Meta-analyses often need to find relevant studies whose abstracts don't report key information in standardized databases. Example: Johnson and Eagly (1990)
painstakingly reviewed over 6000 abstracts to find 162 qualifying studies that compared men and women's leadership style. Abstracts usually had something about "leadership" in the abstracts, but seldom mentioned gender. (One might even guess that those that did mention gender found larger differences, a form of reporting bias.) The only way to find the relevant studies was to obtain the reports and scan through them for reports about gender.
On the face of it, Google Scholar may seem like the solution, but it is limited to items that are on the web, and much, perhaps most, is not publicly available on the web. And there are all those multiple hits that can make searching for studies like the proverbial hunt for a needle in a haystack.
What is really needed is full-text databases that permit scholars to find relevant reports quickly and efficiently.
Here's where I'll ask your input: What are the best full-text databases? Ebsco, Science Direct, and PubMed seem like relevant candidates. However, these database hosts can only do full-text
searches of the articles to which they have full-text access (and not for those articles that are linked out to other places). And of course one must have subscription to search them. Are there better options out there?
Does anyone know of dedicated projects to create comprehensive databases of reports relevant to a research phenomenon?