HookDurkheim counted suicides to prove society is real
In 1897 Émile Durkheim did something that sounds mundane and was actually revolutionary: he collected official statistics on suicide across whole regions of Europe and looked for patterns. He found that suicide rates were remarkably stable year on year within each society, yet differed consistently between groups — Protestants killed themselves more than Catholics, the unmarried more than the married, soldiers more than civilians. His conclusion was startling. The most private, individual act imaginable was in fact patterned by something outside the individual — the level of social integration and regulation in their group. Suicide, he argued, was a social fact: real, external, and exerting force on us like the walls of a room.
That study is the founding argument of this whole unit. Every methods decision a sociologist makes is downstream of one question: is society a set of measurable, external social facts (like Durkheim believed), or a web of subjective meanings you have to interpret from the inside? This section teaches the toolkit — quantitative and qualitative design, the six main sources of data, the primary/secondary and quantitative/qualitative distinctions, the positivism–interpretivism debate that Durkheim's suicide study ignited, and the practical, ethical and theoretical (PET) pressures that decide which method a researcher actually reaches for. Get the PET framework fluent and you can evaluate any method the exam throws at you.
ModelResearch design — turning a curiosity into a study
Good research is built, not improvised. A study begins with an aim (the broad area of interest) which positivists sharpen into a hypothesis — a testable, falsifiable prediction stated before data collection. To test it you must operationalise your concepts: turn an abstract idea like 'material deprivation' into something measurable, such as eligibility for free school meals. A pilot study — a small-scale trial run — irons out confusing questions and logistical problems before the real thing.
Because you rarely study everyone, you draw a sample from a sampling frame (a list of the target population). Random sampling gives everyone an equal chance; systematic takes every nth name; stratified divides the population into groups and samples each in proportion, improving representativeness; quota fills preset numbers of each type; snowball uses contacts to reach hidden populations (drug users, criminals) where no frame exists; opportunity grabs whoever is available. The three quality checks examiners want you to name are validity (does it measure what it claims?), reliability (would repeating it give the same result?) and representativeness (can findings be generalised to the wider population?). These three words are the spine of every methods evaluation.
MechanismThe six sources of data — and what each buys and costs
The spec names six sources. Questionnaires (postal or self-completion) are cheap, wide-reaching, reliable and easy to quantify, but suffer low response rates and impose the researcher's categories, threatening validity. Interviews range from structured (a spoken questionnaire — reliable, comparable) through semi-structured to unstructured (a guided conversation — high validity and rapport, but time-consuming, hard to replicate and vulnerable to interviewer bias). Observation can be participant or non-participant and overt or covert: covert participant observation gives rich, valid insight into a group's real behaviour (avoiding the Hawthorne effect) but raises serious ethical and safety problems and cannot be repeated.
Experiments are rare in sociology: the laboratory experiment offers control and reliability but is artificial and unethical for most social questions, so sociologists use field experiments (Rosenthal and Jacobson's classroom study) or Durkheim's comparative method — treating society as a natural laboratory. Documents — personal (diaries, letters), public (government reports, media) and historical — are cheap and sometimes the only window on the past, but Scott warns to test them for authenticity, credibility, representativeness and meaning. Official statistics are free, large-scale and reliable, but positivists and interpretivists split over them fiercely — for Durkheim they are hard social facts, while interpretivists like Atkinson argue a suicide statistic is merely a coroner's socially constructed label, not an objective count.
ModelPrimary vs secondary, quantitative vs qualitative
Two cross-cutting distinctions organise all data. Primary data is collected first-hand by the researcher for their own purpose — a fresh questionnaire, an interview, an observation. Its advantage is fit: it answers exactly the question asked. Its cost is time and money. Secondary data already exists — official statistics, documents, prior studies, the census. It is cheap and sometimes the only route to the past or to large populations, but it was gathered for someone else's purpose, so it may not fit the aim and its reliability must be interrogated.
Cutting across that is the quantitative/qualitative divide. Quantitative data is numerical — rates, percentages, correlations — prized by positivists for revealing patterns and allowing comparison, but often thin on meaning. Qualitative data is descriptive — words, feelings, observed behaviour — prized by interpretivists for its depth and validity, but harder to quantify, compare or generalise. The two axes combine: a postal questionnaire is usually primary and quantitative; a historical diary is secondary and qualitative. A common exam error is to treat 'primary' as a synonym for 'qualitative' — they are independent dimensions, and naming both correctly for a given method is easy AO1 credit.
