16 / 18 · Reason beyond certainty
Support is not entailment
Separate deduction, induction, and causal claims without dismissing empirical evidence.
Builds on Cases, contradictions, and what they prove
Go to practice ↓A question to keep in mind
A method helped every participant you observed. What can you conclude about the next participant?
Deduction aims at a conclusion that must follow if the premises hold. Induction uses observations to support claims beyond the observed cases. A strong inductive argument can remain defeasible: new evidence may rationally change the conclusion.
An association says two measurements vary together. A causal claim asks whether changing one would change the other under a suitable comparison. Selection, confounding, measurement changes, and chance can all generate associations without the proposed cause.
Random assignment can make alternative explanations less plausible, but does not guarantee perfect balance in a particular sample, correct measurement, no attrition, or generalization to everyone. Report the design, uncertainty, and population rather than simply adding the word 'proven'.
Work through an example
- Teaching example: people who chose a study app improved more than those who did not. App users also studied twice as long before joining.
- The observation supports an association in this sample. Prior habits are a rival explanation; the app's causal effect is not identified by this comparison alone.
- A randomized comparison with the same outcome measure and tracked dropouts would address some gaps. It would still estimate an effect with uncertainty, not a universal guarantee.
Your turn
0 / 3People using a focus timer finish more tasks. They chose whether to use it. Which claim is supported by this observation alone?
Read solution · does not award completion
Timer use and task completion were associated in the observed group.
Insufficient evidence for a causal claim is not evidence of the opposite causal claim.
What is one reason to randomly assign participants to two methods?
Read solution · does not award completion
It makes treatment assignment independent of participant traits by design.
Randomization is a design property, not a promise of perfect outcomes or unlimited external validity.
A well-designed study supports a likely effect, but later evidence could change the estimate. Is that a failure of reasoning?
Read solution · does not award completion
No. Revisable support is normal in empirical reasoning.
Use calibrated language: what was measured, what is supported, and what would change your mind.
Bring it back to your own work
Choose one causal headline. Write one rival explanation and a feasible comparison that would help distinguish it.
How do I retain independent judgment when using AI here?
Keep your own answer before consulting an AI candidate. Check sources, calculations, inference and conditions; explain adoption or rejection. Then solve a new-context task independently and apply the checks to this lesson.
Practice the verification workflow and keep a recordMy learning status
Mark statuses manually; visits do not establish mastery. Languages share statuses while lessons keep their own answers. This backup contains statuses only, not answers or notes.