Research Philosophy and Study Design

Introduction

Every dissertation rests on decisions about what counts as knowledge, how knowledge can be justified, and how evidence should be collected. These decisions are not merely technical. A study of employee engagement, for example, will look different if engagement is treated as a measurable psychological state, a socially constructed experience, a consequence of unequal power relations, or a practical organisational problem requiring several kinds of evidence.

Research philosophy concerns assumptions about reality, knowledge, values and the relationship between the researcher and the researched. Study design concerns the overall plan for answering a research question. A sound design creates alignment among the research problem, philosophical position, reasoning approach, data, methods and conclusions. Misalignment—such as using an interpretive question with a narrowly positivist design—can weaken validity even when data collection is competently performed.

This chapter compares four influential philosophical perspectives, explains deductive, inductive and abductive reasoning, and examines exploratory, descriptive, explanatory, experimental, longitudinal and cross-sectional designs. Examples are drawn primarily from human resource management (HRM).

Learning Objectives

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

1. Explain the meaning and importance of research philosophy.
2. Compare positivist, interpretivist, critical and pragmatic perspectives.
3. Distinguish deductive, inductive and abductive reasoning.
4. Select a research design appropriate to a research question.
5. Evaluate issues of validity, credibility, ethics and feasibility.
6. Justify a coherent philosophy–reasoning–design–method chain in a dissertation proposal.

1. Research Philosophy: The Foundations of Inquiry

A research philosophy is a set of assumptions that guides how a researcher understands the world and produces knowledge. Four dimensions are especially useful:

Ontology includes realist positions, which view phenomena as existing independently of observers, and constructionist positions, which emphasise that meanings are produced through social interaction. Epistemology addresses whether knowledge is best generated through observation and measurement, interpretation of meaning, critique of power, practical problem-solving, or a combination of these. Axiology recognises that research choices may reflect values, including commitments to fairness, employee voice or organisational improvement.

Philosophy does not mechanically determine one method. Surveys are often associated with positivism, and interviews with interpretivism, but either method can be used from different philosophical positions. A critical researcher may use a survey to document inequality, while a positivist researcher may conduct structured interviews to measure clearly defined perceptions. The important issue is coherence and justification.

2. Four Major Perspectives

2.1 Positivism

Positivism assumes that an external reality can be studied through systematic observation. It commonly values measurement, objectivity, replication, hypothesis testing and the identification of regular relationships. Concepts such as job satisfaction, turnover intention and absenteeism are operationalised as variables, and evidence is collected using standardised instruments.

A typical positivist HRM study might hypothesise that perceived supervisor support is negatively associated with turnover intention. The researcher would define both constructs, collect comparable responses from a sample, and test whether the observed relationship supports the hypothesis.

Strengths include clarity, comparability, statistical estimation and potential generalisability. Limitations include the risk of reducing complex human experiences to numerical indicators, overlooking context and treating researcher-defined categories as if they were universally meaningful. Statistical association does not by itself establish causation.

2.2 Interpretivism

Interpretivism holds that social reality is experienced and interpreted through meanings, language, relationships and context. Employees may understand “fairness,” “leadership” or “career success” differently across departments, cultures and life circumstances. Knowledge is therefore developed through close engagement with participants rather than only through detached measurement.

Interpretive designs commonly use semi-structured interviews, focus groups, observations, diaries and document analysis. Quality is assessed through credibility, reflexivity, transparency, depth and sensitivity to context. The researcher acknowledges that interaction with participants influences the knowledge produced.

Interpretivism is valuable when a topic is poorly understood, sensitive or meaning-laden. Its limitations include time-intensive analysis, limited statistical generalisation and the possibility that the researcher’s interpretation will be insufficiently supported by the data. These limitations can be addressed through a clear sampling rationale, reflexive records, careful coding and participant-sensitive interpretation.

2.3 Critical Perspectives

Critical research examines how power, inequality, ideology and institutional arrangements shape social life. It asks not only “what happens?” but also “who benefits?”, “whose knowledge is privileged?” and “how might conditions be changed?” In HRM, critical research may examine discrimination in promotion, insecure work, algorithmic surveillance, unpaid emotional labour or the gap between equality policies and lived experience.

