Researchers are really good at coming up with questions. Sometimes, however, the answers they come up with are inappropriate. Let’s take for example work by Gail Wyatt, Ardis Moe and Don Guthrie called “The Gynecological, Reproductive, and Sexual Health of HIV-Positive Women.”
The authors understandably want to understand how HIV positive women go about meeting their gynecological, reproductive and sexual health care needs. What services support them and what blocks access to services are no doubt critical questions. My problem lies with the answers they develop.
Complicated questions about how and why certain people end up with HIV obviously have complicated answers. The authors contend that ethnic differences exist among HIV positive women that mirror ethnic variation among HIV negative women in terms of sexual behaviors and reproductive health care management. Maybe that’s true, maybe it isn’t. The authors have what researchers call an endogeniety problem. Endogeneity happens when the independent variable is correlated with the error.
Basically all this fancy jargon just means that the researchers didn’t include an important factor that has a relationship with a factor they did include. For example, let’s say you and I were doing a research project on women’s health experiences. We decide to ask women in a hospital waiting room to fill out a survey and do interviews with us. We have and endogeneity problem here because we’re only speaking to women who have the means and spare time to get to the hospital. We might be missing a factor like “easily accessible transportation” or a“job with paid time off” that would ultimately be really helpful in helping us understand women’s health care experiences in New York City.
Authors Wyatt, Moe and Guthrie have a major endogeneity problem because they only work with women already involved with a hospital. They also have some problems with unclear terms and small sample sizes, but endogeneity is their biggest problem. Since we can’t tell what their research means for women in Los Angeles (where their study takes place). Unfortunately they reach problematic conclusion because their approach is filled with errors. Let’s make sure to keep a critical key on research!
By Adrienne Wallace, Research Intern
For a midwifery class, I have been researching the access and obstacles to prenatal care, for undocumented immigrants. My search has resulted in two key factors: a lack of information and, as you mentioned, the endogeneity problem. One article I found, compared birth certificates to Medicaid records as a means to look at the prenatal health and birth outcomes of undocumented immigrants. Sadly, that was the most accurate resource I found. By retrospectively reviewing data, there has a substantial amount of potential errors. For example, this study goes under the assumption that all the reported information was accurate. We cant’ forget the possiblities of language barriers, and under/over reporting. Also, what about all the immigrants who did not have access to Medicaid, or those who had home-births? We are missing a huge sample! Not to mention, how was their ethnicity and legal status determined? If it was by self-reporting and ‘checking’ the Hispanic, Non-hispanic, Mixed-race, White, or Other box then there is definately a potential for error.