Qualitative Data Collection and Analysis

Introduction and context

Qualitative research is concerned with how people interpret, experience and construct social reality. In human resource management (HRM), it is especially useful when the research question concerns meaning, relationships, identity, emotions, organisational culture or processes that cannot be adequately represented by numerical measures alone. Questions such as “How do employees experience hybrid work?”, “Why do staff distrust a performance-management system?” and “How is inclusion enacted in everyday team interactions?” require attention to language, context and lived experience.

Qualitative evidence is not simply informal opinion. Rigorous qualitative research involves a defensible design, careful sampling, ethically conducted data collection, systematic analysis, reflexive interpretation and a transparent record of decisions. The researcher seeks contextual insight rather than statistical representativeness. Quality is therefore assessed through concepts such as credibility, dependability, confirmability and transferability.

This chapter explains four central methods—interviews, focus groups, observation and document analysis—and shows how to transform raw material into codes, categories, themes and evidence-based interpretations.

Learning objectives

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

1. Explain the purposes and assumptions of qualitative HRM research.
2. Select and design appropriate interviews, focus groups, observations and document studies.
3. Apply purposive sampling, ethical safeguards and high-quality recording procedures.
4. Code and analyse qualitative material using thematic or content analysis.
5. Establish credibility, reflexivity, dependability, confirmability and transferability.
6. Present contextual findings without exaggeration, disclosure of identities or unsupported claims.

Qualitative inquiry in HRM

Qualitative research is commonly associated with interpretivist and constructivist traditions. An interpretivist researcher assumes that organisational reality is understood through the meanings people attach to events. A constructivist perspective further suggests that knowledge is produced through interaction between participants, context and researcher. These assumptions do not imply that “anything goes”; they require the researcher to make the interpretive process explicit.

A qualitative project may be exploratory, descriptive, explanatory or evaluative. Exploratory studies investigate an under-researched issue; descriptive studies provide a detailed account of experience; explanatory studies examine how participants understand causes and consequences; evaluative studies assess how a policy or intervention operates in practice.

Qualitative and quantitative methods are not opposites in value. They answer different kinds of questions. A survey may estimate how widespread low engagement is, while interviews may explain what “engagement” means to employees and why it changes. Mixed-methods research can combine these strengths, but qualitative evidence should not be judged solely by quantitative standards.

Important terminology

• Unit of analysis: The entity being studied, such as an employee, team, HR process or organisation.
• Participant: A person who contributes data to the study.
• Informant: A participant selected because of relevant knowledge, such as an HR manager or union representative.
• Purposive sampling: Deliberate selection of information-rich cases.
• Reflexivity: Critical examination of how the researcher’s position, assumptions and relationships influence the research.
• Code: A concise label attached to a meaningful segment of data.
• Category: A group of related codes.
• Theme: A patterned, interpretive meaning that answers the research question.
• Negative case: Evidence that challenges an emerging interpretation.
• Audit trail: A documented record of methodological and analytical decisions.

Designing the qualitative study

Aligning the question, method and evidence

A strong design begins with a question rather than a preferred technique. The following alignment is useful:

One method rarely provides a complete account. For example, employees may describe a policy as empowering in interviews, while observation reveals that supervisors still make unilateral decisions. Using several sources allows triangulation, but differences should be analysed rather than mechanically “resolved.”

Sampling and access

Qualitative sampling is usually purposive. Selection criteria should follow the research question and may include role, tenure, location, employment status, gender, shift pattern or experience of a particular HR process. Maximum-variation sampling seeks contrasting cases; criterion sampling includes everyone meeting specified conditions; snowball sampling uses participant referrals, although it can over-represent connected individuals.

Sample size is determined by the scope and information power of the study, not by a universal numerical rule. Relevant considerations include the specificity of the sample, quality of dialogue, complexity of the topic, number of comparisons and analytical strategy. Researchers sometimes use “saturation” to describe the point at which additional data produce no substantially new conceptual insights. Saturation must be demonstrated through an analytic record, not asserted merely because a target number was reached.

Access through an employer creates special risks. Employees may believe participation is compulsory or fear that criticism will be reported to management. Recruitment should therefore make voluntariness clear, avoid supervisor selection where possible, and explain who will see the data. Separate consent should be obtained for audio recording, quotations and any future use.

Conducting interviews

Interview design

A semi-structured interview combines consistency with flexibility. An interview guide normally contains:

1. An opening explanation, consent confirmation and warm-up question.
2. Broad questions about experience.
3. Probes for examples, sequence, meaning and consequences.
4. Questions addressing variation or alternative interpretations.
5. A closing invitation to add anything important.