Critical approaches may draw on Marxist, feminist, critical race, postcolonial or labour-process traditions. They often treat participants as knowledgeable social actors rather than passive sources of data. Research may seek both understanding and transformation, for example by making marginalised employee experiences visible.

The strength of a critical perspective is its ability to reveal structural causes and ethical consequences that conventional performance measures may hide. A limitation is that openly normative research may be criticised as less neutral. Critical researchers respond that all research involves choices and that pretending to be value-free can conceal existing power relations. Rigorous critical inquiry still requires evidence, transparent reasoning and fair representation of alternative interpretations.

2.4 Pragmatism

Pragmatism begins with the research problem and asks which forms of inquiry will produce useful, credible answers. It does not require commitment to one universal view of reality or one preferred method. A pragmatic HRM study might combine a survey showing that remote workers report isolation with interviews explaining how communication practices create that experience.

Pragmatism supports mixed methods, sequential designs and practical recommendations. It is particularly appropriate for organisational problems involving both measurable outcomes and human meanings. However, “whatever works” is not a sufficient justification. Researchers must explain why particular methods fit the question, how different forms of evidence will be integrated, and how contradictions will be handled.

Comparison of Perspectives

No perspective is automatically superior. Suitability depends on the question, purpose, participants, resources and ethical context.

3. Reasoning in Research

Research reasoning connects concepts, evidence and conclusions. The three principal forms are deduction, induction and abduction.

3.1 Deductive Reasoning

Deduction moves from a general proposition or theory to specific predictions. A researcher may reason:

• Theory proposes that autonomy improves intrinsic motivation.
• Employees with greater perceived autonomy should report higher motivation.
• A measurable hypothesis is tested with data.

Deduction is common in explanatory and experimental research. Its advantages are theoretical discipline and transparent hypothesis testing. It can nevertheless test only what has been specified and may miss unexpected meanings or mechanisms.

3.2 Inductive Reasoning

Induction moves from observations toward patterns, concepts or broader explanations. A researcher interviewing employees about career development may notice recurring themes: informal sponsorship, unequal access to high-profile assignments and uncertainty about promotion criteria. These observations can support a conceptual explanation developed from the data.

Induction is common in exploratory qualitative research. It does not mean that researchers begin without any assumptions; prior literature and experience influence attention. Strong inductive work makes this influence explicit and distinguishes evidence from speculation.

3.3 Abductive Reasoning

Abduction is reasoning to the best available explanation. It begins with an observation that is surprising, incomplete or inconsistent with an existing explanation. The researcher moves between data and theory, revising interpretations as new evidence appears.

For example, an organisation may find that a costly wellbeing programme has high participation but no improvement in absence rates. Interviews may reveal that employees fear being seen as unable to cope. The researcher may revise the initial explanation—low participation or poor programme quality—and develop a new account centred on stigma and trust.

Abduction is valuable for complex organisational phenomena and mixed methods. It combines creativity with disciplined comparison of rival explanations. It should not be confused with guessing: the proposed explanation must be supported, tested or refined through evidence.

4. From Research Question to Design

A study design is the logical structure that links a research question to evidence and conclusions. Before choosing a design, a researcher should clarify:

1. Purpose: Is the study exploring, describing, explaining, predicting, evaluating or changing practice?
2. Unit of analysis: Is the focus an individual employee, team, department, organisation, policy or industry?
3. Time: Is one point sufficient, or must change be observed?
4. Control: Can an intervention be assigned or manipulated ethically and practically?
5. Access and feasibility: Are participants, records, time and skills available?
6. Ethics: Could participation create risk, retaliation, identification or employment disadvantage?

The terms below describe different, sometimes overlapping, design purposes and time structures.

4.1 Exploratory Design

Exploratory research investigates a topic that is insufficiently understood. It seeks concepts, questions, mechanisms or hypotheses rather than precise population estimates. Methods may include literature reviews, interviews, focus groups, observations, document analysis and qualitative case studies.

Example: A researcher explores how employees experience AI-assisted performance appraisal before constructing a survey instrument. Exploration is appropriate when terminology is unsettled or existing measures may not reflect local meanings. It should not be presented as proof of prevalence or causal effect.

4.2 Descriptive Design

Descriptive research characterises a population, situation or distribution. Questions include: What proportion of employees work remotely? What are the reported levels of engagement? Which groups experience the highest training access?

