Electronic letters to:

Research:
Steffanie A. Strathdee, Stephen L. Martindale, Peter G.A. Cornelisse, Mary Lou Miller, Kevin J.P. Craib, Martin T. Schechter, Michael V. O'Shaughnessy, and Robert S. Hogg
HIV infection and risk behaviours among young gay and bisexual men in Vancouver
CMAJ 2000; 162: 21-25 [Abstract] [Full text] [PDF]
*eLetters: Submit a response to this article

Electronic letters published:

[Read eLetter] When Subjects Speak Back
Russell A Westhaver, Simon Franser University   (8 January 2003)

When Subjects Speak Back 8 January 2003
Top
Russell A Westhaver,
PhD Candidate
Department of Sociology and Anthropology,
Simon Franser University

Send letter to journal:
Re: When Subjects Speak Back

rwesthav{at}sfu.ca Russell A Westhaver, et al.

The Vanguard Project is an ongoing prospective study of the social determinants of HIV infection among gay and bisexual men in the greater Vancouver area. The motive for conducting this research is to help tailor intervention and prevention strategies toward those especially at-risk for contracting HIV. The article “HIV infection and risk behaviours among young gay and bisexual men in Vancouver” (Strathdee 2000) is part of this project. In this letter, I reflect on the Vanguard Project and its ostensible motive as a means of critiquing it and some the broader epistemological planks upon which this research is based. I make these reflections on the basis of a relatively unique position. Not only am I a sociologist who studies gay men’s communities and gay men’s health but I was also a participant in the Vanguard Project.

As a way of doing research, the Vanguard Project implicitly assumes a variety of interrelated positivist tenets developed in the natural sciences and transferred to the study of human social behaviour. It assumes that humans as part of the natural world are structured in ways commensurate with natural phenomena. Nature, being regular in most respects, can be expected to respond to external factors in predictable manners. Humans, as part of the natural world, are generally assumed to respond rationally to external stimuli. This research model feeds into and supports a broader paradigm of health promotion that assumes appropriate information about the social determinates of HIV will lead to appropriate responses to the changing and mobile nature of HIV transmission. The struggle with HIV is an information war: sound, appropriate, and timely information about the social determinates of HIV will assist us in curbing its spread. By providing researchers and those interested in promoting health with the right information about the nature and form of HIV transmission we, as mostly rational creatures, become aware of how to avoid it.

Positivist social research is also conducted with the aim of producing knowledge claims that are applicable to an entire population. Very frequently, however, cost and/or logistical concerns present barriers to the study of an entire population. As a result, the bulk of survey research--like the Vanguard Project--rests on a smaller sample of individuals drawn from a larger population. Using sample data to make knowledge claims about a larger unknown population assumes random sampling techniques. In methodological terms, the Vanguard participants represent a non-random sample. In the logic of scientific inquiry, non-random samples cannot be used to make claims about a larger unknown population without violating basic methodological principles of research.

In terms of the Vanguard Project, this fundamental methodological principle brings with it a very precise and clear limitation: the findings and conclusions emerging from an analysis of Vanguard participants cannot be used to make claims about other gay and bisexual men. To their credit, Vanguard researchers do not--in any direct way--extend their findings and conclusions beyond their sample. Researchers are careful to frame findings in terms of the cohort studied. For example, Hogg et al (2001) conclude “A significant increase in new HIV infections was observed between 1995-1999 and 2000 within the entire cohort” (1321, emphasis added). Researchers make, however, many indirect generalising gestures through the use of p- values and confidence intervals. For example, Hogg et al. completed the previous statement: “This remained true when participants who had injected drugs were excluded from the analysis (p<0.05). Among men who have never injected drugs, the rate of new infections has increased from 0.6 per 100 person years in 1995-99 (95% CI: 0.2-1.0) to 3.2 per 100 person years in the year 2000 (95% CI: 0.6-5.8)” (Hogg et al. 1321). In using these generalising gestures, researchers are claiming that data and findings from the Vanguard project are useful for estimating qualities of a larger unknown population of gay and bisexual men in Vancouver. As much as p-values and confidence intervals represent generalizable claims, they are also rhetorical gestures. Reporting p-values and confidence intervals functions as a bridge linking conclusions based on sample data to calls for population wide social/health policy changes and/or interventions strategies. For example, Strathdee et al. (2000) concludes: “Our findings underscore the urgent need for targeted interventions among young gay and bisexual men who remain at high risk for HIV infection” (25). Elsewhere Vanguard Project researchers write, “The results from the present study suggest new avenues for identifying and targeting prevention for young gay and bisexual men who remain at high risk for HIV infection”

One could hardly disagree with the motive directing these conclusions: in light of sample data which suggests a relationship between psychosocial factors and the risk of HIV seroconversion, we need to begin thinking about delivering better preventative health care in relation to these determinates. This is hardly a puzzle. Puzzling is how data and conclusions--based on a non-random sample--are given a very wide scope of application. Generalising these findings beyond those sampled by making conclusions about the needs of a general unspecified population of “young gay men” is methodologically unsound. The issue here is subtle--but no less import for its subtlety: conclusions derived from the study of non- random samples--and the policy implications emerging from these conclusions--cannot, in good scientific faith, be applied to those who are not part of the sample studied. Differently, the only methodologically defensible point of application for the descriptions, discussions, and conclusions emerging from this research--say in the form of policy recommendations or intervention strategies--would be the participants sampled.

