Qualitative approaches to social research
Introduction to Social Research Methodology
The logic of qualitative inquiry
In session 1 we classified research along two axes; this session takes up one end of the first of them. Qualitative research works with words, images, and observed behaviour — data not easily reduced to numbers. It is usually descriptive rather than experimental, in that the researcher observes the world rather than manipulating it, and its reasoning is characteristically inductive: it builds understanding up from observations rather than testing hypotheses fixed in advance. Babbie discusses much of this territory under the heading of field research, and notes that a field researcher seldom approaches the task with precisely defined hypotheses to be tested; more typically, the task is to make sense of an ongoing process that cannot be predicted in advance. This is exactly the situation of Meridian’s board: it knows that turnover has doubled, but nobody yet knows why — and you cannot test an explanation you have not yet discovered.
Two commitments define the qualitative outlook. The first is naturalism: study social life where it actually happens, rather than in the artificial settings of the survey or the laboratory. Behaviour in a store, a warehouse, or a team meeting is best understood in that setting, shaped by its pressures and routines. The second is meaning: take seriously the actor’s own point of view — what the rota means to a shift worker, or what “engagement” means to the people being scored on it. Where a survey imposes the researcher’s categories, qualitative work asks what categories the participants themselves use. By going directly to the phenomenon and observing it as completely as possible, researchers can notice nuances of attitude and behaviour that escape every other method.
Qualitative studies work with small numbers — a dozen interviews, three teams, one organisation — and participants are selected purposively rather than randomly: they are chosen because they are informative, such as the longest-serving employee, the store that bucks the trend, or the customer who left. Small size is a consequence of depth, not a defect: an hour-long interview generates more material than a hundred questionnaires. The trade-off is explicit — qualitative samples do not statistically represent a population, and honest qualitative researchers never claim they do. Data collection often stops at saturation, the point at which new cases keep telling you what you already know.
Because qualitative inquiry typically runs from data to theory, it is the natural approach when a phenomenon is new, sensitive, or poorly understood — when you do not yet know enough to write the right closed questions. Its findings often become the hypotheses that later quantitative work tests, completing the cycle between induction and deduction described in session 1. Immersion in a setting also challenges preconceptions: a researcher may assume Meridian staff leave over pay, only to find after ten interviews that pay has barely been mentioned while unpredictable rotas come up every time. Creswell & Creswell treat this flexibility as a hallmark of the qualitative worldview: the design stays emergent, adapting as understanding grows.
Qualitative approaches are the right choice for topics that defy simple quantification, for attitudes and behaviours best understood in their natural setting, and for processes unfolding over time — how a reorganisation, a conflict, or a new system settles in, watched as it happens rather than reconstructed afterwards. Babbie’s inventory of field-research topics translates directly into organisational life: practices and encounters (how handovers, briefings, and customer interactions are actually done), roles and relationships (how “team leader” or “new manager” is performed and received), and groups and organisations (shift cliques, departments, whole firms). Whenever the question is how or why rather than how many, the qualitative family deserves a look.
The qualitative family
Five approaches make up the core of the qualitative family. They share the logic set out above — naturalism, meaning, purposive sampling, induction — but differ in what kind of evidence they produce, and therefore in which questions they suit. This session is about choosing among them; how to actually conduct interviews and focus groups is the subject of session 12, and how to analyse qualitative data is the subject of session 14.
| Approach | What it gathers | Questions it suits |
|---|---|---|
| In-depth interviews | One person’s account, in their own words | What does this experience mean to individuals? |
| Focus groups | Talk between people — views forming in interaction | How will people collectively react to X? |
| Observation and ethnography | What people actually do, in context | What do people actually do here? |
| Case studies | Everything about one instructive case | How does it all fit together in this one setting? |
| Document and content analysis | What an organisation’s texts reveal | What does the paper trail reveal? |
In-depth interviews
An in-depth interview is a one-to-one conversation designed to gather detailed information about experiences, motives, and interpretations. Interviews sit on a spectrum of structure. Structured interviews use a predetermined set of questions delivered in a consistent order, ensuring uniformity across participants; they produce quantifiable, easily compared results, but at the cost of depth and nuance. Semi-structured interviews combine a topic guide with the freedom to probe and follow up, using open-ended questions within a loose framework — the workhorse of organisational research, as in the pairing “how would you describe the work culture here?” followed by “can you share specific experiences that shaped this perception?”. Unstructured interviews are open and exploratory, without a fixed set of questions; the participant’s narrative leads, which suits life histories and sensitive topics.
The strengths of interviewing are depth (topics explored in detail, in the respondent’s own words), flexibility (questions adapt to the answers, and unexpected themes can be pursued), and clarification and rapport (misunderstandings are fixed on the spot, and trust invites candour). The limits are that interviews are time-consuming and costly to schedule, conduct, transcribe, and analyse; that the interviewer’s presence and phrasing shape responses; and that they yield accounts rather than behaviour — what people say they do is not always what they do. For organisational diagnosis, interviews are unbeatable for individual meaning but weak evidence for how common a view is. At Meridian, exit interviews with leavers and semi-structured interviews with current staff are the most direct route to why people go.
