What makes research methods different?

A research method is a systematic procedure for collecting evidence to answer a research question. What researchers actually do varies with the method. They might change a condition, measure an association, observe behaviour or ask people about their experiences. Those choices determine which conclusions the evidence can support.

Research methodology is the overall approach that connects a research question to the choice and use of methods. Picking a convenient way to collect information isn't enough. The method must fit the question and the participants, as well as the available resources and ethical limits.

Keep one distinction in mind throughout this topic: evidence that two things occur together is not evidence that one causes the other.

Experiments: changing a condition

An experiment investigates a possible causal relationship by comparing outcomes across conditions. The factor whose conditions are manipulated or compared is the independent variable. The dependent variable is the outcome measured to assess differences between those conditions.

In a true experiment, the researcher manipulates the independent variable and uses random allocation to assign participants to conditions. If participants complete every condition, they are assigned to condition orders instead. Random allocation uses chance to determine participants' conditions or condition orders. It reduces systematic differences between conditions, but doesn't guarantee a perfect balance of every relevant characteristic.

For example, a researcher could allocate participants to complete a proofreading task with or without periodic notification sounds, then compare the number of errors. Here, the researcher changes the sound condition. They don't simply ask people how often they receive notifications.

A quasi-experiment compares conditions without random allocation, often because group membership or exposure already exists. A researcher could, for instance, compare proofreading performance among employees already working different shift patterns. That comparison is possible, but shift pattern may not be the only difference between the groups. Their experience, sleep or workload may differ too.

The crucial distinction is how participants enter the conditions. The diagram contrasts conditions created by the researcher with a comparison based on existing groups.

True experiments are especially useful for questions about cause and effect. Researchers can manipulate a factor and compare outcomes while controlling other influences. Standardized procedures also make replication easier. There are limits, though. A tightly controlled task may not represent everyday behaviour, and participants may change what they do because they recognize what is being investigated. Some potentially important variables cannot ethically or practically be manipulated.

Quasi-experiments let researchers investigate differences that a true experiment could not reasonably create. The main limitation is weaker control over pre-existing differences between groups. Different outcomes may reflect the variable of interest, another group difference, or both.

Correlational studies: measuring an association

A correlational study measures variables to investigate whether they are associated, without manipulating them. A correlation is a relationship in which variation in one variable is associated with variation in another.

Suppose researchers measure weekly commuting time and self-reported fatigue. A positive relationship would mean that longer commuting times tend to accompany greater fatigue. A negative relationship would mean that longer commuting times tend to accompany lower fatigue. Neither pattern explains, by itself, why the relationship occurs.

When manipulation would be unethical or impractical, correlational studies can be useful. They identify patterns in naturally occurring behaviour and provide a basis for prediction. They do not establish causality. The direction of influence may be unclear, or another variable may influence both measures. Measuring more variables can help researchers investigate alternative explanations; it does not automatically turn an association into proof of causation.

Case studies: depth within a bounded case

A case study is an intensive investigation of a bounded individual, group, organization or situation, often using several sources of evidence. To understand how a small team adjusts to a major change in its working practices, a researcher might combine interviews with observations and existing records.

Depth is the strength here. A case study can trace connections between experiences, behaviour and circumstances over time, picking up details that a brief questionnaire might miss. Unusual cases that would be difficult to investigate through a large sample can also be studied this way.

A distinctive case, however, may not represent other people or settings. Researchers also have to decide which evidence matters, and those interpretive decisions create opportunities for bias. Rich description can support an explanation, but a case study does not usually isolate a single cause. It can contain several methods: a case study is not simply another name for an interview.

Interviews: asking and probing

An interview is a research method in which a researcher asks participants questions to obtain accounts of their experiences, views or behaviour.

In a structured interview, the interviewer delivers predetermined questions in a standardized way. This improves consistency between participants and makes responses easier to compare. It also leaves the interviewer little freedom to explore unexpected answers.

