can take different values, such as age, purchase frequency, or a scale score. Operationalization specifies how a construct will be observed or measured. For example, perceived value might be operationalized through several questionnaire items rather than a single vague question.

A unit of analysis is the entity about which conclusions are drawn: an individual, household, transaction, store, firm, advertisement, or online post. The population is the complete set of units relevant to the research question. A sample is the subset actually studied. A sampling frame is the practical list or procedure used to access population units.

A parameter is a characteristic of a population, while a statistic is a characteristic calculated from a sample. A hypothesis is a testable proposition about relationships or differences. Reliability concerns consistency of measurement; validity concerns whether the measure supports the intended interpretation. A measure can be reliable without being valid—for example, a consistently miscalibrated scale.

A research design is the overall plan linking questions, evidence, procedures, and analysis. Qualitative research typically examines meanings, experiences, processes, and contexts using methods such as interviews, focus groups, observation, and textual analysis. Quantitative research represents observations numerically to estimate distributions, test relationships, or model outcomes. Mixed-methods research intentionally integrates qualitative and quantitative evidence.

An experiment manipulates one or more conditions and observes outcomes while attempting to control competing explanations. Observational research records conditions without assigning the treatment or exposure. Causal inference concerns whether a change in one factor produced a change in an outcome, rather than merely being associated with it.

3. Evolution of Marketing Research

Marketing research developed alongside modern mass markets. Early commercial information practices included sales records, trade publications, distributor reports, and informal customer feedback. As manufacturers faced geographically dispersed consumers and mass media, standardized surveys, opinion polling, advertising testing, and syndicated market measurement became more important during the twentieth century.

Postwar research expanded through probability sampling, consumer panels, motivation research, statistical analysis, and the professionalization of research agencies. Universities contributed theories of consumer behavior, measurement, experimentation, and decision analysis. Research increasingly became a formal organizational function rather than an occasional activity of sales departments.

The digital era changed both the scale and speed of evidence. Transaction databases, loyalty programs, web analytics, mobile devices, social platforms, connected products, and automated experimentation created continuous streams of behavioral data. Cloud computing and machine learning enabled segmentation, recommendation, anomaly detection, and prediction at a scale previously impractical.

Yet technological change has not removed foundational problems. Digital behavior may be abundant but difficult to interpret. A click is not necessarily interest, and an exposure is not necessarily attention. Platform data may omit nonusers, change definitions without notice, or be inaccessible for independent verification. The evolution of research is therefore not simply a movement from “small data” to “big data”; it is a movement toward more heterogeneous evidence requiring stronger governance, integration, and critical judgment.

3.1 From episodic studies to continuous learning

Traditional research often followed a project cycle: brief, design, fieldwork, analysis, report. Contemporary organizations increasingly use continuous research systems, including brand trackers, customer-experience programs, experimentation platforms, and real-time dashboards. Continuous measurement can identify change earlier, but it may also encourage metric fixation and constant reaction to noise. A dashboard is useful only when its measures are valid, its trend interpretations are appropriate, and its users know what action the measures can support.

4. The Marketing Research Ecosystem

Marketing research is conducted within an ecosystem of interacting actors rather than by a single department.

Clients and commissioning organizations define decisions, allocate resources, provide internal data, and use findings. They may be manufacturers, retailers, service firms, governments, nonprofit organizations, or agencies.

Full-service research firms may provide problem definition, design, fieldwork, analysis, and reporting. Specialist firms focus on activities such as sample provision, qualitative recruitment, conjoint studies, sensory testing, brand tracking, social listening, or advanced analytics. Syndicated-data providers collect standardized information and sell access to multiple clients. Panel providers recruit participants for repeated or one-time studies. Technology vendors supply survey, experiment, dashboard, customer-data, and artificial-intelligence tools.

Media, advertising, and consulting agencies use research to plan campaigns, evaluate communication, identify audiences, and advise on strategy. Data brokers and platform companies may aggregate behavioral, demographic, or contextual information, subject to legal and contractual restrictions. Universities and public institutions contribute theory, methods, training, and independent research.

Participants and communities are not merely data sources. They provide information, bear privacy and reputational risks, and may be affected by decisions based on research. Regulators, professional associations, ethics committees, and standards organizations establish expectations concerning consent, privacy, transparency, quality, and fair treatment.

A client may outsource fieldwork while retaining responsibility for the research question, governance, and use of results. Outsourcing does not transfer ethical accountability.

