Originally Posted by roubs

Going back to your initial question, where cara mentioned looking for gaps in existing research you can really go one step further: most articles explicitly give general "future directions" which is a great place to start. Look for papers in an area that interests you and then operationalize how you would implement one of those "Future studies should look at..." statements. Ideally this would be done in the context of knowing the wider literature in that topic and fitting your idea into a gap as was mentioned.