Questions should be open, neutral and comprehensible. “Tell me about your experience of the appraisal process” is preferable to “Did the fair appraisal process improve your motivation?” The latter presupposes fairness and a positive effect. Useful probes include “Can you give an example?”, “What happened next?”, “How did that affect you?” and “Did everyone see it in the same way?”

The interviewer should listen actively without turning the session into counselling, advocacy or interrogation. Silence can allow reflection. Leading questions, excessive self-disclosure and premature reassurance may distort the account. In sensitive HR topics, participants should be reminded that they may decline a question or stop the interview.

Interview quality

A pilot interview tests wording, sequence, duration, recording and emotional impact. Field notes should record contextual information, notable pauses, non-verbal cues where observable, and the interviewer’s immediate reflections. Transcription may be verbatim or selectively detailed, but the convention must be stated. Removing hesitations can improve readability but may erase analytically relevant uncertainty or emotion.

Conducting focus groups

Focus groups generate data through interaction, not merely through a series of individual answers. They are valuable for exploring shared language, norms and disagreements. A skilled moderator asks questions, manages participation and prevents dominant members from controlling the discussion. A second researcher may observe group dynamics and maintain field notes.

Composition requires careful judgement. Similarity may increase comfort, but excessive homogeneity can suppress useful difference. Mixing supervisors and subordinates is often inappropriate because power may inhibit disclosure. Groups should be small enough for participation and large enough to generate interaction; the appropriate size depends on topic sensitivity, participant availability and the complexity of the guide.

Ground rules should cover respect, one person speaking at a time, voluntary participation and the limits of confidentiality. Unlike an individual interview, the researcher cannot guarantee that other participants will keep disclosures private. This limitation must be explained during consent.

Analysis should attend to both content and interaction. Agreement, hesitation, joking, interruption, silence and challenge may reveal group norms. However, silence must not automatically be interpreted as agreement; it may indicate discomfort, uncertainty or unequal power.

Observation in organisational settings

Observation can be participant or non-participant, overt or—in exceptional circumstances—covert. HR student projects should normally use overt observation because covert observation creates substantial ethical and privacy concerns. The researcher may observe meetings, onboarding, team routines or customer-facing work, provided access and consent are appropriate.

A useful observation protocol distinguishes:

• Descriptive notes: What happened, who was present, where and when.
• Reflective notes: Initial interpretations, questions, emotional responses and possible bias.
• Analytic notes: Emerging concepts, comparisons and links to the research question.

Observation should record the setting as well as speech. Seating, interruptions, technologies, artefacts and the sequence of decisions may be analytically important. The observer should avoid disrupting work and should not record irrelevant personal information. If individuals enter a setting unexpectedly, the researcher needs a plan for privacy and consent.

Document analysis

Documents are socially produced artefacts, not neutral containers of facts. A disciplinary policy, employee handbook, intranet message, meeting minute, grievance letter or diversity report reflects the purposes and power relations of its authoring context.

A document analysis should ask:

1. Who produced the document, for whom and for what purpose?
2. When and under what organisational circumstances was it produced?
3. What concepts, responsibilities and behaviours does it make visible?
4. What is absent, ambiguous or treated as normal?
5. How does the document relate to employee accounts and observed practice?

The researcher should distinguish policy from implementation. A policy may promise employee voice while meeting records show little opportunity for influence. Contradiction is not necessarily a flaw in the data; it may be a finding about the gap between formal and enacted HRM.

From raw data to interpretation

Qualitative analysis is iterative. Researchers move repeatedly between data, codes, developing interpretations and the research question. Analysis can begin during data collection, allowing later interviews to investigate emerging issues, while preserving a clear record of changes.

Preparing the data

Organise recordings, transcripts, field notes and documents using consistent file names and version control. Remove direct identifiers and replace them with pseudonyms. Maintain a data inventory and a log of missing, damaged or excluded material. Software such as NVivo, MAXQDA, ATLAS.ti or a structured spreadsheet can support retrieval and comparison, but software does not perform interpretation automatically.

Coding

A code should capture what a segment means in relation to the research question. Coding may be:

• Descriptive: labels the topic, such as “training access.”
• In vivo: uses a participant’s own phrase, such as “visibility tax.”
• Process: identifies action, such as “avoiding escalation.”
• Emotion: identifies expressed feeling, such as “anxiety.”
• Analytic: records an interpretation, such as “informal norms override policy.”

