Sociology › Section B · Sociological research › Topic B.2
Research methods
Once a sociologist has a question, the next decision is how to gather the evidence. Surveys, field research, experiments and existing data each answer some questions well and others badly, so choosing the method is part of the science.
Choosing a research design
A research design is the plan for answering a question: whom to study, what to collect and how. Sociologists draw on four main families of methods: surveys, field research, experiments and the analysis of existing data. Each has strengths and weaknesses, and the topic often decides. You cannot quietly observe a prison without permission, or run an experiment on divorce. Researchers pick the method that fits the question, protects participants and can be carried out with the time and money available. Many studies combine several methods.
Surveys
A survey asks a group of people a set of questions, through a questionnaire or an interview, in person, by phone, by mail or online. Surveys can reach large samples quickly and are good at measuring what people say they think, feel and do. The General Social Survey has asked national samples of US adults about their attitudes and behaviour since 1972, allowing researchers to track change over decades. But small changes in wording or question order can change answers, and respondents may give socially desirable answers, the ones that make them look good, rather than accurate ones. What people report is not always how they behave.
Samples, questions and data
A survey targets a population, the whole group of interest, such as first-year college students, and usually questions a sample drawn from it. A random sample lets results be generalised. Closed-ended questions, such as yes-or-no or multiple-choice items, produce quantitative data: numbers that can be counted and analysed statistically. Open-ended questions let people answer in their own words and produce qualitative data, richer but harder to sort and compare. An interview is a conversation in which the researcher can follow up, ask for examples and explore a surprising answer, but must avoid steering the respondent towards a particular reply.
Field research and participant observation
Field research gathers data in people's natural settings, such as a workplace or an online community. In participant observation the researcher joins in a group's everyday life while studying it. William Foote Whyte spent several years living in an Italian-American neighbourhood of Boston for Street Corner Society (1943), which showed that what outsiders saw as a disorganised slum had its own clear social structure. Observation can be overt, with the group aware of the researcher's purpose, or covert, which gives access to natural behaviour but raises ethical questions covered in the next topic. Field research is rich in meaning and context but slow, and one setting may not represent others.
Ethnography and case studies
Ethnography is extended immersion in a whole community to describe its way of life as its members understand it. Robert and Helen Lynd lived in Muncie, Indiana, for about eighteen months to write Middletown (1929), a study of an ordinary American city changing under industrialisation; it became a bestseller. The Canadian sociologist Dorothy Smith developed institutional ethnography, which starts from people's everyday experience, especially women's, and traces how institutions organise it. A case study examines a single person, group, event or organisation in depth. It is valuable for rare cases, but one case cannot establish a general pattern.
Experiments
Experiments are the strongest design for testing cause and effect. The researcher manipulates an independent variable for an experimental group, leaves a control group untreated, and compares the dependent variable, ideally after assigning people to groups at random. Laboratory experiments give tight control but can feel artificial. Field experiments take place in real settings. An important sociological type is the audit study, in which researchers send matched applicants, real or on paper, who differ in only one characteristic, such as race or a criminal record, and compare how they are treated. Many sociological questions cannot be studied by experiment, for practical or ethical reasons.
Worked example: reading an audit study
In 2001 the sociologist Devah Pager sent pairs of trained young men to apply for entry-level jobs in Milwaukee, publishing the results in 2003. 1. Design: within each pair, testers had matched résumés, and one in each pair was assigned a fictional criminal record; pairs were either white or Black. 2. Dependent variable: whether the employer called the applicant back. 3. Finding: a criminal record sharply reduced callbacks, and white testers with a record were called back about as often as Black testers without one (roughly 17% against 14%). 4. Why it supports a causal claim: because applicants were matched on everything else, differences in callbacks can be attributed to race and record. 5. Limits: one city, one period and only the first stage of hiring.
The Hawthorne effect
Between 1924 and 1932, researchers at the Western Electric Hawthorne Works near Chicago studied how lighting, breaks and pay affected workers' output. Productivity seemed to rise with many different changes, and one popular interpretation was that workers were responding to being observed. People changing their behaviour because they know they are being studied became known as the Hawthorne effect. Later reanalyses of the original data found the evidence much weaker than the legend suggests, but the warning remains useful: when participants know the purpose of a study, observation itself can alter what is observed.
Secondary data and content analysis
Secondary data analysis uses information collected by others: census data, government statistics, organisational records, historical documents, newspapers or social media posts. It saves time and money, allows study of the past and is nonreactive, or unobtrusive, because it does not disturb the people being studied. Content analysis systematically codes the content of texts or media, for example counting how often women appear in leading roles in a sample of films. The drawbacks: the researcher cannot control how the data were collected, the exact measure needed may not exist, and records reflect the assumptions of the time they were made.
Worked example: why young people leave a small town
Plan a study of why young adults leave a rural town. 1. Secondary data: census figures show how the number of people aged 18 to 29 has changed over twenty years. 2. Survey: a questionnaire to every final-year high school student measures how many intend to leave and their main reasons, producing quantitative data. 3. Interviews: in-depth conversations with a smaller group explore what home, opportunity and belonging mean to them, producing qualitative data. 4. Participant observation: time at community events shows how young people are included or overlooked. 5. Combining methods gives breadth and depth. 6. Check: in a small town people are easily identified, so confidentiality needs particular care.
Common mistakes to avoid
First, treating survey answers as direct records of behaviour: they show what people report. Second, assuming a large sample is automatically representative. Third, claiming cause from field research or surveys alone: experiments are the strongest test of causation. Fourth, generalising from one case study or one ethnographic setting. Fifth, citing the Hawthorne studies as firm proof of a strong effect: the original evidence has been questioned. Sixth, forgetting that existing records were collected for other purposes and carry the assumptions of their time.
Key takeaways
Surveys reach large samples, as the General Social Survey has since 1972, producing quantitative data from closed questions and qualitative data from open ones, but wording and social desirability affect answers. Field research, including participant observation (Street Corner Society, 1943) and ethnography (Middletown, 1929), reveals meaning in context but generalises poorly. Experiments, including audit studies such as Pager's, are the best test of cause. The Hawthorne effect warns that observation can change behaviour. Secondary and content analysis are cheap and unobtrusive but depend on others' data.
Flashcards
Flashcard
The plan for whom to study, what to collect and how
Test your understanding
Question 1 of 5
A researcher wants to know whether employers treat applicants differently by race. Which method gives the strongest evidence of cause?
Question 2 of 5
Survey respondents report exercising far more than fitness trackers show they actually do. This is best explained by:
Question 3 of 5
Which is a key advantage of secondary data analysis?
Question 4 of 5
A sociologist lives for two years in a fishing village, joining in daily work while studying the community's way of life. This is:
Question 5 of 5
What do later reanalyses suggest about the Hawthorne studies (1924-1932)?
