This is a very limited study that is getting too much press. From what I understand, she has researched "perfect use":
>Her studies found that in perfect use -- meaning the man pulls out every time -- withdrawal has a 4 percent failure rate, as compared to condoms, which have a 2 percent failure rate.
>"But nobody's perfect," said Jones, who published her commentary in the June issue of Contraception magazine.
So, if nobody's perfect, I would say we need more research before drawing conclusions one way or the other.