Kind of ballpark estimates, aren't they? "Much of the scientific literature, perhaps half, may simply be untrue..." Well that isn't "most" is it? "Perhaps?" "May be?" This is redolent of B.S.

of "false" research. Still, science is about learning new things. So of course, it is not a stagnant body of knowledge.

The point in science is constantly learning.

Science constantly challenges our assumptions.

The Bible doesn't change over time. Some scholars picked some manuscripts, put them together into a book in a certain order, and that's that. A lot of people want science to be that kind of black and white, true or false certainty.

But science is the opposite. It is a developing understanding of the universe around us.

So??? What's this thread really about?

where one is examining complex organisms with their own contigent histories and must rely on extensive statistical analysis to extract any interpretable data at all.

Most scientific research, particularly in the physical sciences, is on much firmer footing, if for no other reason than that it is much simpler to attempt to reproduce experimental results. Even in fields that are almost purely observational, rather than experimental, the connections between cause and effect are much less tangled, thus more easily observed and more easily verified.

The authors of this piece really should be taken to task for failing to make such important distinctions. The "published research reports" they are referring to are medical studies carried out using the tools and methods of science, not research in science itself. Medicine is still more art than science, and always will be, because every patient is uniquely different from every one previously studied.

In which one is examining complex organisms interacting in a complex system, controls are rare, methodology is often sloppy, changing just one thing at a time is well nigh impossible, the thinking highly "motivated" because we must "think of the children!", and the statistical analysis is formulaic without pondering degrees of freedom and reasonable default hypothesis.

In one metastudy, a researcher and his "lab" looked over 300+ ed research articles on a given subfield. Between misquoted definitions, lack of controls, changing more variables than they were accounting for, and just bad data collection and analysis they found 2 studies that passed their quality check. One found a barely statistically significant positive effect. The other found a barely statistical negative effect.

where people can't agree on basic definitions, and are basically biased because the object of the study is people, and the researchers are themselves people, which makes true objectivity impossible.

There have been complaints about the quality of research designs, hypothesis testing, etc. for many decades. At least today, we have much more access to international research, and large databases of information.

Even though there are certainly issues with funding biased research, publish-if-you-pay, and reviewers who make mistakes; overall the good quality publications are excellent. It's hard to even keep up in some narrow fields.

I'd say the big issue to me is insisting that researchers get grants or produce patents to "make money". This pushes science to the applied side, so good basic ideas may not get attention.

For college faculty, the reliance on "citation indexes" for tenure has also led to researchers playing games for survival, and that may eventually affect research reporting. "Open" journals online is also changing the game.

We live in interesting times!