A model answer applying the quantitative/qualitative distinction shows the analytical move examiners reward: 'A researcher investigating why pupils truant faces a genuine trade-off. Official truancy statistics are secondary and quantitative: they are free, cover the whole country and are reliable, so they can establish that truancy correlates with poverty and low achievement. But they are a positivist's tool — they reveal the pattern without the meaning, and they may under-record truancy that schools have an incentive to hide, threatening validity. To understand why a particular pupil truants, the researcher would need primary qualitative data from unstructured interviews, which sacrifice representativeness and reliability for the verstehen — the subjective understanding — that the numbers cannot supply. The strongest research design therefore triangulates: use the statistics to map the pattern, then interviews to explain it.' The paragraph does not merely define the terms. It applies both to one concrete issue, names the theoretical camp behind each, states the cost of each choice, and lands on triangulation as the resolution — which is precisely the synoptic, evaluative reasoning the top band demands.
CasePositivism, interpretivism and the nature of social facts
This is the theoretical engine of the whole unit. Positivists, following Durkheim, treat sociology as a science modelled on the natural sciences. They believe society is made of objective social facts that exist outside the individual and exert measurable force; the job of the sociologist is to collect quantitative data, uncover correlations and establish cause-and-effect laws, standing detached and value-free. Durkheim's suicide study is the exemplar: patterns in official statistics revealed the hidden causal force of social integration.
Interpretivists, following Max Weber, reject the copy-and-paste of natural-science method. People, unlike atoms, attach meaning to their actions, so you cannot explain society without verstehen — an empathetic understanding of how actors see their own world. They favour qualitative methods that access those meanings and are deeply sceptical of statistics: Douglas and Atkinson argued a suicide 'rate' is not a social fact at all but the accumulated outcome of coroners' interpretive decisions about what counts as suicide — the statistic constructs the reality it claims to measure. Weber's own position was subtler than the caricature: he wanted both causal explanation and interpretive understanding, and realists such as Bhaskar later argued science itself studies unobservable underlying structures, so the gap between sociology and 'real' science is smaller than positivists and interpretivists both assume.
MechanismPET — why researchers choose what they choose
No method is picked in a vacuum. AQA wants you to weigh three sets of factors, remembered as PET. Practical factors are the mundane constraints that decide most real research: time, money and funding (a PhD student and a government-funded team have very different options), access to the subjects, the researcher's own skills, and the requirements of the funding body. Covert observation of a gang is cheap in money but enormous in time and risk; a postal questionnaire is the reverse.
Ethical factors are the moral limits set out in guidelines like those of the British Sociological Association: informed consent, confidentiality and anonymity, protection from harm, avoiding deception, and special care with vulnerable groups such as children. Covert methods breach consent and deceive by design, which is why they need heavy justification. Theoretical factors are the researcher's own methodological stance: a positivist chasing reliability and representativeness reaches for questionnaires and statistics, while an interpretivist chasing validity and verstehen reaches for unstructured interviews and participant observation. The examiner's prize move is to show these three pull against each other — the most valid method (covert observation) is often the least ethical and the least reliable — so every method choice is a compromise, never a clean win.
VocabularyKey terms the mark scheme pays for
TrapsMisconceptions that cost marks
ExamWhat examiners want
Research Methods is assessed for AO1 (knowledge), AO2 (application) and AO3 (analysis and evaluation), and appears in Paper 1 mainly as the 10-mark 'Outline and explain two...' question and as the theoretical backbone of the 20-mark Methods in Context answer. On the 10-marker there is no Item and no evaluation quota — you need two clearly separate points, each developed with a study or a concept and analysed, so make each half a self-contained paragraph rather than one blurred blob.
The evaluation engine for every methods question is PET plus the validity/reliability/representativeness trio. Train yourself to write the sentence 'this method is high in X but low in Y' for every method, because the top band is reached by showing trade-offs, not listing features — covert observation is valid but unethical and unreliable; questionnaires are reliable and representative but low in validity. Anchor points in named sociologists (Durkheim, Weber, Douglas, Atkinson, Rosenthal and Jacobson) for AO1 credit, and when a question ties a method to a topic, apply the abstract point to that concrete topic — a generic 'questionnaires have low response rates' scores far less than showing why that specific issue matters for that specific research aim.