Choosing and Evaluating Research Methods

Choosing a method to answer a psychological question

Psychologists choose research methods according to what they want to discover. Testing whether noise affects memory requires a different approach from exploring how someone experiences memory loss. The first question needs a comparison that separates the effect of noise from other influences; the second needs detailed information about a person’s experiences.

Experiments investigate the effects of an independent variable (IV) on a dependent variable (DV). In laboratory and field experiments, the researcher changes the IV and measures the DV. For example, a researcher could vary background noise and measure the number of words recalled. Other influences, such as the time allowed to learn the words, are extraneous variables: they must be controlled as far as possible so they do not provide an alternative explanation.

Questionnaires, interviews, case studies and observational studies gather information without deliberately manipulating an IV. They can describe behaviour and experiences, but do not establish cause and effect on their own. No method is best for every question: choosing well means balancing control, realism, detail, practicality and ethics.

Laboratory experiments: control over the situation

A laboratory experiment takes place in a controlled, artificial environment. It does not have to involve scientific equipment: a quiet room arranged for a memory task can be a laboratory setting.

In the noise-and-memory example, the researcher could keep the word list, learning time and instructions the same while changing the noise level. A standardised procedure means carrying out the study in the same way for participants, apart from the intended differences between conditions.

High control makes alternative explanations less likely, strengthening the conclusion that the IV caused a change in the DV. Standardisation also makes the procedure easier for another researcher to repeat and check. Laboratory experiments are therefore useful when the aim is to test a specific possible cause under carefully controlled conditions.

However, recalling a word list in a research room may not reflect remembering information in everyday life. This limits ecological validity, meaning how well findings reflect everyday behaviour. Participants may also guess the aim from clues in the procedure. These demand characteristics can lead them to alter their behaviour, so the results may reflect their response to the study rather than the effect being investigated.

Field experiments: changing a variable in everyday surroundings

A field experiment also involves the researcher manipulating the IV, but takes place in participants’ natural setting. For example, a researcher could introduce a kindness programme to some pupils in a school and compare their later prosocial-behaviour scores with those of pupils who did not receive it. The IV is whether pupils receive the programme; the DV is their score.

Field experiments are suitable when researchers want to test an intervention or influence in the setting where it normally matters. Their realistic surroundings can produce more natural behaviour and higher ecological validity. If participants are unaware of the investigation, demand characteristics may also be reduced.

The trade-off is lower control. Differences in teaching, friendships or other events could affect pupils’ behaviour, making it harder to identify the programme’s effect. Being in a natural setting does not remove these alternative explanations.

Studying people without their awareness can also raise ethical problems: they may not have given informed consent or know that they can withdraw. Realism is not a justification for exposing people to harm or distress.

Natural experiments: studying an existing difference

In a natural experiment, the researcher does not create or manipulate the IV. The difference already exists independently of the investigation. Researchers might compare teachers who have already received empathy training with teachers at another school who have not. Unlike the field experiment above, the researcher has not decided who receives the intervention.

Natural experiments are useful when manipulating the variable would be impossible or unethical. For example, researchers can investigate the effects of adverse childhood circumstances without deliberately subjecting children to them.

Because participants already belong to their groups, the researcher cannot randomly allocate them to the naturally occurring conditions. Groups may differ in other important ways: teachers from different schools may have different experience or working conditions. These differences make a cause-and-effect conclusion less secure. An unusual event or circumstance may also be difficult to reproduce, limiting replication.

The key distinction is therefore not simply where the research happens. Laboratory and field experiments have researcher-manipulated IVs; natural experiments have naturally occurring IVs.

Questionnaires: collecting answers from many people

A questionnaire collects self-reported information through written questions. It is useful for surveying the views or experiences of many people quickly and at relatively low cost.

A closed question provides fixed response options, such as asking students to rate exam stress from 1 to 5. These answers are easy to count and compare, but the options may not capture everything a respondent wants to say. An open question, such as ‘What worries you most about exams?’, allows an answer in the respondent’s own words. It can reveal reasons behind a rating, although interpreting many different answers takes longer.

Questionnaires can reach large samples, but a large sample is not automatically representative: who receives and completes the questionnaire still matters. Respondents may misunderstand a question without anyone being available to clarify it, and researchers cannot easily follow up an unexpected answer.