Focus groups
In a focus group, typically five to fifteen people are brought together in a comfortable, private setting for a guided discussion of a topic, led by a moderator. The distinctive data is the interaction: how views are voiced, challenged, defended, and revised in front of peers. Focus groups are a staple of market research — groups evaluating a product, or reacting to a proposal before launch — and of political research on how policies will be understood. Participants do not statistically represent any meaningful population; the purpose of a focus-group study is to explore rather than to describe or explain definitively, and typically more than one group is convened, because any single group of five to fifteen people may be idiosyncratic.
Their strengths are diverse perspectives gathered efficiently in one sitting, interactivity (participants react to one another, surfacing agreements and fault lines that a one-to-one interview cannot show), and rich data including emotions, group norms, and body language. Their limits are group dynamics (dominant voices can silence others, and minority views may go unspoken), conformity and caution (people say less, or say safer things, in front of colleagues), and less control with more complex analysis (discussions wander, and many voices make intricate data). For organisational diagnosis they are excellent for testing reactions and surfacing shared understandings, but risky for sensitive topics where hierarchy is in the room. Before Meridian rolls out a new bonus scheme, two or three staff focus groups would reveal how it will be understood — and misunderstood — on the shop floor.
Observation and ethnography
Observational research involves watching and recording the behaviour and interactions of people in their natural (or controlled) settings, without manipulating the environment or the subjects. Its variants differ by the researcher’s position: in non-participant observation the researcher watches without joining in, while in participant observation the researcher takes part in the life of the group being studied. Ethnography extends participant observation into long-term immersion — weeks or months inside a setting, building trust and understanding the group from an insider’s perspective. Observation suits routines, workarounds, and informal norms: the gap between the process manual and what actually happens.
Its strengths are that it captures behaviour rather than accounts — what people do, not what they remember or admit doing — that it preserves context, and that it is well suited to processes over time and to practices that participants cannot fully articulate. Its limits are observer effects (being watched changes behaviour, at least initially), subjectivity (findings pass through the observer’s interpretation, which careful design and multiple observers can mitigate), and the fact that it is slow, local, and hard to generalise from — deep knowledge of one setting, silence about all others. Ethical care is also needed: consent and privacy are harder to manage than in an interview, especially in semi-public settings. A researcher spending two weeks working alongside a Meridian store team would see how the rota, the scheduling app, and the Saturday rush actually collide — things no one would think to mention in an interview.
Case studies
A case study is an in-depth examination of a single instance — one store, one team, one merger, one crisis — using multiple sources of evidence in combination: interviews, observation, documents, and numbers. The case is chosen purposively, because it is typical, extreme, deviant, or revelatory. Case study is one of Creswell & Creswell’s five core qualitative designs, and it is the natural home of the question “what is going on in this particular place, and what can it teach us?”. It suits questions about how many factors combine in context, and it is the standard way to learn from outliers. Its limits are that case boundaries blur, that depth is costly, and that a single case supports only cautious generalisation. At Meridian, one store has kept turnover below 10% while the company average doubled; a case study of that store — its manager, its rota practice, its team — may be worth more than any average.
Document and qualitative content analysis
Organisations write constantly: policies, emails, announcements, meeting minutes, exit forms, reviews, complaints, chat logs. Document analysis treats these texts as evidence about the organisation that produced them. Content analysis does so systematically: the researcher defines the question and selects a sample of material, develops a coding scheme of categories or themes, pre-tests and revises it, applies it consistently to the whole sample, and then analyses the coded material for patterns, frequencies, and relationships. A classic example is coding a year’s worth of newspaper articles to examine how women in leadership roles are portrayed — recording words such as “strong”, “emotional”, or “intelligent” and themes such as professional competence or personal life, and comparing the patterns with coverage of male counterparts.
The approach is unobtrusive: nobody is interviewed or observed, so nobody reacts to being studied, and the past is as accessible as the present. Its limits are that documents record the official trace rather than the lived reality, and that whatever was never written down is invisible. Three years of Meridian’s internal announcements and exit-interview forms, coded for tone and themes, would show how the company talks about change — and how leavers talk back.
In practice the five approaches combine rather than compete: a case study typically contains interviews, observation, and documents, and ethnographers interview constantly. The matching table above is a starting point for design, not a cage.
