Science you say?
Are the two samples (and statistics derived from them) comparable? (92% of the OkCupid userbase is straight, only 8% are gay).
Are the samples drawn independently from the population? We don't know why people choose to sign up for OkCupid so we cannot say if they are representative of the general population. Possible bias skews in the sample (depending on the observer's prejudices) could include: only slutty people, only people who are not getting laid enough, etc etc.
Is your primary source peer-reviewed, or is it a marketing piece?