|Home | About | Journals | Submit | Contact Us | Français|
The paper by Evans et al,1 which compares a self‐selected internet sample of gay men with a national probability sample of men who have sex with men (MSM), fills an important gap in the field of internet and MSM sampling. A decade ago, the internet was considered a vagary of electronics for the purpose of sampling, but studies such as that of Ross et al2 in Sweden demonstrated that rather than collecting “more of the same” urban and gay‐acculturated gay men as traditional samples, an internet sample appeared to contain more men who were younger, bisexual or heterosexual, from small towns or rural areas and poorly educated, compared with traditional bar and venue samples. More recently, Ross et al3 compared an internet sample (heterosexual, bisexual, MSM) in Sweden with a gold‐standard random sample from a national sexuality study and found that while there were expected differences in age (younger) and occupation (more students and fewer retired people) and a bias towards larger cities in the internet sample, the differences between the two samples were not major.
The comparison of a gay internet sample with a national random study population of MSMs, however, showed a significant gap until this study by Evans et al.1 Their data show trends similar to the Swedish national probability study3 in that internet respondents were more likely to be younger and students, but were also more likely to live outside major metropolitan areas, as Ross et al2 had found in their convenience sample of MSM on the internet in Sweden. Particularly important, however, is Evans et al's findings that the internet sample of MSM was likely to be in worse health (despite being slightly younger on average), over four times more likely to have had a sexually transmitted infection (STI) in the past year and to have major differences in the reports of anal (and unprotected anal) intercourse. These latter findings seem to reinforce the earlier findings of this team,4,5 that the internet may act as a site for people with HIV infection to “serosort”—select other HIV‐infected men as partners for unprotected sex.
The internet has become an electronic venue for partner selection for gay men and other MSM, which is unrivalled in its potential sample size. It has the classic advantages of anonymity, affordability, acceptability and what is called “approximation”—use of the internet for sexual experimentation or cybersex by men who might not otherwise be accessible for study in traditional venues.6 Evans et al confirm that the MSM who were previously most difficult to access (non‐London‐dwellers and younger men) can be reached through the internet. The internet is, according to their data, also significant as a source for accessing MSM whose sexual risk behaviour for HIV and other STI acquisition is high—contrary to the earlier finding in the Swedish MSM convenience samples.2 The internet can also access MSM, some of whom may not be identified as gay, who are spread across the country and away from prevention programmes which are concentrated in larger cities.
HIV/STI prevention programmes using the internet can deliver tailored interventions at any hour of the day or night anonymously and efficiently right into the bedroom (or wherever the computer is located!). And, as Evans et al now demonstrate, the biases within an internet sample, which were assumed to be massive given the usually <1% sample of banner‐generated links to website to completed surveys,7 are quantifiable and are often smaller than many might imagine. Evans and colleagues have put an important capstone on internet sampling studies with regard to MSM. From these data, for this population, internet sample weighting or stratification will remove one of the last methodological barriers to sampling; will aid implementing HIV/STI prevention internet interventions; and will promote generalising from internet data.