Methods of Research
Before you start: research = gathering evidence. To find out about family life of people you know you could ask them, watch them, read their diaries, or look at statistics — each way has strengths and limitations (that is the whole chapter!).
1. Types of Data
① Primary data
Definition: Primary data = information collected personally by the researcher (questionnaires, interviews, observation).
| Strengths | Limitations |
|---|---|
| Researcher has complete control — how data is collected, by whom, for what purpose Greater control over reliability, validity & representativeness |
Time-consuming (esp. interviewing many people) & expensive Difficult access to target group; people may refuse; in historical research respondents may be dead |
② Secondary data
Definition: Secondary data = data that already exists — documents (government reports & statistics, letters, diaries) or previous research by other sociologists.
| Strengths | Limitations |
|---|---|
| Saves time, money, effort (e.g. crime / marriage / divorce stats) Sometimes the only source (e.g. suicide) Good for historical & comparative work Some forms (official statistics) very reliable — collected the same way from same sources |
Not made with sociologists' needs in mind — official definitions of poverty, class, ethnicity may differ from sociological ones Personal documents can be unreliable May reflect one person's view only (not wider opinion) |
③ Quantitative data
Definition: Quantitative data expresses information numerically in 3 ways:
| Form | Meaning | Example |
|---|---|---|
| 1. Raw number | a total count | total people living in a society |
| 2. Percentage | number per 100 | ≈ 80% of Indians follow the Hindu religion |
| 3. Rate | number per 1,000 in a year | birth rate of 1 ⇒ 1 baby per 1,000 people each year |
| Strengths | Limitations |
|---|---|
| Shows relationships statistically — good when you only need to compare numbers, not reasons Summarises sources & allows comparisons; correlations test whether a hypothesis is true/false; tracks change over time (longitudinal) More reliable — easy to replicate (standardised, unchanging questions) Easier to stay objective — no close personal involvement, so bias is less likely to intrude |
Often needs an 'artificial social setting' to control responses (answering a stranger's list of questions, or being watched in a lab) Only captures who, what, when, where — a narrow range Doesn't reveal reasons: lacks depth (more detail = harder to quantify) ⇒ superficial / surface-level Some argue 'normal' behaviour or 'real' answers can't be captured in artificial settings |
④ Qualitative data
Definition: Qualitative data captures the quality of behaviour: the 'why' rather than the 'what, when, where'. Explores feelings, experiences, meanings applied to behaviour.
Venkatesh (2009) studied a young gang from its members' viewpoint · Goffman (1961) studied patients' experiences in a mental institution.
| Strengths | Limitations |
|---|---|
| People can talk and act freely ⇒ complex reasons captured Methods like participant observation build rapport & let researcher experience lives Study people in 'normal' settings ⇒ shows how people really behave & what they really believe (high validity) |
Intensive study of small groups ⇒ may be unrepresentative (can't generalise) Hard to compare across time/place — no two groups are qualitatively the same Depth & detail ⇒ difficult to replicate ⇒ lower reliability |
- Quantitative = numbers · reliable · objective · replicable ✗ superficial · artificial · no 'why'
- Qualitative = meanings · valid · depth · natural ✗ small/unrepresentative · not replicable · lower reliability
2. Secondary Sources
① Official statistics
Created & published by governments — a major source of secondary quantitative data to examine trends and patterns within and between societies.
- Patterns: broad overview of behaviour across local, national & international areas. Durkheim (1897) found distinct patterns of suicide by comparative analysis of official suicide statistics from different societies.
- Trends: data from different years shows how something has changed.
- Comparisons within groups (e.g. middle- vs working-class family size) and between societies.
| Strengths | Limitations |
|---|---|
| May be the only source (e.g. suicide) Costly / slow data (marriage, divorce, crime) readily available — esp. since the internet Representativeness: often large, carefully chosen samples Many recorded by law (crime, unemployment, marriages, births, divorces) |
No great depth / detail Validity problems — depends on what governments include or exclude; only a partial picture: ① info not available — e.g. 'dark figure of crime' (crimes not reported / recorded) ② don't reveal reasons — exact murder numbers say nothing about why people kill |
Interpretation: quantitative data seems more objective, but researchers must interpret what it means. A statistical rise in crime could be:
- a real rise
- a different way of defining & counting crime
- police targeting certain crimes & arresting more people
Governments change definitions; different governments define concepts differently ⇒ reliability questioned — must compare like with like.
② Personal documents, digital content & media sources
Personal documents: letters, diaries, oral histories, autobiographies, photographs. Media: newspapers, books, moving images. Online: blogs & websites. (Strengths/limitations similar, so treated together.)
| Strengths | Limitations |
|---|---|
| Access to data that would cost lots of money, time, effort Secondary data where primary is impossible (the past) Historical / comparative use — track social change; things we take for granted may once have been seen differently Qualitative depth & detail: Samuel Pepys (England 1660s), Anne Frank (hiding from Nazis, Amsterdam) Compare past & present accounts to test validity of current accounts — Pearson (1983) used 100+ yrs of media accounts to show 'hooligan/yobbish' behaviour is not new in the UK Can compare literal meaning vs hidden meaning (hopes, fears, beliefs of the author) — newspaper articles may tell us more about the writer than the topic |
Practical: not always easy to find; where from? Paper docs can be faked/changed — original or copy? Same for digital, photo, video Don't always know why or by whom created ⇒ believable? Did author have first-hand experience or just repeat what they heard? Reliability problems — may be incomplete, inaccurate, unrepresentative (a diary = one individual's view) Digital sources change: old websites vanish, others updated so original lost; old tech (floppy disks) hard to access |
3. Quantitative Methods
① Questionnaires
Written questions in 2 forms:
- Postal questionnaires — completed in private, researcher absent (today: web-based / emailed).
