Entrepreneurial decisions are made under uncertainty. A founder may need to estimate customer demand, understand why users abandon a product, evaluate a competitor, or test whether a new business model is financially and socially viable. Business research provides a disciplined way to reduce that uncertainty. It does not eliminate risk or guarantee correct decisions; rather, it produces credible evidence, makes assumptions visible, and helps decision-makers distinguish facts from opinions.

A research project is more than the collection of data. It is a connected process that begins with a management or entrepreneurial problem, develops a research question, selects a philosophical position and design, identifies appropriate evidence, manages fieldwork, analyses information, and communicates conclusions. Weakness at any stage can undermine the entire project. For example, sophisticated statistical analysis cannot repair a poorly defined population, and a large sample cannot compensate for a question that does not address the business decision.

This chapter explains how to plan a complete research process. Particular attention is given to positivist, interpretivist, and pragmatic approaches; research design; time horizon; units of analysis; mixed methods; feasibility; budgeting; and project management.

Learning Objectives

1. From Business Problem to Research Problem

A business problem is usually expressed in managerial language: “Sales are falling,” “Should we enter this market?” or “Why are customers not renewing subscriptions?” A research problem translates that concern into an answerable question about evidence. The distinction matters because different explanations require different designs.

For example, declining sales may result from lower awareness, poor distribution, price sensitivity, a weak product experience, or a change in competitors’ actions. The research problem might therefore be: “Which factors explain declining repeat purchase among customers in the target segment?” The question is narrower, measurable, and connected to a decision.

Management concern → research problem → objectives → research questions or hypotheses → design → evidence → analysis → conclusions and recommendations.

Research objectives state what the study will accomplish. They should be specific, feasible, and aligned with the decision. Common objectives include describing a market, exploring consumer motivations, measuring relationships between variables, evaluating an intervention, or forecasting demand.

Research questions are interrogative statements, such as “How do first-time users describe barriers to adoption?” Hypotheses are testable propositions, such as “Perceived convenience is positively associated with intention to use the service.” A hypothesis is not required in exploratory or interpretive research, but it is useful when a study tests relationships derived from theory.

Scoping and the literature review

A literature review does not mean listing everything previously written. It identifies concepts, theories, empirical findings, measurement approaches, disagreements, and gaps relevant to the research question. Useful sources include peer-reviewed articles, academic books, industry reports, government statistics, trade publications, and carefully evaluated company data.

The researcher should distinguish primary data—collected for the present study—from secondary data—previously collected for another purpose. Secondary data can be inexpensive and valuable, but its definitions, date, coverage, and quality must be checked. A market-size estimate based on a different geographic area or customer definition may mislead even if the source is reputable.

2. Research Philosophy: How Knowledge Is Understood

Research philosophy concerns assumptions about reality, knowledge, values, and appropriate methods. Philosophy is not an abstract decoration: it influences what counts as evidence, how a question is framed, how the researcher relates to participants, and how findings are interpreted.

Positivism

Positivism is associated with the view that an external reality can be observed and studied systematically. It emphasizes measurement, objectivity, regularities, and the testing of propositions. Typical methods include structured surveys, experiments, panel data, and statistical modelling. A positivist entrepreneur might measure whether delivery speed predicts customer satisfaction while controlling for price and product category.

Strengths include clarity of variables, comparability across respondents, replicability, and the ability to estimate patterns in a population. Limitations include the risk of reducing complex experiences to narrow measures, overlooking context, and treating measured indicators as complete representations of social phenomena. Positivist studies also depend on valid instruments and appropriate assumptions; numerical data are not automatically objective or meaningful.

Interpretivism

Interpretivism emphasizes that social reality is experienced and interpreted by people. Concepts such as trust, identity, fairness, and entrepreneurial legitimacy may have different meanings in different contexts. Interpretivist research seeks to understand those meanings rather than assume that a universal measurement captures them.

Common methods include semi-structured interviews, observation, ethnography, diary studies, and qualitative analysis of documents or digital communities. The researcher is an instrument in the research process and should reflect on how relationships, language, status, and prior assumptions influence interpretation.

