In the last two decades, scientists, government officials, and science policy experts have expressed concerns about the increasing role of financial interests in research. Many believe that these interests are undermining the objectivity of research by causing bias and error, suppression of results, and even outright fraud. This paper seeks to shed some light on this view by (1) explicating the concept of objectivity in research, (2) describing some ways that financial interests can undermine objectivity, and (3) briefly discussing some strategies for mitigating or managing the influence of financial interests. These strategies include the (a) using government funds to counter-balance privately funded research, (b) increasing public input into government funding decisions, (c) disclosing and managing conflicts of interest in research, (d) auditing data, (e) expanding access to data. Since it is neither possible nor desirable to eliminate financial interests from research, the wisest policy is to manage and counter-balance these interests for the good of science and society.

Keywords: objectivity, bias, financial interests, private funding, conflicts of interest, publication, access to data

The influence of financial interests on research and development (R &D) has grown steadily since the 1970s. Prior to World War II, research funding (public or private) was modest, but the government began to make major commitments to R & D during and after the war. From the 1940s to the 1970s the government and private industry both steadily increased their funding of R & D, but the government still funded the majority of R & D as well as most basic research (Dickson 1988). That balance has shifted in the last two decades as private dollars have flowed into industries that place a great emphasis on R & D, such as pharmaceuticals, biotechnology and biomedicine, and computers and information systems. In the United States (US), private industry accounts for more than 60% of all R & D funding or more than $100 billion annually. Privately funded basic research has tripled since 1980 to a total of about $8 billion per year, and 71% of all R & D in the US is conducted in private laboratories (Jaffe 1996). Biomedical R & D has been one of the key factors in this trend toward privatization: the private sector’s contribution to biomedical R & D rose from $2 billion per year in 1980 to $16 billion in 1990 and will probably pass the $30 billion per year mark in 2000 (Beardsley 1994, Resnik 1999).

Many scientists working in the public or private sector also now have significant financial interests in research, such as stock, copyrights, and patents (Krimsky et al 1996, Blumenthal 1997, Bowie 1994, Zolla-Parker 1994). Prior to the 1970s, universities and scientists did not aggressively pursue patents, since patents were not worth the trouble or expense. A scientist might try to obtain a patent to protect his invention or as a symbol of achievement, but he could not expect that the patent would lead to great financial gain (Bowie 1994). But all that has changed in recent years as more and more scientists are seeking patent protection and universities have developed programs to help scientists protect intellectual property (Bowie 1994). One reason why patenting has become more prevalent is that the gaps between "science" and "technology" or "basic research" and "applied research" have closed in the last couple of decades (Bowie 1994). When Watson and Crick discovered the structure of DNA in 1953, the practical applications of their research were not readily apparent. Today, however, new genetic information can be put to practical use in medical diagnosis and treatment, drug development, and genetic engineering (Kitcher 1996). Scientists owning stock in research has followed a similar trend. In years past, most researchers were compensated with a salary, stipend, or honoraria or other set fee. Today, many researchers report that they also receive stock as compensation. One reason why payment in the form of stock has become more popular is that companies that are short on cash, such as new biotech start-ups, often find it easier to pay researchers with stock than with cash (Dickson 1988).

Although the influx of private money has been a windfall for many researchers and universities, many scientists, government officials, and science policy experts have expressed concerns about the increasing role of private interests in research. One of the most prominent and frequently voiced fears is that private interests are undermining the objectivity of research by causing bias and error, suppression of results, and even outright fraud (Ziman 1996, Broad and Wade 1993, Bowie 1994). Although there are no data that establish a clear link between fraud and private funding, some recently published studies suggest that sources of funding can bias or influence research procedures and outcomes (Easterbrook et al 1991, Friedberg et al 1999). There have also some well-publicized cases of private corporations suppressing undesirable results, publishing biased results, and failing to reveal important health concerns to regulatory agencies (Wadman 1996, Hilts 1994, Associated Press 1993, 2000).

The purpose of this essay is to develop a philosophical framework for thinking about financial interests and objectivity in research. In order to accomplish this task, the paper will (1) explicate the notion of objectivity in research, (2) describe some ways that financial interests can undermine objectivity, and (3) briefly discuss some strategies for mitigating or managing the influence of financial interests. These strategies include the (a) using government funds to counter-balance privately funded research, (b) increasing public input into government funding decisions, (c) disclosing and managing conflicts of interest in research, (d) auditing data, (e) expanding access to data. The overall thesis of this paper is that financial interests can threaten the objectivity of research but that there are some steps that we, as a society, can take to control or mitigate the influence of individual and corporate interests. Since it is neither possible nor desirable to eliminate financial interests in research, the wisest strategy is to shepherd these interests for the good of science and society.

During this century alone, many thoughtful philosophers, historians, and sociologists have addressed questions about the objectivity in science. It is not my aim to answer all of these questions in this essay or develop a robust and detailed theory of scientific objectivity. However, I will attempt to sketch what I take to be a reasonable and mainstream approach to these issues, which will be based on the work of several writers, most notably Longino (1990), Kosso (1992), and Shrader-Frechette (1994). Most importantly, I believe the view I will sketch can makes sense of some of the concerns expressed by people who fear that private industry undermines or threatens scientific objectivity. While I will not provide necessary and sufficient conditions for defining "scientific objectivity" or "objectivity in research" I will discuss some key characteristics of objectivity and distinguish between different ways of viewing objectivity.

I shall follow a usage of the term "objective" that is fairly common among practicing scientists. When scientists speak of objective research, they have something in mind like "unbiased research." For example, many of the policies and professional codes for collecting, reporting, and analyzing data and for disclosing financial interests refer to "objectivity" in contrast to "bias" (Sigma Xi 1986, Panel on Scientific Responsibility and the Conduct of Research (PSRCR) 1992, National Academy of Science (NAS) 1994, Department of Health and Human Services (DHHS) 1994, Food and Drug Administration (FDA) 1998, American Statistical Association (ASA) 1999, Office Research Integrity (ORI) 1999). This seems simple enough, but the idea of "bias" is a complex notion that encompasses many different types of biases (Longino 1990). Some biases result from political ideologies, religious doctrines, or moral concerns, i.e. human values. For example, Stephen Jay Gould (1981) has argued that many of the studies of human intelligence were influenced by racial and ethnic biases. Women’s health researchers have provided evidence to show that research on women’s medical conditions was for many years shaped by sexist biases that were used to exclude women from clinical trials (Dresser 1992). Soviet research on genetic and development was profoundly affected by the Communist Party’s endorsement of Lysenkoism and rejection of Mendelianism (Joravsky 1970).

