Yeah, raggin' on Mythbusters again.
Pheromones: basic structural problem. You want to test the subtle effects of human pheromones, you don't hit the subject with a concentrated nose-full and ask for her immediate, conscious opinion. You give her a low dose for several minutes, and monitor physiological responses (no, I don't know which ones, let alone how to measure them).
Aside from that... the "choose your favorite" is a poor mode of voting, especially with many options. I suggest having a slider on each station, initially set to 0, which the subject can move anywhere in the -10 to +10 range.
Blondes: again, basic problem. A proper blonde is going to be generally melanin-deficient, and will have pale skin and, probably, blue eyes. Get too far from that and what you have is not a blonde but a woman who has inexplicably bleached her hair. You can take a pale, blue-eyed woman, put her in a dark wig, and have a convincing effect; going the other way just screams fake!
Also, how many of the test subjects were actual gentlemen?