Interpretivism can reveal motivations and processes that standardized surveys miss. However, studies may be time-consuming, context-specific, and difficult to generalize statistically. Quality is supported through transparent sampling logic, rich contextual description, reflexivity, systematic coding, and attention to contradictory evidence.

Pragmatism

Pragmatism begins with the research purpose and the practical consequences of alternative explanations. It does not require allegiance to one method. A pragmatic researcher may combine a survey to estimate the prevalence of a problem with interviews to understand its causes, provided the combination addresses the decision coherently.

Pragmatism is especially useful in entrepreneurship, where questions often concern action: whether to launch, pivot, invest, segment, or redesign. Its danger is methodological opportunism—selecting methods merely because they are convenient. A pragmatic design still requires a clear rationale, valid evidence, ethical safeguards, and explicit integration of findings.

Comparing the philosophies

Dimension	Positivism	Interpretivism	Pragmatism
View of reality	Relatively external and measurable	Socially constructed and context-dependent	Judged through practical consequences and problem context
Main purpose	Explanation, prediction, testing	Meaning, understanding, process	Actionable understanding and problem solving
Typical evidence	Structured numerical data	Textual, visual, and observational data	Quantitative, qualitative, or combined
Researcher role	Seeks distance and control	Interprets and reflects on involvement	Selects methods appropriate to the question
Typical quality concern	Reliability, validity, replicability	Credibility, reflexivity, transferability	Usefulness, coherence, and integration

These are ideal types rather than rigid categories. A survey can include open-ended questions, and an interview study can use some numerical description. The key is to make philosophical assumptions and methodological choices consistent.

3. Research Design and Methodological Choice

Research design is the overall plan linking questions to evidence and conclusions. It specifies what will be studied, from whom, when, how data will be collected, and how claims will be supported.

Exploratory design

Exploratory research is appropriate when the problem is poorly understood or concepts are not yet clear. It may use expert interviews, focus groups, observation, secondary-data analysis, or a small pilot. Its output is often a refined problem definition, conceptual framework, customer typology, or set of propositions. Exploratory findings should not be presented as precise population estimates unless the design supports that claim.

Descriptive design

Descriptive research answers questions such as who, what, where, when, and how much. A structured survey can estimate the proportion of potential customers who recognize a brand, provided the sample and measurement are appropriate. Descriptive research may be cross-sectional or longitudinal.

Explanatory or causal design

Explanatory research investigates relationships and possible causes. Experiments are particularly useful for causal claims because random assignment can balance observed and unobserved differences between treatment groups. For instance, users may be randomly assigned to two versions of a landing page, with conversion rates compared afterward.

Observational studies can identify associations but generally require stronger assumptions and controls before causal language is justified. A relationship between income and purchase frequency does not by itself show that income causes purchase frequency; age, location, access, or product need may also matter.

Evaluation and action-oriented designs

Evaluation research assesses whether a program, product, or intervention achieved its intended outcomes. It may examine implementation, outputs, outcomes, costs, and unintended effects. Action research combines inquiry with cycles of planning, action, observation, and reflection, often involving organizational participants in solving a practical problem.

Case study design

A case study examines a bounded case—such as a start-up, platform, market entry, or partnership—in depth and context. Multiple sources, including interviews, documents, observations, and performance data, can strengthen the analysis. A case study supports analytical insight and transfer of lessons to similar contexts, but it does not automatically establish statistical generalizability.

4. Time Horizon and Research Sequence

A cross-sectional study collects data at one period or a short defined window. It is efficient for describing current attitudes or comparing groups, but it cannot observe change well and is limited for establishing temporal order.

A longitudinal study collects data at multiple time points. A panel repeatedly studies the same units, whereas a repeated cross-sectional design surveys different samples from the same population over time. Longitudinal research can examine retention, adoption, or changes after a campaign, but it costs more and faces attrition, history effects, and changes in measurement or population composition.

