Illusions of Objectivity
Paradigms and prejudices
Many non-scientists are awed by the power and seeming certainty of scientific knowledge. So are most students of science. Textbooks are full of apparently hard facts and quantitative data. Science seems supremely objective. Moreover, a belief in the objectivity of science is a matter of faith for many modern people. It is fundamental to the world-view of materialists, rationalists, secular humanists and all others who uphold the superiority of science over religion, traditional wisdom and the arts.
This image of science is rarely discussed explicitly by scientists themselves. It tends to be absorbed implicitly and taken for granted. Few scientists show much interest in the philosophy, history or sociology of science, and there is little room for these subjects in the crowded curiculum of science courses. Most simply assume that by means of the scientific method, theories can be tested objectively by experiment in a way that is uncontaminated by the scientists' own hopes, ideas and beliefs. Scientists like to think of themselves as engaged in a bold and fearless search for truth.
Such a view now excites much cynicism. But I think it is important to recognize the nobility of this ideal. Insofar as the scientific endeavour is illuminated by this heroic spirit, there is much to commend it. Nevertheless, in reality most scientists are now the servants of military and commercial interests. Almost all are pursuing careers within institutions and professional organizations. The fear of career setbacks, rejection of papers by learned journals, loss of funding, and the ultimate sanction of dismissal are powerful disincentives to venture too far from current orthodoxy, at least in public. Many do not feel secure enough to voice their real opinions until they have retired, or won a Nobel Prize, or both.
Popular doubts about the objectivity of scientists are widely shared, for more sophisticated reasons, by philosophers, historians and sociologists of science. Scientists are part of larger social, economic and political systems; they constitute professional groups with their own initiation procedures, peer pressures, power structures and systems of rewards. They generally work in the context of established paradigms or models of reality. And even within the limits set by the prevailing scientific belief system, they do not seek after pure facts for their own sake: they make guesses or hypotheses about the way things are, and then test them by experiment. Usually these experiments are motivated by a desire to support a favourite hypothesis, or to refute a rival one. What people do research on, and even what they find, is influenced by their conscious and unconscious expectations. In addition, feminist critics detect a strong and often unconscious male bias in the theory and practice of science.
Many practicing scientists, like doctors, psychologists, anthropologists, sociologists, historians and academics in general, are well aware that detached objectivity is more an ideal than a reflection of actual practice. In private, most are prepared to acknowledge that some of their colleagues, if not they themselves, are influenced in their researches by personal ambition, preconceptions, prejudices and other sources of bias.
The tendency to find what is being looked for is deep-seated. It has a basis in the very nature of attention. The ability to focus the senses in accordance with intentions is a fundamental aspect of animal nature. Finding what is looked for is an essential feature of everyday human life. Most people are well aware that other people's attitudes affect the way they interact with the world around them. We are not surprised by such biases in politicians, nor by the differences in the way people see things within different cultures. We are not surprised to find many everyday examples of self-deception in members of our families and among friends and colleagues. But the 'scientific method' is generally supposed to rise above cultural and personal biases, dealing only in the currency of objective facts and universal principles.
Biases in science are easiest to recognize when they reflect political prejudices, because people of opposing political views have a strong motive to dispute the claims of their opponents. For example, conservatives like to find a biological basis for the superiority of dominant classes and races, explaining their differences as largely innate. By contrast, liberals and socialists prefer to see environmental influences as predominant, explaining existing inequalities in terms of social and economic systems.
In the nineteenth century, this 'nature-nurture' debate focused on measurements of brain size; in the twentieth, on measurements of IQ. Eminent scientists who were convinced of the innate superiority of men over women and of whites over other races, were able to find what they wanted to find. Paul Broca, for example, the anatomist after whom the speech area of the brain is named, concluded that: 'In general, the brain is larger in mature adults than in the elderly, in men than in women, in eminent men than in men of mediocre talent, in superior races than in inferior races.' He had to overcome many factual obstacles to maintain this belief. For example, five eminent professors at Goettingen gave their consent to have their brains weighed after they died; when these cerebral weights turned out to be embarrassingly close to average, Broca concluded that the professors hadn't been so eminent after all!
Critics of a more egalitarian political persuasion have been able to show that generalizations based on different brain sizes or IQ scores rested on the systematic distortion and selection of data. Sometimes the data themselves were actually fraudulent, as in the case of some of the publications of Sir Cyril Burt, a leading defender of the view that intelligence is largely innate. In his book The Mismeasure of Man, Stephen Jay Gould traces the sorry history of these purportedly objective studies of human intelligence, showing how persistently prejudice has been clothed in scientific garb. 'If- as I believe I have shown - quantitative data are as subject to cultural constraint as any other aspect of science, then they have no special claim on final truth.'
