In a nutshell, it is the practice of making a subject seem scientific, but skipping or masking important stuff by using common scientific terms, and also perhaps coming out with some theories and diagrams which are pushed straight to the general consumer without any review by other scientists to see how valid the work done really was.
For example, one common form of this misdirection is the use of testimonial data as proof of the usefulness of a product.
Why isn't testimonial data good data?
1) People do not understand what the placebo effect is. Our own bias in assessing or hoping for an outcome will affect our judgement of that outcome. If its a subjective outcome like "do you feel better?", all the worse.
2) How do you stop someone from picking just the good results and not showing you the bad(cherry picking)?
3) Then there is the human ego. Many folks tend to put away that 200 dollar bio-energy pendant and consider it $$ lost to bad investment, rather than come out and explain they were foolish in falling for the scheme in the first place.
Things like that make testimonials a little shaky to rely on, but they are most used by snake-oil merchants... No proper testing documentation on what it does, just fear mongering and the like.
Sometimes the best tool we have is good old common sense to look around these things.
Other times, the net helps too ;)
I found a neat article online which i would like to try to apply to local context next.