Science communication: the dangers of oversimplifying microbiome studies
Lauren Freeman and Saray Ayala spoke to us about the dangers of oversimplification and decontextualization in microbiome research - and why they think scientists should be careful before recommending behavioural changes
You’ve raised concerns about the blurring between correlation and causation, especially in microbiome research. Why is this distinction important?
This is a critical distinction in science that often gets distorted, especially when scientific findings are reported in popular media. Correlation between two variables can be indicative of a causal relation, but it can also be spurious, like the correlation that exists between the divorce rate in Maine and per capita consumption of margarine in the United States (a correlation of 99.7%!). (This website has a great collection of other spurious correlations: http://www.tylervigen.com/spurious-correlations).
What we argued in our article about the placental microbiome is that we must be careful about conclusions that we extract from (mere) correlations. Even if they are not spurious correlations, we still need to determine whether one variable has a causal influence on the other, or whether they are related through a third variable that is causing both. This is something of which both scientists and philosophers are very aware.
But, there are also additional dangers with blurring the line between correlation and causation other than compromising the rigor of scientific studies. As Tania Lombrozo warns, we commonly think of causal claims as being value-free descriptions of how the world is, whereas in fact, causal attributions are also influenced by social norms and our beliefs about how things should happen (See http://www.npr.org/sections/13.7/2014/08/25/343121679/using-science-to-blame-mothers-check-your-values; http://philpapers.org/rec/HITCAN; http://philpapers.org/rec/KNOCJA).
For example, dominant gender norms dictate that mothers are the parents who ought to nurture their children and secure their well-being, while fathers are not held to the same expectations. When such norms are violated – for example, if a mother is not nurturing, or is not as nurturing as expected – it often results in a causal attribution directed at the mother and she is doubly blamed: first, she’s blamed for not being a “good mother” and second, she’s blamed for causing (potential or actual) harm to her child(ren). But this causal attribution is more closely related to our values of how things ought to be (gender norms about mothers) rather than to how things actually play out in the world.
The way Aagaard et al.’s study was reported in the media is an example of this dangerous kind of reasoning, namely, of jumping too quickly to a causal connection (namely, between a mother’s periodontal health and preterm birth), which seems to be motivated more by social norms than by the facts.
There are many public health studies that lead to experts calling for behavioural change. You assert that we need to also look at societal changes – why?
This is a great question that really gets to the heart of our work on the microbiome and our respective interests in questions of social justice. One of the main points that we wanted to convey in our article is that to even try to separate “behavior changes” from “societal changes” is to start off on the wrong note: such a separation is misleading and does not accurately represent reality. This is because what individuals do is so deeply entrenched in social structures, values, norms, and expectations. It is all too common to focus on the individual without considering the relational structures in which individuals are embedded, which constrain what they can do (and can think of doing) in different ways.
Focusing on individual behaviors to the exclusion of their social embeddedness is problematic for a number of reasons. First, it obscures the complex network of social factors that are out of the individual’s control. Moreover, it facilitates blaming individuals for their problems without recognizing that the issues are much more complex and often out of their control. Furthermore, it is often the case that those who are blamed for, for example, their poor health, are often in the most disadvantaged positions in our society who have limited resources. In fact, for many people who suffer from poor health, it’s not that they choose not to see a doctor on a regular basis, not to take their medications, and not to exercise or eat healthy foods. Rather, it’s often the case that they do not have access to physicians, cannot afford medications, do not have leisure time for exercise, and cannot afford (or do not have access to) fresh food.
Our criticism of the Aagaard et al. study falls in this area. Aagaard et al. called for women to keep good periodontal health yet failed to consider contexts in which it might not be possible for them to do so (through no fault of their own). We then drew out the dangerous consequences of blaming those women for potentially causing harm to their future fetuses. To be clear, we do not mean to say that public health studies should stay quiet about which behaviors promote health. Rather, what we argue is that guidelines on behavioral changes need to be sensitive to the reality that individuals do not live isolation from and unaffected by social dynamics, and importantly, that we are not all placed in the same position in society. The resources, services, and opportunities to which we each have access, and the norms and expectations that apply to us, vary according to our race, sex/gender, socioeconomic status (SES), ability, immigration status, and other factors. Therefore, calling for behavioral changes needs to be done within this broader picture. If we only focus on individuals, we aren’t able to see structural issues of injustice that operate beyond the level of the individual. When each act of injustice is seen as a one-off, patterns of injustice are missed.
What are your suggestions to scientists and clinicians when making health recommendations based on their research - or when communicating science to the public?
Our suggestion is that when making health recommendations, and in general when communicating scientific studies, scientists and reporters must avoid oversimplification and decontextualization. Although it is true that some simplification is needed when communicating science to the broad public, we need to be careful not to omit or downplay those details that can distort the results. This often occurs in catchy headlines like the following, quoted in a Nature commentary by Richardson et al., which convey very misleading ideas of the facts: ‘Mother’s diet during pregnancy alters baby’s DNA’ (BBC), ‘Grandma’s Experiences Leave a Mark on Your Genes’ (Discover). One example of how reporting could go right is this: when discussing the effects of something on pregnant women, cast the net more broadly to include discussions of both maternal and paternal roles, in addition to environmental, economic, and social factors that could be contributing to the issue at stake.