Descriptive studies require clear definitions, appropriate sampling and reliable measurement. A sample percentage describes the sample directly; generalising it to a wider population requires a defensible sampling design and attention to sampling error and nonresponse. Descriptive evidence can establish the scale of an issue without explaining why it occurs.

4.3 Explanatory Design

Explanatory research investigates relationships, mechanisms or causes. It may test whether leadership style predicts engagement, whether workload mediates the relationship between understaffing and burnout, or why a policy produces different outcomes across departments.

Explanatory designs may use statistical modelling, comparative case analysis, process tracing or mixed methods. Researchers must distinguish association from causation. Causal claims require temporal ordering, plausible mechanisms and attention to alternative explanations; experiments provide stronger control than most observational designs.

4.4 Experimental and Quasi-Experimental Design

An experiment manipulates an independent variable, compares conditions and measures an outcome. Random assignment, when feasible, helps balance observed and unobserved differences between groups. For example, teams could be randomly assigned to receive a structured coaching intervention or a comparison condition, with engagement measured before and after implementation.

A simple pretest–posttest comparison can be represented as:

Here, “Outcome” might be a validated engagement score. This difference describes change within a group but does not by itself prove that the intervention caused it. History, maturation, concurrent policy changes and regression to the mean may also explain change.

A stronger design compares intervention and control groups. A difference-in-differences estimate is commonly expressed as:

Y_T is the outcome for the treatment group, Y_C is the outcome for the comparison group, and “before” and “after” identify measurement times. The estimate is most credible when groups would otherwise have followed parallel trends, treatment exposure is clearly defined, spillovers are limited, and measurement is consistent. In organisational research, randomisation may be constrained by fairness, managerial authority or contamination between teams; quasi-experiments using matched groups or policy timing may then be more realistic.

4.5 Cross-Sectional Design

A cross-sectional study collects data at one point or during a short period. It is efficient for describing current conditions and examining associations. An employee survey conducted in March to assess perceived fairness and intention to leave is cross-sectional.

Its principal limitation is weak evidence about temporal order. Employees who intend to leave may perceive management less favourably, rather than poor management causing intention to leave. Common-method bias can also arise when the same respondent reports both predictor and outcome at the same time.

4.6 Longitudinal Design

A longitudinal study collects data at multiple times. It can examine development, stability, transitions and temporal ordering. Designs include panel studies, repeated cross-sectional studies, cohort studies and time-series studies.

A panel follows the same individuals; a repeated cross-sectional study samples different individuals from the same population at each wave. Longitudinal research can identify whether engagement changes after a restructuring, but it faces attrition, changing organisational conditions, panel conditioning and increased cost. Researchers should document missing waves and avoid assuming that observed change is caused by the event under study.

Design Selection Guide

5. Ensuring Alignment and Quality

A coherent dissertation can be represented as a chain:

Research problem → research question → philosophical perspective → reasoning approach → design → sampling → data collection → analysis → claims.

For example, the question “How do newly promoted supervisors learn to manage conflict?” supports an interpretivist, inductive, exploratory design using purposive sampling and semi-structured interviews. The question “Does a conflict-management course reduce formal grievances within six months?” supports a pragmatic or positivist, deductive, explanatory evaluation design, potentially using a quasi-experimental comparison.

Quality criteria vary by design but should always be explicit. Quantitative studies consider reliability, construct validity, internal validity, external validity and statistical conclusion validity. Qualitative studies consider credibility, dependability, confirmability, transferability and reflexivity. Mixed-methods studies must also show how datasets are connected—for example, whether qualitative findings explain quantitative patterns.

Sampling should match the claim. Probability sampling supports statistical generalisation when properly implemented. Purposive sampling seeks information-rich cases and is appropriate for interpretive or critical inquiry, but the resulting claims should concern meanings, processes or theoretical insight rather than population percentages. Convenience sampling may be practical but increases risks of selection bias.

6. Ethical and Governance Considerations

HR research involves power asymmetries. Employees may fear that honest answers will affect promotion, pay or job security. Ethical practice therefore requires informed consent, voluntary participation, confidentiality, secure data storage and clear separation between research data and managerial appraisal wherever possible.

Researchers should minimise coercion, avoid collecting unnecessary personal data and explain limits to confidentiality. Small departments may make individuals identifiable even when names are removed. Sensitive topics such as discrimination, harassment, health or immigration status require additional safeguards. Institutional ethics approval, data-protection compliance and transparent retention and deletion procedures are essential.

