Chapter 3 How Do We Compare?

Covariation does not prove causality. We can think about some examples of covariation without causality. For exmaple, suppose the relationship between the crime rate in South African Republic and the military spending in the U.S. The crime rate in South African Republic increases annually, and the military spending increases annually. However, can we say that the crime rate in South African Republic increases the military spending in the United States?

3.1 Five causal hurdles

We have to overcome five hurdles to argue causality.

3.1.1 A credible causal mechanism

You have to answer the “how” and “why” questions. Only if “yes,” we can move on.

3.1.2 Temporal order

The cause, \(X\), must come before the outcome, \(Y\), in time.

3.1.3 Covariation

More of \(X\) is associated with more or less of \(Y\). We can explain variables with only variables. Thus, when \(Y\) should vary, and \(X\) also should vary-they have to covary in regular patterns.

3.1.4 Endogeneity

\(Y\) might cause \(X\). However, can you answer the question that “do the strong survive or are survivors strong?”

3.1.5 Spuriousness

Some unidentified factor is responsible for the relationship between \(X\) and \(Y\) (omitted variable bias).

3.2 Concepts you should know

You should know the five huddles for causality.

• Credible causal mechanism

• Temporal order

• Covariation

• Endogeneity

• Spuriousness