Is there a way to statistically control for a variable (attractiveness) in a within-subjects design?

Let's say I have presented 3 men to participants (a given participant sees all 3 conditions): Andrew is 5"9; Bob is 6"; and Craig is 6"3. I ask participants about the desirability of each man, and also to rate his physical attractiveness. I want to find out if height influences a man's desirability. I want to somehow 'control for' the man's physical attractiveness. How do I go about this - some kind of ANCOVA perhaps?