Causal Reasoning & Correlation vs. Causation

What You Need to Know

Causal reasoning is everywhere on the LSAT: arguments that claim X causes Y, or that changing X will change Y. The test loves exploiting the gap between:

• Correlation: X and Y occur together.
• Causation: X produces Y.

Core rule: A correlation (or sequence in time) does not by itself prove causation.

The basic causal claim forms you’ll see

• Direct causal claim: “X causes Y.”
• Causal explanation from an effect: “Y happened, so X must have caused it.”
• Causal prediction/control: “If we do X, Y will (increase/decrease).”
• Causal comparison: “Group A has more Y than group B, so A has more X (the cause).”

Why it matters

Most causal questions are really about spotting what’s missing: other ways the evidence could be true without the claimed causal relationship being true.

Critical reminder: To justify a causal conclusion, you generally must rule out the main alternatives: reverse causation, common cause, and coincidence/selection effects.

Step-by-Step Breakdown

Use this same routine for Strengthen, Weaken, Flaw, Necessary Assumption, and Evaluate questions involving causation.

1) Translate the argument into “cause → effect” language

• Identify the cause candidate (what’s supposedly doing the causing).
• Identify the effect (what changed / differs between groups).
• Identify the evidence type (correlation, before/after, group comparison, testimonial, etc.).

Mini-annotation example

• Claim: “Using standing desks reduces back pain.”
• Cause candidate: standing desks
• Effect: less back pain
• Evidence: employees with standing desks report less pain (correlation)

2) Ask: what’s the evidence, logically?

Most LSAT causal evidence is one of these:

• Correlation: X and Y go together.
• Temporal sequence: X happened before Y.
• Group difference: Group with X has more/less Y.

None of these automatically proves “X → Y.”

3) Run the “3 classic alternative explanations” checklist

For a causal conclusion based on correlation/group difference, ask:

1. Reverse causation: Could Y cause X?
2. Common cause (confounder): Could Z cause both X and Y?
3. Coincidence / selection / measurement: Could the relationship be an artifact of who got counted or how measured?

4) Match answer choices to the gap

• Strengthen: rules out alternatives, adds mechanism, shows the effect changes when you manipulate the cause, or shows no difference when the cause is absent.
• Weaken: introduces a plausible alternative cause, shows reverse causation is likely, shows the effect occurs without the cause, or shows the cause occurs without the effect.
• Necessary assumption: often “no alternative explanation of type X is responsible.” If that assumption fails, the conclusion collapses.
• Evaluate: asks for a fact that would help decide between the causal claim and an alternative explanation.

5) Use “control group / hold constant” thinking

You’re not doing real science, but the logic is similar:

• If we hold everything else constant and only change X, then a change in Y supports causation.
• If Y changes while X doesn’t, or X changes while Y doesn’t, causation is in trouble.

Exam mindset: The LSAT doesn’t require you to prove the true cause; you just need to see whether the argument earns its causal leap.

Key Formulas, Rules & Facts

Correlation vs. causation (high-yield distinctions)

The “usual suspects” in causal flaws

What strong causal support looks like (in answer choices)

Examples & Applications

Example 1: Classic correlation → causation (Flaw / Weaken)

Stimulus (pattern): “Cities with more police officers have higher crime rates. Therefore, adding police officers increases crime.”

Key insight: This screams reverse causation (high crime causes more police) or common cause (big cities cause both).

Weaken answers you’d love:

• “Police staffing levels are increased in response to rising crime.” (reverse causation)
• “Larger cities have both more officers and more crime regardless of changes in staffing.” (common cause)

Example 2: Before/after policy (Strengthen / Evaluate)

Stimulus (pattern): “After the school introduced uniforms, disciplinary incidents fell. Therefore, uniforms caused the reduction.”

Key insight: Timing alone isn’t enough; other changes might have happened.

Strengthen:

• “No other disciplinary policies changed during that period.” (rules out confounders)
• “A similar school that did not adopt uniforms saw no decline.” (comparison/control)

Evaluate question angle:

• “Did the method of recording incidents change when uniforms were introduced?” (measurement shift)

Example 3: Self-selection (Weaken)

Stimulus (pattern): “People who take yoga classes report less stress. Thus yoga reduces stress.”

