Correlation vs. Causation

A common reasoning error is mistaking correlation for causation.

Correlation

Tells us the relationship between two or more variables, can describe relationships but NOT explain them (no manupulation or control)

What are the strengths of correlation?

Preliminary

Simple (record what exists)

Identify relationships of interest

Investigate variables difficult to manipulate

High external validity

What are the weaknesses of correlation?

Not clear

Unabiguous explanation

Low internal validity

What are the limitations of correlation?

3rd variable problem (a + b, NAH, influenced by x)

Directionality problem (what variable influences which)

Illusory Correlation

The tendency to overestimate the link between variables that are only slightly, or not at all related

Causation

Relationships that show association or relationships

What are the criteria for causation?

Covariation (very together)

Time-order (cause first, effect after)

Plausible alternative explainations (controls for extraneous & confounding variables)

What are the common errors in causal reasoning?

Third variable problem

Directionality problem

Post hoc fallacy

Third Variable Problem

Illusory correlation, two things tend to change at the same time, but not together because a third thing is impacting both of them

Directionality Problem

You don’t know which variable is leading to the other

Post Hoc Fallacy

After this, therefore because of this; concluding that a second event was caused by a first event when it only followed it

Why can correlation not always be used to infer causation?

A statistical association does not prove one causes the other

Relationships may be coincidental, caused by a third variable, bring about a directionality problem, or a post hoc fallacy

Only shows that variables move together, not that one directly causes the other

What is the third variable problem & provide an example

This problem is an illusory correlation where two things tend to change at the same time, but not togtehr because a third thing is impacting both of them

An example of this could be the correlation between ice cream sales & drowning

These two variables are impacted by the hot weather, not each other