5. Research, Intelligence, Analytics, and Decision Making

These terms overlap in practice but refer to different activities.

The distinctions are functional rather than absolute. A research project may use analytics; an intelligence system may include commissioned surveys; a manager may combine all three with experience. Confusion arises when a predictive model is treated as a decision, when intelligence is treated as unbiased research, or when a dashboard is assumed to explain causes.

For example, a retailer may use intelligence to monitor a competitor’s store openings, analytics to predict which customers may reduce visits, research to understand perceptions of the retailer’s value proposition, and management judgment to choose whether to alter prices or service. Each activity contributes different evidence.

6. Evaluating Research Value

Research has value when it improves a consequential decision enough to justify its cost, time, risk, and organizational burden. Value is not equivalent to statistical sophistication or report length.

A practical evaluation asks:

1. Decision relevance: Will findings influence a real choice?
2. Uncertainty reduction: Does the study address an important unknown?
3. Actionability: Can findings be translated into feasible action?
4. Credibility: Are design, measurement, sampling, and analysis appropriate?
5. Timeliness: Will the evidence arrive while alternatives remain open?
6. Economic and social cost: Do expected benefits justify financial, operational, privacy, and reputational costs?
7. Reversibility: Is the decision small and reversible, or costly and difficult to undo?

A useful conceptual expression is:

Here, EVSI denotes the expected value of sample information; a represents an available action; U(a) is the utility or payoff of that action; and E denotes an expectation. The expression compares the expected payoff when a decision uses new information with the best expected payoff without that information. In a practical business appraisal, the expected value of information should be compared with research cost and risk. The formula assumes that decision alternatives, utilities, and probability judgments can be represented meaningfully; it is a conceptual decision-analysis tool, not a guarantee of monetary precision.

Research is especially valuable when a decision is important, uncertain, and difficult to reverse. It may be less valuable when the decision is low-risk, information is already sufficient, or time has made further inquiry impractical. Sometimes the correct conclusion is not to conduct a study.

7. Sources of Error and Limitations

Research evidence is always imperfect. A useful framework distinguishes total survey error and, more broadly, total research error. The categories below may overlap, but they help identify where quality can fail.

7.1 Problem-definition error

If the management problem is misunderstood, even flawless fieldwork may answer the wrong question. A decline in sales may reflect distribution, availability, price, competition, or changing needs. Treating it only as an advertising-awareness problem creates specification error.

7.2 Coverage and sampling error

Coverage error occurs when the sampling frame excludes relevant population members or includes inappropriate units. Sampling error is the difference between a sample result and the population value arising because only a sample was observed. Probability sampling permits statistical estimation of sampling variability under specified assumptions; nonprobability samples may still be useful, particularly for exploratory or hard-to-reach research, but their generalizability must not be overstated.

7.3 Nonresponse and participation error

People who do not respond may differ systematically from those who do. Incentives, topic sensitivity, survey length, recruitment channels, and trust can affect participation. Increasing sample size does not automatically remove nonresponse bias.

7.4 Measurement error

Measurement error arises when observed responses differ from the intended construct. Sources include ambiguous wording, leading questions, poor scales, recall limitations, social desirability, interviewer effects, translation problems, device effects, and inappropriate proxy measures. Behavioral data also have measurement error: missing events, bot activity, duplicate users, changing cookies, and inconsistent platform definitions.

7.5 Processing and analysis error

Errors can enter through coding, data integration, missing-data treatment, weighting, model specification, multiple comparisons, or software implementation. A statistically significant relationship may be practically trivial, unstable, or produced by chance. Conversely, a meaningful effect may be missed because the study lacks precision or uses a poor measure.

7.6 Interpretation and reporting error

Researchers and managers may confuse correlation with causation, generalize beyond the population studied, ignore contradictory evidence, or present uncertain estimates as exact facts. Selective reporting, inappropriate visual scales, undisclosed sponsorship, and “precision theater” can make weak evidence appear authoritative.

A useful discipline is to separate four statements:

• what was observed;
• what the design supports as an explanation;
• what remains uncertain; and
• what action is reasonable given the uncertainty.

7.7 External validity and temporal instability

Findings may not transfer across countries, segments, channels, periods, or economic conditions. Consumer attitudes and competitive structures change. A result can be internally credible in one context yet have limited external validity elsewhere. Replication, benchmarking, sensitivity analysis, and post-launch monitoring can help assess stability.