A semi-structured interview uses a planned set of questions but allows follow-up questions and changes in sequence. There is a common focus, with room for clarification. The trade-off is that interviews may differ between participants. Choices made by the interviewer can influence the evidence collected.

A focus group is a group interview in which a moderator guides discussion about a selected topic. Participants respond to one another. Their exchanges can reveal shared meanings or disagreements, as well as ideas that an individual interview might not generate. But dominant speakers may steer the discussion. Quieter participants may contribute little, and people may avoid expressing an unpopular view.

When a question concerns how people interpret an experience, interviews are particularly useful. They offer detailed accounts, not direct recordings of behaviour. Memory and willingness to disclose information can affect answers, as can the relationship with the interviewer. Recording and analysing detailed interviews takes time too.

Surveys and questionnaires: collecting comparable reports

A survey collects information from a sample through a standardized set of questions. A questionnaire is a data-collection instrument made up of written or electronically presented questions. Surveys commonly use questionnaires, but the method and the instrument are not quite the same thing.

Using questionnaires, researchers can gather comparable information from many participants relatively efficiently. Closed questions offer fixed response options that are straightforward to compare. Open questions let participants answer in their own words, providing detail but requiring more interpretation.

The main limitation lies in what participants understand, remember and choose to report. Wording can steer an answer. Fixed options may not fit someone's experience, and those who respond may differ from those who don't. A large number of responses does not remove these problems on its own.

When writing a questionnaire, ask one clear thing at a time. Avoid wording that suggests a preferred answer. Asking separately about frequency of use and perceived usefulness, for example, is clearer than asking whether a service is 'frequent and helpful'. Trying the questions with a small group first can reveal ambiguity before the main data collection.

Observations: recording behaviour

An observation systematically records behaviour as it occurs. This method is useful for questions about what people actually do, rather than what they say they do. The distinctions below describe separate features of an observation, so they can be combined.

• Naturalistic observation records behaviour in its usual setting. Controlled observation records behaviour in a setting or situation arranged by the researcher.
• Overt observation takes place when participants know they are being observed. Covert observation takes place when participants do not know they are being observed.
• Participant observation involves the researcher taking part in the activities of the group being studied. Non-participant observation involves the researcher recording behaviour without joining those activities.

Naturalistic observation preserves everyday context. Researchers have less control over other influences, though, and may have to wait for the behaviour of interest. With controlled observation, situations are more consistent and researchers can create opportunities to observe particular behaviours. The arranged setting may change how people behave. It does not automatically make the observation an experiment: an experiment requires a comparison directed at a possible causal relationship.

Overt observation allows more transparent participation, but people may change their behaviour because they know they are being watched. Covert observation may reduce that particular influence while raising concerns about consent and privacy. Taking part through participant observation can give the researcher an insider's understanding. That involvement may also affect the group and make detached interpretation difficult. Non-participant observation keeps the researcher out of the activity, although some behaviour may be hard to understand without insider knowledge.

Define observable behaviour before collecting evidence so that it can be recorded systematically. 'Offers another person assistance' is more usable than 'is kind'. If several observers are involved, shared definitions help them record behaviour consistently. Observation can show that an action occurred; it cannot directly reveal the person's private motive.

Selecting a method for a psychological question

Start with the evidence the question needs. If you want to investigate whether changing a condition changes behaviour, an experiment is a strong candidate. For an association without manipulation, use a correlational approach. An interview or case study may be more suitable for understanding a detailed experience. Consider a survey to compare many people's reports, or observation to record behaviour in a setting.

You don't have to treat these choices as mutually exclusive. A correlational study might obtain its measurements through questionnaires; a case study might use interviews and observations. The overall design and the tool used to collect information answer different questions about the research.

Next, ask whether the method can provide relevant evidence from the participants available. Interviewing a few people can provide depth, but it does not establish how common their experiences are in a population. A questionnaire can gather many reports. It cannot replace observation when accurate behavioural records are needed. What is feasible also depends on access and time, the researcher's skills, and the sensitivity of the topic.

