So the men and the women talk, and then then the men’s voices were erased, leaving just the women’s voices. A group of neutral observers heard the recording (of just the women’s voices) and indicated via several questions whether the women sounded beautiful.

Bradley: good comments, I’ll make sure Ori sees them so he can reply. I’ll also give them a bit more thought, the paper you link to is interesting, and only going meta will allow us to integrate both perspectives.

I agree both with your article’s conclusion that “The fact that I can identify a particular bias in those I disagree with is only very weak evidence that I am more right than they” and Ori’s point that “the biggest step is to recognize how often we get swayed.” He is not talking about how not to sway other people but about how not to be swayed by other people. And the alternative to awareness is, of course, blissful ignorance.

Will sleep on this…

http://www.overcomingbias.com/2008/06/meta-is-max—b.html

“The kicker is that the women who the men thought were pretty ended up sounding beautiful on the phone. And the women who the men thought were less attractive ended up sounding less beautiful. We take on the roles others ascribe to us.”

When you say “ended up sounding beautiful on the phone,” do you mean the women sounded beautiful as rated by the men, or that the women sounded beautiful based on some ‘objective’ measure of how beautiful their speech was (e.g. based on the tonal qualities of their voice or coding of their speech acts).

If you mean the former, that the women sounded more beautiful as rated by the men, then it seems that the data is better interpreted as showing that we ’see’ people as fulfilling the roles we ascribe to them, rather than people actually fulfilling the roles we ascribe to them. Rather than the women actually acting more attractive, the men are likely just interpreting the women’s actions as the actions of a more attractive woman. The women are not acting any different (i.e. are not fulfilling the roles the men ascribe to them); the men simply see (or hear) the women as fulfilling those roles. This would seem like the much more accurate interpretation of the data.

It seems, however, that you’re suggesting the second interpretation–that women ‘objectively’ sounded more beautiful. You say that the women fulfill the roles ascribed to them, which seems to mean that they actually behaved differently in the two conditions (when men thought they were more versus less beautiful). And your example of employees and children fulfilling the roles ascribes to them also reveals that you favor this second interpretation. But for this interpretation to be right, there would have had to have been data recorded on the women’s behavior (viz. their speech). And then the researchers must have examined if the women’s behavior (e.g. the tone of their voices or their speech acts) differed significantly between the conditions–when the men had seen a more attractive picture versus when the men had seen a less attractive picture. (And they would have needed a coding system to label certain tonal qualities or speech acts as indicating a “more beautiful sounding woman” versus a “less beautiful sounding woman”). I would be very surprised if this were the case; but this is what would have been needed to support your interpretation.

I am not familiar with the study you are discussing, so perhaps your interpretation is valid. And the idea that, to a significant extent, we fulfill the roles ascribed to us is almost certainly true (although I’m not familiar with that literature). But based on how you’ve described the study here, it’s ambiguous as to whether or not it supports that idea (since it’s not clear if the women “sounded” more beautiful to the men–which would not support ‘the idea’–or if the women sounded more beautiful ‘objectively’–which would support ‘the idea’).