8. Ethics, Privacy, and Governance

Ethical research respects autonomy, minimizes harm, protects confidentiality, and represents evidence honestly. Participants should normally receive understandable information about the purpose, procedures, risks, data use, incentives, and voluntary nature of participation. Consent must be meaningful rather than hidden in dense legal language.

Privacy governance includes data minimization, purpose limitation, access controls, retention limits, secure transfer, de-identification where appropriate, and procedures for responding to breaches. Legal obligations vary by jurisdiction and context; organizations should obtain current legal and compliance advice rather than assume that publicly accessible data are unrestricted for every use.

Important issues include:

• collecting sensitive information without a necessary purpose;
• re-identification of supposedly anonymous datasets;
• covert observation or undisclosed recording;
• vulnerable participants and unequal bargaining power;
• discriminatory targeting or exclusion;
• dark patterns that manipulate consent;
• transferring data across jurisdictions;
• conflicts of interest and sponsor influence; and
• publishing findings in a way that stigmatizes communities.

Professional codes and data-protection laws provide frameworks, but ethical quality also requires judgment. A legally permissible study may still be exploitative or misleading.

9. Technology and Artificial Intelligence

AI can assist with questionnaire programming, transcription, translation, text classification, sentiment analysis, synthetic data exploration, sample quality checks, personalization, and prediction. These applications may reduce cost and increase speed, but they introduce risks of automation bias, opaque models, fabricated outputs, privacy leakage, data drift, and unfair performance across groups.

Good practice includes documenting data provenance, validating outputs against human-reviewed standards, testing subgroup performance, preserving human accountability, and disclosing material use of automated methods. Synthetic respondents or generated text should not be treated as substitutes for evidence from real participants unless their purpose and limitations are explicitly established. AI-generated summaries require source checking because fluent language is not proof of valid inference.

The most valuable role of technology is often augmentation: helping researchers search, organize, compare, and analyze evidence while leaving problem definition, ethical judgment, causal interpretation, and accountability to qualified people.

10. A Framework for Professional Research Practice

A disciplined research process can be expressed as a sequence, although real projects may iterate:

1. Clarify the decision and context. Identify the decision, alternatives, constraints, and consequences.
2. Audit existing knowledge. Review internal records, prior studies, market intelligence, theory, and relevant external evidence.
3. Translate the decision into research questions. Specify constructs, populations, outcomes, and possible explanations.
4. Choose a design. Match exploratory, descriptive, causal, predictive, qualitative, quantitative, or mixed methods to the question.
5. Plan measurement and sampling. Define operational measures, recruitment, inclusion criteria, quality controls, and analysis before fieldwork where feasible.
6. Collect and manage evidence ethically. Document procedures, protect participants, monitor quality, and preserve a reproducible audit trail.
7. Analyze in relation to the design. Use methods appropriate to the data and avoid claims that exceed identification or measurement assumptions.
8. Interpret alternatives and uncertainty. Consider rival explanations, subgroup differences, practical significance, and transferability.
9. Communicate for action. Present findings, limitations, implications, and recommended decisions clearly to the intended users.
10. Learn after implementation. Compare outcomes with expectations and update the organization’s knowledge base.

A high-quality research report commonly includes an executive decision statement, background, methodology, findings, uncertainty and limitations, implications, recommendations, technical appendices, and disclosure of sponsorship or conflicts.

Applied Case: Should Northstar Foods Launch a Plant-Based Snack?

The following is an illustrative fictional case.

Background

Northstar Foods, a regional packaged-food company, observed rapid growth in plant-based snack products in specialist retailers. The sales director proposed a national launch. The chief financial officer warned that the category might be concentrated among a small group of enthusiastic buyers and that a national launch would require a costly production commitment.

Problem

The initial request was “Find out whether consumers like the concept.” That question was too weak for the decision. Management needed to assess target demand, product fit, willingness to pay, reasons for rejection, competitive alternatives, and whether a limited launch could generate evidence before national distribution.

Analysis and approach

The research team first audited retailer data, syndicated category estimates, competitor prices, search trends, and prior customer research. It then conducted qualitative interviews with current category users and occasional snack buyers to explore usage occasions, language, barriers, and reactions to alternative product ideas. The team used these findings to develop a structured survey measuring purchase frequency, need states, concept appeal, perceived benefits, price expectations, and competitor consideration.

Because stated purchase intention can overstate actual behavior, the team recommended a controlled regional pilot. Stores were selected to represent different customer and channel conditions. The pilot tracked distribution, trial, repeat purchase, realized price, returns, and substitution. A matched comparison region provided a benchmark, while the team documented differences that could threaten causal interpretation.

