Judging Evidence: Relevant, Sound, Sufficient

A claim is a sentence someone wants you to believe. Evidence sits underneath it, and the link between them can snap in three different places even when every word on the page is true. Relevance asks whether the evidence is even about the same thing as the claim. Soundness asks whether the evidence itself is solid: accurate, logically put together, not undercut by its own numbers. Sufficiency asks whether the evidence, even if relevant and sound, covers enough ground to carry the whole claim. A passage can satisfy two of these and still fail the third, and that is almost always where the trap is built.

The most common failure is relevance dressed up as support: a true fact about durability sitting under a claim about speed of payback, or a true fact about one town sitting under a claim about a nation. The fact survives fact-checking. It just answers a different question than the one the claim asked. The second common failure is sufficiency: scope, sample size, or time span too narrow to stand for the bigger thing the claim names. More detail about Ashford does not make Ashford a better stand-in for the whole country; it just makes you more certain about Ashford.

This sits right next to counterclaims and causal reasoning, because both are really sufficiency problems in disguise. A counterclaim often works by finding a case the original evidence does not cover. A causation claim resting on a correlation fails sufficiency in a specific way: the evidence is relevant and the numbers may be sound, but a correlation alone never rules out a third cause behind both things, so it cannot carry a causal claim by itself.