- Researcher-administered — completed with researcher present, answered verbally ⇒ these are structured interviews.
| Question type | What it is | Example |
|---|---|---|
| Closed-ended / pre-coded | researcher gives set answers to choose from; respondent can't go beyond the list | Do you own a sociology textbook? Yes [1] / No [2] / Don't know [3] |
| Open-ended | respondent answers in own words ⇒ limited qualitative data | "What do you like about studying sociology?" |
Many questionnaires mix both, but the main aim is usually to quantify responses.
| Strengths | Limitations |
|---|---|
| Pre-coded answers are easy to quantify, count & code (a simple count) Contact large numbers quickly & efficiently — respondents do the time-consuming work Highly reliable — everyone answers the same questions ⇒ easy to replicate Anonymity improves validity (esp. personal questions); less risk of biased answers / respondent guessing what researcher wants |
Low response rate ⇒ carefully designed sample becomes unrepresentative (it selects itself) Can't do anything if respondents ignore questions or answer wrongly (e.g. tick two boxes) Hard to examine complex issues & opinions Researcher must decide at the start what is significant — no chance to change later Can't know if question was understood or means the same to all; someone else may complete it (anonymity ⇒ validity & representativeness hit). Pilot studies reduce — not remove — these problems |
② Structured interviews
Researcher asks questions in person; same questions, same order every time — no flexibility ⇒ consistent, comparable results.
| Strengths | Limitations |
|---|---|
| Researcher can fix reliability problems (misunderstanding / not answering) No unrepresentative sample problem — response rate = 100% |
Assumptions (pre-judgements) about behaviour; can contain unintentionally biased questions No anonymity ⇒ two related problems ↓ |
② Researcher effect — the relationship between researcher & respondent biases responses: aggressive interviewers may intimidate people into answers they don't believe; status (gender, age, class, ethnicity) may bias data — e.g. a woman embarrassed to answer personal questions from a man (and vice versa).
③ Experiments
Test the relationship between variables (things that can change under controlled conditions). Researcher changes (manipulates) the independent variable (IV) and measures any change in the dependent variable (DV) — not changed by the researcher; any change must be caused by the IV.
| Relationship | Meaning |
|---|---|
| Correlation | 2+ things happen at roughly the same time — only suggests a relationship (could be chance) |
| Causation | when one action occurs, another always follows ⇒ powerful: allows prediction |
Hard to tell correlation from causation in real research (coincidence). Two ways to separate them:
- Test and retest — the more times a test is replicated with the same result, the more likely the relationship is causal.
- Use groups with exactly the same characteristics: an experimental group (behaviour manipulated) + a control group (not manipulated).
| Type | Description |
|---|---|
| Laboratory | Closed environment, conditions precisely monitored/controlled ⇒ no outside variables. Unusual in sociology — artificial; findings unlikely to apply to the 'real' world; participants know they're studied (behave differently); consent issues. |
| Field | Not in a controlled setting ⇒ more appropriate for sociology. Very hard to control all IVs in a natural setting ⇒ tends to establish correlation, not causation. Same principles: IV/DV test a hypothesis or answer a research question. |
| Strengths | Limitations |
|---|---|
| Lab experiments easier to replicate than field (more control of conditions & variables); standardised ⇒ high reliability Powerful, highly valid statements about cause-and-effect Field experiments manipulate real-world situations to find underlying reasons for everyday behaviour |
Hard to control all influences, even in a lab — just being studied adds an uncontrolled variable Hawthorne (observer) effect — Mayo (1933), Hawthorne factory, Chicago: lighting & temperature changed in various ways, yet work rate always rose — because workers knew they were being studied |
④ Content analysis
Has quantitative & qualitative forms; both study texts (TV, written documents…). Quantitative: statistical techniques to categorise & count frequency of behaviour using a content analysis grid (Table 2.1).
| Character | Gender | Age | Place & purpose | On screen |
|---|---|---|---|---|
| Azir Khan | Male | 25 | Office (employee) | 30 seconds |
| Safiq Dhonna | Male | 56 | Office (customer) | 43 seconds |
| Angelique Basson | Female | 37 | Shop (customer) | 84 seconds |
Table 2.1: simple grid recording behaviour of characters in a TV programme.
- Meehan (1983): identified & analysed stereotypical roles of women in US daytime soap operas.