Other biases result from psychological prejudices. For example, "self-deception can occur in science when researchers fail to critically examine their work because they want to believe that their research is accurate or truthful. Despite emphasizing the norm of "self-criticism," scientists can be highly gullible (Broad and Wade 1993). The "observer effect" is a well-documented phenomenon where observers see what they want or expect to see (Broad and Wade 1993). For many years astronomers claimed that they could see canals on Mars and Galenic physicians claimed to observe pores in the septum of the heart. In the Cold Fusion debate, some have argued that psychological factors, such as the observer effect or some other type of self-deception, corrupted Stanley Pons and Martin Fleischmann’s experimental design and made their result results impossible to reproduce (Huizenga 1992). When interpreting probabilities, most human beings are influenced by a variety of psychological biases, such as the anchoring bias, where one estimates a new probability based on a familiar one, and the availability bias, where one overestimates a probability that receives a great deal of publicity (Kahneman, Slavic and Tversky 1982). For example, many people overestimate the risk of a patient contracting HIV from a health worker based on a few well-publicized cases.

Some biases result from the social, institutional, cultural, or economic conditions of the research environment or research enterprise. For example, many of the drugs that are used on children have not been adequately tested in pediatric populations and this creates a biased (and poor) understanding of drug therapy in children. Without adequate evidence about the effects of drugs in children, physicians who want to prescribe medications for children often must treat them as "little adults" and use body mass as the only variable for drug dosing (Tauer 1999). There are several reasons why this happens. First, human subjects research regulations are designed to protect vulnerable subjects, such as children. Thus, there are ethical, legal and social barriers to research on children. Second, institutions and clinical researchers are therefore hesitant to design or implement protocols that test drugs in pediatric populations. Third, pharmaceutical companies often develop drugs for conditions common in adults, such as depression, because these drugs give a good return on investment. Even though these drugs may be prescribed for children as an "off-label" use, these companies often do not plan to test the drugs in pediatric populations because they are targeting adult consumers (Tauer 1999). Many of the biases examined in this paper, i.e. financial biases, will involve social and economic conditions of the research environment.

Finally, some biases are due to faulty assumptions used in hypothesis development and testing, experimental design, or data analysis. For example, the science of craniometry made the faulty assumption that intelligence is linked to the size and shape of the human head (Gould 1981). We now know that this is a faulty assumption though it seemed quite reasonable to many researchers at one time. Sigmund Freud held that sex (or psychosexuality) played a crucial role in normal psychological development as well as psychosis and neurosis, but contemporary psychologists argue that his theories overemphasized sex and were based on the unproven assumptions about the importance of sex. Others have questioned Freud’s methodology and have argued that he did not subject his theories to rigorous tests (Grunbaum 1985).

In thinking about objectivity and bias in research, it is also useful to distinguish between two different ways of conceiving of objectivity. One way of viewing objectivity is to see it as an outcome, product, or aim of research. For instance, many research ethics books, reports, and codes of conduct mention objective (or unbiased) data or results (Sigma Xi 1986, American Physical Society (APS) 1991, PSRCR 1992, LaFollette 1992, DHHS 1994, NAS 1994, Resnik 1998a, ASA 1999, ORI 1999). Philosophers and science scholars often talk of objectivity and bias in terms of objective beliefs, objective facts, objective truths, objective statements or objective knowledge (Popper 1959, 1972, Kosso 1992, Shrader-Frechette 1994, Ziman 1996). In any case, these notions of objectivity are all goal-directed or teleological notions in that they view objectivity as an aim or goal of inquiry.

A different way of thinking about objectivity is to conceive of it as a process, procedure, or method for conducting research. Research ethics publications frequently address objectivity in research design, data analysis, methods and procedures, and communication (DHHS 1994, ASA 1999, FDA 1998, ORI 1999, NAS 1994). Philosophers and science scholars frequently address objectivity in methods, rules of inductive inference, criteria of theory-choice, and in the social structure of science (Howson and Urbach 1989, Longino 1990, Kosso 1992, Solomon 1994). These notions of objectivity focus not on the outcomes or goals of inquiry but on the act, process, or procedure of inquiry. Hence, I we can characterize these ideas as "procedural" accounts of objectivity.

According to a standard view of scientific methodology, the methods of science are designed to achieve the goals of science (Laudan 1984). If the goal is objectivity, then the methods should be procedures that are likely to bring about this result. So, "objective procedures" are methods that are designed to achieve objective outcomes. Thus, many scientists and philosophers of science view rules of research design as procedures that enable one to produce objective data or outcomes (Giere 1991, Grinnell 1992). Many philosophers and science scholars have argued that scientific methods are rules that help one to acquire or develop objective theories, truths, or beliefs in the short-term or long-run (Popper 1959, Lakatos 1978, Pierce 1955).

To summarize this section, we can say that objective research is research that is not affected by (or is independent of) political ideologies, religious doctrines, moral concerns; psychological prejudices; social, institutional, cultural, or economic conditions; or faulty assumptions. Objectivity is a goal of the research enterprise and scientific methods should promote this goal.

Given all of the different kinds of biases that can affect research, objectivity seems to be a pretty lofty ideal. Indeed, many sociologists and historians of science have argued that it is unachievable. The watershed event in the critique of scientific objectivity was the publication of Thomas Kuhn’s The Structure of Scientific Revolutions (1962). In this book, Kuhn argued that science is not a slow and progressive march to the objective truth but that it involves revolutions and changes in scientific worldviews. Kuhn also argued that human values as well as social, economic, historical, and psychological factors influence the changes that occur during scientific revolutions. Although Kuhn did not reject the notion of objectivity in science, he claimed that scientists and science scholars should rethink this notion (Kuhn 1977).

Since 1962, many of the great debates in the history, philosophy, and sociology of science since have focused on questions about objectivity. Many science scholars have attempted to show that objectivity is either an illusion or an unreachable goal since human values as well as psychological, social, economic, and political, and other factors influence (or bias) research (Collins 1985, Latour and Woolgar 1986, Harding 1986, Fuller 1988, Longino 1990, Bloor 1991). On the other hand, many science scholars and scientists have rushed to defend science against charges of relativism, constructivisim, and bias (Laudan, 1990, Kitcher 1993, Gross and Levitt 1994). The so-called "science wars" have spilled out of the halls of academia and onto the public stage when the media covered physicist Alan Sokal (1996) parody of cultural studies of science, which he published in a Social Text, a journal that addresses social and cultural factors in science.