Technology creates further issues. AI tools may assist transcription, coding or survey design, but researchers remain responsible for accuracy, confidentiality, bias and disclosure. Personal or identifiable employee data should not be entered into an external AI system without appropriate authorisation. Automated coding should be checked against the original data, and participants should not be misrepresented by opaque classification systems.

7. Illustrative Case: Designing a Study of Hybrid-Work Fairness

Background

A fictional professional-services organisation introduces hybrid working. Senior managers believe flexibility improves retention, while some employees report that remote workers receive fewer developmental opportunities.

Problem and questions

The organisation needs to know both whether a disparity exists and how it is produced. Two questions are proposed:

1. Do office attendance, access to mentoring and promotion outcomes differ across employee groups?
2. How do employees interpret and experience fairness in hybrid-work decisions?

Analysis of design choices

A single survey could describe perceived fairness but might not explain informal allocation of opportunities. Interviews could reveal mechanisms but could not estimate the scale of differences. A pragmatic mixed-methods design is therefore appropriate. The first phase uses administrative records and a confidential survey to describe patterns. The second uses purposive interviews with employees at different grades, locations and caregiving circumstances. Abductive reasoning allows the researcher to move between statistical patterns and employee explanations.

If records show that remote employees receive fewer promotions, this is an association, not automatic proof of discrimination. Job tenure, role type, performance, manager assignment and self-selection into remote work may be relevant. A longitudinal follow-up could examine promotion and mentoring over 12 months. If a transparent mentoring intervention is introduced to selected comparable teams, a quasi-experimental evaluation could assess whether access improves.

Ethical approach

The research team should prevent managers from seeing identifiable responses, report sufficiently aggregated results, protect small groups from re-identification and provide participants with information about data use. Interviews about discrimination require careful safeguarding and referral information if distress occurs.

Outcome and learning points

The organisation receives a profile of disparities, employee accounts of informal decision-making and recommendations for transparent mentoring allocation. The case demonstrates that design should follow the problem: descriptive evidence identifies the pattern, interpretive evidence explores meaning and mechanism, and longitudinal or quasi-experimental evidence can assess change.

8. Common Errors and Best Practices

Common errors include choosing a method before formulating a question, treating a philosophical label as a substitute for justification, claiming causation from a cross-sectional correlation, generalising from a convenience sample, ignoring contradictory evidence and presenting participant quotations without context. Another error is assuming that mixed methods automatically produce stronger research; integration must be planned rather than merely placing a survey and interviews side by side.

Best practice involves stating assumptions explicitly, defining the unit of analysis, aligning design with the claim, piloting instruments, documenting decisions, protecting participants and distinguishing evidence from interpretation. A dissertation proposal should explain not only what will be done, but why the design is the most defensible feasible response to the question.

Chapter Summary

Research philosophy provides assumptions about reality, knowledge and values. Positivism emphasises measurement and testing; interpretivism emphasises meaning and context; critical perspectives examine power and inequality; pragmatism selects methods according to the practical research problem. Deduction tests theory-based predictions, induction develops concepts from observations, and abduction seeks the best explanation by moving between theory and evidence.

Exploratory designs investigate poorly understood phenomena, descriptive designs portray conditions, explanatory designs examine relationships and mechanisms, and experimental or quasi-experimental designs provide stronger evidence about interventions. Cross-sectional studies offer efficient snapshots, whereas longitudinal studies examine change and temporal order. Effective dissertations align question, philosophy, reasoning, design, sampling, analysis, ethics and claims. No design eliminates all limitations; rigour depends on transparent justification and appropriate interpretation.

Key Terms

Ontology: Assumptions about the nature of reality.

Epistemology: Assumptions about how knowledge can be produced and justified.

Axiology: The study of values and their role in research.

Methodology: The overall logic and organisation of inquiry.

Positivism: A perspective emphasising observable, measurable and systematically testable knowledge.

Interpretivism: A perspective emphasising socially produced meanings and contextual understanding.

Critical research: Inquiry concerned with power, inequality, ideology and social transformation.

Pragmatism: A problem-centred perspective that evaluates methods by their usefulness and consequences.

Deduction: Reasoning from general theory to specific predictions.

