Something I posted here many years ago about this exact question, rings true still today. You have to really know what you're reading, how to read it, and what is being studied.
Like
@FridayJones outlines, many studies are biased the moment they begin, because they have selectively chosen their sample within specifics in which they hope to find their outcome. I used PTSD to demonstrate my point years ago, and will now.
An EMDR study using 250 people who have endured trauma and resulted in PTSD. When you read the study participants and they excluded complex and combat trauma because they're found to be treatment resistant, then they have just tainted the studies outcome results of effectively treating PTSD with EMDR.
When they conclude an effective treatment of 70% no longer meeting PTSD diagnosis, that isn't accurate for the treatment at all, because they excluded two of the largest groups of trauma, being complex and combat trauma. If they simply added participants based on meeting PTSD, regardless of the trauma types, then the results would be accurate, but their selection criterion really should only conclude, EMDR is effective for a 70% approximate outcome of treatment to no longer meet PTSD on mild traumatic cases that result in PTSD.
You have to read studies for what they are, and ensure you don't miss the underlying issues. The media will cherry pick pieces from a study and exacerbate the outcome, because they sell ad slots based on people watching their program. The more watching, the higher the ad price. The same thing happens here, the more topical and controversial the thread topic, the more views a thread gets, thus the more ad exposure and clickthroughs those exposed ads get, whether the content is factual or not.
The best studies are meta-analysis' as they review a whole bunch of studies in on a given topic and experts then define a conclusion based on a large cross section, including discarding studies for their invalid / biased / manipulated conclusions, thus assessing purely what could be defined empirically.
Meta analysis are the first thing I typically look for in a subject, as they will give the most accurate information.
