Thoughts some of you might be interested in this.
The IAT measures implicit biases on various subjects. One of the tests measures whether you have an implicit preference for thin or fat people, and to what degree, or no preference at all. According to their summary, 70% of respondents showed an automatic preference for thin people over fat people to some degree, and only 12% showed any degree of preference for fat people over thin people. 19% of respondents showed no clear preference.
You can take the sample test here. I'd be interested to see what results people actively involved in fat-positive spaces get; I personally got a slight preference for fat people over thin people.