Well they're taking data from studies already done, but it's not just like they put a bunch of papers in a bucket and stir - if it's a smart meta-analysis (and

Well they're taking data from studies already done, but it's not just like they put a bunch of papers in a bucket and stir - if it's a smart meta-analysis (and there are other kinds) they'll be looking for effects suggested by data but not looked for in the original studies. Like if you did a meta-analysis of a hundred experimental cancer treatments that took down detailed demographics and found that people living in certain areas of the country were more likely to develop cervical cancer or something. It's definitely "original research" but it's not experimental research.