A counter to your point about politeness skewing the results is that the women who were shown the sketches of themselves afterwards clearly seemed to react in such a way that indicated that they realized they were too harsh on themselves, that they had negative views of their appearance which was unjustified, and that they were relieved/flattered to see how strangers saw them. So, it's not a proof of anything, but I'm just pointing out that womens' description of themselves seemed to be pretty reflective of how they genuinely viewed themselves.
That's why I was so frustrated that they just went with the publicity stunt when this could have been real. it would have been a bit harder to spin into a slick documentary, but it would have been much more valuable.
The artist knows perfectly well what the intention is, and which group is which.
ALSO, the results are going to be skewed by standard rules of politeness -- even given a person with a perfectly balanced view of themselves, the way they describe themselves to a stranger surely should avoid being too praising (or they'd seem narcissistic), whereas someone else's description of them would be skewed to the positive side, also out of politeness -- who knows if the person being drawn is going to hear this description later, or is a friend of the artist, etc.?
It would be easy to do this better.
Imagine if this were (seemingly) a study about the value and accuracy of police sketches. Use multiple sketch artists, and let them know their sketches are going to be tested by IDing from a large lineup.
Some sources will be self-describing (i.e., someone very familiar with the face) and others will be someone with only a short encounter with the subject.
This is a plausible (fake) study, and the remaining problems (with the real intent) are much, much fewer.