- Page (2005): 'concept mapping' — tracked how media professionals portrayed global warming as 'natural' vs 'social' causes.
| Strengths | Limitations |
|---|---|
| Finds hidden themes & patterns (e.g. how far women in TV/magazine ads are linked with housework & cleaning) Quantifying lets researchers draw complex conclusions from simple data-collection Standardised grid ⇒ data can be checked & replicated |
Reliability may be limited — subjective judgements about what to count, which categories to use, which behaviour fits which category Can all behaviour be neatly categorised? Doesn't tell us how or why audiences receive, understand, accept or ignore the themes found |
4. Qualitative Methods
① Unstructured interviews
Researcher gains an understanding of how people think & feel; respondents talk freely about what they feel is important (like a conversation).
| Strengths | Limitations |
|---|---|
| Researcher's limited input ⇒ data reflects the respondent's interests; accurate & detailed expression of beliefs; no researcher pre-judging what's relevant Strong understanding / rapport ⇒ distrustful people open up; sensitive issues explored in depth Informal — can be somewhere respondent is at ease (e.g. own home) |
Needs considerable skill — must resist temptation to influence, encourage or interrupt Little control ⇒ may drift into irrelevant areas Time-consuming, as is analysing & transcribing recordings Low reliability — non-standardised ⇒ impossible to replicate Interview effects — respondents try to please; influenced by what they know of the interviewer |
② Semi-structured interviews
Combines strengths of unstructured & structured. Researcher is free to ask questions in any order & any wording. Has an interview schedule (areas to focus on) but no fixed list of questions — different respondents may get different questions on the same topic. Aim: understand from the respondent's viewpoint, not generalise. Uses open-ended questions (some prepared, some arising naturally).
| Strengths | Limitations |
|---|---|
| Less risk of researcher predetermining what's discussed Respondent can raise things interviewer never thought of — new knowledge informs later interviews & further questions Discovers what someone really means / believes ⇒ greater depth ⇒ higher validity (research achieves its real aims) |
Needs skill: right questions, rapport, thinking quickly More time-consuming than questionnaires; huge data to analyse — much may be useless Imperfect recall: past events hard to remember & memory is selective; can't know if respondent tells the truth Lacks standardisation ⇒ hard to analyse & generalise |
③ Group interviews (focus groups)
Respondents gather to discuss a topic decided in advance by the researcher. Group may be a representative sample (cross-section of society) or just a group the researcher wants to explore. Focus groups often same-sex & similar backgrounds (stops gender/class variables affecting reliability/validity).
Success depends on: clear guidelines to avoid arguments · predetermined questions · interaction in the group (Gibbs 1997: gives 'unique insights into shared understandings of everyday life') · skill of researcher (stop 1–2 people dominating).
| Strengths | Limitations |
|---|---|
| Researcher can control pace & scope; plan a schedule to focus / refocus; ask questions, stop or change focus Reflects how people naturally share & discuss ideas — more realistic: we form views by talking to and listening to others, not alone |
Must control group behaviour while keeping focus (skill) Representativeness: if 1 of a carefully chosen 10 doesn't show up, sample becomes unrepresentative 'Groupthink' — pressure to reach desired outcomes (say what researcher/group wants to hear) ⇒ group consensus, not individual beliefs |
5. Observation
Based on the idea that data is more valid if gathered by seeing how people behave, rather than trusting what people say they do. Two techniques: non-participant & participant.
① Participant observation
Researcher takes part in the behaviour studied. Based partly on Weber's (1922) verstehen — 'to understand by experiencing' (Mead 1934: take the part of the other & see through their eyes = empathy). Two forms: overt & covert.
Overt: group knows it is studied; researcher joins openly, usually with permission & cooperation of the group (or key members).
Overt PO — strengths
| Recording data is easy — group knows researcher's role; can ask questions, take notes openly |
| Hierarchical groups (e.g. big business): access to all levels (boardroom and shop floor) |
| Sponsorship — trust & cooperation of an important member gets you in; makes it easier to separate participant & observer roles |
| Reduces risk of 'going native'; clear role ⇒ less risk of unethical / criminal / dangerous behaviour — can pull back from risky situations |
| Highly detailed picture; sees what people actually do vs what they say ⇒ ↑ validity |
Overt PO — limitations
| If group refuses permission ⇒ research can't happen |
| Huge time, effort & money (Venkatesh ≈ 8 years on one gang) |
| Observer / Hawthorne effect — awareness of researcher may make people behave unnaturally |
| Level of involvement: too superficial without full participation · ethics (can't join illegal acts) limits true experience · risk of 'becoming the story' (special treatment) |
| Impossible to replicate; must trust researcher; can't record everything ⇒ selection, interpretation, retelling |
Problems: sometimes acted 'like one of them' (lost objectivity) · given special treatment · witnessed a fake punishment beating — higher-level members 'putting on a show' (observer effect).
Question: how might it differ if he'd been (a) white American or (b) African-American?
Covert: researcher secretly joins the group — subjects unaware. Aim: see behaviour in its natural setting. Must balance researcher & participant roles without revealing the true role.
