1) This is all part of the process, and shouldn't be too much cause for concern. Interesting results will be re-tested for confirmation. If these results can't be replicated, it casts doubt on the original study and the reputation of the authors. Negative results are rarely published except when they challenge a prior paper with a positive result.
2) The citation index is another way that bad data gets ignored. The average citation for a paper is less than 1. This means that most papers are never cited and a few good ones are cited heavily. It is true that if wrong ideas become generally accepted, they may remain in the literature unchallenged for some time and can be hard to root out.
3) I suspect that research in biomedical and other highly profitable fields bias these results, since whenever there is huge money to be made, there is pressure for a positive finding. Many of these studies are even bankrolled by the drug companies who want to show that they have the next promising cure or, better yet, treatment--since if you cure someone, you lose your market ;) It's always good to look at who's funding the work.