Time horizon should match the question. “What do customers think now?” may require a cross-sectional design. “How does trust develop during the first six months of use?” requires repeated measurement or retrospective qualitative inquiry. A short academic project should not promise a longitudinal design unless follow-up access and resources are genuinely available.

The stages may overlap and iterate. Pilot results can revise a questionnaire; early interviews can refine later survey items. Such iteration should be documented rather than hidden.

5. Units, Population, and Sampling

The unit of analysis is the entity about which a conclusion is made: an individual customer, household, firm, transaction, store, product, or country. The unit of observation is the entity from which data are actually recorded. These may differ. A study may analyse firms using financial records collected from several managers per firm.

The target population is the full set to which the researcher intends to apply findings. The accessible population is the portion that can realistically be contacted. A sampling frame is the operational list or procedure used to identify population members. Frame undercoverage occurs when eligible units are missing; duplicate or outdated records can create other problems.

Stratification divides a population into relevant subgroups and samples within them, which can improve representation. Cluster sampling selects groups such as branches or communities and then studies units within selected groups; it may reduce costs but often increases sampling variability because members of a cluster resemble one another.

Sample size should be justified by the research design, desired precision, expected variability, analysis plan, access, and resources—not by a universal rule. For a proportion estimated from a simple random sample, a commonly used planning expression is:

Here, n₀ is the preliminary sample size, z is the standard-normal critical value for the selected confidence level, p is the anticipated population proportion, and e is the desired margin of error. The formula assumes a probability sample, an independently selected population of sufficient size, and a proportion estimate under the usual large-sample approximation. If p is unknown, 0.5 gives the largest planning variance. Nonresponse, design effects, subgroup analysis, and finite-population corrections may require increasing or adjusting the target.

The preliminary target is therefore 385 completed responses, before allowing for nonresponse or complex sampling. This calculation does not make a convenience sample representative.

6. Measurement and Data-Collection Quality

Operationalization translates an abstract construct into observable indicators. “Customer loyalty” might be represented by repurchase intention, actual repeat purchase, recommendation intention, and switching resistance. Researchers should distinguish formative measures, where indicators combine to form a construct, from reflective measures, where indicators are effects of an underlying construct.

Reliability concerns consistency. Validity concerns whether the measure supports the intended interpretation. Content validity asks whether important aspects are covered; construct validity examines alignment with the theoretical concept; criterion validity considers association with a relevant external standard. A pilot can identify ambiguous wording, excessive burden, missing response categories, technical problems, and interviewer inconsistency.

Questionnaire design should use clear language, one idea per item, balanced response options, and an appropriate recall period. Leading, double-barrelled, loaded, and unnecessarily technical questions should be avoided. Interview protocols should provide structure without preventing participants from explaining unexpected issues.

7. Mixed-Methods Research

Mixed methods deliberately integrates quantitative and qualitative evidence within one study. It is not simply using two methods in parallel. Integration occurs when one strand informs the other, when findings are compared, or when both contribute to a joint interpretation.

Mixed methods can provide breadth and depth, reveal contradictions, improve instrument development, and strengthen practical recommendations. It also requires more time, methodological competence, integration planning, and careful handling of conflicting findings. The researcher should specify the priority of each strand, the timing, connection points, and integration procedure before data collection.

For example, a start-up might survey 400 potential users to estimate interest in a refill service, then interview 20 users with high interest and 20 with low interest. The survey estimates patterns; interviews investigate why stated interest does or does not translate into willingness to pay. The final conclusion should not merely report two separate sections; it should explain how the findings jointly inform segmentation and product design.

8. Feasibility, Budgeting, and Project Management

A method is appropriate only if it is feasible. A feasibility assessment considers:

A research budget should connect each cost to an activity. Typical categories include personnel, participant incentives, travel, recruitment, survey or panel fees, transcription, translation, software, equipment, data storage, dissemination, and contingency. Costs may be fixed, variable per participant, or shared across activities. A simple planning identity is:

This is a budgeting identity, not a statistical formula. It assumes that costs have been classified consistently and that the number of completed cases—not merely invitations—is used for variable fieldwork costs.

