For the record, not all scientists accept the research regarding birth control and mate preference as sound science. There's not even good evidence that humans can detect (even unconsciously) the MHC genotypes of another person (most of the papers don't show a statistically significant effect), which the referenced paper takes as one if its premises.

In addition, in the paper referred to in the article, there isn't a lot of difference between those on the pill and off (the error bars tended to overlap):

http://www.ncbi.nlm.nih.gov/pubmed/18700206?ordinalpos=1&...

It was also not a random or blind trial: "allocation to the pill/control group was entirely the decision of the volunteers, not the experiments."