There's a review and meta-analysis somewhere of these kinds of studies (I should be working so am not going to look it up) estimating that about 1/3 of the vari

There's a review and meta-analysis somewhere of these kinds of studies (I should be working so am not going to look it up) estimating that about 1/3 of the variance in attractiveness ratings can be attributed to the person being rated, 1/3 due to rater, and 1/3 due to the interaction.