I'm not sure "underestimate" is the right description here; my opinion (as an androphile) is that the male attractiveness distribution is heavily skewed, basically in the way that women think it is, if the 1-5 scale measures the underlying strength of attraction rather than quintiles. (3s, 4s, and 5s all fall in the top quintile of male attractiveness, but it seems that there are much larger gradations there than there are in the top quintile of female attractiveness.)

I'm not an androphile myself, but that's my impression too, for various reasons (see e.g. the paragraph starting with “Similarly” in this post).