Covert PO — strengths
| May be the only way to study groups who wouldn't be researched: • Criminal / deviant: Ward (2008) — member of the rave / dance-drugs culture; 5 yrs in London clubs, parties, pubs, houses; her knowledge & friendships gave easy access • Closed groups: Lofland & Stark (1965) — religious sect • Defensive groups: Ray (1987) — Australian environmental groups, suspicious of open research |
| Avoids observer effect — behaviour largely unaffected by researcher's presence |
| Personal experience ⇒ insights into meanings, motivations, relationships ⇒ can explain why; sociological insight makes sense of behaviour even where members don't fully understand it themselves |
Covert PO — limitations (Goffman: 3 problems)
| ① Getting in — invitation-only groups; entry requirements (need qualifications to study doctors / accountants); researcher's characteristics must match (a man can't join a group of nuns). Way round: Goffman took a manual job in the institution — neither a doctor nor mentally ill. |
| ② Staying in — no access to all areas (a 'pupil' can't enter the staff room); risk of discovery by gatekeepers; must learn the culture fast, keep a convincing back-story, think on feet; lack of 'insider knowledge' ⇒ found out. Roles clash: must choose participate vs observe (criminal acts) — or go native ⇒ validity doubts. |
| ③ Getting out — hard to stop (can't just leave a gang); ethics of leaving people who trust you; right to pretend / use members for own purposes? |
| Also: can't replicate · must trust the researcher · can't take notes / ask many questions / record openly |
Non-participant observation: watching from a distance, subjects don't know; researcher not personally involved (e.g. shopping mall, sports crowd). Consent often not a problem (can't ask everyone). Unlike PO, can produce quantitative data (count how often an act occurs).
| Strengths | Limitations |
|---|---|
| Access — people who don't want to be studied (illegal / secret / embarrassing) Studied objectively in a natural setting — 'everyday behaviour' |
Not easily replicated — group characteristics/structure change over time Misses depth, richness & intimate detail Ethics — observed without permission |
| Type | Overt / covert | Participation | Characteristics |
|---|---|---|---|
| Non-participant | Overt/covert | None/minimal | Complete observer |
| Covert participant | Covert | Full | Complete participant |
| Overt participant | Overt | Full | Participant-as-observer |
Table 2.2: Types of observation (Brewer, 2000)
6. Stages of Research Design
- Planning — decide strategy (what to research, how) & formulate hypotheses / questions.
- Information gathering — identify a sample, run a pilot study, apply methods to collect data.
- Information processing — analyse & interpret the data's meaning.
- Evaluation — internal (was the method appropriate?) + external (report conclusions to a wider public for analysis & criticism).
① The research problem
Decide the general topic, then the specific aspect. Influenced by: personal interests · current interest (sociologists/society) · funding · practical factors (can respondents be contacted?). Then review previous research — gives ideas, whether to replicate, how to avoid earlier errors.
② Hypothesis or research question
| Hypothesis | Research question | |
|---|---|---|
| What | Statement to be tested ⇒ needs suitable methods | Method must generate high levels of descriptive data |
| Linked to | Quantitative | Qualitative |
| Example | Ginn & Arber (2002): "The effect of motherhood on full-time employment is minimal for graduate women" | Conway (1997): "Does parental choice help to strengthen the advantage of the middle classes over the working class?" |
③ Collecting data — sampling
Ideal = study everyone in the target population — usually impossible (e.g. 'doctors in India'). So choose a sample (small number from the target population), e.g. 1000 doctors, and draw conclusions about all doctors — but only if the sample is representative.
Representativeness (may matter more than size) = sample's characteristics accurately reflect the population's. If 60% of doctors are male ⇒ 60% of the sample should be. Then findings can be generalised.
Access may be denied for: legality (names can't be revealed by law) · confidentiality (business refuses payment records) · privacy (some groups don't want to be studied).
Sampling techniques
| Technique | How it works | Notes |
|---|---|---|
| Random (simple) | random pick from sampling frame; everyone has an equal chance — like a lottery | based on probability ⇒ representative |
| Systematic | direct from frame, e.g. 25% of 100 names ⇒ every 4th name | for very large populations; not truly random (5th name never chosen) but random enough |
| Stratified random | divide population into groups (strata) with known characteristics (e.g. age); each group = separate random sample | avoids chance over/under-representation of small groups |
| Stratified quota | know characteristics only; opportunity basis — e.g. need 20 men; ask until 20 agree, then quota full | sampling frame not strictly necessary |
| Non-representative | researcher doesn't want to generalise; interested in the group itself | Venkatesh's gang: only representative of itself — didn't matter |
| Opportunity | when a representative sample isn't possible. Best opportunity or snowball | useful when no frame & little known about population |
2. Snowball — find one willing member of the target population → they suggest more → … until sample is usable. Unrepresentative but may be the only option.
④ Pilot studies
A trial run before the full study — a research tool used for one of two reasons:
- Mini version of the full study — test feasibility (spot problems like access), work out resources (staff, finance), and show funding bodies the full study is feasible & worthwhile.
- Pre-test a method (e.g. questionnaire) — test question types, check the data gives what's needed, remove bias / unreliability (leading or ambiguous questions).
