Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Objective science should not be without values
Debate over whether mammography screening saves lives has become front page news again in recent months. The opening salvo came in October 2001 from Gotzsche and Olsen, who argued in the Lancet that five of the seven randomised trials of screening mammography were of poor or flawed quality and that the remaining two fail to show that the benefits outweigh the risks.1 In the US, scientific experts remain divided in their recommendations regarding the benefits of mammography for women in their 40s, although the practice has been strongly defended by political leaders.2 This controversy, like many in the realm of disease prevention, raises an important question: when science is uncertain, what else drives an individual investigator’s interpretation of the evidence?
In the midst of a public debate over prevention recommendations, it often seems that even in the best of circumstances (that is, when there is a randomised clinical trial) there is little objective knowledge to be found. Indeed, prevention science is frequently charged with being subjective and value laden. The investigator’s scientific judgment may be polluted by influences other than the evidence itself. The list of suspected polluters is long, including commitment to a favourite theory, type of training (physician, epidemiologist, biologist), personal habits (smoking, diet), concern about the financial costs of screening, and attitudes towards balancing false positives versus false negatives, to name just a few.
Such influences seem to threaten traditional notions of scientific objectivity. At the start of the scientific revolution in the 17th century, the character of a scientist was believed to be as important as any experimental technique. A scientist, it was believed, should exhibit selflessness, disinterestedness, and a willingness to change her views.3 In other words, the scientific investigator should be a neutral observer with no personal interest in their theories or results. Today, in comparison, we are somewhat jaded. No one claims that scientists can be wholly objective and impervious to outside influences, but, at the same time, few of us are willing to give up on scientific objectivity.
Some epidemiologists have responded to this challenge by insisting that individual investigators try, as much as possible, to insulate their scientific work from the realm of policy making.4,5 They urge that investigators should not be active in influencing policy when they are conducting conducting scientific research on a related topic. In other words, they should try to be neutral and disinterested towards their research conclusions. Thus, a researcher who advocates for legislation requiring insurers to pay for mammography should not conduct a meta-analysis or write a review of clinical trials on the effectiveness of mammography. By separating policy (and morality and politics) from science, the argument goes, objectivity will be enhanced. This idea has previously drawn support in other scientific disciplines that are prone to scientific controversy. For example, in the 1920s, when social scientists were accused of being political and unscientific, sociologist Max Weber responded by warning his colleagues: “It is the duty of the man of science to remain silent . . . on value questions upon which he is so freely encouraged to expound.”6
However, others have responded that values cannot be isolated from the conduct of research, because many scientific theories and practices themselves are based on value judgments rather than empirical data and epidemiologists actually carry a professional obligation to participate in policy decisions.7,8 Historians and sociologists of science have marshalled much evidence to support this view, demonstrating how implicit individual prejudice can drive scientific debates.9
Rigorous methods are frequently touted as the best means of preserving objectivity, but they are no panacea. The ascent of the randomised controlled trial as the gold standard for testing new treatments was driven by a need for objective assessments of the claims of drug manufacturers.10 R A Fisher originally touted randomisation as the only means of generating an objective measurement of the risk of inferential error.11 However, as the mammography debate vividly illustrates, even a randomised controlled trial (or seven) is not sufficient to enforce objectivity. The methodological benefits of randomisation apply only to a single study and not to the synthesis of evidence from several studies. In contrast, the practice of causal inference, for example, entails implicit judgments and choices on the part of investigators, and our methods here are underdeveloped.12 Evidential synthesis, as in a meta-analysis, is the most important stage in the scientific process for developing prevention recommendations. We should certainly heed calls for more robust methodologies for evidential synthesis, as methods are essential, if not sufficient, for objective science. However, at the same time, we can benefit from a fuller understanding of the role of values in the interpretation of scientific findings.
Most importantly, we ought to distinguish between objectivity at the level of the individual and objectivity at the level of the scientific community. If objective science can be achieved at all, it will be at the level of the scientific community. For example, the peer review process, that cornerstone of quality science, provides oversight aimed at the community level. It is not the neutrality of individual investigators that keeps us honest, but the diversity of opinions and critical outlook of the scientific community as a whole. In fact, adversarial debate about methodology, and about values, drives advocates to marshal stronger evidence and clarify their reasoning. It is because opinions of the value of mammography differ that randomised trials and rigorous data analyses have been pursued. Thus, individual neutrality is not the key to objective science.
Moreover, values themselves are not always subjective or a matter of individual inclination. Values can be held by individuals or pursued by the scientific community as a whole. Those who insist that values are dangerous to scientific objectivity tend to focus on those values belonging to the individual, such as individual ideological allegiances.5,6 But the scientific community also pursues common values, such as truth, simplicity, and explanatory power, which are constitutive of scientific activity as we know it.13 Additionally, epidemiologists follow common ethical values in their professional practice.14 Such values are fundamental to the practice of science and ought not be isolated from it. In fact, much more could be done to incorporate them into daily scientific practice.
As prevention science is currently in the hot seat, we should resist the temptation to retreat to a position of neutrality, otherwise we run the risk of losing the rationale for doing research at all. Not only is neutrality an unrealistic goal, it is an undesirable one. While the debate over mammography is unlikely to be resolved soon, avoiding the policy implications and values behind it will not resolve it any sooner.
Objective science should not be without values