Ravi:
>
> In case someone has not already mentioned it, underage does
> not mean "not grown woman" (from a sexual attractiveness
> perspective). We have come up with a number (18 in many
> parts) that is considered not "underage", but a girl/woman
> is both sexually attractive, and active, before that age.
> You really don't need to get into rape fantasies!
>
[WS:] It is more complicated than that. Consent age varies among jurisdictions and the age of the perpetrator is usually a factor
http://www.cga.ct.gov/2003/olrdata/jud/rpt/2003-R-0376.htm
Miles: Sampling 101 - you can draw conclusions from a non-random sample if you do proper weighting. All that is required is that probability of selection be known (and thus unequal probabilities compensated by weights.) Many sampling techniques deviate from random selection.
Another point - biased sample is a problem only if there is a correlation between probability of selection and variables under investigation e.g. if people of certain sexual orientation are more likely to be in a sample than people of a different sexual orientation. Sometimes that correlation is obvious, sometimes it is not, but if you have a probability sample (e.g. when the probability of selection of element of the sample is known) you can compensate under- or over-representation by weights.
Wojtek