Self-report also depends on honesty. Social desirability bias occurs when someone gives an answer that makes them look good rather than an accurate answer. For example, a person may overstate how much exercise they do. Anonymous responses and neutral wording can reduce this risk, although they cannot guarantee truthful answers.

Interviews: exploring answers through conversation

An interview gathers self-reported information through spoken questions. It is particularly useful when researchers want to understand experiences in detail or clarify what a participant means.

A structured interview uses the same questions in the same order. This makes responses easier to compare and the procedure easier to repeat, but restricts opportunities to explore an interesting answer. An unstructured interview is a flexible, guided conversation. The interviewer can ask follow-up questions, producing richer detail, but different conversations are harder to compare and analyse.

An interviewer can build rapport—a comfortable, trusting relationship—which may help participants explain their experiences. However, interviews take time, making it harder to investigate as many people as a questionnaire can. The interviewer’s manner or wording may also influence answers. Social desirability can be especially problematic face to face, when participants feel that another person is judging them.

For a broad survey of exam-stress ratings, a questionnaire is practical. For understanding why particular students feel stressed, an interview may be more useful. The choice depends on whether breadth or depth matters most.

Case studies: understanding a particular case in depth

A case study investigates a particular person, group or event in detail. It often combines methods, such as interviews, observations and psychological tests, and may follow the case over a long period.

For example, investigating a person with severe memory loss through repeated tests and interviews can reveal a detailed pattern of abilities and difficulties. Case studies are especially suitable for unusual conditions or experiences that would be difficult to investigate in a large sample.

Their strength is depth: combining information can reveal details that a short test or questionnaire would miss and suggest ideas for further research. However, an unusual person may not represent others, so findings cannot safely be generalised to everyone. The researcher’s close involvement may also lead to biased interpretation. Detailed knowledge of one case does not by itself establish what caused its difficulties.

Observational studies: recording what people do

An observational study involves watching and recording behaviour rather than relying on what participants say about themselves. It is useful when the research question concerns visible actions, such as how children interact in a playground.

A naturalistic observation takes place in the usual setting. Playground behaviour may be realistic, giving high ecological validity, but weather, group size and other influences are difficult to control. A controlled observation takes place in a situation arranged by the researcher. This makes conditions more consistent, but the setting may affect how people behave. Arranging a setting does not necessarily make the study an experiment: an experiment investigates an IV’s effect on a DV.

Observers can remain separate from the group (non-participant observation) or join it (participant observation). Participants may know they are being observed (overt observation) or not know (covert observation). Awareness can change behaviour, while covert observation raises concerns about consent and privacy.

Observations show actions directly, avoiding some inaccuracies of self-report. However, watching someone does not necessarily reveal their thoughts or reasons. The observer’s expectations may also influence what they notice and record.

Behavioural categories and agreement between observers

To record behaviour systematically, researchers break it into behavioural categories: clearly defined, observable actions. For a playground investigation, ‘pushing another person’ and ‘calling another person an insulting name’ are more useful categories than ‘being aggressive’. Observers can tally each occurrence instead of making a vague overall judgement.

Categories should be distinct enough that observers can decide consistently where an action belongs. They should also cover the behaviours relevant to the research question. A brief trial can reveal unclear definitions before the main observation begins.

Interobserver reliability is the extent to which observers agree when recording the same behaviour. Two observers watch the same activity at the same time and record independently, using the same categories. They then compare their records. Close agreement suggests that the recording system is consistent rather than dependent on one person’s interpretation.

If agreement is poor, researchers can clarify the categories and train observers before checking again. Agreement does not prove that every relevant behaviour was recorded, but it strengthens confidence in the consistency of recording.

For an ethical classroom activity, observe only with appropriate consent and school approval, protect identities, and do not provoke aggression or distress to obtain data. Use an agreed observation period, then stop recording at its end and compare the independently completed records.

Reaching a justified choice

A suitable method must answer the particular question. A laboratory experiment offers control for testing an effect; a field experiment tests a researcher-made change in everyday surroundings; a natural experiment investigates a difference the researcher cannot create. Questionnaires offer breadth, and case studies offer depth, and observations record actions directly.

A justified choice recognises the cost of the advantage. For example, choosing interviews for detailed accounts means accepting smaller samples and possible interviewer influence. Evaluation is strongest when it explains why a method’s strengths matter for the aim and whether its limitations make another method more suitable.