Solution and decision

Survey results suggested broad curiosity but a narrower high-potential segment: consumers seeking convenient savory snacks with credible nutritional benefits. Qualitative evidence showed that “plant-based” alone was not persuasive; taste, texture, and portability mattered more. The pilot showed strong trial in selected stores but weak repeat purchase where the product was placed away from the snack aisle. Northstar delayed national expansion, improved the product and merchandising, and ran a second pilot.

Outcome and learning points

The research did not produce a simple yes/no answer. It changed the decision from an expensive national launch to a staged learning strategy. Key lessons were:

• The original management question needed conversion into answerable research questions.
• Different methods addressed different uncertainties.
• Stated attitudes were insufficient evidence of repeat behavior.
• A pilot improved decision quality but did not create perfect causal certainty.
• Research value came from changing the timing, scope, and conditions of action.
• The organization treated limitations as decision information rather than as a reason to conceal uncertainty.

Challenges and Best Practices

Graduate-level research practice requires resisting several recurring temptations: beginning with a preferred conclusion, treating large datasets as representative by default, confusing engagement metrics with business outcomes, using complex models without a clear decision purpose, and reporting subgroup patterns without adequate evidence.

Best practice is to use the simplest design capable of answering the question credibly; triangulate important conclusions across appropriate sources; predefine key outcomes and decision rules when feasible; distinguish exploratory findings from confirmatory tests; document exclusions and transformations; test interpretation with stakeholders who understand the market and with reviewers who can challenge assumptions; and preserve a transparent record of methods and changes.

Researchers should also ask who benefits, who bears risk, and whose voice is missing. Market research is not neutral merely because it uses numbers. Choices about populations, constructs, categories, incentives, and reporting shape which realities become visible to management.

Chapter Summary

Marketing research is systematic, decision-oriented inquiry that links markets and stakeholders with evidence. Its purposes include exploration, description, diagnosis, prediction, and prescription, but each purpose supports different claims. Core concepts include constructs, variables, operationalization, populations, samples, validity, reliability, research design, and causal inference.

The field evolved from records and surveys into an ecosystem involving clients, agencies, panels, data providers, platforms, technology vendors, universities, regulators, and participants. Research differs from marketing intelligence, analytics, and management decision making, although the activities interact. Research value depends on decision relevance, uncertainty reduction, actionability, credibility, timing, cost, and reversibility.

Evidence can be weakened by problem-definition, coverage, sampling, nonresponse, measurement, processing, analysis, interpretation, and external-validity errors. Ethical practice requires informed participation, privacy protection, fair treatment, transparent sponsorship, secure governance, and responsible use of technology. AI can augment research but does not remove the need for human accountability and critical judgment. Professional research is an iterative process that converts decisions into questions, questions into defensible designs, evidence into qualified interpretations, and interpretations into learning and action.

Key Terms

Marketing research: Systematic inquiry that provides information for identifying problems, opportunities, evaluating actions, and improving marketing decisions.

Research question: A focused question about what must be learned to support a decision.

Construct: An abstract concept that requires conceptual definition and operational measurement.

Variable: A characteristic that can take different values across units or observations.

Operationalization: Specification of how a construct will be observed or measured.

Population: The complete set of units relevant to a research question.

Sample: A subset of the population that is studied.

Sampling frame: The practical list or procedure used to access population units.

Reliability: Consistency or stability of a measurement procedure.

Validity: The extent to which evidence supports the intended interpretation or use of a measure.

Research design: The plan connecting questions, evidence, data collection, and analysis.

Qualitative research: Research examining meanings, experiences, processes, and contexts, usually with nonnumeric or textual evidence.

Quantitative research: Research representing observations numerically for estimation, comparison, or modeling.

Mixed-methods research: Deliberate integration of qualitative and quantitative approaches.

Marketing intelligence: Ongoing monitoring and interpretation of the external marketing environment.

Marketing analytics: Use of data and analytical methods to identify patterns, estimate outcomes, or evaluate effects.

Causal inference: Reasoning about whether a factor produced a change in an outcome.

Measurement error: Difference between an observed measure and the intended construct or true value.

External validity: The extent to which findings transfer to other populations, settings, times, or conditions.

Expected value of sample information: The expected improvement in decision payoff from obtaining additional information.

