The vowels in Alan are totally at the front

I love that the internet has made possible all sorts of new experiments that come in the guise of slightly lascivious time-wasting. Having posted the same picture with different names on Hot or not, an MIT linguist has found that the position of vowels in names is correlated with the sexy rating given by voters. From the New Scientist article:

She found that men labelled with names including “front vowels,” such as the “aaa” sound in Matt were rated as more attractive by website viewers than photos labelled with “back vowel” names, such as the “aw” sound in Paul. The opposite was true for women’s names.

Perfors speculates that front vowels are perceived as slightly “smaller” than back vowels, and that this might suggest slightly feminine features, and that “Maybe women are subconsciously looking for more sensitive or gentle men.” They not out for obviously female names, however, as those with clearly feminine names were ranked lowest by the intrepid Hot or Not rating crew. Sorry, Leslie and Sandy.

Aside from affirming my nominative sexiness, the interesting aspect of the article is the use of the “in the wild” internet. One commeter on the article says that this is “a very clever example of how to use the tremendous resources of the web,” and I suppose that’s true. But I’m also a little surprised to see Hot or Not used in this way. The debate over how researchers should use online communities/web sites/mailing lists isn’t a new one: Anthropologists especially, but also sociologists and others, have put a lot of thought into the issues of privacy, methodology, and ethics that come with online research.

For comparison’s sake, imgine you were a social scientist who wanted to have undergrad students (there’s a generally unlimited supply of them) rate peoples’ attractiveness based on the names given to them. Before beginning any experiments, you would have to navigate your institutions human subjects protection office, carefully signing waivers declaring that you will not be poking anybody with sticks, forcing them to take part, exploiting them in any way, and so forth. This could take weeks to accomplish, because there is sharp scrutiny on any research involving people.

Having completed human subjects review, you would carefully assemble multiple sets of photos in order to have control and experimental groups; recruit participants; collect some information from each one; disclaim them appropriately as to the not-being-poked-with-sticks factor of the research; and then, finally, you could show them the pictures.

Online, the ethical issue is that there’s no way to disclaim users of Hot or Not—they won’t know that they’re participating. Now, I happen to think that human subjects protections in some areas of research are far too zealous (in my own case, I’m endlessly frustrated that journalists can simply call up the people they want to talk with, while I have to justify the fact that I even know their names); however, considerations of how to go about online research collection usually seem too shallow. My sense of the ethical resolution, such as it is, to questions of collecting data online is that online sources of data should be understood not as static records—the way we might think of some secondary sources—but as dynamic places in which people are invested to varying degrees. It’s impossible to determine from brief overview of this project to what extent human subjects protection was considered; giving Perfors the benefit of the doubt, it was probably approved, but the meaning of that approval is in this context is unclear: Participants certainly didn’t know that their votes were being watched; would they have participated if they did? Did the operators of Hot or Not know?

Methodologically, of course, online research continues to be problematic: Control and experimental groups seem impossible; sampling of subjects is also very problematic.

Ultimately, I agree that this is an innovative use of the internet. This is neat research. But, it raises issues that are worth considering (if not critical) for the way in which we approach our work, be it online or offline.