⑤ Operationalisation
Sociological concepts may be unclear to non-sociologists — researchers must decide how to put them into practice = operationalise. e.g. 'social class': people won't agree what 'middle-' / 'working-class' means or how to measure it; asking 'which class are you?' gives answers from different understandings. So ask occupation or income and let the researcher judge the class.
⑥ Interpreting results
Findings don't 'speak for themselves' — analyse for common themes & trends; does data support the hypothesis? Present: findings · conclusions about hypothesis/question · limitations (sample, response rate, validity…) · suggestions for further research · improvements to the design.
7. Approaches to Research
① Case studies
Study a particular group / 'case' — e.g. Westwood (1984): 12-month participant observation of women workers in the 'Stitchco' factory. Not really a method but a technique in which different methods generate data; usually qualitative.
| Strengths | Limitations |
|---|---|
| Single group studied over time ⇒ great depth & detail, greater validity than simple quantitative studies Small self-contained cases (factory — Westwood; school — Lacey 1970; gang — Venkatesh) ⇒ lots of data cost-effectively Uncovers meanings of everyday behaviour (often uses PO but other methods too) Small case studies can be pilot studies — develop hypotheses, test methods, spot problems |
Large in-depth studies take time, effort, money High demands on researchers' skills (months/years living with subjects) & on respondents (extensive questioning/observation) Hard to generalise — small groups representative only of themselves; but comparisons between similar groups possible |
② Social surveys
Cross-sectional survey = a 'snapshot' of behaviour at one time.
- Qualitative form: generally descriptive — illustrate a type of behaviour (e.g. suicide, income, poverty in one country / area / different countries).
- Quantitative form (more common): analytic — analyse correlations & causations. Durkheim's suicide study used cross-sectional surveys from different societies for comparative analysis ⇒ theory of types of suicide.
Both normally need representative samples (goal = generalise). They compare standardised groups (income, education, gender) on a single variable (e.g. suicide rate): do highly educated people have higher suicide rates than less educated?
Task: find out whether your country has a census, the latest date and what it asked.
③ Ethnography
(Also fieldwork.) Tries to achieve a detailed, in-depth understanding of a group / social situation. Began in anthropology — researchers took part in daily lives of (often traditional, pre-modern) societies to grasp their worldview. In sociology strongly linked to participant observation (overt or covert) + other methods: unstructured interviews, qualitative documents, even some quantitative data.
④ Longitudinal studies
Comparative analysis tracking changes in a representative sample over time (months → many years). Same group analysed at stages of life using methods from questionnaires to non-participant observation. Carried out in 'waves' at intervals; researcher stays removed, limited contact.
| Strengths | Limitations |
|---|---|
| Track personal & social change over long periods — reveals hidden trends Large representative samples ⇒ suggest correlations & causal relationships |
Sample attrition — people drop out (lose interest, move without contact details, die) ⇒ representativeness falls, worse the longer the study Only ever a 'quick look' at a given moment ⇒ criticised for lacking depth & validity |
⑤ Mixed methods — pluralism & triangulation
Methodological pluralism = using more than one method (can mix quantitative & qualitative) because every method has different strengths/weaknesses — questionnaires: reliable but low validity; covert PO: the reverse. Instead of a 'design problem' (test hypothesis = positivism / answer question = interpretivism), think: how to get the highest reliability & validity whatever the method.
Triangulation = putting pluralism into practice — ways to improve reliability & validity.
| Type | What it is |
|---|---|
| Methodological | 2+ methods — Denzin (1970): offset one method's weakness with another's strength. e.g. questionnaires assume respondents tell the truth ⇒ use participant observation to check they do what they say. Or compare two methods on the same people (semi-structured interview + focus group) — same conclusions ⇒ confirms reliability & validity. Combos: 2+ researchers/same technique · 1 researcher/2+ techniques · 2+ researchers/2+ techniques. |
| Researcher | for studies relying on interpretation. Same method ⇒ same results ⇒ confirms reliability. Researchers of different ethnicity, age, gender, class ⇒ check observer / interviewer bias. |
| Data | different sampling strategies — different times, contexts, people; both the individuals' accounts and the researcher's own experience. Venkatesh could make sense of drug dealing & 'being black and poor' because he was intimately involved and gathered data from others. |
- Barker (1984) — Unification Church ('Moonies'): overt PO + questionnaires + semi-structured interviews.
- Hey (1997) — girls' friendships in 2 London schools: PO + personal documents (diaries, passed notes).
8. Positivism vs Interpretivism
① Positivism
It is possible & desirable to study social behaviour with similar methods to the natural world — sociology can be like a science. Social systems = structures existing independently of individuals. Institutions = macro level; we feel structures as forces pushing our choices (choice is limited). So study the causes (structural forces), not the effects (choices). Structures = real, objective forces — like gravity / electromagnetism, unseen but observable by: systematic observation · accurate testing · quantitative measurement ⇒ reliable knowledge ⇒ theories that explain & predict.
Scientists must be personally objective — don't participate/influence; prefer quantitative methods; value reliability (replication). Questionnaires, structured interviews, experiments, comparative & observational studies = higher reliability & let researcher 'stand apart'. Research should not depend on subjective interpretation ⇒ value-free approach.
