Originally Posted by Libertine

Just read the "study" mentioned in the article. As it turns out, the study isn't a study but another article. The article gets the 4% from a book, which presumably does get it from a study.

So this article references another article which references a book which references some unknown and untraceable study - which happens to be the one that gives a better result than any study that can actually be found.

Most studies that can actually be found give numbers that tend to range from 72%-82% - and yes, the 82% is the "more realistic" one quoted here.

On the other hand, condoms have an observed effectiveness of about 83-95%. And loe and behold, instead of choosing the top end of observed effectiveness (not the same as the 98% perfect use stat), as they did in the case of withdrawal, for condoms they choose the bottom end.

In other words, the "study" is based on an extremely biased selection of data. My guess is that the researchers in question decided to go for publicity instead of accuracy - since a study that concludes "pulling out still not effective" doesn't give much chance of getting headlines.