It is not my aim in this essay to stop the "science wars" or find a solution to the academic debate about the objectivity of science. However, since my main goal is to consider how private funding affects the objectivity of research and what we should do about this, I will offer a response to the critique of objectivity that explains why objectivity is worth pursuing.

As I see it, the debate about objectivity in science consists of two different types of questions, (1) conceptual questions about the nature of science and the nature of objectivity, and (2) empirical questions about whether and how various social, economic, and other factors affect objectivity. It may turn out that empirical studies provide us with strong evidence that science as it is practiced is indeed highly biased. Unless one can show through some kind of conceptual, logical, or linguistic analysis that science is, by its very nature, objective (or not objective), then these questions about objectivity cannot be settled a priori (Kitcher 1992). At this point in time at least, it is still an open and factual question as to whether objectivity is to some degree achievable in practice. These questions are similar to questions about the feasibility of other abstract concepts or idealizations, such as justice and rationality. Even if it is impossible for a society to be perfectly just, or for a person to be perfectly rational, it may still be possible for a society to be or become more (or less) virtuous or for a person to be more (or less) rational in comparison to some ideal standards. Thus, I will take it as a working assumption of this paper that the goal of objectivity is neither unrealistic nor futile: research can be or become more (or less) objective in comparison to some ideal standard.

If it is at least possible to make progress toward objectivity, if policies and decisions can in fact make a difference in whether science becomes more (or less) objective, then the question we need to ask is, "should scientists strive for objectivity in research?" I will briefly outline some reasons why scientist should strive for objectivity or why objectivity has some value or worth.

First, objectivity has moral value because biased research can have adverse effects on physical and mental health, safety, environmental cleanliness, economic growth, and other social values. For example, if a pharmaceutical company sponsors research on a new drug and the study provides a biased analysis of the data that underestimates the risk of certain side-effects, such as liver damage, then this bias could cause discomfort, disability, disease, or death in patients. In medical research, it is important to eliminate biases in order to protect widely accepted moral values, such as individual and public health and social welfare (Brody 1995). A similar point applies to other types of research where scientific applications have an impact on things that we hold to be of moral value, such as engineering, agriculture, nutrition, psychology, economic development, education, and environmental management. The general point is that research biases should be avoided because they can lead to undesirable social consequences. Objectivity is valuable because it promotes research with socially valuable consequences (Shrader-Frechette 1994).

Second, objectivity has political value in public policy debates and in the administration of justice. In policy debates, policymakers rely on politically neutral reference points or "facts," provided by scientists and other experts, to develop and implement legislation and administrative rules and guidelines (Jasanoff 1990). Opposing sides may provide different interpretations of the facts and they may challenge each other’s facts, but it is important that opposing sides recognize some points upon which they can agree, otherwise these debates would be impossible to resolve in any rational manner (Thompson and Gutman 1996). Without politically neutral reference points to serve as a basis for agreement, disagreement, and argument, public policy debates can only be resolved through rhetoric or sheer force. In the administration of justice, it is important for judges and juries to have some legally neutral reference points, provided by experts, in order to settle legal questions of guilt, innocence, and liability. If the facts used in court cases were tainted through their association with one side or the other, then legal questions would also be impossible to resolve in a rational manner (Jasanoff 1995).

Third, objectivity has symbolic or cultural value in that the public expects science to be objective. Although the image of the dedicated scientist toiling away in a laboratory in pursuit of truth has taken a beating since it heyday in the 1950s, objectivity still is a key plank in the public’s understanding of science and a significant reason why the public supports science (Nelkin 1995). Indeed, revelations about fraud, corruption, and bias in research have probably played an important role in the anti-science attitudes that have become more popular since the 1960s (Broad and Wade 1993, Haerlin and Parr 1999). Of course, it is important for the public to have a realistic understanding of science and the public was not well served by the old, naïve images of researchers who are insulated from human values and other biases. However, images that portray researchers as completely biased and corrupt can also be harmful. In order to trust and support science, people need to have a realistic but also optimistic view of the research process. Objectivity has value as a counter-balance to excessive cynicism and skepticism about research.

Fourth, objectivity has epistemic value in that many people, including science scholars and scientists, would like to have objective knowledge about the world (Kosso 1992, Popper 1972). People would like to learn the "truth" about nature and have some grasp of an "independent reality." Although the goal of obtaining an objective understanding of the world has a long and storied history dating back to at least Plato, one could argue that this is the weakest of all reasons for valuing objectivity because this goal is more difficult to achieve than the other goals and because it is more contentious. Hence, I will acknowledge but not stress the epistemic value of objectivity.

To summarize my argument thus far, objectivity in research may indeed by very difficult or even impossible to achieve, but researchers should strive for objectivity because it has moral, political, and cultural value. Objectivity is an ideal, but it is an ideal that is worth pursuing.

Having set the stage for a discussion of objectivity in research, we can now ask how financial interests can undermine objectivity and what we should do about this problem. By "financial interests" I mean corporate, economic interests, such as those of private industry, as well as the individual economic interests of researchers. As a preface to this discussion, it is worth coming to terms with the fundamental reasons why financial interests can (and often do) lead to biases in research. The main reason why corporate, financial interests can undermine the objectivity of research is that for-profit companies and non-profit companies sponsor research in order to obtain a return on investment (Brody 1995). Companies view R & D funding decisions as like other business decisions in which profit (or a prudent use of resources) is the dominant goal (Bowie 1994, Dickson 1988). Although many companies may make strong public statements about social responsibility, their commitment to research, and other values, these other values rarely take precedence over economic motives. Indeed, companies who do not give high priority to profit (or at the least the prudent use of resources) will find themselves soon out of business. None of this implies that private corporations place no value on objectivity in research, since objective results can be very useful in developing and marketing new products and in protecting companies from legal liability. For example, drug companies now usually cite published studies in the flyers they distribute to physicians and in their magazine advertisements. However, companies tend to view objectivity in research as a means to an end, not as an end-in-itself. Objectivity is useful when it serves the organization, but it is not useful when it does not (Wade 1994).

Individual researchers with financial interests can have similar motivations. Most researchers recognize the importance of objectivity as means to their career goals and many value it for its own sake. Many scientists value objectivity even before they study science in college or enter the profession. During their college and university studies, researchers become indoctrinated into scientific values and traditions, which emphasize the intrinsic worth of truth and objectivity (Merton 1973). Scientists come to recognize the extrinsic value of adhering to objective procedures as they enter the scientific profession (Fuchs 1988). Researchers who do not follow these procedures and methods do not get papers published, obtain grants, or have academic appointments. However, as long as researchers have other goals besides objectivity—and many do these days and always have—then these other motivations can erode their commitment to objectivity. Thus, the academic ethos, which places a high priority on objectivity, can be eroded by financial interests (Wade 1994).