Start with initial coding of a manageable subset. Codes may be deductive, based on a framework, inductive, developed from the data, or abductive, developed through a dialogue between evidence and theory. A codebook should define each code, include inclusion and exclusion rules, and provide an example.

Thematic analysis

A practical thematic-analysis sequence, consistent with widely used approaches, is:

1. Familiarisation: Read and reread the material while recording initial observations.
2. Initial coding: Code relevant data systematically across the dataset.
3. Theme development: Group related codes into candidate patterns.
4. Theme review: Test whether each theme is coherent internally and distinct from others; return to the full dataset.
5. Theme definition: State the central organising concept, boundaries and implications of each theme.
6. Writing the report: Build an argument using concise evidence and interpretation.

A topic is not automatically a theme. “Communication” may be a broad topic; “communication is available but not safe” is a potentially interpretive theme if supported by patterned evidence.

Content analysis

Qualitative content analysis focuses on the meanings and patterns in communications. Conventional content analysis allows categories to emerge from the data; directed content analysis begins with a theoretical framework; summative approaches may examine the presence and use of words but must still interpret context. Counting coded segments can be useful for transparency, but frequency does not equal importance. A rarely mentioned event may be highly consequential.

Comparing cases and sources

Create matrices that compare roles, sites, time periods or data sources. For example, a matrix might place “voice in performance reviews” in rows and employees, managers and policy documents in columns. Such displays expose convergence, contradiction and missing evidence. Negative cases should be actively sought. An interpretation that explains only supportive participants is weaker than one that accounts for both supportive and resistant accounts.

Credibility, reflexivity and auditability

Qualitative rigour is multidimensional.

• Credibility: The interpretation is plausible and well supported by the data. Strategies include prolonged engagement where feasible, triangulation, participant clarification, peer debriefing and attention to negative cases.
• Transferability: Readers can judge whether findings apply in another context because the report provides thick description of setting, participants and conditions. The researcher does not claim universal generalisation.
• Dependability: The process is logical, traceable and documented. An audit trail records changes to the guide, sampling decisions, coding revisions and analytic meetings.
• Confirmability: Findings are grounded in evidence rather than personal preference. Reflexive notes, quotation-context checks and explicit consideration of rival explanations help.

Member checking requires care. Participants can clarify factual errors or comment on whether interpretations are recognisable, but they are not the sole arbiters of analysis. A manager may reject a finding because it is uncomfortable; this does not automatically invalidate the evidence.

Reflexivity should be continuous. A researcher who is also an employee, HR professional or student may possess insider access but also carry assumptions and loyalties. A reflexive memo can ask: What do I expect to find? Whose account do I trust most? How might my status influence what was disclosed? What did I overlook? Reflexivity is not a confession; it is an analytical resource that makes influence visible.

An audit trail may contain the proposal, ethics approval, recruitment records, interview guide versions, field notes, codebook revisions, analytic memos, case-comparison tables and decisions about excluded data. It should not contain unnecessary identifiable information.

Ethics, governance and data protection

Qualitative HR research often involves power, employment security and sensitive personal information. Ethical practice requires informed consent, voluntary participation, confidentiality, minimisation of harm and secure data management. Participants should know the purpose, procedures, risks, benefits, recording arrangements, withdrawal conditions and dissemination plan.

Confidentiality is a promise about how researchers handle data; anonymity means that identity cannot reasonably be inferred. In a small department, a combination of job title, tenure and quotation may identify a person even after the name is removed. Use broad descriptors, alter non-essential contextual details where ethically permissible, and avoid presenting a recognisable “unique” case without explicit justification.

Data protection obligations vary by jurisdiction, but responsible practice generally includes data minimisation, access control, encryption, retention limits and secure deletion. Employers should not receive raw transcripts merely because they funded access. Mandatory reporting limits—such as imminent serious harm—must be explained before data collection.

Generative AI may assist with transcription, translation or preliminary organisation, but it creates risks involving confidentiality, inaccurate transcription, fabricated summaries, bias and ownership. Do not upload identifiable employee data to an unapproved system. Verify every automated transcript, retain human responsibility for coding and disclose material AI assistance according to institutional policy.

Illustrative case: hybrid work and perceived fairness

Background

The following is a fictional illustrative case. A medium-sized professional-services organisation introduced hybrid work. Senior leaders believed that the policy improved flexibility, but an employee survey showed declining confidence in promotion fairness.

