Causal reasoning · 6/6
Write a causal explanation that could be wrong
Combine observations, mechanisms, alternatives and a verification plan.
About 15–20 minutes. The following cases are fictional.
Record your first judgment
After a library extends its hours, monthly loans rise from 800 to 1,040. School also resumes and new books arrive. Managers want immediate expansion.
Understand the judgment process
A 30% loan increase is an observation; attributing it to hours is a causal claim; expansion is a decision. Costs, access, other benefits and displacement matter to that decision.
Later hours might serve readers previously unable to visit. School demand, new books and counting changes are alternatives. Loans during new hours support a mechanism but could have displaced loans from old hours.
Compare libraries with similar trends that did not extend hours, checking calendar and purchasing differences. If feasible, randomize phased rollout and inspect cross-library movement. Prespecify incremental loans per extra service hour, costs and staff burden.
State what would weaken the claim: comparable libraries growing equally, or new-hour loans mainly replacing old-hour loans. A reversible pilot can be reasonable under uncertainty without declaring universal effectiveness.
Checks you can perform
- Separate observation, causal claim and decision goal.
- List a mechanism, two alternatives and population limits.
- Design a comparison that distinguishes explanations.
- Prespecify outcomes, harms and revision conditions.
Guided practice
Write a six-sentence review including evidence that would change your recommendation.
Explore the explanation
Loans rose 30%; the goal is affordable access; later hours might help evening readers, but school and books are alternatives; compare contemporaneous libraries and purchasing; track incremental loans, burden and timing shifts; reduce or reverse extension if little extra access accompanies high costs. Other well-supported plans are possible.
Apply it in a new context
A sports venue changes its booking interface and complaints fall, while anonymous complaints are also removed. Draft a provisional conclusion.
Compare after attempting
Recorded complaints fell; improved experience is not established. Reporting probability may have changed. Use consistent satisfaction, completion and assistance measures, and compare interfaces under the same complaint process.
Preserve a revision record
A review is not automatic causal certification. Update with evidence and separate effect size from decision value.
Check whether you supplied inspectable evidence, a feasible next step and revision conditions. Revealing an answer is not mastery.
Source and scope
Hernán & Robins — Causal Inference: What If (chapters 1–3, 7–8)
The source provides conceptual or methodological background. Cases, steps and exercises are authored here; they are not the original experiment or evidence of this course’s effectiveness.
Completion is a personal record, not proof of mastery or certification.