One study says red wine is good for you, another says it's not. Medical research is often bewildering.
Scarce funding, competition and the pressure to publish are leading to sloppy science, according to science reporter Richard Harris.
Scientists are often faced with a terrible choice between doing what needs to be done to keep their labs running and doing what's right, he says.
Harris investigates the problem in his new book, Rigor Mortis: How Sloppy Science Creates Worthless Cures, Crushes Hope, and Wastes Billions
Harris says he expected scientists wouldn't be keen to talk about these issues, but many wanted to share their stories.
"They're very concerned about this and these are widely held beliefs in science."
Currently, scientists in the United States are rewarded for the volume they publish and for "flashy results", he says.
Bad ingredients, poorly designed studies, poor statistical analysis and the pressure of a hyper-competitive environment are also sending people off in the wrong direction.
Only one out of every five scientific grant proposals in the United States gets funded and often those with the flashiest ideas succeed, Harris says.
"So if you come along and say I'm very curious to know if this seemingly important piece of research is reproducible, they'll say 'Do it on your own time'.
"It's better to be exciting than to be right."
A lot of work can be based on an incorrect study before its findings are revealed to be false, he says.
In 1999, a study found that if you took cells out of bone marrow, you could essentially convert these into any kind of cell you wanted to – "a fabulous discovery".
A lot of scientists "jumped on that bandwagon" and produced results and papers using the technique described. But several years later, scientists at Stanford University pointed out that no-one had ever tested the initial study.
After a some careful experiments, including stitching two mice together, it was discovered that actually no cell change was no happening.
"It took several years to sort out, a huge amount of excitement and as it turned out a huge amount of wasted effort."
Clinical scientists are supposed to write down in advance what they expect to find, but if they find something unexpected, this is often itself treated as evidence, rather than a new hypothesis.
And medical and science journals don't often enforce that rule, he says.
Drug companies and universities are supposed to be open about their failures and declare their results, but this often doesn't happen and contributes to the slowness of scientific research.
"Sometimes there are competitive reasons for companies not to tell people where the dead ends are. The system is set up so they're supposed to, but, in particular, scientists at universities are not very good at following up on it."
Failure is an essential part of science, he says.
"Even if you're the world's best scientist, most of the ideas you try will fail in the lab … Most ideas don't pan out."
The question is how many of these failures can be avoided.
"If you're gonna fail, fail for the right reasons."