Correlation, association and claims about cause
Correlation describes a relationship or pattern; however, we must keep the calculation of this pattern separate from the causal explanation of why the relationship exists.
What Pearson’s r describes
Pearson’s r measures linear association between two numerical variables from −1 to +1. Positive values mean larger values tend to occur together; negative values mean larger values of one tend to occur with smaller values of the other.
Pearson’s r is unitless. Positive changes in measurement units, such as metres to centimetres, leave it unchanged. If either variable has zero variation, correlation is undefined because its denominator is zero.
Inspect the pairs first
A scatter plot can show curves, subgroups or outliers hidden by a single coefficient. A coefficient near zero still does not rule out a strong nonlinear association.
Each pair of values must belong together. With repeated measures or time-series data, an independent-pairs significance calculation may misrepresent uncertainty. The descriptive measure and its inferential analysis are separate components of the question.
A causal claim needs more evidence
A correlation analysis is unable to establish if there was a cause-and-effect from one variable on the other or if there is some other factor that influenced both variables. Therefore, it is important to consider possible alternate explanations such as selection bias or measurement errors when evaluating a correlation.
An observed association does not, by itself, establish a direct causal relationship. A positive association could reflect age, work schedule or health differences. Adjusting for measured factors does not rule out unmeasured confounding. Description, prediction and causation are distinct aims: understanding patterns, forecasting outcomes and assessing intervention effects. Appropriate randomised designs and analyses can support causal comparisons; observational causal claims need explicit assumptions.
Worked example: direction without a causal conclusion
For five paired data points, X = 1, 2, 3, 4, 5 and Y = 2, 1, 4, 3, 5.
Pearson's r = 0.8.
Together, the pairs show a positive linear association. The numbers alone do not identify which variable affected the other, or whether an additional variable affected both.
Correlation is describing a linear relationship. To draw conclusions regarding causal relationships, one will need to provide additional evidence based on the research design and its underlying assumptions.