Judging quality
Sceptics of qualitative research ask a fair question: with small samples, no statistics, and interpretation everywhere, is this not just anecdote? The honest answer is that bad qualitative research is exactly that, but good qualitative research is not — and there are criteria for telling them apart. The criteria from session 1 — validity (are we measuring what we intend to measure?) and reliability (would the same procedure give the same result again?) — were built with quantitative measurement in mind. Babbie’s verdict on field research is instructive: compared with surveys and experiments, its measurements typically have more validity but less reliability. Watching real behaviour in context gets closer to the truth of a particular setting than a fixed questionnaire item can, but another observer might see that setting differently. Qualitative researchers therefore use a parallel vocabulary that asks the same underlying questions in a form that fits the method.
| Question | Quantitative term | Qualitative term |
|---|---|---|
| Is the account accurate? | Validity | Credibility |
| Do the findings travel? | Generalisability | Transferability |
| Was the process consistent? | Reliability | Dependability |
Credibility is the qualitative counterpart of validity: are the findings believable as an account of this setting? It is strengthened by prolonged engagement (enough time in the field), triangulation (multiple sources, methods, or observers pointing the same way), and member checking (asking whether participants recognise the account as true to their experience).
Transferability is the counterpart of generalisability: could these findings illuminate other settings? The qualitative researcher does not claim statistical generalisation. Instead, thick description — rich detail about the case and its context — allows readers to judge how far the findings travel. It is worth noticing who does the work here: in a survey, the sampling design carries the burden of generalisation; in qualitative work, the description does.
Dependability is the counterpart of reliability: was the process consistent, transparent, and traceable? Since a qualitative study cannot be exactly repeated — the setting moves on, and the researcher is part of the instrument — the standard becomes an audit trail: documented decisions about who was selected and why; recorded and carefully transcribed data, with systematic notes that separate what was observed from what was inferred (record what you know happened and what you think happened, and keep the two apart); and a coding process another researcher could follow and check. The underlying discipline is the same as in session 1: explicit, transparent, repeatable-in-principle procedure is what separates research from impression.
The practical test to apply when reading or commissioning a qualitative study: how long did the researcher spend in the field? Do different sources point the same way? Did participants recognise the account? Are the analytic steps documented? If those answers are missing, scepticism is deserved.
Qualitative research in organisations
Consider first the problem that opened the course. Meridian’s annual engagement survey (a CAWI questionnaire sent to all 450 staff) reports store-employee engagement down from 3.8 to 3.2 out of 5. The score is genuinely useful: it is comparable across years and stores, and it triggered the alarm. But it cannot say what “engagement” means to a shift worker, which of a hundred possible causes is at work, or what to change. The qualitative follow-up is a programme of semi-structured interviews across stores and roles, sampled purposively — long-servers, recent joiners, leavers. This division of labour is the standard pattern in organisational research: quantitative monitoring detects; qualitative inquiry diagnoses. The board’s question changes shape accordingly — from “why is the score 3.2?” to “what is it like to work here, and what has changed?” — a question only qualitative methods can answer.
Second, understanding how teams actually work. Every organisation has two structures: the formal one in the org chart and the process manuals, and the informal one in daily practice. Diagnosis usually needs the informal one, and people cannot fully report it, because routine practice is largely invisible to those inside it. Observation and mini-ethnography answer questions such as: where does the shift handover actually lose information? Who do people really go to when something breaks? Which rules are followed, bent, or quietly ignored — and why? Good field notes record both empirical observation and interpretation, clearly separated.
Third, customer experience research, where the same logic applies outward. Meridian’s customer satisfaction scores are falling, and the scores alone cannot say why. The qualitative toolkit maps directly onto customer experience: interviews with lapsed customers about what switching away meant from their side; focus groups reacting to a store refit or loyalty scheme before rollout; observation of real shopping trips to see where customers hesitate, queue, or give up; and document analysis of complaints and online reviews, coded for recurring themes. Market researchers have used every one of these methods for decades — qualitative approaches are not academic luxuries but standard commercial practice.
Choosing qualitative approaches honestly means knowing their limits. They cannot provide statistical descriptions of large populations — no percentages, no margins of error, no “23% of staff intend to leave”. They cannot establish how widespread a discovered pattern is: ten interviews can find a mechanism, but not measure its prevalence. They are labour-intensive, and their conclusions depend on interpretation that must be documented and defended. The remedy is not to abandon them but to pair them: qualitative work to discover and understand, quantitative work to measure and generalise. The next session turns to the quantitative side of that partnership.
Conclusion
Qualitative inquiry rests on a coherent logic — naturalism, meaning, small purposive samples, and inductive discovery — in which depth is bought at the price of breadth. The family offers five main approaches: in-depth interviews, focus groups, observation and ethnography, case studies, and document analysis, distinguished by what they gather and therefore by the questions they suit. Quality is judged by credibility, transferability, and dependability — the same underlying demands as validity, generalisability, and reliability, translated for a different kind of evidence. For organisational diagnosis, qualitative methods answer the questions that dashboards raise: not whether the engagement score fell, but why — and what to do about it. Sessions 12 and 14 return to this family for technique: how to interview well, and how to analyse what you gather.