Dude a meta analysis is just collected data and research being expanded on. How is that bogus? A meta analysis is crucial in creating and recreating experiments and studies that can be repeated or confounded upon with new information based on prior, recent and new research.
I'm glad you asked that. When you conduct an experiment, you design the test in order to reveal the data that will either support or refute your hypothesis. So for example if I wanted to study the health of smokers past 40 I might design a test to track the health of 2 groups of 40+ people, smokers and non-smokers. Now, in order to conduct my experiment I might track several variables: heart health, sex life, diet and so on, and I might gather data on my test groups for a long time, perhaps years. Studies like that are expensive, difficult, fraught with potential error points, and unfortunately, the best that can be done in many instances.
So now, later on, some broke ass phd who needs to publish research in order to keep their tenured position but has 0 research budget comes along, and wants to use my data to create a study regarding the sex life of people past 40. Seems fair enough right? Except maybe, most likely even, the data I gathered in regards to people's sex lives only has a bearing w.r.t. smoking, and doesn't apply well to people's sex lives in general. So maybe there's some significant information missing. At least, the data will be skewed, and it's fundamentally impossible to say exactly *how* it's skewed without conducting original research to answer the question at hand.
Which demonstrates the problem, but I think in practice it's actually worse, because these meta analysis' typically draw from more than one data set. So the skew gets worse, or even larger chunks of important information is missing. At the end of the day, the usefulness of the data is so diluted the whole meta-study amounts to nothing more than a bunch of hand waving, which is what we've got here. Essentially, the data becomes so meaningless any conclusion can be supported, or not. In other words, total crap science.