How does “no detectable increase versus control, in our test data” not prove there is no connection (errors in the study aside). And why does that not prove any

How does “no detectable increase versus control, in our test data” not prove there is no connection (errors in the study aside). And why does that not prove anything, but "yes detectable increase versus control, in our test data” does? Because, sure there can be errors either way. But a study produces new knowledge, not just knowledge or "just as much a mystery"