I'll take a stab at this.

The Scientific Method, as I was taught it from middle school to college:

1. Observe a phenomenon.
2. Raise a question about said phenomenon.
3. Research the topic in question.
4. Form a hypothesis as to the nature of the phenomenon.
5. design an experiment to test that hypothesis against a control.
6. Analyze the data yielded by experiment.
7. Repeat the experiment to verify it isn't a fluke.
8. Publish all of the above in sufficient detail that other scientists may examine your work for flawed methodology and repeat your experiments to further verify it isn't a fluke.
9. Conclude whether your hypothesis is or is not supported by experimental evidence.

THIS WORKS

What is being done all over the world right now:

1. Get hired by a multinational corporation traded on the Dow Jones.
2. Be assigned a fact to prove, probably about an existing product.
3. Research the topic in question.
4. Design an experiment that will support the fact you're looking to prove.
5. Use a very small sample size.
6. Conclude something wishy-washy like "there's a statistically significant correlation".
7. Publish a densely written paper with a very convoluted title in some obscure sketchy journal somewhere.
8. Cite that paper in your own press releases with headlines that blow the conclusion way out of proportion.
9. No one ever follows up on any of this, the experiment is never really peer reviewed, or is reviewed by others engaged in similar nonsense, and the public only ever reads the headline.