② Discover the 'general rules' (structures) deciding individual behaviour
③ Researcher personally objective (no values/beliefs/prejudices) & systemically objective (objective methods)
④ Able to quantify & measure behaviour
② Interpretivism
Difference between society and nature: social reality is formed by interaction of conscious people. We can act (free will), not just react to structural stimulation ⇒ people are unpredictable ⇒ positivism not appropriate. Unpredictability is built through meanings: society isn't objective 'out there' but created daily by interaction and experienced subjectively — something 'in here' to be experienced and understood.
So causal relationships are impossible in theory or practice; the researcher can only describe reality from the viewpoint of those who define it (classroom, family…) = empathy / verstehen (Weber). Aim: help respondents 'tell their story' ⇒ understand & explain their choices.
② Rules are context-bound — change subtly with the situation
③ Needs close study & understanding of context ⇒ often participant observation
④ Participation gives deeper insight; objective detachment rejected — sociologists should acknowledge their values, not claim to be value-free
⑤ Reliability matters but validity matters more
| Positivism | Interpretivism | |
|---|---|---|
| Society | real structures 'out there' | created by interaction & meaning |
| People | shaped by structure (limited choice) | free will; unpredictable |
| Aim | explain causes, find laws, predict | understand meanings (verstehen) |
| Methods | quantitative: questionnaires, structured interviews, experiments, stats | qualitative: participant observation, unstructured / semi-structured interviews |
| Prefers | reliability | validity |
| Values | objective, value-free | acknowledge values |
9. Is Sociology a Science?
Science = a way of producing knowledge that is factual & objective (not opinion, guesses or faith): identify a problem → collect information → offer an explanation. It is a methodology with 2 qualities:
- Reliable — can check accuracy by repeating (replicating) and getting same/similar results.
- Valid — data measure/describe what they claim to. e.g. govt crime stats measure crime, but validity is limited if they only record crimes reported to police.
② Communal — 'public knowledge', freely shared; others build on it ⇒ cumulative; peer review — can't be taken 'on trust', others must be free to replicate
③ Disinterested — aim is knowledge; no personal (financial etc.) interest in the outcome (else researcher bias ⇒ validity doubts)
④ Sceptical — nothing beyond criticism; continual evaluation. This is how science differs from faith: science is 'true' only because it's not yet disproved; faith is considered self-evidently true & can't be disproved
| Positivist view (early sociology) | Interpretivist view |
|---|---|
| Social world explored like the natural world; laws of behaviour to discover (like gravity, thermodynamics); same principles & methods as chemistry, biology, physics; objective, values kept out. Took hold in the Enlightenment — belief that science would give complete understanding & build better societies. Later positivists: people think & decide ⇒ methods not always appropriate; be as objective as possible, accept complete objectivity is unachievable (choice of topic reflects what's thought worthwhile). |
Scientific methods not appropriate. Purpose = understand why people behave as they do — see through their eyes, uncover meanings. Different methods, assessed differently. Positivists say these lack reliability (can't replicate) & objectivity. Interpretivists: validity > reliability; qualitative methods capture what actions mean. Since full objectivity is impossible, be open about values so readers can judge their influence. |
10. Research Issues
Three kinds of consideration shape choice of topic, choice of method(s) & conduct of research: theoretical · practical · ethical.
① Theoretical considerations
Ackroyd & Hughes (1992): methods aren't simply 'tools' (like a hammer for a picture) that are right/wrong for a task — they have no single clear purpose and research is surrounded by theoretical beliefs about the nature of the social world & how to study it. Must decide what counts as data: statistical or descriptive? test a hypothesis or report what respondents say?
| Topic choice | Choice of method |
|---|---|
| • Audience may influence/dictate (Jessop 2003 'Governance and Meta-governance…' fine for academics, meaningless to the public) • Purpose: testing a hypothesis ⇒ narrower topic than a descriptive account • What's 'worthy of study' depends on values: Personal — researcher's interest (crime vs education) Institutional — universities & governments fund research (UK: ESRC, mainly govt-funded). Government may want to know more to make policy — or block research whose findings would criticise its policies |
• Perspective: Interactionists avoid statistical methods (not seeking causality) · Positivists take the opposite view (not interested in descriptive accounts) • Beliefs about reliability & validity of methods; value judgements about how something should be studied — if you think covert participation is unethical and invalid, you won't choose it |
② Practical considerations
| Topic choice | Choice of method — 'fitness for purpose' |
|---|---|
| • Large / long-term research is expensive; those who commission & pay may influence topic, method & conduct • UK & USA: govt agencies fund research to help policy-makers; research not supporting this may not win funding • Key issues: access to subjects & their co-operation — may be denied. Options: covert research (Goffman posed as cleaning staff — ethically questioned) or sponsorship (member 'protects' researcher) • Explains why much research studies the powerless (can't say no), not the powerful (can resist) |
• Dunican (2005): fitness for purpose = how well method suits the context & answers the research question. Venkatesh began with a questionnaire, asked one question, subjects said he must experience it • Topic suits method: quantitative for statistical relationships (Kessler 2000: do sponsored people do better in small business?) • Time: PO may take years (Venkatesh ≈ 8 yrs) • Funding: questionnaires cheapest < in-depth interviews < participant observation; also affects team size • Group size/structure: questionnaires ⇒ large, dispersed groups · PO ⇒ small, localised groups |
No access / time / funding ⇒ other considerations are irrelevant. After that come ethical issues. Activity 2.12: practical issues of studying (a) suicide (b) lifestyles of the rich & famous (c) police culture (d) family life of teachers — which is hardest? (powerful groups → access; suicide → no direct subjects, sensitive).