Problem

The research question was: “How do employees and managers experience fairness in promotion decisions under hybrid work?” The study used twelve semi-structured interviews, one focus group with employees from different departments, observation of two promotion-panel meetings, and analysis of the hybrid-work policy and six anonymised promotion communications.

Analysis

Initial codes included “visibility,” “criteria unclear,” “manager advocacy,” “documentation burden,” “flexibility benefit” and “fear of being forgotten.” These codes were compared across sources. The policy presented hybrid work as performance-neutral, while interviews described informal visibility during office days. Observation showed that panel members frequently referred to “who has been seen taking initiative,” although this phrase was absent from the formal criteria. One employee, however, reported that remote work increased concentration and improved performance. This negative case prevented the researchers from treating hybrid work itself as universally harmful.

Three themes were developed:

1. Formal neutrality, informal visibility: Official criteria were neutral, but evaluators used visibility as an unofficial signal.
2. Flexibility with unequal costs: Hybrid work benefited some employees but created different access to sponsorship and information.
3. Evidence work as a hidden requirement: Employees undertook additional documentation to make remote contributions visible.

Approach and outcome

The researchers recommended publishing behavioural examples for promotion criteria, recording evidence of outcomes rather than physical presence, training panel members to identify proximity bias, and auditing promotion decisions by work pattern and role. The report used pseudonyms, carefully selected quotations and a transparent account of contradictory evidence.

Learning points

The case demonstrates why multiple methods are valuable, why policy documents cannot be treated as evidence of practice, and why themes should explain patterns rather than merely repeat interview topics. It also shows that a qualitative conclusion can be practically significant without claiming that every employee has the same experience.

Common challenges and best practices

Qualitative projects face practical limitations. Participants may provide socially desirable answers, access may be controlled by gatekeepers, group dynamics may silence minority voices, and the researcher may collect more material than can be analysed well. Small samples limit breadth, while transcription and coding are time-intensive. These are not reasons to abandon qualitative research; they are conditions to manage openly.

Best practice includes narrowing the research question, piloting instruments, keeping a recruitment log, recording context during collection, coding consistently but flexibly, searching for disconfirming evidence, separating description from interpretation, and linking every major claim to adequate evidence. Reports should explain what was studied, how it was studied, what was found, how the interpretation was developed and what remains uncertain.

Professional and employability relevance

HR professionals routinely interpret exit interviews, engagement comments, grievance narratives, consultation feedback, meeting records and employee-relations cases. The skills in this chapter support evidence-informed practice: asking non-leading questions, recognising power, identifying patterns without stereotyping, protecting confidentiality and translating employee experience into fair organisational action. These skills are relevant to HR analytics, employee relations, organisational development, learning and development, diversity and inclusion, recruitment research and change management.

Chapter Summary

Qualitative research provides contextual understanding of how employees and organisations experience and construct HR practices. Interviews generate detailed individual accounts; focus groups reveal interaction and shared norms; observation examines practice in context; and document analysis investigates formal meanings, purposes and omissions. Rigorous studies align questions, sampling, methods and evidence; protect participants from employment-related risks; and maintain a transparent audit trail.

Analysis proceeds through familiarisation, coding, categorisation, theme development, comparison and interpretation. Codes are tools for disciplined thinking, not substitutes for judgement. Credibility, transferability, dependability and confirmability are strengthened through triangulation, negative-case analysis, reflexivity, thick description and documented decisions. Ethical practice requires informed consent, confidentiality, secure data management and cautious use of technology, including generative AI. The final objective is not merely to report what participants said, but to develop a defensible explanation of patterned experience and its implications for HRM.

Key Terms

Purposive sampling: Deliberate selection of participants or cases that can illuminate the research question.

Semi-structured interview: An interview guided by prepared questions but allowing probing and flexibility.

Focus group: A moderated group discussion in which interaction is itself a source of data.

Observation: Systematic attention to behaviour, interaction and context in a naturally occurring setting.

Document analysis: Critical interpretation of written, visual or digital organisational artefacts.

Reflexivity: Examination of how the researcher’s position and assumptions shape the study.

Code: A concise label assigned to a meaningful segment of data.

Codebook: A documented set of code definitions, boundaries and examples.

Category: A grouping of related codes.

Theme: An interpretive pattern of meaning relevant to the research question.

Thematic analysis: A systematic process for developing and interpreting patterned meanings.