A persistent and pervasive source of the illusion of objectivity is the style in which scientific reports are written. They give the appearance of coming from an idealized world in which science is an entirely logical exercise, free from all human passion. 'Observations were made ...','It was found that ...','The data show ...', and so on. These literary conventions are still taught to budding scientists at school and university: 'A test-tube was taken.. .'.
Scientists publish their findings in technical articles, called papers, in specialized journals. In a justly famous essay called 'Is the scientific paper a fraud?', the immunologist Peter Medawar pointed out that the standard structure of these papers gives 'a totally misleading narrative of the processes of thought that go into the making of scientific discoveries'. In the biological sciences, a typical paper starts with a brief introduction, including a survey of previous relevant work, then a section on 'Materials and Methods', followed by 'Results' and finally a 'Discussion'.
The section called 'results' consists of a stream of factual information in which it is considered extremely bad form to discuss the significance of the results you are getting. You have to pretend that your mind is, so to speak, a virgin receptacle, an empty vessel, for information that floods in from the external world for no reason which you yourself have revealed. You reserve all appraisal of the scientific evidence until the 'discussion' section, and in the discussion you adopt the ludicrous pretence of asking yourself if the information you have collected actually means anything.
In fact, of course, the hypotheses that the experiments were designed to test generally come first, rather than last. Since Medawar wrote this passage there has been a greater conscious recognition of this sequence of events, and an increasing tendency to mention hypotheses in the introductions to papers. But the same conventions remain: passionless prose, the use of the passive voice and the pretence that data are unembellished facts. Practising scientists are well aware that this style is a kind of make-believe; but it has now become obligatory for anyone with pretensions to objectivity, and has been adopted by technocrats and bureaucrats as well.
Deceit and self-deception
The illusion of objectivity is most powerful when its victims believe themselves to be free of it. Along with a laudable sense of honour, a tendency to self-righteousness has been present in experimental science right from the outset.
With Galileo, the desire to make his ideas prevail apparently led him to report experiments that could not have been performed exactly as described. Thus an ambiguous attitude towards data was present from the very beginning of Western experimental science. On the one hand, experimental data was upheld as the ultimate arbiter of truth, on the other hand, fact was subordinated to theory when necessary and even, if it didn't fit, distorted.
A similar vice afflicted other giants in the history of science, not least Sir Isaac Newton. He overwhelmed his critics with an exactness of results that left no room for dispute. His biographer Richard Westfall has documented how he adjusted his calculations on the velocity of sound and the precession of the equinoxes, and altered the correlation of a variable in his theory of gravitation to give a seeming accuracy of better than 1 part in 1,000.
Not the lease part of the Principia's persuasiveness was its deliberate pretense to a degree of precision quite beyond its legitimate claim. If the Principia established the quantitative pattern of modem science, it equally suggested a less sublime truth - that no one can manipulate the Fudge factor so effectively as the master mathematician himself.
Probably the commonest kind of deception - and of self deception - depends on the selective use of data. For example, from 1910 to 1913, the American physicist Robert Millikan was engaged in a dispute with an Austrian rival, Felix Ehrenfeld, about the charge on the electron. Both Millikan's and Ehrenfeld's early data were rather variable. They depended on introducing oil drops into an electric field and measuring the strength of the field needed to keep them suspended. Ehrenfeld claimed that the data showed the existence of subelectrons with fractions of a unit electron charge. Millikan maintained there was a single charge. To rebut his rival, in 1913 he published a paper full of new, precise results supporting his own view, emphasizing in italics that 'this is not a selected group of drops but represents all of the drops experimented upon during sixty consecutive days'.
A historian of science has recently examined Millikan's laboratory notebooks, which reveal a very different picture. The raw data were individually annotated with comments such as 'very low, something wrong' and 'beauty, publish this'. The 58 observations published in his article were selected from 140. Ehrenfeld meanwhile went on publishing all his observations, which continued to show a far greater variability than Millikan's selected data. Ehrenfeld was disregarded while Millikan won the Nobel Prize.
Millikan was no doubt convinced that he was right, and did not want his theoretical convictions to be disturbed by messy data. Probably the same was true of Gregor Mendel, the results of whose famous pea-breeding experiments were, according to modern statistical analysis, too good to be true.