Suppose fixed costs are $2,000, variable cost is $8 per completed survey, 300 surveys are required, and contingency is 10% of the pre-contingency subtotal. Then the subtotal is $2,000 + ($8 × 300) = $4,400, and the planned budget is $4,840. The contingency basis should be stated; a project using 10% of fixed costs would produce a different amount.

Project management tools include a work breakdown structure, responsibility matrix, milestones, risk register, version control, and decision log. A simple schedule should show dependencies: ethical approval precedes recruitment; piloting precedes full launch; data cleaning precedes analysis; and analysis precedes final recommendations. Risks can be assessed qualitatively by likelihood and impact, with mitigation such as backup recruitment channels, alternative interview modes, secure data storage, and reserve time.

9. Ethics, Governance, and Responsible Data Use

Ethical research respects autonomy, minimizes harm, protects privacy, and treats participants fairly. Informed consent should explain purpose, procedures, duration, foreseeable risks, benefits, voluntary participation, withdrawal rights, data use, and contact details. Consent must be meaningful, especially where participants are employees, students, customers, or economically dependent individuals.

Researchers should collect only data necessary for the stated purpose, restrict access, use secure storage, define retention periods, and remove or minimize identifiers where possible. Sensitive data require stronger safeguards. Legal requirements vary by jurisdiction, but compliance is not a substitute for ethical judgment. Research involving children, vulnerable groups, deception, health information, or covert observation may require additional review.

Entrepreneurs should be especially cautious when researching their own customers. Existing trust can create pressure to participate, and commercial incentives may bias interpretation. Participants should not be promised outcomes that the research cannot guarantee. AI tools can assist with transcription, coding suggestions, translation, or questionnaire drafting, but researchers remain responsible for confidentiality, accuracy, bias checking, copyright, disclosure, and human review. Identifiable data should not be uploaded to an unapproved system.

10. Case Study: Testing a Refill-Service Opportunity

Background

GreenCup, a small venture selling takeaway beverages near a university, is considering a reusable-cup subscription. Management believes environmentally conscious customers will adopt the service, but it does not know whether convenience, deposit size, hygiene concerns, or price will determine uptake.

Problem and objectives

The decision is whether to pilot the subscription at two locations. The research objectives are to estimate potential adoption, identify barriers, understand user meanings of “hygienic” and “convenient,” and determine which offer attributes should be tested.

Analysis and design

The founders initially proposed an online convenience survey. The researcher identified two weaknesses: students who already follow GreenCup online would be overrepresented, and closed questions might conceal important concerns. A pragmatic, sequential exploratory design was selected. First, 18 semi-structured interviews were conducted with a purposive mix of current customers, noncustomers, commuters, and staff. Analysis identified four recurring issues: return-point visibility, cleaning credibility, deposit affordability, and fear of forgetting the cup.

These findings informed a survey with clear definitions and a short choice experiment comparing alternative deposit levels, return locations, and reminder features. The sampling plan used student mailing lists and intercept recruitment at both locations, with quotas by location and broad customer status. Because this was not a probability sample, the report described estimates as indicative rather than population-representative.

Approach and governance

Participants received an information sheet and could decline without affecting service. Interview recordings were transcribed, de-identified, and stored in restricted folders. The team piloted the survey with 15 respondents, discovering that “return convenience” was interpreted inconsistently; the item was rewritten to specify walking time and available return points.

Outcome

The survey suggested strong stated interest but lower willingness to pay when the deposit was high. Interviews explained the gap: customers supported the environmental idea but distrusted claims about cleaning and expected reminders. GreenCup launched a limited pilot with visible return stations, a published cleaning protocol, a moderate refundable deposit, and automated reminders. The pilot measured actual enrolment and return behaviour rather than relying only on intentions.

Learning points

The case demonstrates that a research question should serve a decision, not merely generate attractive statistics. Qualitative work improved measurement; quantitative work assessed the prevalence and relative importance of issues; and behavioural pilot data provided a stronger basis for action than stated preference alone. It also shows why sampling limitations, operational definitions, privacy, and feasibility must be reported honestly.

