'This programme isn't about pets," said Jolyon Jenkins at the start of Everything We Know Is Wrong (BBC Radio 4, Tuesday, 11am). "That's just a way in."

He was referring to what we'd just heard, an archive clip of a news story about how owning pets can lower blood pressure and therefore help cardiac patients.

But his real subject was the scientific research that report was based on, one of thousands conducted every year.

You know the kind of thing: red wine is good for you, mobile phones are not. In fact, with pets being the second most common cause of conflict between neighbours in America, any link between them and blood pressure is likely to be detrimental to the health of cardiac patients, at least according to anthrozoologist Hal Herzog.

Jenkins's wider point - and a source of some fascination to scientists like Stanford University's Dr John Ioannidis - is that a great deal of the published research is just plain wrong. Ioannidis even wrote a paper about it, titled Why Most Published Research Findings Are False.

And why are they mostly false? Because the most important factor in any research conclusions - that the results they're based on be reproducable - often doesn't stand up to scrutiny.

It's known in the trade as "the decline effect": the more people study something, the less true it seems.

Take mice, for instance, the animal of choice for many scientific experiments.

When Oregon-based neuroscientist John Crabbe commissioned the same experiment on the same strain of mice at the same time and under the same conditions at laboratories in Portland, New York and Edmonton, the results varied.

His best guess? The mice reacted differently to the people handling them. Pheromones were at the root of it, he thought.

Another problem is with the academic industry itself, which needs splashy, headline-grabbing results. So they divide funding into lots of small pots, but that means each study is necessarily small - and the smaller the study, the greater the chance of errors.

People have studied this, too, though who's to say their findings are right? "It's like seeing faces in shadows," said Jenkins, wonderingly.

"It's a very good analogy," Ioannidis replied. Worrying stuff that should make us ask far more questions about the ethics of science.