December 8, 2017

While the central reason for the existence of Unskewed is to dissect shoddy, one-sided, peer-reviewed research as soon as it appears online, another reason it exists is to dwell — at least occasionally — on the quirks, oddities, and irrationalities of our politicized research era today. And that is the case with a new article appearing in the June 2017 issue of Perspectives on Sexual and Reproductive Health. It’s entitled “Barriers to Abortion Care and their Consequences,” and explores trends — insofar as such things could be assessed in such data — in the experiences of 29 women who “had traveled either across state lines or more than 100 miles” to seek an abortion. My concerns about it are four-fold. The authors publish a study that: 1) Samples on the dependent variable, a basic mistake; 2) Tells us nothing with certainty; 3) Hijacks emotional terms in order; to 4) Bolster a politically and legally effective literature.

First, the sampling criterion itself is a key piece of the conclusion — a basic mistake that seems to have been overlooked during the peer review process. Gabriel Rossman, a talented cultural sociologist at UCLA, defines the problem as: “Sampling on the dependent variable is when you select cases on the basis of meeting criteria and then use those cases as evidence for the criteria.” Jenna Jerman and three of her fellow Guttmacher Institute researchers searched for people who had gone out of state, or else over 100 miles, in order to document — as the heart of the study — that “barriers to abortion … can have consequences.”

But the typical consequences interviewees talked about (that is, the dependent variable or outcome) were grouped into five categories, all of which have to do with the fact that the interviewees had to travel to seek an abortion (in order to be interviewed in the first place). The authors listed the consequences: “[T]ravel-related logistical issues, system navigation issues, limited clinic options, financial issues, and state or clinic restrictions.” Each is a function of the sample criterion. It cost interviewees money to travel. It involved logistics and making arrangements. Limited clinic options — sometimes prompted by restrictions — was a key reason why they traveled. In other words, their sample criterion guaranteed they would come to the conclusions they did. This is not how high-quality social science research is supposed to be conducted.

I don’t really have problems with what the study reveals, but that is largely because of its utter inability to tell us something — anything — new. I am not opposed to exploratory research using a few dozen interviews. But since it appears in print, it ought to be informative. Other studies they cite have already noted the challenges of barriers to getting an abortion. So what can interviews with 29 women tell us that is new? It cannot, as its authors claim, do any sort of “deductive analysis.” What it delivers is modest. It simply notes that some of the interviewees faced more than one barrier. The closest the authors come to saying something distinctive is in their claim that it is the

… intersection of multiple barriers to abortion care (that) creates consequences for women who travel for abortion, and the effects may be greater than those of individual barriers previously identified.

“Intersectionality” meets the study of abortion here; that is its key claim. Two things stand out about this quote. First, the term “care” is, or has already been, hijacked. “Care,” a term emotionally associated with the good, appears in print in this study — linked to abortion — a whopping 46 times. Abortion equals care here. Words matter, of course, and the political and legal debate over this subject reminds us of this in spades. Second, the word “may” appears 25 times, as in:

… encountering a lack of information may have resulted in the need to jump through hoops to receive care, which in turn may have generated further gaps in information and more hoops.

The “lack of information” may have resulted in the need to jump through hoops to receive an abortion — that is, care — but it may not have. It may have “generated further gaps in information,” but it may not have. Social science is being overrun by possibilities today, when what it ought to deliver is probabilities. “The possibilities are endless,” the cliché goes. They are indeed. Probabilities are not. Social scientists write the words “may,” “can,” “might,” and “could” alarmingly often today. What we ought to want to know is what does affect an outcome (and how and why). Which makes me wonder why this study in print at all, especially given this admission:

We found no direct link between any one barrier and any one consequence (or any minimum number of barriers that will have consequences), and no clear patterns of barriers and consequences according to women’s demographic characteristics.

Its mundane conclusion — as stated in the abstract — is that “barriers to abortion … can have significant consequences …” while at the same time, “ …the effect of any individual barrier was unclear.” That’s your tax money hard at work.

I suspect this study is in print in order to reinforce “the literature,” that pool of studies that can and is used as a political and legal hammer when mustered. It was, for example, used to great effect in the US Supreme Court’s decision on Whole Women’s Health v. Hellerstedt, something the authors make sure to note and openly align with:

Indeed, the Whole Women’s Health v. Hellerstedt ruling affirms that the cumulative effects of multiple abortion restrictions result in a violation of a woman’s constitutional rights: “Increased driving distances do not always constitute an ‘undue burden,’ but they are an additional burden, which, when taken together with others caused by the closings … help support the District Court’s ‘undue burden’ conclusion.” Future research attempting to establish that state restrictions on abortion access represent an undue burden for women and focusing on individual barriers may not produce sufficient evidence; highlighting the concept of compounding barriers may be especially salient.

In sum, this study seems tailored to help buttress legal issues around abortion access, and to help the Supreme Court’s decision in Whole Women’s Heath v. Hellerstedt remain in place. Why do I think that? Because the authors admit it: “The mechanism by which any one barrier — or any combination of barriers — triggers the judgment of an unconstitutional undue burden is unclear.” That is, they sense that the difference between what constitutes an “undue burden” and what doesn’t is subjective and fragile in the minds of judges. Given that tenuous situation, the more of them that appear in peer-reviewed journals — genuine quality, novelty, and evidence for its conclusions be damned — the better, they surmise.

Expect to see this study cited in amici briefs soon.

About the Author

Mark Regnerus is an associate professor of sociology at the University of Texas at Austin and a senior fellow at the Austin Institute for the Study of Family and Culture. His research is in the areas of sexual behavior, family, marriage, and religion. Mark is the author of over 40 published articles and book chapters, and three books.

About our Editors