Key insight: Selection bias: less-stressed people (or more health-focused people) may be more likely to take yoga.

Weaken:

• “People with lower baseline stress are more likely to enroll in yoga.”

Strengthen (best kind):

• “Participants were randomly assigned to yoga vs. no yoga, and the yoga group’s stress decreased more.”

Example 4: Causal prediction (Necessary assumption)

Stimulus (pattern): “To reduce traffic accidents, the city should lower the speed limit, because high speed causes more severe accidents.”

What’s being assumed (often necessary):

• Lowering the posted speed limit will actually lower driving speeds (a key link between policy and behavior).

How LSAT tests it: A necessary assumption choice often plugs a missing link like “drivers will comply” or “enforcement will occur.” If that link fails, the recommendation falls apart.

Common Mistakes & Traps

1. Mistake: Treating correlation as proof
   What you do: See “X and Y are associated” and accept “X causes Y.”
   Why wrong: The same data fits reverse causation, common cause, or selection bias.
   Fix: Automatically generate at least one plausible Z and a reverse direction.

2. Mistake: Forgetting reverse causation is often the easiest weaken
   What you do: Hunt for complicated alternatives when the stimulus basically hands you “Y → X.”
   Why wrong: Many LSAT stimuli are written so reverse causation is the cleanest gap.
   Fix: When you see correlation/group difference, ask: “Could the effect be driving the cause?”

3. Mistake: Confusing “evidence consistent with” with “evidence proves”
   What you do: Treat supportive facts (like temporal order) as conclusive.
   Why wrong: Being consistent is a low bar; many stories are consistent with the same facts.
   Fix: Look for answer choices that exclude alternatives, not merely restate the correlation.

4. Mistake: Over-demanding scientific certainty on Strengthen
   What you do: Reject good strengthen answers because they don’t prove causation.
   Why wrong: Strengthen only needs to make the conclusion more likely.
   Fix: Prefer answers that attack the biggest gap (often the most plausible alternative cause).

5. Mistake: Missing “measurement change” and “definition shift”
   What you do: Accept “Y increased” without checking whether Y was measured the same way before/after.
   Why wrong: A reporting change can mimic a real change.
   Fix: On before/after claims, ask: “Did anything about how they counted change?”

6. Mistake: Ignoring the denominator / rate vs. raw number
   What you do: Buy causal claims from raw totals (e.g., “more accidents”) without considering population or exposure changes.
   Why wrong: More drivers can raise total accidents even if risk per driver fell.
   Fix: Check whether the argument should be about a rate (per capita, per mile, per customer).

7. Mistake: Mixing up what the conclusion actually claims
   What you do: Attack the wrong link (e.g., mechanism) when the conclusion is a policy recommendation requiring compliance.
   Why wrong: The vulnerable point may be implementation, not biology/psychology.
   Fix: Restate the conclusion precisely: “Do X and Y will happen,” then ask what must be true for that to work.

8. Mistake: Falling for “causal overreach” from limited samples
   What you do: Accept broad causal generalizations from one city, one school, one month.
   Why wrong: Small/atypical samples increase coincidence and hidden-variable risk.
   Fix: Weaken with “this sample is unrepresentative” or “other factors unique to this case.”

Memory Aids & Quick Tricks

Warning: Don’t assume “mechanism” is always required. On the LSAT, eliminating a confounder often strengthens more than adding a plausible story.

Quick Review Checklist

• You can instantly label the claim as correlation vs causation.
• You can restate the argument as X → Y and spot what evidence is actually given.
• You automatically check RCC: Reverse causation, Common cause, Chance/selection/measurement.
• For Strengthen, you look for: ruling out confounders, control groups, random assignment, dose-response, cause/effect dependency.
• For Weaken, you look for: alternative cause, reverse direction, effect without cause, cause without effect, measurement change.
• For Necessary assumption in causal/policy arguments, you find the missing link that must hold for the plan to work.
• You watch for rates vs totals and before/after counting changes.

One clean causal checklist used consistently is worth more than trying to “feel” which answer sounds scientific—stick to the logic and you’ll be fine.