Planning research: aims, hypotheses and pilot studies
Research starts with a topic (for example, how teenagers use social media). The researcher then writes either an aim (what they want to find out) or a hypothesis (a statement that can be tested, such as "girls spend more time on messaging apps than boys").
Next comes a pilot study: a small trial run with a few people. It checks that questions are clear, that the method works and how long it takes. Fixing problems now saves time and money later.
The main steps are:
- Choose a topic and read what is already known.
- Write an aim or hypothesis.
- Choose a method and run a pilot study.
- Choose a sample.
- Collect the data.
- Analyse and interpret the results.
- Write up conclusions and say how to improve the study.
Sampling
The population is the whole group being studied. Studying everyone is rarely possible, so we take a sample. A sampling frame is the list the sample is chosen from (a school register, a voters' list).
- Random sampling: everyone has an equal chance (names from a hat or a computer).
- Systematic sampling: every nth name on the list (every 10th student).
- Stratified sampling: split the population into groups (age, gender) and pick from each in the right proportion.
- Quota sampling: interviewers fill set numbers from each group, but they choose who (quicker, less random).
- Snowball sampling: one person introduces the next; useful for hard-to-reach groups such as homeless people.
- Opportunity sampling: whoever is available; quick but often biased.
A representative sample has the same mix as the population, so results can be generalised to everyone. A larger random sample is usually more accurate, but a biased sample stays biased however big it is.
Methods and types of data
Quantitative data are numbers that can be counted and shown in charts. Qualitative data are words, pictures and meanings that show how people feel and why.
- Questionnaires: written questions. Closed questions have fixed answers (yes/no, tick boxes) and give quantitative data; open questions let people write freely. Cheap and quick for large samples; low response rates are a problem.
- Structured interviews: the same questions read out in the same order; reliable and easy to compare.
- Unstructured interviews: a guided conversation; rich, detailed qualitative data, but slow and hard to repeat. Interviewer bias can happen when the interviewer's manner affects answers.
- Observation: watching behaviour. Participant (joining the group) or non-participant; overt (the group knows) or covert (hidden). Covert observation avoids the Hawthorne effect (people change behaviour when watched) but raises ethical problems.
- Experiments: controlling variables to test cause and effect. Rare in sociology because real social life is hard to control and it raises ethical issues; field experiments happen in real settings.
- Longitudinal studies follow the same people over years; case studies look deeply at one group; mixed methods (triangulation) combine methods to check one against another.
Primary and secondary data
Primary data are collected by the researcher for this study (surveys, interviews, observation): up to date and fit for purpose, but costly. Secondary data already exist: official statistics (census, crime figures, exam results), documents (diaries, letters, websites), media and earlier research. They are cheap and can cover large numbers or the past, but they were made for another purpose and may be biased or incomplete.
Evaluating research: reliability, validity and representativeness
- Reliability: if another researcher repeats the study the same way, they get the same results. Standardised methods (closed questionnaires, structured interviews) are usually more reliable.
- Validity: the data give a true picture of what is being studied. In-depth interviews and participant observation are usually more valid because people explain in their own words.
- Representativeness: the sample reflects the population, so findings can be generalised.
Reading graphs and tables
When reading data, check: the title and source (who collected it and when), what the units are (numbers or percentages), the size of the sample, the biggest and smallest values, any trend over time, and anything missing. Say what the data show before saying what they might mean. Example: "In the table, the share of households with internet rose from 20% to 75% in ten years; this suggests…"
Choosing a method: practical, ethical and theoretical issues
Practical issues: time, money, access to the group, the researcher's skills and personal characteristics, and funding bodies' wishes.
Ethical issues: get informed consent; keep identities confidential and anonymous; avoid harm (physical or emotional); avoid deception; take extra care with vulnerable groups such as children; follow the law and professional guidelines.
Theoretical issues:
- Positivists see sociology as a science. They want objective, reliable, quantitative data to find patterns and causes: questionnaires, structured interviews, official statistics.
- Interpretivists want to understand meanings and feelings from the inside (verstehen). They prefer valid, qualitative data: unstructured interviews, participant observation, personal documents.
Methods in context: researching education
Schools bring special issues: pupils are young (consent from parents and the school is needed), teachers are busy, classrooms are watched by gatekeepers, and pupils may give answers they think the teacher wants. A study of classroom labelling might use non-participant observation of lessons plus interviews with pupils, while a study of results by gender would use official exam statistics.
Key formulas and definitions
- [object Object]
- [object Object]
- [object Object]
- [object Object]
- [object Object]
- [object Object]
Worked examples
1. A researcher hands out a closed questionnaire to 500 students about hours of homework. Is the data quantitative or qualitative, primary or secondary?
Quantitative (numbers of hours) and primary (the researcher collected it).
2. A school has 600 boys and 400 girls. Using stratified sampling, how many of each should be in a sample of 50?
Boys are 600 ÷ 1000 = 60%, girls 40%. 60% of 50 = 30 boys; 40% of 50 = 20 girls.
3. Why might covert participant observation of a street gang be valid but raise ethical problems?
Valid: members do not know they are watched, so they act naturally and the researcher sees real behaviour from the inside. Ethical problems: no informed consent, deception, the researcher may witness or be pressured into crimes, and it could be dangerous if discovered.
4. A table shows that 72% of 1,000 surveyed adults trust doctors and 34% trust journalists. Write two correct statements.
Out of the adults surveyed, more than twice as many trust doctors as trust journalists (72% vs 34%). About 340 of the 1,000 adults said they trust journalists.
Common mistakes
- Thinking a bigger sample always fixes everything. A large biased sample is still biased; how people are chosen matters as much as how many.
- Mixing up reliability and validity. Reliable = repeatable; valid = true. A method can be reliable but not valid.
- Calling all interviews qualitative. Structured interviews with fixed answers give quantitative data.
- Assuming official statistics are always the full truth. They are made by organisations for their own purposes and can leave things out (for example unreported crime).