✦ScienexySCIENCE KNOWLEDGE LIBRARY
learning-resource · SCI-PSY-002-115

Correlation and Causation

Explains how correlation describes relationships between variables, why association does not automatically establish causation, and how research designs investigate causal relationships.

Explanation

A fuller explanation is being developed. Verify this concept with cited sources before relying on it.

Example or application

A worked example has not yet been added.

Conceptual Coverage

Correlation

Correlation is a statistical relationship between variables. It describes whether values tend to vary together and, for a specified correlation measure, the direction and strength of that association.

Causation

Causation means that changes in one factor produce changes in another under specified conditions. Establishing causation requires evidence beyond a simple association.

Types Of Correlation

  • Positive correlation: higher values of one variable tend to occur with higher values of another.

  • Negative correlation: higher values of one variable tend to occur with lower values of another.

  • Zero or near-zero linear correlation: no meaningful linear association is detected, although a nonlinear relationship may still exist.

Causal Requirements

  • Association: the proposed cause and outcome show an appropriate relationship.

  • Temporal precedence: the proposed cause occurs before the outcome.

  • Alternative explanations: plausible confounding factors are considered and addressed.

  • Appropriate design: the research method supports the strength of the causal claim.

Common Causal Patterns

  • Direct causal influence

  • Indirect influence through a mediator

  • Confounding by a third variable

  • Bidirectional relationships

  • Spurious association

  • Nonlinear relationships

Scientific Foundation

Scientific Basis

Correlation and causal inference are central to psychological research because many psychological variables are related without being experimentally manipulated.

Historical Development

Statistical correlation developed through the work of researchers including Francis Galton and Karl Pearson. Modern causal inference uses experimental design, statistical modelling, and explicit assumptions about possible causal relationships.

Psychological Example

A study may find that stress scores and sleep difficulties are associated. This does not establish whether stress affects sleep, sleep difficulties increase stress, or another factor influences both.

Research Designs

  • Randomized experiments can provide strong evidence about causal effects when appropriately conducted.

  • Longitudinal studies establish temporal patterns but do not automatically eliminate confounding.

  • Natural experiments and quasi-experimental designs can investigate causal questions when random assignment is not possible.

  • Cross-sectional correlational studies describe associations at a particular time.

Deep Understanding

Why Correlation Is Not Causation

Two variables may be associated because one influences the other, because the second influences the first, because both are influenced by another factor, or because the association is coincidental or measurement-related.

Three Classic Problems

  • Directionality problem: it may be unclear which variable influences the other.

  • Third-variable problem: another factor may explain the association.

  • Spurious correlation: an association may appear without a meaningful direct relationship.

Worked Example

Observation

People who spend more time studying may also obtain higher examination scores.

Possible Explanation One

More study time may improve learning and performance.

Possible Explanation Two

Students with stronger prior knowledge may both study more effectively and obtain higher scores.

Possible Explanation Three

Access to resources or other environmental factors may influence both study patterns and performance.

Conclusion

The association alone cannot determine which explanation is correct.

Causal Reasoning

Researchers should specify a causal question, identify possible confounders, establish the time order, select an appropriate design, and evaluate whether the assumptions needed for causal interpretation are reasonable.

Mathematical Coverage

Pearson Correlation

Formula

r = Σ[(xᵢ − x̄)(yᵢ − ȳ)] ÷ √[Σ(xᵢ − x̄)² × Σ(yᵢ − ȳ)²]

Variables

R

Pearson correlation coefficient.

Xᵢ

An individual value of variable X.

Yᵢ

The corresponding individual value of variable Y.

X̄

Mean of variable X.

Ȳ

Mean of variable Y.

Σ

Summation across observations.

Range

The Pearson correlation coefficient ranges from −1 to +1.

Interpretation

The sign indicates the direction of a linear relationship, while the absolute magnitude indicates its strength in the sample.

Worked Example

Dataset

Study HoursScore

1

50

2

60

3

70

4

80

Result

The four observations form a perfectly positive linear pattern, so Pearson's r = +1.

Interpretation

This small illustrative dataset shows a perfect positive linear association. It does not establish that study hours alone caused the scores.

Additional Methods

  • Spearman's rank correlation assesses monotonic relationships using ranked data.

  • Partial correlation examines association while statistically accounting for selected variables.

  • Regression analysis estimates relationships between predictors and outcomes.

  • Causal diagrams can clarify assumed causal pathways and potential confounders.

Caution

Correlation coefficients can be strongly affected by outliers, restricted ranges, measurement error, and nonlinear patterns.

Practical Learning

Applications

  • Studying relationships between stress and sleep

  • Investigating study habits and academic performance

  • Examining social support and well-being

  • Analyzing attention and task performance

  • Evaluating relationships between psychological measures

Example

A researcher investigates whether self-reported stress is associated with sleep duration. The study can describe the association, but causal conclusions require additional evidence and an appropriate design.

Exercise

Create three examples of positive and negative correlations in psychology. For each example, identify one possible third variable that could influence both variables.

Critical Thinking

Limitations

  • Correlation does not establish causal direction.

  • Unmeasured confounders may remain.

  • Measurement error can weaken or distort observed relationships.

  • Outliers may substantially influence Pearson's correlation.

  • A zero linear correlation does not rule out nonlinear relationships.

Common Misconceptions

  • A strong correlation does not automatically mean causation.

  • A weak correlation may still be scientifically or practically meaningful.

  • A negative correlation does not mean that one variable is harmful or undesirable.

  • Correlation does not necessarily imply a direct relationship.

Evaluation

Interpret correlations by examining the scatterplot, sample size, measurement quality, confidence interval, potential confounders, and research design.

Knowledge Assessment

Questions

QuestionAnswerExplanation

What does a positive correlation indicate?

Higher values of one variable tend to occur with higher values of another.

The variables tend to move in the same direction.

What is the directionality problem?

The difficulty of determining which variable influences the other.

An observed association does not identify causal direction.

If Pearson's r is −0.90, what does the sign indicate?

A strong negative linear association.

The negative sign indicates opposite directions of variation, while the magnitude indicates a strong linear relationship.

Application Exercise

A study finds a positive correlation between social-media use and stress. Provide three possible explanations, identify a potential confounder, and explain what additional research could investigate causation.

Further Learning

Prerequisites

  • Variables in Psychological Science

  • Basic descriptive statistics

  • Research Design

Related Topics

  • Independent, Dependent and Control Variables

  • Regression Analysis

  • Confounding Variables

  • Experimental Research

  • Causal Inference

Advanced Directions

Study causal diagrams, longitudinal modelling, quasi-experimental methods, randomized trials, and modern causal inference.

Sources and verification

Sources are shown as provided in the project data. Not every source has yet been independently verified. Please report errors or missing citations.

  • Source verification pending; check an authoritative textbook, university, standards body, or peer-reviewed source.

Suggest a correction or source · More in Foundations of Psychology