The tendency to publish only the 'best' results and to tidy up data is certainly not confined to famous figures in the history of science. In most if not all areas of science, good results are likely to advance the career of the person who produces them. And in a highly competitive and hierarchical professional environment, various forms of improving the results are widely practised, if only by omitting unfavourable data. This practice is indeed normal. Apart from anything else, journals are disinclined to publish the results of problematical or negative experiments. Little professional credit results from unclear data or seemingly meaningless results.
I know of no formal study on the percentage of research data that are actually published. In the fields I know best from personal experience, biochemistry, developmental biology, plant physiology and agriculture, I estimate that only about 5-20 per cent of the empirical data are selected for publication. I have asked colleagues in other fields of enquiry, such as experimental psychology, chemistry, radio astronomy and medicine, and come up with similar results. When the great majority of the data are discarded in private processes of selection - often 90 per cent or more - there is obviously plenty of scope for personal bias and theoretical prejudice to operate both consciously and unconsciously.
The selective publication of data creates a context in which deception and self-deception become a matter of degree. Moreover, scientists usually regard their research notebooks and data files as private, and tend to resist any attempts by critics and rivals to go through them. True, it is usually assumed that a researcher will, within reason, make his or her data available to any colleague who might express a desire to see them. But in my own experience, this ideal is far from the reality. On the several occasions I have asked researchers if I may see their raw data, I have been refused. Maybe this says more about me than about prevailing scientific norms. But one of the very few systematic studies of this cherished principle of openness gives little ground for confidence. The procedure was simple. The person conducting it, a psychologist at Iowa State University, wrote to thirty-seven authors of papers published in psychology journals requesting the raw data on which the papers were based. Five did not reply. Twenty-one claimed that their data had unfortunately been misplaced or inadvertently destroyed. Two offered access only on very restrictive conditions. Only nine sent their raw data; and when their studies were analysed, more than half had gross errors in the statistics alone.
Those who refuse to expose their raw data to scrutiny may well have nothing to hide; they may simply find it inconvenient to explain their notes to someone else, or suspect the motives for the request, or resent an implied slur on their honour. It is not the purpose of this discussion to suggest that scientists are particularly prone to deliberate fraud and deception. On the contrary, most scientists are probably at least as honest as most members of other professional groups, such as lawyers, priests, bankers and administrators. But scientists have greater pretensions to objectivity, and at the same time a culture which encourages the selective publication of results. These conditions are favourable for the deliberate deception of others, but I do not see this as the most important threat to the ideal of objectivity. Self-deception is the greatest danger, especially collective self deception encouraged by dominant assumptions about objective reality.
Many scientists recognize the potential for wishful thinking in others, and are quick to dismiss results of research in unorthodox fields such as parapsychology and holistic medicine as due to self deception, if not deliberate fraud. And indeed some of those who challenge orthodox ideas may well deceive themselves. But they do little harm to the progress of science because their results are either ignored, or else subjected to extremely critical scrutiny. Organized groups of Skeptics, such as CSICOP, the Committee for the Scientific Investigation of Claims of the Paranormal, are always ready to challenge results that do not fit into the mechanistic world-view, and try their best to discredit them. Parapsychologists are so accustomed to these critical responses that they are unusually aware of the pitfalls of experimenter effects and other sources of bias. But conventional science is not subject to a similar degree of sceptical scrutiny.
Peer review, replication and fraud
Scientists, like doctors, lawyers and other professionals, generally resist attempts by outside agencies to regulate their conduct. They pride themselves on their own system of controls. These are threefold:
1) Applications for jobs and grants are subject to peer review, ensuring that the researchers and their projects meet with the approval of established professionals.
2) Papers submitted to scientific journals have to pass the critical scrutiny of expert referees, usually anonymous.
3)All published results are potentially subject to independent replication.
Peer review and refereeing procedures do indeed act as important quality checks, and are no doubt often effective, but they have a built-in bias. They tend to favour prestigious scientists and institutions. And independent replication is in fact rarely performed, for at least four reasons. First, in practice it is difficult to replicate a given experiment exactly, if only because the recipes are incomplete or fail to communicate practical knacks. Second, few researchers have the time or resources to repeat other people's work, especially if the results come from a well-funded laboratory and involve expensive apparatus. Third, there is usually no motivation for replicating the work of others. And fourth, even if exact replications are performed, it is difficult to get them published because scientific journals favour original research. Replication of other people's results is usually attempted only under special conditions, for example when the results are of unusual importance or when fraud is suspected on other grounds.