In the long history in which social scientific research--like epidemiology or sociology--has tried to legitimate itself, incorporating the logic and mechanics of natural science into methods and theories of social inquiry has been the most predominant strategy. An overarching goal of social research informed by this positivist dream is the discovery of universal and general laws about human behaviour. While few social researchers are likely to identify themselves as positivists, and few would willingly argue that a perfect social explanation is possible, most would agree that regular and general associations between human behaviours and the social world do exist and can be found. In this atmosphere, social scientists are rewarded for producing results that parallel the logic of the natural sciences. In light of the fact that confidence intervals and p -values function as a mechanism allowing the researcher to generalise sample findings to a wider unknown population, they represents tools enabling researchers to propose/discover general human laws. It is also a means by which researchers can situate a project within a value horizon dominated by a legacy of positivism--and in so doing accrue accolades and prestige. Research and researchers that are able to make general and genealizable claims are rewarded and supported through funding opportunities and prestige in ways that situated research is not. Insofar as the Vanguard Project is based on a non-random sample, it represents, however, a situated research project and generalizable claims are not methodologically defensible. And yet, researchers make just these generalising claims and feed them into a reward system that values generalizable conclusions. The effect is significant: it disproportionately benefits researchers.

In 2000, the Vanguard Project distributed a progress report to all participants (Vanguard Project, 2000). The first third of what is ostensibly a lay document charts out the Project’s organization, introducing the research staff, along with a few references to rewards received for meritorious research. The majority of the remaining two- thirds summarises research findings. What is troubling is that the document charts out peculiarly academic accomplishments--highlighting papers published in peer review journals, and talks, posters, and abstracts accepted at academic conferences. With a focus on accolades and references to peer reviewed journals and conferences, there are few mechanisms through which important information could reach those men who need it most--the participants. Researchers who are able to publish and circulate their research through journals and conferences are, however, quite likely to secure both prestige and promotion as journals and conferences are the main vehicles through which researchers are rewarded for their work. This speaks volumes about where funding, tenure, promotion, placement in graduate schools, and employment are situated in relation to the needs of participants.

Not only is there a disproportionate benefit accrued by researchers, but this system of dissemination functions to isolate participants from the information and conclusions derived from this research. While journals and conferences are a cornerstone of good scientific inquiry, there are few Vanguard participants who are able to pick up the latest issue of the Journal of Acquired Immune Deficiency Syndrome and Human Retroviruses or the Canadian Medical Association Journal. Fewer still have the resources to travel to Durban South Africa to attend a conference. Moreover, journals and conferences are not likely to produce information that is accessible and actionable by those most in need. There is very little that is actionable in beta coefficients or p-values. Vanguard Project researchers do, of course, disseminates findings to Vancouver’s gay community through other mechanisms: community newspapers, email, and mail releases to study participants. Informing this method of dissemination, however, is an increasingly outdated health promotion paradigm of HIV intervention. The assumption that HIV research can contribute to HIV intervention by adding to the stock of knowledge about safe sex is increasingly in doubt. In light of this, I make a strong claim: this research project and its dissemination represent a misuse of information and resources. How are the conclusions and findings of this research benefiting those whose need for these conclusions and findings is greatest? This is particularly vexing given that the Vanguard Project constructs itself as a “community-based study” (Stratdhee et al. 2000: 22). In a very real sense, this project functions as a form of exploitation: researchers benefit from the research data while participants do not.

Intervention necessarily hinges on the allocation of resources. Time and money make intervention possible and I could not call for the Vanguard project to provide these resources--especially if such an allocation would undermine the project’s research mandate. The Vanguard Project is, however, uniquely position to become the resources so frequently called for in the Project’s studies' conclusion while filling its primary research objective. The Vanguard Project has at its disposal an enormous body of information in the form of statistical summaries. The challenge in using these resources for the purposes of both research and intervention lies in following a research course that steers away from collecting data for the purposes of representativeness. This is also a course that steers away from the kinds of academic accolades and professional recognition the Project seems to court. By bearing in mind that Vanguard data is applicable only to Vanguard participants, Vanguard researchers and other similarly structured projects can shift the analytical focus from representeativeness to relevance. I challenge the Vanguard Project--and other similarly structured research projects who claim to be “community based studies”--to seriously engage with the subjects upon whom their project, research, and careers depend. Bringing data into contact with those for whom it has most relevance through true community based research strategies could create the conditions necessary to translated these findings into actionable--as opposed to merely academically valuable-- knowledge about HIV prevention.

Hogg, R. et al. (2001). “Increasing incidence of HIV infections among young gay and bisexual men in Vancouver” AIDS, 15(10): 1321-1322

Strathdee, S. et al. (2000) “HIV infection and risk behaviours among young gay and bisexual men in Vancouver” Canadian Medial Association Journal, 162(1): 21-25

Vanguard Project (2000). Five Year Report: Summary of Research Findings 1995 to 2000.

Endnotes

1.http://www.cahrconf.hivnet.ubc.ca/Vanguard/homeframe.html, December 12, 2002 2.More engaging methods of community involvement are already part and parcel of Health Canada’s approach to HIV/AIDS. See http://www.hc- sc.gc.ca/hppb/hiv_aids/can_strat/research/overview.html