To help understand some specific ways that financial interests can affect the objectivity of research, it will be useful to describe a standard model of the steps of scientific research (Giere 1991, Grinnell 1992, NAS 1994, Resnik 1998a) so that we can see how private interests may affect these different steps. Among researchers, these steps have other names, such as the "research protocol," "research plan," "research design,"or "research proposal." For my purposes a label these steps as follows::

Select a problem to study or a question to answer.

Step 2: Conduct a literature search in order to have a better understanding of the problem or question and the relevant methods, tests, and experiments.

Step 3: Develop hypotheses to test in the research project.

Step 4: Review and propose and materials and methods used in the research project.

Step 5: Design tests or experiments that will confirm or disconfirm the hypotheses.

Step 6: Perform tests or experiments and collect and record data.

Step 7: Analyze data and results.

Step 8: Interpret data and results.

Step 9: Publish data and results in peer reviewed journals or media.

Step 10: Communicate data and results to non-expert audiences, such as the public and government agencies.

Step: 11: Store, manage, and share data and results.

These steps are an idealization in that real research may deviate from this model with regard to time order of some of the steps, length of time on each step, etc. However, the model will provide us with a useful framework for thinking about private funding and research bias. I will discuss several common problems that can arise as a result of private interests and show how they can affect various stages of the research process.

As we noted earlier, private companies sponsor research in order to make money. Some research projects are lucrative, but others are not. Thus, for example, pharmaceutical companies spend a great deal of money on diseases and medical conditions that affect many people, such as cardiovascular disease, hypertension, diabetes, cancer, obesity, and impotence. But they spend relatively little money on diseases that affect only a few people, such as Tay-Sachs disease or Marfan’s syndrome. In medicine, this is known as the "orphan disease" problem and it is a direct result of the economics of biomedical research. Drug companies also develop drugs (known as "copy cat" drugs) that are very similar to existing, highly profitable medications while they may decide to not develop unprofitable (or "orphan") drugs. Drug companies are constantly searching for a new "blockbuster" drug that will earn reap in billions of dollars in profits (Angell 2000, Brody 1995). Similar biases can arise in other industries, such as engineering (Whitbeck 1998). Companies invest R & D dollars in profitable designs and abandon unprofitable ones. Since private firms cannot survive without making a profit or without investing their resources wisely, it should come as little surprise the businesses tend to invest their R & D funds in the most profitable research problems (Zolla-Parker 1994). In response to this problem, some have suggested that the government take steps to sponsor research in areas that private companies will tend to ignore, such as basic research, or unprofitable applied research, such as research on orphan diseases or orphan drugs. However, other biases, such as political interests, can affect government funding decisions. For example, various interest groups have lobbied for research on HIV/AIDS, palliative care, cancer, substance abuse, alternative medicine (Agnew 1999).

These economic and political biases relating to the funding of research can also affect individual scientists who need to obtain funding in order to do research. Individual researchers have other private interests as well, such as tenure, promotion, career advancement or prestige in a scientific discipline, which can affect their financial interests (Hull 1988). Even if you are very interested in studying Marfan’s syndrome or some other rare disease, you will have little success and you can’t have a career in research unless you can secure private or public funding for your research. Thus, macro-level conditions relating to the politics and economics of research funding ultimately affect the micro-level decisions of individual researchers.

Even if a company has decided to sponsor R & D on a problem that appears to be lucrative, the company may decide to withdraw funding if it determines that the results of the research are not likely to promote its interests. For example, Victor DeNobel and Paul Mele conducted research on nicotine addiction for Philip Morris in the 1980s. The two scientists discovered a substance that increases the addictive effects of nicotine when added to cigarettes and they submitted their findings to the journal Psychopharmacology. When Philip Morris found out about this, the company asked DeNobel and Mele to withdraw the paper. The company also stopped funding their research and shut down their laboratory (Hilts 1994a, 1994b). In general, it makes sense, from an economic perspective, for a company to fund research that tends to promote its interests, and several studies support this hypothesis. Davidson (1986) showed that results favoring a new therapy were more likely if the research is supported by the manufacturer of the new therapy. Friedberg et al (1999) reported that only 5% of new cancer drug studies sponsored by the company developing the new drug reached unfavorable conclusions as compared to 38% of studies on the same drugs sponsored by other organizations.

Since funding affects the first step of any research plan—the selection of a problem or question to study—biases in funding influence the entire research process as well as research outcomes (Longino 1990). The results of research reflect in a very profound way the economic, political, social, and other factors that determine research funding. Thus, this is a very important source of bias in research that merits attention and scrutiny. However, there is probably no way that one could eliminate this source of bias. Funding decisions and decisions about problem selection have always been and always will be influenced by human values as well as social, economic, political, and other factors (Longino 1990). Moreover, we should not attempt to eliminate this source of bias, since these biases have moral and political legitimacy. As long as we live under a democratic system where private companies are allowed to sponsor research, then principles of free market economics and freedom of expression imply that private companies should be able to sponsor research that promotes their interests. Moreover, democratic principles also support the idea that the public should have some say in how the government funds research. If the people decide that the government should spend more money on HIV research and less on cancer research, then it is their right, in a democratic society, to voice this preference and influence government decisions (Dickson 1988, Jasanoff 1990, Folz 1999, Dresser 1999). Thus, funding priorities, and by implication problem selections, should reflect human values as well as economic and political circumstances. We can strive to make these funding decisions fair, well balanced, in the public’s interest, but we cannot and should not eliminate these sorts of biases.

Once researchers have found a problem to study, biases can affect other parts of the research process, including the literature search, study design, methods, data gathering, and even data analysis. For better or worse, many researchers these days have financial interests, i.e. conflicts of interest, that may affect their ability to conduct research (Resnik 1998b, Krimsky et al 1996, Friedberg et al 1999). Conflicts of interest can affect scientific judgment, decision-making, motivation and behavior (Resnik 1998b).¹ These interests may include stock options or other investments, private contracts with companies, and intellectual property interests, such as patents. For example, consider a researcher developing a new drug for a pharmaceutical company. He or she may be receiving private funding for the research as well as other compensation from the company, such as stock, salary, or a flat fee. He or she may also be pursuing some share of possible patent rights with the company. All of these financial interests give the researcher a financial stake in the outcome of the research: if the research shows that the drug is safe and effective (or causes people to believe that the drug is safe and effective), then the researcher may stand to earn a great deal of money. Although these interests may be best promoted by conducting ethical and objective research, this need not be the case. Researchers can also benefit economically from unethical and biased research, and many well-publicized cases of fraud and misconduct in science attest to this fact (Broad and Wade 1993).

