You can never test a single scientific claim completely on its own
Any real experiment doesn't just test the hypothesis you're interested in — it also silently relies on your instruments working correctly and your background assumptions holding up. So when a prediction fails, logic alone can't tell you whether your hypothesis is wrong or one of those unstated assumptions is. Pierre Duhem noticed this for physics in 1906; W.V.O. Quine extended it to knowledge in general in 1951.