Back to lesson

CAUSAL ARGUMENTS

Cheat Sheet

START HERE

Correlation and causation

CORRELATION

Two things tend to show up together. For example: people eat more ice cream during summer. Summer and ice cream are correlated.

CAUSATION

One thing actually produces the other. The cause brings the effect about.

THE CORE PROBLEM

Correlation is not causation

The problem that shows up most often: correlation is not causation.

Two things happening together doesn't mean one caused the other. In our example above, just because summer and ice cream are correlated, happening together, doesn't mean summer caused the increase in ice cream usage. Maybe it's being out of school that causes kids to have the time to eat more ice cream (alternate cause).

SPOT THE LANGUAGE

Correlation words vs causation words

CORRELATION (usually the premise)

associated withlinked toconnected to more common intends to occur withfound together with

CAUSATION (usually the conclusion)

caused byleads toresults in due toresponsible forproduces / makes
THE FRAMEWORK

What supports a causal claim

1

Correlation

Two things happen together in a predictable pattern.

2

Timeline

The cause comes before the effect.

3

Mechanism

A plausible pathway from cause to effect.

4

Block alternate causes

Rule out other possible explanations.

STRENGTHEN vs WEAKEN

Same four pillars, opposite directions

STRENGTHEN
WEAKEN
Correlation
The same pattern shows up elsewhere.
No additional correlation
The correlation doesn't appear anywhere else.
Timeline (cause first)
The cause comes before the effect.
Timeline reversed
The effect came before the cause.
Mechanism
A plausible pathway from cause to effect.
No mechanism
The cause couldn't have produced the effect.
Block alternate causes
Other possible causes are ruled out.
MOST COMMON STRENGTHENER
Alternate cause
Something else could explain the effect.
MOST COMMON WEAKENER
—
Reverse causation
B might be causing A instead of A causing B.
NECESSARY ASSUMPTION

What the argument assumes

  • Correlation. The pattern would replicate.
  • Timeline. The cause came before the effect.
  • Mechanism. There's a plausible pathway.
  • No alternate cause. Nothing else explains it.

NA assumes the strengthen side. Negate any one and the argument collapses.

FLAW

What the argument missed

  • Correlation as causation. The umbrella flaw on causal arguments.
  • Timeline reversed. Fails to consider the effect came first.
  • Alternate cause. Fails to consider another explanation.
  • Reverse causation. Fails to consider B might cause A.

These are the trap-answer phrasings on causal Flaw questions.

WATCH OUT

Outliers don't weaken

  • Causal claims are statistical, not conditional.
  • "Some smokers don't get cancer" doesn't weaken "smoking causes cancer."
  • Anchor language: tends to, generally, usually, increases the likelihood.
  • Trap on Weaken: "Person X did the cause but didn't get the effect." Usually wrong.
ONE MORE THING

Chaining causal claims

If A causes B, and B causes C, then A causes C.

Smoking causes tar in the lungs. Tar in the lungs causes cancer. So smoking causes cancer.

Caveat: each link has to hold. The chain is only as strong as its weakest link.

RECAP

Test every causal claim against four pillars: correlation, timeline, mechanism, and blocked alternate causes. Strengthen shores one up. Weaken breaks one (or reverses the causation). Outliers don't weaken, because causal claims are about tendencies, not absolutes.