In our view, it is more desirable to air on the side of precision and caution and to create a situation in which readers are compelled to inquire further into the question at hand, than to oversimplify an account which too easily leads to false and often dangerous beliefs. Oversimplification of results invites handy, distorting interpretations, which likely consist of reading the new results through the lens of dominant social and moral norms. As we said above, our causal judgments are influenced by social norms, so we are always (unconsciously) eager to assimilate new results and candidate causes into our deep-seated ideologies. This is especially pernicious to women and other disadvantaged, marginalized groups since existing social norms place unjust burdens on them. So, we are calling for precision in the reporting of scientific details.
Furthermore, results need to be situated within a broader social context and within the many structural factors that have a causal influence on health and behavior. Sarah Richardson and six other researchers put forward these and other recommendations in the wonderful Nature commentary mentioned above, which discusses some of the ways in which mothers are blamed for the health outcomes of their children (http://www.nature.com/news/society-don-t-blame-the-mothers-1.15693). Specifically, they place the burden of responsibility on scientists, educators, and reporters to be mindful of their capacity to affect and influence the public and to be aware of how their conclusions, recommendations, and headlines might be received, interpreted, and taken up in popular discussions. This responsibility involves being attentive to how scientific research can reinforce and perpetuate morally problematic beliefs. Though they focus on mothers and gender norms that can disadvantage women, their recommendations can apply more broadly to any field of research.
As we note in our article, Aagaard et al.’s study is an especially problematic case of irresponsible recommendations and the way that it was reported in the media was both reductive and distorted the results. Aagaard et al. suggest that women of all ages should be vigilant about keeping good periodontal health in order to prevent premature births. Though on the surface this might seem like a benign recommendation, as we argue, it is anything but. This is because it targets not only pregnant women or those who plan to get pregnant, but also women in their early reproductive years and even young girls. Their recommendations are part of a trend of policing women, their bodies, and their choices at younger and younger ages. If we take Aagaard et al. at their word, then we should be targeting young women and girls to begin thinking about what counts as a good (potential) mother. Doing this reinforces the idea that the value of young women/females lies in them being a (healthy) substrate for pregnancy. This is not the message that we should be sending.
In sum, we are calling for more caution in reporting. We must remember that all research is both produced and applied in a social context – one in which poverty goes unabated, women are viewed as potential pregnant bodies, and health care is not available to all. This reality requires us to take responsibility: for researchers both to acknowledge the bigger picture and to reflect on how their conclusions fit within it, and for the media to exercise more nuance in their reporting
About Lauren and Saray
Lauren Freeman is an Assistant Professor of Philosophy at the University of Louisville. She is currently working on a series of articles that consider microaggressions within the context of clinical medicine. Microaggressions are brief and commonplace verbal, behavioral, and environmental indignities – either intentional or unintentional – that are rooted in implicit or explicit prejudice and/or stereotypes (racial, ethnic, gender, sexuality, religious, disability) that convey hostility, negative slights, or insults to the targeted person. She is considering how microaggressions committed by health care workers can result in a variety of harms to patients: specifically emotional harm, epistemic harm, and harms to self-identity. Insofar as these harms result from microaggressions experienced in medical encounters, she is arguing that they have the further consequence of undermining physician-patient relationships, precluding relationships of trust, and therefore of compromising the kind and quality of care that is received.
Saray Ayala is an Assistant Professor in the Philosophy Department at California State University, Sacramento. She is working on several projects that consider the notion of explanation. The first one articulates an account of what makes something stand in need of explanation. A second project contrasts psychological explanations of unjust practices and discriminatory behaviorwith structural explanations. While a psychological explanation would explain some forms of injustice by appealing to mental states (e.g. women and people of color suffer discrimination because most people have negative implicit attitudes towards them), a structural explanation appeals instead to social factors beyond the individuals’ minds (e.g. the positions women and people of color occupy in our society limit their opportunities in several ways). One of the goals of this project is to clarify the notion of structural explanation, which is still underdeveloped both in philosophy and in the social sciences.
Together, we are also working on a paper entitled “Sex Categorization in Medical Contexts: A Cautionary Tale” which problematizes the use of the female/male dichotomy to categorize people in medical contexts. There is a growing body of research that calls into question the female/male dichotomy. It is the norm, rather than the exception, for substantial individual variation to exist in all of the so-called sex markers (e.g., chromosomes, gonads, hormones, secondary sex characteristics, external genitalia, internal genitalia). In other words, human beings are not naturally divided into only two sexes. Our paper outlines various problems with the current system of sex categorization; we focus on problems that arise around the health care of trans people and people who do not fit squarely into the sex binary. We propose shifting the focus from sex-categories to the relevant sex-related properties of the patient (for example, whether they have a uterus instead of whether they are female, whether they have a prostate instead of whether they are male).