The danger of extrapolation from a convenience sample is that the sample of people who are convenient may be strikingly different than the full group of people you're looking to make a statement about. For example, I know a number of women in the adult entertainment industry - strippers, adult film actors, professional dominatrixes, and the like. They are uniformly smart, empowered, well educated about STD-risk, and intelligent about the need for safe sex. They are also, I am reminded by a study recently published in Women and Health, not necessarily representative of sex workers as a whole.
The study compared a sample of 69 female adult entertainment workers from a club in Rhode Island with an age-matched group from the Behavioral Risk Factor Surveillance System in the same area. It found that the club employees were at significantly increased risk for HIV compared to the control women in the sample -- as they were more likely to have a number of risk factors, including injection drug use, trading sex for drugs or money, having unprotected anal sex, and having a history of STDs. They were also less likely to have undergone HIV testing or have regular healthcare than the other women.
The study was based on a convenience sample too, but it was a far more useful one. Furthermore, it provides an important reminder that although there are many smart, empowered, risk-averse sex workers, there are also a lot of sex workers who do not have the skills, money, or knowledge to protect their health. (Mind you, the same thing could also be said about many people in the general population.) This is, as sex worker advocates would happily step in to remind me, one of the problems with the criminalization and stigmatization of sex work. It makes it much harder for the women and men employed by the industry to get the help and protection they need.