Back in the days when I was involved with some research, I saw some iffy work done. The pressure to come up with some result was powerful, and people kinda sorta convinced themselves that it meant something.
Here's how it worked: You spent six or twelve months accumulating data to test an hypothesis. You ran some basic statistical tests on the data, and it turned out that there was no statistical correlation to support your hypothesis. (Negative results are wrongly and rarely published.) THEN, so as not to waste your pile of data and all of the time spent (and with computers this is easy to do), you would ask the computer to find some pattern in your pile of data. Often it would find one - and that would be your published paper. Data-mining. You just re-write your hypothesis after the fact, and nobody ever knows. It's a form of lying, really.
That ain't science - that's the Texas Sharpshooter Fallacy. Attack Machine has a nice example:
Sagittarians are 38% more likely to break a leg than people of other star signs.
You can read his entertaining piece on the subject at What's Good for you Is Now Bad.
In fact, clustering does occur on a random basis, to confound scientifically- and statistically-illiterate reporters and readers. Not to mention juries.
Dr. Ioannidis of Tufts Medical School says in the WSJ:
There is an increasing concern that in modern research, false findings may be the majority or even the vast majority of published research claims," Dr. Ioannidis said. "A new claim about a research finding is more likely to be false than true."
The hotter the field of research the more likely its published findings should be viewed skeptically, he determined.
That whole article is here.
A great example of a compelling linear relationship below, from Conspiracy proving, beyond doubt, that the gradual loss of pirates is the cause of globalistical warmening.