Correlation is a connection between two events; e.g., when two events occur together. It should be distinguished from causation, a situation when one of the events makes the other happen. When there is a causal relationship between two events, there is also a correlation, but the opposite is not always true (Goldin, 2015).
It is often possible to make the mistake of stating that there is causation based on the fact that there is a correlation when there is no causation in reality (Goldin, 2015). Another common type of mistake is confusing the cause and effect (Grinthal, 2015).
Overall, correlation does not imply causation (i.e., correlation is not a sufficient circumstance to state that there is causation); but correlation might be a sign of causation. On the whole, when events A and B are correlated, several options are possible:
- there is one-way causation between the events, i.e. A→B or B→A;
- there is two-way causation between the events, i.e. A→B and B→A;
- the events are the results of a common cause, i.e., C→A and C→B;
- the causation is mediated by another factor (i.e., A→C, C→B);
- the events are not connected, and the correlation is coincidental.
A correlation can typically be interpreted as evidence for a causal connection between two variables “if the effects are extremely notable and there is no reasonable explanation that challenges causality” (Goldin, 2015). More formally, for the statement about the existence of causation to be corroborated, three criteria need to be met (Trochim, 2006):
- temporal precedence, i.e. the cause happened before the effect (this may be hard to establish, e.g., when studying lengthy and/or cyclic processes, such as inflation and unemployment);
- covariation of the cause and effect (i.e., when one variable changes, the other also does so);
- no plausible alternative explanations (or all the plausible alternative explanations have been ruled out).
References
Goldin, R. (2015). Causation vs correlation. Web.
Grinthal, T. (2015). Correlation vs. causation. American Scientist, 103(2), 84.
Trochim, W. M. K. (2006). Establishing cause & effect. Web.