Content analysis: Analysis of the meanings and patterns within communications or documents.

Triangulation: Comparing multiple sources, methods, researchers or perspectives.

Negative case: Evidence that challenges or qualifies an emerging interpretation.

Credibility: The plausibility and evidential support of qualitative findings.

Transferability: The extent to which readers can assess relevance to other contexts.

Dependability: The consistency and traceability of the research process.

Confirmability: The extent to which findings are grounded in data rather than researcher preference.

Audit trail: A record of methodological, ethical and analytical decisions.

Thick description: Detailed contextual description enabling informed interpretation and comparison.

Review Questions

Short Answer Questions

1. What is the difference between a code, a category and a theme?
2. Why is purposive sampling commonly used in qualitative HRM research?
3. Identify two limitations of focus groups.
4. What is the purpose of a reflexive memo?
5. Distinguish confidentiality from anonymity.

Long Answer Questions

1. Explain how interviews, focus groups, observation and document analysis produce different kinds of qualitative evidence.
2. Describe the stages of a rigorous thematic-analysis process and explain the purpose of each stage.
3. Discuss how credibility, transferability, dependability and confirmability can be established in an HRM dissertation.
4. Explain the ethical challenges of researching employees through an organisational gatekeeper.
5. Critically assess the value and limitations of using generative AI in qualitative data collection and analysis.

Analytical / Application Questions

1. An organisation’s policy promises equal access to promotion, but remote employees report reduced visibility. Design a qualitative study using at least two methods and justify your sampling and analysis strategy.
2. During a focus group, one senior employee answers every question while junior employees remain silent. Analyse the methodological and ethical problem and propose corrective action.
3. A researcher finds that most interviews support an emerging theme, but two participants strongly contradict it. Explain how the researcher should analyse and report these negative cases.

Multiple Choice Questions with Answers

Question 1

Which method is most suitable for exploring how employees interact with one another during a team meeting?

A. Observation

B. A closed-ended survey

C. Payroll analysis

D. A financial audit

Correct Answer: A. Observation

Question 2

What is the main purpose of purposive sampling?

A. To give every person an equal probability of selection

B. To select information-rich participants relevant to the research question

C. To eliminate the need for ethical approval

D. To guarantee statistical generalisation

Correct Answer: B. To select information-rich participants relevant to the research question

Question 3

Which question is least leading in a semi-structured interview?

A. How did the fair appraisal process improve your motivation?

B. You agree that hybrid work is beneficial, don’t you?

C. Tell me about your experience of the appraisal process.

D. Why did your manager fail to support you?

Correct Answer: C. Tell me about your experience of the appraisal process.

Question 4

In thematic analysis, what normally follows initial coding?

A. Immediate statistical generalisation

B. Grouping related codes into candidate themes

C. Deleting contradictory evidence

D. Publishing the raw transcripts

Correct Answer: B. Grouping related codes into candidate themes

Question 5

Which concept refers to documenting methodological and analytical decisions?

A. Audit trail

B. Anonymity

C. Sampling frame

D. Inter-rater frequency

Correct Answer: A. Audit trail

Question 6

Why should a researcher examine negative cases?

A. To remove participants who disagree

B. To test and qualify an emerging interpretation

C. To increase the number of quotations automatically

D. To replace the research question

Correct Answer: B. To test and qualify an emerging interpretation

Question 7

Which statement best describes transferability?

A. The statistical probability that findings are correct

B. The extent to which readers can judge relevance to another context

C. The guarantee that no participant can be identified

D. The number of codes created by a researcher

Correct Answer: B. The extent to which readers can judge relevance to another context

Question 8

What is a major ethical risk when an employer recruits its own employees for research?

A. Participants may perceive participation as compulsory

B. The sample will always be too large

C. Documents cannot be analysed

D. Observation becomes impossible

Correct Answer: A. Participants may perceive participation as compulsory

Question 9

Which practice best protects confidential qualitative data?

A. Uploading identifiable transcripts to any free AI tool

B. Giving raw transcripts to all managers

C. Using access controls, pseudonyms and secure storage

D. Publishing job titles with distinctive quotations

Correct Answer: C. Using access controls, pseudonyms and secure storage

Question 10

Why is a policy document not necessarily evidence of actual HR practice?

A. Policies contain no language

B. Documents reflect formal intentions that may differ from implementation

C. Documents are always less reliable than interviews

D. Policy analysis cannot identify organisational power

Correct Answer: B. Documents reflect formal intentions that may differ from implementation