One way that financial interests can affect research is that they may make researchers hesitant to share preliminary ideas, data, or results prior to submitting a publication or patent application, especially when intellectual property interests are at stake. A researcher who shares information about a new invention could jeopardize any possible or pending patents rights if that information is disclosed before she obtains a patent. Several studies have shown that concerns about intellectual property rights now play a key role in contributing to secrecy in science (Altman 1996, Gibbs 1996, Blumenthal 1997, Bowie 1994). Even researchers who do not have patents at stake may be concerned that they will lose the race for priority and proper credit if they share information (Hull 1988). Researchers who are receiving private funding may be legally obligated to not share information prior to publication if they sign contracts limiting their rights to disclose data or results. Even researchers who submit their work for publication may decide to leave out some key items of information, usually specific details crucial to running an experiment or conducting a test, in order to protect their work. Peer reviewers may not be able to see that these key details (or "tricks of the trade") are missing (Grinnell 1992, Broad and Wade 1993). In the dispute over Cold Fusion, some commentators have speculated that the researchers did not disclose all of the important details of their experiments in communications with scientists and the media and that this contributed to problems with replicating their results (Huizenga 1992).

What does secrecy have to do with bias? Although the mere fact that a person is unwilling to share information about his or her research does not automatically make that research biased, openness is one of the most important ways of avoiding bias in science (Longino 1990, Resnik 19998a, Munthe and Welin 1996). Sharing ideas is a way of exposing them to criticism and feedback. People who are not members of the research team may be able to spot problems with hypotheses, assumptions, research designs or methods that members of the team are not able to recognize, due to their own gullibility or prejudice. Indeed, peer review and replication both can not take place without openness. In order to review a scientific study or replicate the results, one must have all the information that is required to understand and implement the study (LaFollette 1992).

Although the evidence shows that financial interests are contributing to the climate of secrecy in research, we should not forget that the conflict between secrecy and openness in science is not new and would exist even if researchers had no financial interests at stake (Resnik 1998a). Researchers would still have several have reasons to avoid sharing their data or ideas. Some would hesitate to share information in order to protect undeveloped research from premature exposure to criticism. For example, Darwin waited over twenty years to expose his theory of natural selection to public criticism. Others would refuse to share ideas in order to avoid having them stolen. For instance, Leonardo Da Vinci wrote in mirror writing to protect his ideas. Researchers working on secret military projects would have deal with issues of secrecy, openness, and national security (Kennedy 2000). Researchers working with human beings need to concern themselves with protecting confidentiality information about research subjects (Brody 1995). Thus, financial interests most certainly contribute to a climate of secrecy in research, but they have not created all of these threats to openness.

Financial interests can also affect many other judgments and decisions in the research process (Resnik 1998b, Friedberg et al 1999). Many steps in research require scientists to make judgments or decisions based on complex information. For example, in conducting a literature search, a researcher must decide which articles to read and how to interpret them. In choosing research methods, a researcher must decide which methods are appropriate and whether they require some special modifications in the context of the research. In experimental design, a researcher must control research biases and be sensitive to unanticipated problems with a study design. When analyzing data, a researcher must decide which statistical techniques would provide the best analysis of the data.

Since these judgments and decisions often require a great deal of experience, skill, and critical reflection, it is easy to for researchers to make erroneous judgments or decisions. Since human reasoning and cognition is fallible, scientists may make mistakes or succumb to subtle biases when they have a financial stake in the outcome of research (Resnik 1998b). Although financial interests do not automatically undermine scientific judgments and decisions or invalidate them, the still affect their reliability and trustworthiness (Resnik 1998b). Even scientists with good motives may fall prey to self-deception and other subtle biases when a great deal of money is at stake. For instance, many commentators have argued that financial interests probably had an effect on the error and self-deception that occurred in cold fusion research (Huizenga 1992).

Of course, not all researchers have honorable or "pure" motives, and financial interests may also corrupt motivation and behavior. Even if a researcher knows how to develop and implement a protocol that will promote objectivity and avoid bias, he or she may decide to deviate from the protocol (or change the protocol) in order to promote a particular outcome. In some cases, researchers have committed various types of scientific misconduct, such as fabrication or falsification of data or plagiarism for financial gain (Broad and Wade 1993). For example, financial interests played an important role in fraud that occurred in the Second European Stroke Prevention Study (ESPS2). An independent audit of the trial showed that researchers had fabricated data for 438 patients. The company sponsoring the trial, Boehringer Ingleheim, paid participating centers $1500 for each patient enrolled in the trial (Enserink 1996). Money also probably played a role in William Summerlin’s fraudulent research on skin transplantation. Summerlin claimed to have developed a technique that would allow one to transplant skin patches from black mice to white mice. To prove that his method worked, he used a black marker to color black patches on the white mice (Broad and Wade 1993). Any technique that could increase the success rates of organ or tissue transplantation would earn its inventor millions of dollars and career in research.

Although published data do not show that fraud is very common in science, research misconduct does occur, and the mere fact that it does occur can have a profound effect on the culture of research and the public’s perception of science (Broad and Wade 1993, Resnik 1998a). The Office of Research Integrity (ORI) received about 1,000 allegations of misconduct from 1993 to 1997 and concluded that misconduct occurred in 76 cases. A survey of 2,600 scientists and students showed that 50% were aware of ethically questionable research practices and 9% knew of cases of fabrication, falsification, or plagiarism (Swazey et al 1993). A different survey of researchers indicated that about 20% were aware of serious breaches of research ethics (Abbott 1999). Although many factors contribute to scientific misconduct, such as the pressure to produce results and poor supervision of subordinates, most commentators agree that financial interests play a role in many cases (PSRCR 1992, LaFollette 1992). Even if researchers do not engage in forms of research bias that the Federal Government defines as "misconduct," they still may use statistical techniques to manipulate, trim, fudge, or cook data (Resnik 2000, DeMets 1999, Swazey et al 1993).

