Psychology › Section B · Psychological research › Topic B.2

Approaches to research

A psychology question arrives, and the researcher must choose a vehicle: the case study that knows one person deeply, the survey that knows thousands shallowly, the observation that watches, the archive that never forgets, and the experiment that decides. Each has a trade-off - and matching method to question is the researcher's first skill.

A toolkit with trade-offs

No research method is best; each is best FOR something. Observation-based methods produce rich, deep information from small numbers - hard to generalise. Surveys reach large, generalisable samples but collect shallow, self-reported data. Archival research mines existing records cheaply but surrenders all control over what was collected. Experiments (the next topic) can support cause-and-effect conclusions but require control that some real-world questions cannot grant. The researcher's first decision - before any data - is matching the question to the method whose weaknesses the question can survive.

Case studies: depth without breadth

A case study studies one person (or a very few) in extraordinary detail - a rare patient, an unusual brain injury, a remarkable memory. The richness is unmatched by any other single method: no survey learns what a good case study learns. The price is generalising - applying findings to the larger population. Cases are chosen BECAUSE they are unusual, so the case's lessons may not transfer to ordinary people. Case studies are the field's microscopes: incomparable resolution, tiny field of view.

Naturalistic observation

Naturalistic observation watches behaviour in its natural environment, with the researcher keeping distance so as not to disturb what is being watched. Pupils act differently once they notice an inspector at the back of the classroom, and the same is true of anyone who knows they are being studied. The method is also central to animal research: Jane Goodall began watching wild chimpanzees at Gombe, Tanzania, in 1960 and studied them for decades. Some scientists criticised her for naming the chimps instead of numbering them, arguing that names undermined objective detachment (McKie, 2010). The method's costs: time, money, luck - and no control over what happens in front of the observer.

Structured observation and observer bias

Observation can be structured: people are watched performing set, specific tasks. Mary Ainsworth's Strange Situation (developed in the 1970s) is the classic example - a procedure evaluating attachment between infant and caregiver, with phases in which a stranger enters, the caregiver leaves, and the caregiver returns; the reunion behaviour is most telling for attachment style (developed fully in a later chapter). Every observational method risks observer bias: observers, being human, may unconsciously skew what they record toward what they expect or hope to find. The protection is clear, pre-established criteria for what counts as a recorded behaviour - and, where possible, multiple observers who agree.

Surveys: breadth without depth

Surveys trade depth for reach: hundreds or thousands of respondents, questionnaires or interviews, samples large and diverse enough to generalise to a population - the case study's mirror image. Their weakness is self-report: people may lie, misremember, or shade answers to look good - claiming more exercise or fewer hours on their phones than they really manage, for instance. A real example of a survey's reach: the US General Social Survey has asked large national samples about attitudes and behaviour since 1972, letting researchers track how public opinion shifts across decades - something no case study could show.

Archival research

Archival research analyses existing records - census data, hospital files, historical documents, social-media archives - and its economics are its charm: no recruiting, no data collection, just analysis. The costs are structural. The researcher never interacts with participants (an ethical simplification, too), but also has zero control over what was originally collected, so questions must be tailored to fit the data rather than the reverse; and there is no guarantee records from different sources were gathered consistently, making cross-source comparison treacherous. Archival research is rummaging through a library someone else catalogued.

Longitudinal research

To see how people change over time, test the SAME people repeatedly across months, years or decades: longitudinal research - for example, measuring the same people's sleep habits at 15, 25 and 35. A real example is the Dunedin Study in New Zealand, which has followed about 1,000 people born in 1972-73 from birth into middle age. The price is high: time (results take years), money, and attrition - the loss of participants who move, fall ill, die, or quit. Attrition can build up over time and bias a sample, so many studies recruit extra participants; Dunedin is unusual in keeping over 90% of its living members.

Cross-sectional research

The cheaper alternative for studying change: cross-sectional research compares different segments of the population AT THE SAME TIME - 20-year-olds, 40-year-olds, and 60-year-olds surveyed once, their differences read as an estimate of how people change with age. The shortcut carries a confound - a second difference mixed up with the one being studied: the 60-year-olds differ from the 20-year-olds not only in age but in generation, history, upbringing - the cohort effect. Is the difference aging, or being born in a different era? Cross-sectional data cannot fully separate the two; only following people through time can.

Worked example: choosing the method

Match each question to a method. 1. "What is daily life like for a person with a rare memory disorder?" Case study - rarity demands depth, and generalising is not the point. 2. "What percentage of voters support the new policy?" Survey - breadth is the entire question. 3. "How do toddlers behave when briefly separated from a caregiver?" Structured observation - a controlled scenario, predefined behaviours, multiple observers. 4. "Did crime rates change over the last 50 years?" Archival research - the records already exist. 5. "Does watching violent media cause aggression?" Experiment - only random assignment can support "cause." Check: every answer is a trade-off, and a different question flips the answer - the method serves the question, never the reverse.

Worked example: spotting the confounds

A cross-sectional study finds that 65-year-olds report more satisfaction with their social lives than 25-year-olds, and concludes that contentment grows with age. Read it sceptically. 1. Cohort confound: the 65-year-olds grew up with different norms, different economies, different technologies - their satisfaction may reflect their generation, not their age. 2. Survivorship: people in poor health or isolated lives are less likely to reach 65 or to answer a survey, so the older group may be a more contented subset. 3. Self-report: both age groups may shade their answers. 4. Verdict: an interesting difference between groups; only a longitudinal study, tracking the same people as they age, could show whether individuals actually become more satisfied.

Common mistakes to avoid

First, asking a breadth question with a depth tool - generalising from a case study, or demanding depth from a survey. Second, ignoring observer bias - define behaviours and use multiple observers. Third, trusting self-report blindly: people underreport what embarrasses them. Fourth, comparing archival records gathered under different rules. Fifth, forgetting that longitudinal participants drop out, and that those who leave may differ from those who stay. Sixth, reading cross-sectional age differences as pure aging - the cohort effect is always in the room. Seventh, choosing the method by convenience rather than by the question.

Key takeaways

Every method is a trade-off. Case studies: unmatched depth, no generalisation (the microscope). Naturalistic observation: real behaviour in its setting, at the risk of observer bias (guard with clear criteria; Goodall and her naming controversy). Structured observation: Ainsworth's Strange Situation. Surveys: breadth and generalisability at the cost of depth and self-report honesty (the General Social Survey, since 1972). Archival research: cheap, but shaped by other people's choices. Longitudinal: same people over time, expensive and prone to attrition (the Dunedin Study). Cross-sectional: fast, but cohort effects ride along. The question chooses the tool - and every finding inherits the tool's weaknesses.

Flashcards

Flashcard

Case study

One or few cases, in extraordinary depth

The microscope: resolution without field of view.

Test your understanding

Question 1 of 5

A researcher spends months documenting the daily life of one person with an extremely rare memory condition. This is:

Question 2 of 5

The major limitation of case studies is:

Question 3 of 5

A study surveys 25-, 45-, and 65-year-olds once each and concludes that patience increases with age. The main confound is:

Question 4 of 5

High attrition is a particular problem for which design?

Question 5 of 5

Observer bias is best reduced by:

← Previous topicNext topic →