③ Ethical considerations
Ethics = the morality of doing something — applies to the type of research (is it ethical to study people without their knowledge?) and researcher behaviour (deceiving people about purpose?). Guides how people are persuaded to participate and how they're protected physically & psychologically during/after. Legal & safety issues affect topic, method & conduct.
| Legal | Safety |
|---|---|
| Esp. when observing / joining illegal behaviour — e.g. Ditton (1977) workplace theft. Is it ethical to research crime at all? May avoid immersive methods (PO) to avoid dilemma. Responsibilities to criminals & victims. Participants told of consequences (e.g. negative media publicity). Gain informed consent. Relationships based on trust & honesty — if you promise anonymity, revealing identities to authorities/media is unethical. | Physical & psychological safety of everyone. Deep-involvement methods (covert PO): don't cause upset/distress to vulnerable people at the end — e.g. regularly visiting elderly people then simply breaking contact is unethical (they may see it as friendship). Unprepared researchers should choose a different method. |
Ethical practice — sociologists follow a code. Ask: ① Is it true? (no fabricating data / falsifying results) · ② Is it fair? — ownership (who's author?), plagiarism (passing off others' work) · ③ Is it wise? (can it be morally justified; would another topic/method be better?).
Informed consent: participants usually asked to consent, fully informed of purpose & nature of the research. Morally ambiguous areas:
| Issue | Study |
|---|---|
| Studying people who don't want to be studied | Wallis (1977) — Scientology leaders refused access to current members ⇒ he used former members' opinions & experiences |
| Tricking people into co-operating | Rosenhan (1973) — suspected doctors couldn't accurately diagnose schizophrenia; sent students faking symptoms into hospitals |
| Experimenting on people unaware / causing distress | Milgram (1974) — respondents believed they were giving electric shocks to 'learners' for wrong answers; no shocks given, 'victims' acted — but some respondents broke down |
Activity 2.13: identify & assess ethical considerations in studying suicide / the rich & famous / police culture / family life of teachers. Are any areas off-limits?
11. Bias, Values & Funding
Sociological knowledge differs from journalism, personal experience & everyday talk because it deals in facts: data collected & analysed objectively = value-free (more correctly value-neutral — no one can truly 'act without values'). Best we can do: recognise where values intrude and adjust strategy to limit or cancel their effect.
② Funding — payers influence what and how ⇒ ethical Q: is the researcher responsible for how findings are used?
③ Choice of method — beliefs about reliability & validity; quantitative vs qualitative
④ Questions asked — who to question, what, how respondents may answer: positivists → closed (easy to quantify); interpretivists → open-ended own words
⑤ Data analysis — what data to include / exclude
Getting funded: approval from a university/institution or a funder (UK: ESRC). Must convince them it's worthwhile & fits their current priorities; projects with a clear practical outcome more likely funded. Where govts control funders, research that is embarrassing / inconvenient or suggests policies are wrong is unlikely to be funded.
12. Assessing Methods: Key Concepts
No general agreement on how to collect data about the social world — different methodological approaches develop their own ideas of scientific sociology.
Ecological validity = how far methods reflect the world being studied (closer to natural environment ⇒ more valid). Lab experiments = very low (artificial); covert PO = high (real life).
Qualitative (PO, unstructured interviews) usually higher validity — closer to experience & perceptions. In surveys, respondents may give socially approved answers ⇒ invalid (though reliable — same answers to another researcher).
Positivists: social structures are solid, permanent, real, independent of observer ⇒ discovery; researcher stands 'outside' the behaviour as a removed observer.
Interpretivists: many realities, depending on how people see the social world — it exists only as interpretations ⇒ aim = subjective understanding; researcher = an objective channel for people to 'tell their story'. Harder than in natural science ⇒ be open about values so others judge.
e.g. postal questionnaires: low response ⇒ ① higher chance of sampling error from an unrepresentative self-selected sample · ② survey results lack validity — can't be generalised to the target population.
13. Exam-style Practice
Describe two… [4] — name each technique/example, then briefly describe it. Short; no long development needed.
Explain two reasons… [8] — choose two reasons, explain each using sociological terms (closed, pre-coded, valid…). Repeating the question's wording in your opening sentence helps you answer directly.
Explain one strength and one limitation/weakness… [6] — one of each, explained.
Evaluate the view that… [26] — argue both sides and reach a judgement.
Model (technique 1): One sampling technique is random sampling. This is when everyone in the target population has an equal chance of being selected, similar to a lottery.
Point 1: identifies a technique by name, then briefly describes it.
