It's easier to get a desired result in fields like psychology and educational sciences, since it's so hard to control the experimental setting. To add insult to injury, this makes that even when done correctly, studies often don't generalize.