It should also be mentioned that individual researchers should not been held entirely accountable for financially related biases that occur in designing and implementing protocols. Private corporations, such as pharmaceutical companies, can also rig experiments in order to produce outcomes that favor their products (Crossen 1994). For example, a company might develop a study design that does not generate data on an important side effect of a drug. A company might also use a statistical technique that omits some data from the analysis in order to produce more favorable results. The FDA is acutely aware of this kind of industry bias, and this is one reason why the agency requires independent data and often conducts its own independent studies (Brody 1995). Even the financial interests of government agencies may have an impact on research. For example, researchers working on the Strategic Defense Initiative (SDI) faked some crucial tests to convince Congress that funding for the program should be continued (Weiner 1994).

Like biases in problem selection, it is also not realistic to expect that we can eliminate financial biases that can occur in designing and implementing research. If we take a historical perspective on this problem, we see that it is not really new. For example, Antoine Lavoisier earned money from his work in chemistry, James Watt profited from his design of the steam engine, Michael Faraday earned money from work in electromagnetism, and Thomas Edison profited from his many inventions (Meadows 1992). Since the 1800s, private companies, such as Ford, Westinghouse, DuPont, IBM, General Electric, Microsoft, and Burroughs-Wellcome, have profited greatly from their investments in R & D. Today, a good R & D program is the key to success in many high-tech industries, such as pharmaceuticals, software, and genomics. As long as we live in a society where (1) research costs money, (2) and the outcomes of research have economic impacts, then it will be impossible to eliminate financial interests that may affect the design and implementation of research protocols.

However, there are some features of today’s research environment that are new or at least especially vexing. First, more money is at stake now than in the past. For all of his inventions, Edison was not extraordinarily rich. Today, a researcher can become an instant multi-millionaire from one patent on a chemical process or new drug (Dickson 1988). Second, more scientists and universities today are seeking patents. Third, few researchers in the past received stock or stock or other investments. Fourth, as noted earlier, industrial sponsorship of research has increased dramatically. Not only are companies spending more money, but more companies are sponsoring R & D.

Finally, the link between financial interests and objectivity has been broken. In the recent past, e.g. 1950-1980, it was not as easy to make money in research through fraud or bias. Inventions and other economically lucrative research products had to pass some kind of test that would provide some measure of objectivity, such as peer review or success in application, before they could generate money. This is still the case with patented inventions, which cannot be patented without some proof that they work. Today, however, it is possible to make a great deal of money by merely getting people to believe that a research product works. For example, if a company can publish a study demonstrating the superiority of its drug or other product, then this study may still generate millions of dollars in revenue or stock appreciation, even if it later turns out to be fraudulent or biased. The same thing may happen when a company publishes an ethical and unbiased study, of course, but what matters, from a purely economic point of view, is how the study affects "the bottom line." This point was mentioned earlier but it bears repeating now: in today’s research environment many companies and researchers value objectivity as a means to an end but not as an end in itself.

Financial biases can also occur after a study has been completed. Many companies that sponsor research require investigators to sign contracts giving the company the right to review and approve research prior to publication (Hess 1999). If the research is unfavorable to the company, then it may block or delay publication or insist on writing its own interpretation of the data. As mentioned previously, Philip Morris blocked the publication of DeNoble and Mele’s paper on nicotine addiction. In another well-known case, Boots Company made Betty Dong withdraw a paper on drugs used to treat hypothyroidism that had been accepted for publication in the Journal of the American Medical Association (JAMA). Boots had funded her research so that she could show that its product, Synthroid, was superior to alternative drugs. But Dong reached the opposite conclusion: she showed the several alternative medications were as safe and effective as Synthroid and also less expensive. Boots spent several years trying to discredit her research and threatened to sue Dong if she published the paper. Dong had signed an agreement with Boots not to publish the results without written consent from the company (Wadman 1996).

Companies can suppress research for other reasons (Hess 1999). A drug company may decide that it does not wish to share data or results in order to prevent a competing company from gaining an advantage. For example, a drug company might not want other companies to know about its research methods, protocols, plans, techniques, or undeveloped drugs because this information could be useful to other companies. If the company decides to take steps to protect this information and this information has a business value, then it has the right, under trade secrecy law, to not share this privileged information. Scientists that disclose trade secrets can also face fines or imprisonment (Bowie 1994). For example, Philip Morris treated much of its tobacco research as a trade secret and the company had no intention of sharing this research or publishing it (Wadman 1996).

Suppression of research is an obvious type of bias in dissemination, but other, more subtle biases can occur as well. There are several different types of biases that can occur in publication. First, as we noted earlier, there are financial biases in the funding of research. These biases can lead to biases in research results as well as biases in the published research record (Easterbook et al 1991, Misakian and Bero 1998, Resnik 2000). For example, if a drug company sponsors and publishes ten studies demonstrating the effectiveness of its new medication and the NIH sponsors and publishes two studies with mixed results, it will appear to those who read and summarize the literature that 80% of the studies show that the drug is effective. If half of those studies had been funded by an independent agency, instead of the drug company, then the publication record might look very different.

Editorial practices can also encourage biases in publication. First, many journals prefer to publish original research instead of research that attempts to replicate or review previous work. However, all "negative" studies are, by definition, not original, since they attempt to repeat previous studies in order to disprove them. If a drug company sponsors "positive" studies on its new drug, and journals are less inclined to publish negative findings, then this can lead to a bias in favor of the company (Eastebrook et al 1991). Second, most journals also prefer to only publish research that meets an acceptable level of statistical significance, such as P-value = 0.05 or less. Companies can take advantage of this tendency by carefully selecting studies and publishing only positive results that are statistically significant (Resnik 2000, DeMetes 1999). For instance, suppose a company sponsors 12 studies, 6 of which achieve statistically significant results. If 5 of the significant studies yield favorable results and 5 of the insignificant studies yield unfavorable results, and the company publishes only the significant results, then many unfavorable (though insignificant) studies will not be published. Once again, this skews the research record in favor of the company.

The net effect of these publication biases is that they have a profound impact on the publication record. It is not difficult at all for a company to skew the research record through its strategies for disseminating information. Moreover, if the publication record is biased, then even a thorough literature review or well-designed meta-analysis will not be able to overcome this problem, since these methods for analyzing research must rely on the data present in the published record (Resnik 2000). It’s a case of "bias in, bias out," so to speak.

