Ideological Signaling Has No Role in Research

Positionality statements, in which scholars declare identity categories to which they belong and indicate how such categories might bias their research, have become commonplace in the social sciences in recent years. They are required by some journals, such as the Journal of Women and Minorities in Science and Engineering, and “encouraged” by others, such as the Journal of Social and Personal Relationships. For instance, a white man writing about race might acknowledge how his ethnicity prevents him from taking the perspective of nonwhite participants. Like diversity, equity, and inclusion statements, positionality statements are consistent with a progressive worldview that emphasizes power, oppression, and conflict, and conceives the research endeavor as an unfair process that disadvantages minority groups.

Recently, there has been some pushback. In February a research group led by Jukka Savolainen, a professor of criminal justice at Wayne State University, published a paper critical of the statements in Perspectives on Psychological Science, a major journal in the field. The authors argued that it is impossible to construct credible positionality statements because people are hindered by their own subjectivity; that the statements are unnecessary because scientific credibility depends on methods, not identity; and that asking people to focus on their identities undermines norms of impartiality in research.

Supporters of positionality statements, however, view them as important correctives of a research process that exploits objectivity to privilege the perspectives of white men. The statements originated in a true observation about the scientific method: Though science seeks objective truths, it is conducted by subjective human beings who inevitably possess a range of biases, and our backgrounds and experiences may contribute to those biases. As a scholar whose career has focused in part on bias in science, I find that observation uncontroversial. Why then do I agree with the view that we shouldn’t be declaring identity categories in our research?

First of all, it is not clear that positionality statements are the best means to reduce bias in science. Over the past decade scholars have taken a hard look at this question, as fields ranging from psychology to medicine have realized that many previously vaunted results actually proved difficult to replicate when more rigorously tested. Typically referred to as the “replication crisis,” this dismaying phenomenon demonstrated how bias had affected scientific knowledge in negative ways. In this case, publication bias — the pressure to publish positive findings only — had driven scientists to nudge (or, more rarely, fabricate) their findings to support their hypotheses, thus creating a distorted evidence base.

If we understand the default human condition to be selfish — a condition to which scientists are, alas, not immune — the results were predictable. The problem was the incentive structure. Scientists were given incentives — career advancement, grant funding, and newspaper headlines — to produce exciting results, not failures. Many scientists, rather than sticking to the truth and watching their careers flounder, gave in to the temptation to massage their results to make them more impressive.

Racial identity can also affect research bias. Consider, for example, the well-discredited notion that Black medical patients experience less pain than white patients do. The origins of this idea are hard to know for sure, though it appears to have originated among physicians in the 1800s. This incorrect medical assumption, grounded in unscientific prejudice, caused Black patients to be undermedicated for pain for generations.

It is undeniable that our social identities can blind us to the experiences of others, and lead us to ignore or discredit evidence that contradicts our prejudices, particularly when we have social incentives to do so. Yet the debunking of this racist nonsense about Black patients’ pain thresholds came from the scientific process itself, not from any putatively radical challenge to it. For example, one 2012 study noted that, accumulated over 20 years, evidence suggested Black medical patients were undermedicated for pain, compared with patients of other ethnicities. Other studies suggested the disparities originated in false beliefs about biological differences between races, such as that Black people have thicker skin. The identity categories of the scholars conducting this important debunking work was irrelevant: What was critical was that good science destroyed racist fables.

If we accept that our identities can sometimes blind us to the experiences of others, it’s unclear that leaning into identity more is the appropriate fix. It’s also something of a double-edged sword. Let us imagine that an English researcher decides to conduct research on 20th-century Irish terrorism. His Irish colleague complains, “Well, you’re English, the people who oppressed the Irish; you can’t understand what our people went through.” It’s a fair point, of course, but so is the Englishman’s retort: “Sure, but as an Irishman, you’re equally likely to be defensive about your own history.” Identities are complex. Do my identities, as an American with both Irish and English ancestry, balance out within me, or am I doubly conflicted? That may explain my moments of inner turmoil.

