Causal reasoning · 3/6
Reverse arrows and missing people
Distinguish reverse causation from selection and define who a result covers.
About 15–20 minutes. The following cases are fictional.
Record your first judgment
A headline says stressed people use planning apps more, so apps cause stress. Another interviews only six-month users and claims all users feel better.
Understand the judgment process
Stress might encourage adoption rather than result from it. A timeline before and after adoption helps distinguish the possibilities, though temporal order alone cannot exclude common causes.
The second report excludes quitters. If dissatisfied people leave, retained users may look unusually satisfied. Track eligibility, enrollment, withdrawal and survey response, not only the remaining mean.
Whether an effect exists and whom it applies to are different questions. A result for retained users may not describe all registrants. Missing outcomes can shift a result either way; a plausible direction is not an observed fact.
Checks you can perform
- Establish the timing of exposure and outcome.
- List sample entry and exit conditions.
- Report missingness and compare known baseline characteristics.
- Limit conclusions to the covered population and examine missing outcomes.
Guided practice
Of 100 trial users, 40 leave. Of the remaining 60, 48 are satisfied. Can you report 80% satisfaction?
Explore the explanation
Yes, among the 60 retained users only. With the other 40 outcomes unknown, total satisfaction could range from 48% to 88%. This is a possible range, not an estimate or confidence interval.
Apply it in a new context
A company interviews successful founders who often worked late and recommends late nights to all founders. What is missing?
Compare after attempting
Failed founders, founders who did not work late, and other influences on success. Growing business might also cause late nights. A survivor sample cannot establish their benefit.
Preserve a revision record
Additional responses help, but responders can still differ from nonresponders. State residual uncertainty.
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.