The final type of bias in dissemination worth mentioning concerns marketing. Most drug companies spend as much (and often more) on marketing as they do on R & D. For example, Pfizer spent 39.2% of its revenues on marketing and administration in 1999 and other companies also spend a great deal of their money on marketing (Angell 2000). How can marketing affect objectivity? Let’s take the pharmaceutical industry as an example. The goal marketing is to increase sales. Drug companies increase sales by educating (and some would say manipulating) consumers, e.g. patients and health care professionals. If a drug company manufactures a new drug for sexual dysfunction and runs a successful marketing campaign, then physicians will prescribe the drug and patients will demand it. Successful sales will lead to profits for the company which it can reinvest in R & D on the same drug or on similar drugs, and these R & D investments therefore play a key role in future drug development and research. Other companies may attempt to profit from the company’s success and they may develop similar drugs (Brody 1995). All of these events can occur even if the drug is not as safe or effective as other treatments; what matters, from a business perspective, is that the drug sells. Thus, successful marketing campaigns can indirectly contribute to research biases by affecting R & D investments of drug companies.

Like the other previously mentioned financial biases, it is difficult or impossible to eliminate biases relating to dissemination. As long as we live in a democratic system that operates on free-market principles, then companies and researchers will be free to make contractual arrangements that allow the company to review or approve research. Individual researchers may refuse to sign these agreements, but there is little basis for making them illegal since these contracts meet legal standards concerning fairness and disclosure. Even if all scientists refuse to sign these contacts and all results are published, companies can still find ways of biasing the publication record through funding decisions. Companies can stop funding research unfavorable research and they can fund favorable research. It also would seem to be difficult to exercise much control over marketing practices. Government agencies, such as the Federal Trade Commission (FTC) and the FDA have only limited control over marketing practices (Brody 1995). The FTC can prevent companies from running fraudulent or deceptive advertisements, but companies do not need to lie in order to influence consumer decisions. The FDA can regulate claims made on labels about the therapeutic, diagnostic, or prognostic uses of foods and drugs, but companies do not need to make such claims on product labels in order to market their goods. Thus, for better or worse, financial biases relating to the dissemination of information are probably here to stay.

Section 3 of this paper has painted a rather dismal picture of the affects of financial interests on scientific objectivity. The main point has been that these biases can affect 1) problem selection, 2) development and implementation of a protocol, and 3) dissemination of results and that we should not expect that these biases can be eliminated. Many of these biases have existed been present for hundreds of years and most of them are here to stay. In Section 2, I argued that objectivity is possible and that it is worth pursuing. Has Section 3 undermined this claim? Given what I have said about financial biases in research, one might conclude that a) objectivity is not achievable and b) objectivity is not worth pursuing, even as an ideal. It is better to come to terms with the financial biases in research and to stop maintaining the illusion of objectivity.² There is no need, therefore, to take any steps to mitigate or manage these biases since they are inevitable.

I am not that willing to give in to this cynical view. While I am painfully aware of how financial biases can affect research, I still think objectivity is worth pursuing as an ideal for the reasons articulated in Section 2. Given the moral, political, and cultural importance of objectivity, we need to take steps to at least try to mitigate or manage these biases. We live in an imperfect world, but we can work toward making it better. Although we all recognize that a perfectly "just state" will never exist, very few people we therefore conclude that we should not enact any policies that are designed to promote justice. Likewise, we should not conclude from the impossibility of "objective research" that we should not enact any policies to promote objectivity.

Very briefly, I will mention a few policies relating to the financial aspects of research that I think can help to mitigate or minimize biases. I will also explain why I think they help promote objectivity. Many of these policies are already in place are under consideration. By "mitigate" I do not mean eliminate, since biases will continue to exist in any case. By "manage" I mean "direct or steer in a desired direction." We cannot expect that all aspects of the research process will be free from biases, but we can hope that objectivity will emerge as a result of the interplay and interaction of different interests and biases in research. Moreover, in formulating our science policy we can assume that the social structure of research, as manifested in informal rules, official guidelines, social roles, and institutional arrangements, can have an impact on the objectivity of research (Longino 1990, Solomon 1994).

Policy: Use government funding to counterbalance private funding. For example, if a drug company sponsors research on a new drug, then the government should also sponsor research on that drug to help insure that studies that might be unfavorable will be funded or published. The FDA already requires independent research on new drugs, foods, and medical devices, and it sponsors some of its own studies. This organization’s funding has been threatened and cut back in recent years (Resnik 1999). If we want to promote objective research on drugs, it is crucial that the FDA have a budget sufficient to promote independent research and review. Moreover, other government agencies that sponsor research, such as the NIH, NSF, and EPA, should have funding adequate to provide independent research and review, since financial biases can affect engineering studies, environmental impact statements, ecological research, and many other aspects of industry-sponsored science.

Policy: Allow for adequate scientific and public input into government funding decisions. This policy is useful in order to help overcome some of the biases that can infect the process of evaluating research proposals and grants. Thorough scientific review is required to insure that proposed projects meet standards of scientific acceptability, but public input is also needed to insure that the financial and professional interests of a small group of researchers (i.e. an "old boys" network) do not dominate the whole funding process. For example, cancer researchers have financial and professional interests in sponsoring more cancer research, aerospace scientists have financial and professional interests in promoting more aerospace research, and so on. The general idea here is that public input and openness can help balance or some of the biases that can occur in government funding (Dresser 1999). Funding decisions will still be biased, but we can still take steps to make these biases are fair or well-balanced.

As an aside, one might object that the use of government funds and public decision-making may introduce other biases into the research process, depending on how different researchers and interest groups attempt to influence funding priorities. All of our experiences with government sponsored research, from cancer research to HIV research, to human embryo research, provide us with evidence that the process of allocating public research dollars is far from objective (Resnik 1998a, 1999). I admit this much. Objectivity may not be found in any real world process. But the hope here is that these other political and social biases will counter-act or counter-balance the biases created by private industry and will therefore make research more objective than it would have been otherwise.

Policy: Disclose and manage conflicts of interest in research. Many different organizations, agencies, universities, and journals have adopted conflict of interest policies (Resnik 1998b). Most of these policies involve some type of disclosure of financial interests and sources of funding, and some may even include some type of recusal, i.e. eliminating the situation that creates the conflict, when conflicts of interest pose a dire threat to the objectivity of research. The rationale for disclosure is that it will allow people who are interested in research to understand the financial aspects of the research and to take these into account when interpreting and evaluating the research (Resnik 1998b). For example, it is useful to know that a drug company has funded a study on its new drug and that several of the researchers have stock in the company. While these interests do not invalidate the study, they may be reasons to give it more careful scrutiny. When disclosure does not suffice to handle a conflict, recusal may be the best strategy. For example, some journals do not allow researchers with significant financial interests to write review articles about drugs or medical devices. Some commentators have suggested that scientists should not accept stock but should only accept salary or a fee as financial remuneration (Resnik 1998b). Although I think conflict of interest policies are promoting objectivity in research, we need to expand the use of these policies. Most journals and funding agencies have conflict of interest polices, but these polices should be extended to all areas of research that are affected by financial interests, such as peer review, expert testimony, and hiring and promotion decisions.