There are effective ways to combat bias in science. The best way to increase confidence in research results is to increase the rigor and transparency of studies through processes such as the preregistration of hypotheses and methods and the open availability of materials and data. Those procedures are offered by the open science school of thought, specifically designed to address the replication crisis. Though such approaches increase the validity of research findings and may be practiced by scholars of any background, advocates of positionality sometimes appear hostile to open science.

For instance, in one recent paper published in the Journal of Communication, a team of authors accuse open science of “prioritiz[ing] openness while insufficiently addressing essential ethical principles: respect for persons, beneficence, and justice.” Later, the authors appear directly censorious when they suggest that “practicing reflexivity may pose uneasy challenges, such as questioning the appropriateness of a scholar conducting some types of research.” Does this imply that it’s safe for me to conduct that research on Irish terrorism but that a Latino or Black scholar should be forbidden to do the same? This appears to call for something akin to research segregation. That’s dangerous.

The authors appear convinced that open science will lead to harm to nonwhite scholars, although little supporting evidence is presented. And yet there is considerable evidence that open science is consistent with values of diversity and justice, and is effective in increasing the rigor of the scientific method and public trust in science. Further, positionality statements frequently adopt the default assumption that science is (to quote the authors of the above paper) “white, male, elite, and Western,” an accusation that itself appears biased, given that modern science is a comprehensively global phenomenon and by no means an exclusively male one. Indeed, one potential criticism of the recent progressive lurch in academe is that it has abrogated the quest for universalism.

This raises another significant concern: that positionality statements will actually increase, rather than decrease, bias in science. If identity is a potential source of bias, leaning further into identity rather than reaching for universal human principles is likely to make the problem worse. Inherent in the idea of positionality statements is the belief that some identities are credible and others suspect, an assumption that merely inverts the racism of previous generations of scientists.

If one is to be reflective about one’s identity biases, one need not do so publicly, and doing so publicly inevitably sets up incentive structures that are themselves biasing. If we remember the lessons of the replication crisis — that scientists, like all human beings, are by default selfish — then setting up perverse incentives increases bias. But this should not prevent scholars from investigating their own limitations privately. It is always fair to ask ourselves, “Do I have the right expertise for this question?” or “Is there a risk I might be tempted to use this data to make some group I value look good, or another group I don’t value look bad?”

We can consult with other scholars, particularly those not invested in our own work, to inquire about our blind spots. We can make a purposeful effort to collaborate with scholars of different backgrounds and worldviews, sometimes referred to as adversarial collaboration. We can make efforts to learn the perspectives of scholars with varying views, even those with whom we may instinctively disagree. Sometimes we might rightly decide, “Huh, maybe I’m a little too emotionally close to this issue to be objective.”

By contrast, public displays of positionality will be understood as rhetorical claims to credibility that, in fact, have no actual evidentiary value and may serve mainly to bolster some bad science and preclude some good science from ever existing.

The result is that such statements are inherently politicized. Indeed, those coming from scholars with identities that are, from the point of view of the progressive academic left, suspicious can take on a confessional, self-excoriating tone. In their article, Savolainen and colleagues presented several such examples. One article included the following positionality statement: “Both authors are middle- to upper-middle-class white women — one is a mother, the other is not. A commitment to antiracist, intersectional, and feminist principles guides our research efforts, and we conducted this work with an awareness of the politics, dangers, and limitations of affluent white academics writing about the lives of low-income Black Americans.” Despite those “dangers,” they published their research anyway. Positionality statements such as those do little other than signal the authors’ political allegiance. If we suspected the quality of their research, transparent open data could allow other researchers of any background to thoroughly fact-check it. But the authors’ assurances that they are antiracist, intersectional feminists tell us only that they are on the “right” team.

Like DEI statements, positionality statements are rhetorical documents, functioning less to decrease bias than to establish a scholar’s adherence to an approved ideology. To suggest that these public displays of identity are integral to good science is simply untrue. Quite the opposite, in fact. What is needed to combat bias is more rigorous and transparent research, not political loyalty oaths.