There's a lot of good information here. Short summary/excerpt on next pages, and I recommend watching the video if you are interested. Total of 8 pages on this card.
The "chocolate reduces weight" study was an intentionally badly designed study to show how poor the quality of science journalism can get. And you probably already know the results.
Another important thing is that being "statistically significant" doesn't necessarily mean the result is important. It is not so much the measure of impact, but rather certainty.
It is also important to check the sources, but it can be difficult to determine because research can be expensive, so that only non-neutral organizations would fund the research.
Clickbaits are also a threat to science journalism, and journalism in general. The publishers must compete to get views, which means the headlines must be catchy, even if it is less accurate or informative.
"Correlation not causation" is thrown out a lot, but often it gets overlooked that sometimes correlation can mean a causal link, when the studies are designed properly, such as by using random samples and control groups, among other things.
Sometimes, the study is done in vitro (in non-living environment), which means that the test may or may not be applicable in vivo (in living body). Hydrogen peroxide may kill cancer cells in vitro, but it also kills healthy cells in vivo.
Here's the link again, so you don't have to go back to the first slide.