Module 29 ยท Avoiding Common Mistakes
Say what CNA actually found, and not one inch more
By the end of this module you will be able to
- Avoid over-claiming causation from one study.
- Use careful language like 'consistent with the data'.
- Report the whole bundle, not one cherry-picked condition.
The most common interpretation error is over-claiming causation from a single configurational study. CNA identifies dependency structures consistent with a causal reading. It does not, on its own, prove that one thing causes another. Keep your language honest:
| Over-claiming | Honest phrasing |
|---|---|
| "A broken cold chain causes low coverage." | "A broken cold chain appears as a difference-making condition consistently associated with low coverage in this data." |
One factor is never the whole story
Because conjunctivity is the whole premise of CNA, yanking one condition out of a bundle and crowning it "the cause" quietly contradicts the method you just spent ten modules learning. Report the bundle, not the brick.
Reflect
Check your own phrasing.
- Are you saying 'X causes Y', when 'X is part of a configuration linked to Y' is more honest?
- Have you pulled one brick out of a bundle and called it the cause?