In biology and some parts of chemistry, I would say yes 80% of papers have a wrong conclusion (experiments are ok but not interpreted correctly) or completely wrong. But in other fields it is not the case. I'm pretty sure mathematics are much better. Physics it will depend on which branch some are closer to science-fiction anyway so are not verifiable, but most physics is good quality. It is hard to consider science as a whole because it is absolutely not an homogenous lump and practices in reviewing, publishing and peer pressure are radically different.