I don't think depression is that bad of an idea. I guess the important part is that all people who take your survey don't have to write their name on it. A few of my classmates, a few years ago, did use that subject and got representative and not biased results. Just don't ask directly in the survey if they suffer from depression "yes/no", but instead use more global questions to deduce if they do have in fact depression or not.
However, if you really don't want to play in that area, I think food problems and self esteem are something you can definitely get good results for and would be rather fun/easy to make a survey with!
@
(you need an account to see links)