For the course's class practicals, the method–context pairings are fixed rather than interchangeable:

• Health and well-being: an interview, using a structured, semi-structured or focus-group format; at least one participant, or three to eight participants in one focus group.
• Human development: a naturalistic or controlled, overt or covert observation; at least one participant.
• Human relationships: a survey/questionnaire; at least ten participants.
• Learning and cognition: a true or quasi-experiment; at least five participants if each completes all conditions, or ten if separate groups complete different conditions.

These minimums are practical requirements. They don't guarantee strong evidence: the question and procedure still need to match the method.

How ethical considerations shape research

Ethical considerations are judgments about researchers' responsibilities to protect participants and others affected by a study. They shape which questions researchers can investigate and which procedures are acceptable. They also affect what information can be collected and how findings can be reported.

Informed consent is voluntary agreement to participate based on adequate information about the research and its foreseeable implications. Researchers must consider whether participants can understand that information and choose freely. Agreement under pressure is not genuinely voluntary. Participants should be able to withdraw without penalty; any limits on removing already-anonymized data should be explained beforehand.

Protecting participants from harm can rule out an experiment that would deliberately expose people to serious distress. Instead, researchers may investigate naturally occurring differences through a quasi-experiment or correlational study. The research becomes more acceptable, but usually offers less control over alternative explanations. The ethically preferable method is not necessarily the one with the strongest causal inference.

Transparency can affect behaviour too. People may behave differently when they know they are being observed, yet hiding observation may violate their privacy or deny them a meaningful choice. Deception is a research practice that deliberately misleads participants about an aspect of a study. A potential methodological benefit does not remove the need for ethical justification or safeguards, nor for an appropriate explanation afterwards. Debriefing cannot undo serious harm. It also cannot automatically make an unacceptable procedure acceptable.

Confidentiality protects identifiable information from unauthorized disclosure. Anonymity is the absence of a link between information and an identifiable participant. A researcher can keep an interview confidential while still knowing who gave it. That interview is not anonymous. Detailed case studies can be particularly difficult to anonymize, since distinctive circumstances may reveal identity. In a focus group, researchers can protect their own records but cannot guarantee that other participants will keep the discussion private.

Ethics changes methodology in concrete ways. Researchers may remove sensitive questions or avoid unnecessary identifying information. They may alter recruitment, restrict access to recordings or choose a less intrusive method. These protections may reduce detail or convenience, but responsibility to participants is part of sound research, not an optional extra.

External variables and conclusions about causality

Causality is a relationship in which a change in one factor produces a change in another. An external variable is a factor outside the relationship being investigated that may influence the results. A confounding variable is an external variable that varies systematically with the independent variable, offering an alternative explanation for a difference in the dependent variable.

Return to the notification experiment. Suppose the notification condition is tested late in the day and the quiet condition early in the day. Time of testing is then mixed up with sound condition: more errors could reflect fatigue rather than notifications. An incidental influence that varies unsystematically, by contrast, may make results more variable without consistently favouring one condition. Not every external influence is a confound.

In a correlational study, an external variable may influence both measured variables. Demanding work schedules, for example, might be associated with both longer commuting time and greater fatigue. The causal diagram shows why a commuting–fatigue association alone cannot establish which factor causes which.

Where appropriate, researchers strengthen causal inference by manipulating the proposed cause and allocating participants randomly. They also standardize procedures and prevent other factors from changing systematically between conditions. Having each participant complete all conditions can reduce differences between people. However, completing one condition may influence performance in the next, so varying condition order helps address that problem.

Even good controls have limits. Random allocation reduces the risk of systematic group differences; it does not prove that none remain. A standardized procedure controls the features specified by the researcher, not every possible influence. Quasi-experiments need particular attention to pre-existing group differences. Correlations leave both the direction of influence and third-variable explanations open.

The conclusion must match the design. Well-controlled experimental evidence can support a causal explanation, whereas an observed association, detailed account or behavioural pattern generally supports a more limited claim. Strong research makes those limits explicit.