Non-Experimental Methods and Correlation
Beyond the Experiment
Experiments are powerful tools for establishing causation, but they are not always possible or appropriate. Some questions in psychology cannot be answered experimentally — it would be unethical to manipulate certain variables, or the topic is too complex to reduce to a single controlled setting. For these situations, psychologists use non-experimental methods: interviews, questionnaires, case studies and observation studies. Each has its own strengths and weaknesses, and each is suited to different kinds of research questions.
Quantitative and Qualitative Methods
Non-experimental methods can produce either quantitative data (numerical results, such as scores, frequencies or percentages) or qualitative data (detailed descriptions of experience, such as what participants say or how they behave). Most methods can produce either, depending on how they are used — a questionnaire can include both numerical rating scales (quantitative) and open-ended questions inviting longer responses (qualitative).
An interview involves a researcher asking questions and recording the participant’s responses. Interviews can be structured (every participant asked the same predetermined questions in the same order, producing quantitative data) or unstructured (open conversation following the participant’s responses, producing rich qualitative data).
Strength of interviews: they can explore complex experiences in depth; unstructured interviews allow unexpected themes to emerge. Weakness: interviewer bias — the researcher’s manner, appearance or phrasing can influence participant responses; time-consuming and difficult to analyse consistently, especially unstructured data.
A questionnaire is a set of written questions completed by participants, often without the researcher present. Questions can be closed (ticking a box, rating on a scale — produces quantitative data) or open (written response — produces qualitative data).
Strength: large numbers of participants can be reached cheaply and quickly; no interviewer bias if completed independently; easy to analyse if questions are closed. Weakness: participants may not answer honestly (social desirability bias — giving the answer they think they should give rather than their true answer); ambiguous questions can be interpreted differently; low response rates may produce a biased sample.
A case study is an in-depth investigation of a single individual, group or event. It typically uses a combination of methods (interviews, observation, psychometric tests) to build a detailed picture of the case. Case studies are particularly useful when the case is unusual or extreme — for example, studying an individual with a rare neurological condition to understand normal brain function by contrast.
Strength: provides rich, detailed qualitative data that captures complexity; unique cases can generate hypotheses for further research; can study variables that could not be manipulated experimentally. Weakness: findings cannot be generalised to other people because the case may be unique; the researcher may develop a subjective attachment to the participant; results are not replicable.
In an observation study, the researcher watches and records participants’ behaviour without manipulating any variables. Observations can be naturalistic (watching behaviour in its natural setting) or structured (in a controlled setting with predetermined categories to record).
A key technical requirement of observation studies is using categories of behaviour — a predetermined list of specific, clearly defined behaviours to record (e.g. ‘number of times participant initiates conversation’ rather than ‘social behaviour’). This makes data more objective and consistent.
Interobserver reliability is a crucial check: two or more observers independently record the same behaviour, and their records are compared. High agreement indicates the observation categories are clear and consistently applicable. Strength: naturalistic observations have high ecological validity. Weakness: observer bias — what one person notices may differ from another; participants who know they are being watched may behave differently (Hawthorne effect).
Correlation
A correlation measures the association between two variables. It tells us whether, as one variable increases, the other tends to increase (positive correlation), decrease (negative correlation) or show no consistent pattern (no correlation).
Correlations are displayed using scatter diagrams. Each participant provides two scores (one for each variable), which are plotted as a single dot on the graph. A positive correlation produces dots trending from bottom-left to top-right. A negative correlation produces dots trending from top-left to bottom-right. A weak or no correlation produces a scattered cluster with no clear trend.
A correlation shows association, not causation. Two variables may be strongly correlated because a third, unmeasured variable causes both. This is called a third variable problem. For example, ice cream sales and drowning rates both increase in summer, but ice cream does not cause drowning — hot weather (the third variable) drives both.
Strengths and Weaknesses of Correlations
Strength: correlations allow researchers to study the relationship between variables that cannot be manipulated ethically (e.g. the association between stress and illness). They are also useful as a starting point for identifying patterns that may be worth investigating with a controlled experiment.
Weakness: correlations cannot establish causation. Because neither variable is manipulated, we cannot conclude that one variable causes the other. This is the most important limitation of correlational research.
Non-experimental methods summary: interviews (verbal, structured or unstructured), questionnaires (written, closed or open), case studies (in-depth single case), observations (behaviour categories, interobserver reliability checked). None of these establish causation — only a controlled experiment can do that.