Correlation is not causation: a worked verbal reasoning question about a pilot scheme

Worked solution

Correct answer: Cannot say

The principle

A comparison group that shows the same change without the intervention removes the evidence that the intervention worked, but it does not prove the intervention had no effect. 'No evidence of an effect' is not 'evidence of no effect'.

Steps

1. On the pilot wards, readmissions fell from 14% to 11% while faster letters were in use. On its own that is a before-and-after change, not proof of cause.
2. The comparison wards kept the five-day letters yet saw an equally large fall, which the passage attributes to a winter-discharge initiative rolled out to every ward, including the pilot wards. The whole pilot fall can therefore be explained without the letters, so the passage gives no support for 'True'.
3. That does not show the letters did nothing. The initiative might have worked less well on the pilot wards and the letters made up the difference, or the two sets of wards might have differed at the start. The passage never states that the letters had no effect, so 'False' is not established either.
4. The claim is neither confirmed nor contradicted by the passage, so the answer is Cannot say.

Heuristic to carry forward

When a control group changes as much as the treated group, conclude that the passage gives no evidence for the cause in question, not that it disproves it. Answer False only when the passage itself rules the claim out.

Why each other answer is wrong

True

Treats the fall that followed the scheme as proof that the scheme caused it, ignoring the comparison wards that fell by the same amount without faster letters.

False

Upgrades 'no evidence of an effect' into 'proof of no effect': the equal fall elsewhere shows the winter initiative can explain the change, but the passage never says the letters contributed nothing.