Induction: Reasoning from observations toward patterns or concepts.

Abduction: Reasoning toward the best explanation through interaction between theory and evidence.

Exploratory design: A design used to investigate an insufficiently understood issue.

Descriptive design: A design used to portray characteristics, distributions or conditions.

Explanatory design: A design used to examine relationships, mechanisms or causes.

Experimental design: A design involving manipulation and comparison, ideally with random assignment.

Quasi-experiment: An intervention study lacking full random assignment or experimental control.

Cross-sectional design: A design collecting data at one point or short period.

Longitudinal design: A design collecting data across multiple time points.

Internal validity: The credibility of a causal conclusion within the study.

Reflexivity: Critical examination of how the researcher influences the research process.

Review Questions

Short Answer Questions

1. What is meant by research philosophy?
2. State one major difference between positivism and interpretivism.
3. What is abductive reasoning?
4. Why is a cross-sectional design weak for establishing temporal order?
5. Name two ethical risks in HR employee research.

Long Answer Questions

1. Compare positivist, interpretivist, critical and pragmatic perspectives in relation to HRM research.
2. Explain deduction, induction and abduction, using an HRM example for each.
3. Discuss the strengths and limitations of exploratory, descriptive and explanatory designs.
4. Explain how experimental and quasi-experimental designs support causal inference.
5. Develop a philosophy–reasoning–design justification for a dissertation on employee wellbeing.

Analytical / Application Questions

1. An organisation finds that remote employees receive fewer promotions. Design a study that distinguishes description, explanation and causal evaluation.
2. A survey shows low trust in an AI-based appraisal system, but participation is high. Which reasoning approach and follow-up design would best investigate this apparent contradiction? Justify your answer.
3. Critically assess whether a pragmatic mixed-methods design is appropriate for studying discrimination in recruitment. Address ethics, sampling, data integration and the limits of causal claims.

Multiple Choice Questions with Answers

Question 1

Which philosophical perspective primarily emphasises socially produced meanings and context?

A. Positivism

B. Interpretivism

C. Behaviourism

D. Functionalism

Correct Answer: B. Interpretivism

Question 2

A researcher begins with a theory that autonomy increases motivation and tests a specific hypothesis. Which reasoning approach is being used?

A. Induction

B. Abduction

C. Deduction

D. Reflexivity

Correct Answer: C. Deduction

Question 3

Which design is most suitable for investigating a poorly understood employee experience before developing a survey?

A. Exploratory design

B. Descriptive design

C. Experimental design

D. Repeated cross-sectional design

Correct Answer: A. Exploratory design

Question 4

What is a principal limitation of a cross-sectional study?

A. It cannot collect numerical data

B. It always requires random assignment

C. It provides weak evidence about temporal order

D. It cannot describe a population

Correct Answer: C. It provides weak evidence about temporal order

Question 5

Which perspective explicitly focuses on power and inequality?

A. Critical perspective

B. Positivist perspective

C. Pragmatic perspective

D. Descriptive perspective

Correct Answer: A. Critical perspective

Question 6

What is the defining feature of an experiment?

A. The use of interviews

B. Manipulation of an independent variable and comparison of conditions

C. Collection of data from an entire population

D. Analysis at only one point in time

Correct Answer: B. Manipulation of an independent variable and comparison of conditions

Question 7

A study follows the same employees every six months for two years. What type of design is this?

A. Cross-sectional

B. Longitudinal panel

C. One-shot case study

D. Purely descriptive snapshot

Correct Answer: B. Longitudinal panel

Question 8

Which statement about a correlation in an observational study is most accurate?

A. It automatically proves causation

B. It proves temporal order

C. It may reflect confounding or reverse causality

D. It eliminates measurement error

Correct Answer: C. It may reflect confounding or reverse causality

Question 9

What is the main purpose of abductive reasoning?

A. To calculate a sample percentage

B. To select participants randomly

C. To identify the best explanation for an observation

D. To remove all researcher values

Correct Answer: C. To identify the best explanation for an observation

Question 10

Which practice best protects employees participating in sensitive HR research?

A. Allowing managers to inspect individual responses

B. Collecting unnecessary identifying information

C. Separating research data from appraisal and protecting confidentiality

D. Avoiding informed consent to reduce response bias

Correct Answer: C. Separating research data from appraisal and protecting confidentiality