11. Common Challenges and Best Practices

Research projects commonly suffer from scope creep, vague objectives, overreliance on convenience samples, leading questions, low response rates, unplanned analysis, and recommendations that exceed the evidence. Confirmation bias is particularly dangerous in entrepreneurship because founders may seek support for an idea they already favour.

Best practice is to write a concise research protocol before fieldwork. It should state the decision, questions, philosophical rationale, design, population, recruitment, instruments, analysis plan, ethics, budget, schedule, risks, and limitations. Predefine important outcomes where possible, retain an audit trail of changes, pilot procedures, triangulate sources when appropriate, and distinguish descriptive findings from causal claims.

A professional report should connect each recommendation to evidence and uncertainty. It should identify what is known, what is inferred, what remains unknown, and what low-cost experiment or additional research would reduce uncertainty next. Research is valuable not because it sounds certain, but because it makes decision quality better.

Chapter Summary

Research design is the architecture connecting a business problem to defensible evidence and action. A strong process moves from a clearly scoped management concern to objectives, questions, philosophy, design, sampling, measurement, collection, analysis, and communication. Positivism emphasizes measurement and testing; interpretivism emphasizes meaning and context; pragmatism selects and integrates methods according to the practical problem. Exploratory, descriptive, explanatory, evaluation, case study, and action-oriented designs serve different purposes. Cross-sectional and longitudinal horizons answer different kinds of time-related questions. Careful definition of the unit of analysis prevents invalid conclusions, while sampling and measurement decisions determine the credibility of findings. Mixed methods can combine breadth and depth when integration is planned. Finally, feasibility, ethics, budgeting, risk management, and project scheduling are not administrative afterthoughts; they are essential parts of research quality.

Key Terms

Research philosophy: Assumptions about reality, knowledge, values, and appropriate inquiry.

Positivism: An approach emphasizing observable measurement, regularities, and systematic testing.

Interpretivism: An approach emphasizing meaning, context, and socially constructed experience.

Pragmatism: An approach that selects methods according to the research problem and practical consequences.

Research design: The plan linking research questions, evidence, procedures, and conclusions.

Exploratory research: Inquiry used to clarify poorly understood problems or concepts.

Descriptive research: Research that characterizes people, events, conditions, or distributions.

Target population: The full set to which the researcher intends to apply findings.

Sampling frame: The operational list or procedure used to identify eligible units.

Mixed methods: Deliberate integration of quantitative and qualitative approaches.

Feasibility: The practical possibility of completing a study to an adequate standard.

Research ethics: Principles governing autonomy, welfare, fairness, privacy, and responsible conduct.

Triangulation: Using multiple sources, methods, or perspectives to examine a question.

Review Questions

Short Answer Questions

Long Answer Questions

Analytical / Application Questions

Multiple Choice Questions with Answers

Question 1

Which research philosophy most explicitly emphasizes practical consequences and choosing methods according to the research problem?

Question 2

A study interviews users to understand what “trust” means to them in a particular cultural setting. Which approach is most consistent with this purpose?

Question 3

Question 4

Which design is most appropriate when a researcher first conducts interviews and then uses the findings to construct a survey?

Question 5

Using 95% confidence, z ≈ 1.96, p = 0.5, and e = 0.05, the preliminary sample size from n₀ = z²p(1 − p)/e² is approximately:

Question 6

Correct Answer: C. It may be useful for exploration, but statistical generalization is limited

Question 7

A study collects data from the same customers every month for one year. This is best described as:

Question 8

Correct Answer: B. Checking whether indicators represent the intended construct

Question 9

If fixed costs are $2,000, variable cost is $8 per completed case, and 300 completed cases are required, what is the subtotal before contingency?

Question 10

Which ethical practice is most appropriate when using AI to assist with research transcription?

Correct Answer: C. Use an approved secure system and check accuracy and confidentiality

Chapter 3 Research Process and Design

Research Process and Design

Introduction