Activity 2.15 — technique 2 (my answer): Another technique is stratified random sampling. The target population is divided into groups (strata) with known characteristics, such as age or gender, and a random sample is taken from each group in proportion to its size, so no group is over- or under-represented.
Model (reason 1): Difficult in questionnaires (the most common survey) to obtain detailed information. Questions are often closed & pre-coded — useful for collating & analysing, but respondents must choose from a list and can't give the answer they'd like, nor explain it, so the researcher gains no valid insight.
Point 2: clear reason + explained with sociological terms (closed, pre-coded, collating, valid).
Activity 2.16 — reason 2 (my answer): Respondents may not give truthful answers. Because of the interview effect / wanting to make a good impression, they may give socially approved answers rather than what they really think, so the data doesn't measure what it claims to ⇒ invalid (even if reliable, since the same answers would be repeated). Alternative: unintentionally biased questions — leading, ambiguous or vague options.
| Set | Q1 Describe [4] | Q2a Explain two reasons [8] | Q2b Strength + limitation [6] | Q3 Evaluate [26] |
|---|---|---|---|---|
| 1 | two sampling techniques | why social surveys may not give valid data | unstructured interviews | value-free sociology possible & desirable |
| 2 | two types of personal documents | why sociologists use different sampling techniques | social surveys | sociologists should discover laws of human behaviour |
| 3 | two ways practical considerations influence method | why covert PO is not used often | official statistics | sociology should use methods of natural sciences |
| 4 | two ethical considerations | why lab experiments are not used often | personal documents | practical considerations most important in topic & method |
2.1 best data type? Neither alone — quantitative (reliable, comparable) + qualitative (valid, depth) ⇒ mix. 2.2 personal docs (diary, photos, messages): rich qualitative depth but one person's view; ethics of use. 2.3 questionnaire on how students study: mix closed (hours per week? Yes/No) & open (what helps you most?), avoid leading/vague wording, then pilot it. 2.6 overt vs covert PO — differences: subjects know / don't know; observer effect likely / avoided; consent given / not; easy / hard note-taking. School/college study: overt (access, ethics) vs covert (less observer effect) — justify. 2.7 watching better: behaviour people don't notice or won't admit (queuing, littering); asking better: motives, feelings, attitudes. 2.8 concepts to operationalise: social class (occupation, income), poverty, educational achievement (exam grades), deviance. 2.9 poster: draw the four stages (Planning → Gathering → Processing → Evaluation) with the sub-steps. 2.12 / 2.13 see §10: access, funding, consent, harm, anonymity.
14. Full Chapter Cheat-Sheet
| Method | Data | Best for / strengths | Main weaknesses |
|---|---|---|---|
| Postal questionnaire | Quant (some qual) | large samples, reliable, anonymous, cheap | low response, self-selected sample, no depth, can't check understanding |
| Structured interview | Quant | 100% response, can clear up confusion, reliable | interview & researcher effects, no anonymity, pre-judgements |
| Lab experiment | Quant | control, replicable, cause-effect | artificial, Hawthorne effect, consent |
| Field experiment | Quant | real setting, reasons for everyday behaviour | can't control variables ⇒ correlation only |
| Content analysis | Quant / qual | hidden patterns, replicable grid | subjective categories, says nothing on audience |
| Official statistics | Secondary quant | cheap, large, representative, trends | validity (what is counted), no 'why', dark figure |
| Personal docs / media | Secondary qual | past, depth, hidden meanings | faked, incomplete, unrepresentative, authorship unknown |
| Unstructured interview | Qual | validity, rapport, sensitive topics | skill, time, low reliability, interview effect |
| Semi-structured interview | Qual | depth + some structure, new ideas | time, lots of useless data, imperfect recall, not standardised |
| Group interview | Qual | natural interaction, researcher control | groupthink, dominance, absentees |
| Overt PO | Qual | detail, easy recording, validity | access, time/money, observer effect, going native |
| Covert PO | Qual | closed/deviant groups, no observer effect | getting in/staying in/getting out, ethics, can't record |
| Non-participant obs. | Quant / qual | access, natural setting, can count | shallow, not replicable, ethics |
| Case study | Qual | depth, validity | can't generalise, costly |
| Longitudinal | Quant (+qual) | tracks change, causation hints | attrition, 'snapshots' lack depth |
- Data: primary / secondary · quantitative / qualitative — all have strengths & weaknesses
- Design: planning → gathering (sample, pilot) → processing → evaluation · hypothesis ⇒ quant; research question ⇒ qual
- Sampling: random · systematic · stratified random · quota · opportunity (best / snowball) · non-representative; need a sampling frame
- Approaches: case study · social survey · ethnography · longitudinal · triangulation (methodological / researcher / data)
- Positivism = science, structure, objectivity, reliability, quantitative, value-free · Interpretivism = verstehen, meaning, subjectivity, validity, qualitative
- Science = reliable + valid · Merton: Universal, Communal, Disinterested, Sceptical
- Issues: theoretical · practical (access, funding, time) · ethical (consent, legal, safety, true/fair/wise) · values can bias at every stage
- Assess methods with: validity · reliability · objectivity · representativeness · ethics