Under these circumstances deceptions can easily pass unchallenged as long as the results are in accordance with prevalent expectations.
Acceptance of fraudulent results is the other side of that familiar coin, resistance to new ideas. Fraudulent results are likely to be accepted in science if they are plausibly presented, if they conform with prevailing prejudices and expectations, and if they come from a suitably qualified scientist affiliated with an elite institution. It is for the lack of all these qualities that new ideas in science are likely to be resisted. Only on the assumption that logic and objectivity are the sole gatekeepers of science is the prevalence and frequent success of fraud in any way surprising.... For the ideologists of science, fraud is taboo, a scandal whose significance must be ritually denied on every occasion. For those who see science as a human endeavor to make sense of the world, fraud is merely evidence that science flies on the wings of rhetoric as well as reason.
One of the few areas of science subject to a limited form of external supervision is the testing for safety of new foods, drugs and pesticides. In the United States, every year many thousands of test results are submitted by industry for review by the Food and Drug Administration (FDA) or the Environmental Protection Agency (EPA). These agencies have the power to send inspectors to the laboratories that provide the data. They continually unearth falsified results.
The cases of fraud uncovered in the great unpoliced hinterlands of science are rarely brought to light by the official mechanisms of peer review, refereeing of papers and the potential for independent replication. And even if attempts to replicate an experiment fail, this is usually ascribed to a failure to reproduce the conditions of the experiment precisely enough. There is a big psychological and cultural barrier against accusing colleagues of fraud - unless one has strong personal reasons to suspect their integrity. Most known cases of fraud come to light as a result of whistle-blowing by immediate colleagues or rivals, often as a result of some personal grievance. When this happens, the typical response of laboratory chiefs and other responsible authorities is to try to hush the matter up. But if the charges of fraud do not blow over, if allegations are made persistently enough, and if the evidence becomes overwhelming, then an official enquiry is held. Someone is found guilty and dismissed in disgrace.
Most professional scientists deny that these incidents shed doubt on institutional science as a whole; rather, they are seen as isolated aberrations by individuals who have become temporarily unhinged under pressure, or who are rare but inevitable psychopaths. Science is purified by their expulsion. They are scapegoats in the biblical sense. On the Day of Atonement the high priest confessed the sins of the people while laying his hands on a goat. The guilt-laden scapegoat was then expelled from the community into the wilderness, bearing away their iniquities.
Scientists generally feel the need to preserve an idealized self-image, not just for personal and professional reasons, hut also because this image is projected on to them by others. There are many people who put their faith in science rather than religion, and need to believe in its superior, objective authority. And to the extent that science replaces religion as the source of truth and values, then scientists become a kind of priesthood. As with priests in general, there is then a public expectation that they will live up to the ideals they preach: in the case of scientists, objectivity, rationality and the quest for truth. Some scientists, in their public appearances, can be noticed playing up to this role, which seems to invest them as cardinals of reason propounding salvation to an irrational public. There is also a strong disincentive for them to admit that there is anything fundamentally wrong with the beliefs and institutions that legitimize their own position. While it is relatively easy to admit that individuals may err, and to purify the community by expelling them, it is much harder to question the beliefs and idealizations on which the whole system depends.
Philosophers of science tend to idealize the experimental method, and so do scientists themselves. In their insightful study of fraud and deceit in science, William Broad and Nicholas Wade were led to enquire what actually happens in laboratories, as opposed to what is supposed to happen. They found that the reality was far more pragmatic and empirical, involving much trial and error:
The competitors in a given field try many different approaches but are always quick to switch to the recipe that works best. Science being a social process, each researcher is trying at the same time to advance and gain acceptance for his own recipes, his own interpretation of the field.... Science is a complex process in which the observer can see almost anything he wants provided he narrows his vision sufficiently.... Scientists are individuals and they have different styles and different approaches to the truth. The identical style of all scientific writing, which seems to spring from a universal scientific method, is a false unanimity imposed by the current conventions of scientific reporting. If scientists were allowed to express themselves naturally in describing their experiments and theories, the myth of a single, universal scientific method would probably vanish instantly.
I agree with this analysis. This present book is an argument for more democratic and pluralistic scientific research, liberated from the conventions imposed upon institutional science by its role as the Established Church of the secular world order. But whatever forms science takes, it will still depend an experiments.http://www.gaiaguys.net/illusions.htm