Review Questions

Short Answer Questions

1. Define marketing research and explain why systematicity matters.
2. Distinguish a management question from a research question.
3. What is the difference between reliability and validity?
4. Identify two differences between marketing research and marketing intelligence.
5. Why does a large dataset not automatically provide representative evidence?

Long Answer Questions

1. Explain the evolution of marketing research from traditional surveys and records to continuous digital and AI-supported research. Discuss what has changed and what foundational problems remain.
2. Compare exploratory, descriptive, diagnostic, predictive, and prescriptive research, using a marketing example for each.
3. Describe the major actors in the marketing research ecosystem and explain how responsibility is distributed among clients, suppliers, platforms, participants, and regulators.
4. Analyze the principal sources of research error and show how a researcher could reduce each one.
5. Discuss the ethical and governance requirements of using behavioral data and artificial intelligence in marketing research.

Analytical / Application Questions

1. A company observes declining sales and immediately commissions a customer satisfaction survey. Diagnose the weaknesses in this approach and propose a staged research plan.
2. An online experiment reports a statistically significant increase in clicks after a new advertisement is introduced. What additional evidence would be needed before recommending a large media investment?
3. Using the Northstar Foods case, explain why the combination of qualitative research, survey evidence, and a pilot was more useful than relying on any single method.

Multiple Choice Questions with Answers

Question 1

Which statement best describes the primary purpose of marketing research?

A. To replace managerial judgment with automated decisions

B. To provide systematic evidence for identifying problems, opportunities, and actions

C. To collect the largest possible volume of customer data

D. To guarantee that a marketing decision will be successful

Correct Answer: B. To provide systematic evidence for identifying problems, opportunities, and actions

Question 2

Which activity is most clearly marketing intelligence?

A. Conducting interviews to explore reactions to a new product concept

B. Running an experiment to estimate the causal effect of an advertisement

C. Monitoring competitor store openings and changes in channel strategy

D. Choosing a launch price after considering objectives and constraints

Correct Answer: C. Monitoring competitor store openings and changes in channel strategy

Question 3

A measure gives nearly identical results each time but consistently fails to represent the intended construct. It is best described as:

A. Reliable but invalid

B. Valid but unreliable

C. Both valid and reliable

D. Neither measurable nor systematic

Correct Answer: A. Reliable but invalid

Question 4

Which research purpose asks what is happening, to whom, where, and when?

A. Exploratory

B. Descriptive

C. Prescriptive

D. Causal

Correct Answer: B. Descriptive

Question 5

Which is an example of coverage error?

A. A respondent selects an incorrect answer because a question is ambiguous

B. A data analyst incorrectly codes a response category

C. A sampling frame excludes customers who purchase only through mobile channels

D. A manager overstates the practical importance of a small effect

Correct Answer: C. A sampling frame excludes customers who purchase only through mobile channels

Question 6

Why may a predictive churn model fail to show that a retention intervention will work?

A. Prediction and causal inference answer different questions

B. Predictive models cannot use customer data

C. Causal effects are always smaller than predictions

D. A model with many variables is necessarily descriptive

Correct Answer: A. Prediction and causal inference answer different questions

Question 7

Which consideration most strongly increases the potential value of research?

A. The decision is consequential, uncertain, and difficult to reverse

B. The report will contain many pages and visualizations

C. The study uses the newest available software

D. The sample is large regardless of how it was recruited

Correct Answer: A. The decision is consequential, uncertain, and difficult to reverse

Question 8

In the illustrative Northstar Foods case, why was a regional pilot useful?

A. It guaranteed national demand would be identical to regional demand

B. It provided behavioral evidence under market conditions before a larger commitment

C. It eliminated all possible confounding factors

D. It made qualitative research unnecessary

Correct Answer: B. It provided behavioral evidence under market conditions before a larger commitment

Question 9

Which practice is most consistent with responsible AI-supported research?

A. Accepting fluent automated summaries without checking source data

B. Using synthetic respondents as unquestioned substitutes for real participants

C. Validating outputs, testing subgroup performance, and retaining human accountability

D. Removing documentation because automated systems are objective

Correct Answer: C. Validating outputs, testing subgroup performance, and retaining human accountability

Question 10

What is the principal distinction between a statistic and a parameter?

A. A statistic describes a sample, whereas a parameter describes a population

B. A statistic is always qualitative, whereas a parameter is always quantitative

C. A statistic proves causation, whereas a parameter proves association

D. A statistic comes from intelligence, whereas a parameter comes from analytics

Correct Answer: A. A statistic describes a sample, whereas a parameter describes a population