Causality is a relationship in which a change in one factor produces a change in another. In psychology, showing that two things occur together is different from showing that changing one helps bring about the other.

The guide gives related terms for this concept, rather than separate assessable content statements. Use these terms to explain and evaluate causal claims. They aren't a checklist of separate theories to learn.

Agency and motivation

Agency is the capacity of a person to make choices and act intentionally. People do not simply receive biological or environmental influences passively. They can make plans, choose between alternatives and try to change their circumstances. Their choices still face constraints.

Motivation is a psychological process that initiates, directs and sustains goal-oriented behaviour. Two people in a similar situation may behave differently because they have different goals or place different importance on the same outcome.

Imagine two students offered an optional practice session. One attends to improve a skill; the other declines to keep time free for another commitment. To explain their behaviour, we need to consider their goals and choices, not just whether the session was available. But identifying a goal does not establish that it caused the behaviour. A person's explanation of their own action is evidence to consider, not automatic proof.

Influence, interaction and complexity

An influence is a factor that contributes to shaping an outcome. It need not determine what happens in every case. Saying that tiredness influences concentration makes a more limited claim than saying that tiredness inevitably causes poor concentration.

An interaction is a relationship between factors in which the effect of one depends on another. Background noise, for example, might impair performance more strongly on a demanding task than on a simple one. It isn't just that noise and task difficulty both matter. The effect of noise changes with task difficulty.

Complexity is a property of a system in which multiple connected factors contribute to its outcomes. Psychological behaviour often reflects biological, cognitive and sociocultural influences acting together. Over time, some of these influences may feed back on one another. Difficulty concentrating might lead to frustration, which then makes concentration harder.

A useful causal explanation identifies the contribution being claimed without pretending that one factor explains the whole person. A cause can make an outcome more likely without being sufficient to produce it alone. Nor does it have to be necessary in every case.

Correlation is not causation

Correlation is an association between variables in which their values vary together. Establishing an association does not, by itself, establish a causal relationship.

Bidirectional ambiguity is uncertainty about the direction of a possible causal relationship between associated variables. Knowing that stress and sleep disruption are related does not tell us which influences which. Both directions might operate, but the association alone demonstrates neither.

To reason about causes, we need evidence that helps distinguish these alternatives. Establishing that a proposed cause comes before an outcome helps with direction. Controlling alternative explanations helps establish whether that proposed cause actually contributes to the outcome. A confident-sounding description of the correlation cannot replace either step.

Controls and alternative explanations

A control is a research procedure or comparison that limits or assesses alternative explanations for a result. Controls strengthen causal reasoning: they make it harder for something other than the intended factor to account for an observed difference.

Extraneous variables are variables other than the intended explanatory factor that could affect the measured outcome. Suppose groups completing a concentration task receive different instructions as well as different noise conditions. The instructions then offer an alternative explanation. Keeping them consistent helps isolate the contribution of noise. A control must address a plausible competing explanation, not just make the procedure look tidy.

A placebo is an inactive treatment or procedure designed to resemble the intervention being investigated. Comparing an intervention with a placebo helps distinguish the intervention's own effects from those associated with receiving treatment and expecting improvement. It does not guarantee identical expectations, or remove every alternative explanation.

A double-blind procedure is a research arrangement in which neither participants nor the researchers interacting with them or assessing their outcomes know the participants' assigned conditions. This gives expectations less opportunity to influence behaviour, treatment or assessment. Blinding may be difficult when an intervention is obvious to those receiving it.

Wait-listing is a control arrangement in which participants receive an intervention after a delay, while their earlier outcomes provide a comparison with participants receiving it immediately. Access is delayed rather than permanently denied. Waiting, though, is not equivalent to receiving a convincing placebo. Besides immediate access to the intervention, the groups may differ in their expectations, the attention they receive and their experiences.

Internal and external validity

Internal validity is the extent to which a study supports a causal conclusion without plausible alternative explanations accounting for its findings. Effective controls strengthen internal validity. When several relevant features differ between conditions, it is less convincing to attribute the outcome to just one.

External validity is the extent to which findings can be generalized beyond the people, settings and circumstances studied. A carefully controlled concentration task may support a causal conclusion within that study. It may tell us less about concentration in a busy workplace.

These judgements are different. A finding can be convincing within a study but have limited wider application. Conversely, studying behaviour in an everyday setting does not automatically establish what caused it. Strong causal evidence needs justification, as does broad generalizability; neither should be assumed from the other.

Statistical significance

Statistical significance is a judgement that a result meets a specified criterion for incompatibility with a statistical model that assumes no effect or association. It concerns how surprising the evidence would be under that model. It does not prove that a proposed explanation is true.

Even a statistically significant correlation leaves the direction of influence and alternative explanations unresolved. A statistically significant difference between conditions must also be considered alongside the study's controls and validity. Statistical significance alone tells us neither that an effect is large nor that it matters in everyday life.

Reductionism and causal explanation

Reductionism is an explanatory approach that accounts for a complex phenomenon in terms of simpler components or a narrower level of explanation. Explaining concentration only through a biological process, for example, isolates one level instead of considering goals, interpretations and social circumstances together.

Narrowing an explanation can be useful: it makes a proposed causal contribution easier to investigate precisely. Reductionism becomes limited when a partial explanation is treated as complete. Evidence that one process contributes to behaviour does not show that agency, motivation or interacting influences are irrelevant. A causal claim is only as strong as what the evidence isolates—and whether the conclusion stays within those limits.