Policy: Audit data. In the business world, independent audits help prevent fraud, bias, and error by providing an objective review of financial bookkeeping. Some writers have argued that independent audits of research data can have a similar impact on research. Data auditing involves (1) the careful examination of original data contained in laboratory notebooks and other records and (2) comparison of the original data with processed data as well as published data. An audit is not perfect—it cannot stop fraud or other forms of intentional deception—but it can catch discrepancies, errors, statistical biases, and other problems that can be inferred from the data record. Data audits can also help promote an atmosphere of honesty and responsibility in that they make researchers more accountable for their research (Shamoo 1989, Glick and Shamoo 1993).

Policy: Increase access to data and results. As mentioned earlier, sharing of data and results is an important strategy for overcoming bias but researchers may refuse to share information prior to or after completing a study in order to protect or promote individual or corporate financial interests. It may be very difficult to take steps to encourage researchers to share data/results prior to completing a study without overhauling the copyright and patent system, since one of the main reasons why researchers refuse to share data/results is to protect intellectual property interests.³ It may also be difficult to force companies to share research that they treat as confidential business information without revising trade secrecy laws (Resnik 1998c). However, there are some simple and modest steps that one could take to increase access to data. First, one could require that all publicly funded research be accessible once a study is complete. Indeed, Congress passed a bill last year requiring federal granting agencies to ensure that all data produced under a federal grant are available to the public through the Freedom of Information Act. This bill still allows researchers to protect confidential information pertaining to human subjects and does not allow the public to have access to data/results prior to completion of a study (Frankel 1999). Second, one could establish databases for research that (for one reason or other) is not published in peer-reviewed journals, such as negative findings or studies that do not achieve acceptable levels of statistical significance (Resnik 2000). Making these data available would help overcome some of the publication biases discussed earlier. One might argue that these databases could do more harm than good because they would permit the dissemination of sub-standard research. However, as long as researchers who access these databases are aware that they have not been peer-reviewed, they might be able to use the data without being misled. They might find some diamonds in these rough databases, and the only way to find these potential diamonds is to collect many rocks.

This paper has attempted to show how financial interests can create biases in all stages of the research process, from selecting a problem to implementing a protocol to publishing results. It has also defined the concept of objectivity and argued that there are moral, political, and cultural reasons why scientists should still strive to be objective, even if it is impossible to eliminate all biases, including financial ones, from the research process. In order to promote objectivity, scientists should follow policies to mitigate and manage financial interests. Some useful policies include using government funding to counter-balance private research, increasing public input into funding decisions, disclosing and managing conflicts of interest, auditing data, and increasing access to data and results.

In closing, it also worth noting that the whole question of financial interests in research would benefit from further empirical studies on the financial interests of individual researchers and corporations and the affects of these interests on objectivity. Additional critical studies of conflict of interest policies, intellectual property, publication policies, or funding policies would also aid any progress we make on issues relating to financial interests. I encourage sociologists, anthropologists, economists, political scientists, and other science scholars to explore these issues in more depth.

Notes

1. An individual "conflict of interest" is a situation where a person has financial or other interests that undermine her ability to meet her obligations. A person may fail to meet her obligations by making a mistaken judgment or decision or by committing an unethical or illegal act. In science, researchers have an obligation to act objectively, and conflicts of interests can therefore interfere with their ability to fulfill this duty. Institutions may also have conflicts of interest. For further discussion see Resnik (1998b).
2. One might also respond to this problem by attempting to remove all financial interests from research. The only way to do this would be to "socialize" science making all scientists government employees and by letting the government make all funding and publication decisions. This option would not eliminate biases, however, since political and social biases would continue to affect research. Countries that attempted to socialize science, such as the Soviet Union, have introduced ideological and other biases in research (Joravsky 1970).
3. For example, one could extend patent protection to the period prior to filing a patent in order to give researchers extra protection. A scientist could file a claim of "intent" to patent which would protect his research if he or she later decides to file a patent.

References

Abbott, A. 1999. Science Comes to Terms with the Lessons of Fraud. Nature 398:13-17.

Agnew, B. 1999. NIH Invites Activists into the Inner Sanctum. Science 283: 1999-2001.

Altman, L. 1996. Secrecy is Hurting Medical Research, a US Official Says. New York Times (10 February 1996): A11.

Angell, M. 2000. The Pharmaceutical Industry—To Whom Is It Accountable? New England Journal of Medicine 342: 1902-4.

American Physical Society (APS). 1991. Guidelines for Professional Conduct. Washington: APS.

American Statistical Association (ASA). 1999. Ethical Guidelines for Statistical Practice. Alexandria, VA: ASA.

Associated Press. 2000. Report: Diet Drug Firm Accused of Funding Favorable Journal Articles. The Daily Reflector, 24 May 2000: A3.

Associated Press. 1993. Drug Make Lilly Accused of Cloaking Toxic Reaction. The Denver Post, 10 December 1993: A22.

Beardsley, T. 1994. Big-Time Biology. Scientific American 271, 5: 90-94.

Bloor, D. 1991. Knowledge and Social Imagery, 2^nd ed. Chicago: University of Chicago Press.

Blumenthal, D. 1997. Withholding Research Results in Academic Life Science: Evidence from a National Survey of Faculty. JAMA 277: 1224-8.

Bowie, N. 1994. An Evaluation of University-Business Partnerships. In: Bowie, N. (ed.). University-Business Partnerships: an Assessment. Lanham, MD: Rowman and Littlefield, pp.45-99.

Broad, W. and Wade, N. 1993. Betrayers of Truth, new ed. New York: Simon and Schuster.

Brody, B. 1995. Ethical Issues in Drug Testing, Approval and Pricing. New York: Oxford University Press.

Collins, H. 1985. Changing Order. London: Sage Publications.

Crossen, C. 1994. Tainted Truth. Simon and Schuster, New York, N.Y.

Davidson, R. 1986. Source of Funding and Outcome of Clinical Trials. Journal of General Internal Medicine 1: 155-8.

DeMets, D. 1999. Statistics and Ethics in Medical Research. Science and Engineering Ethics 5: 97-118.

DHHS. 1994. Objectivity in Research. Federal Register 59, 123:33242-33251.

Dickson, D. 1988. The New Politics of Science, 2^nd ed. Chicago: University of Chicago Press.

Dresser, R. 1992. Wanted: Single, White Male for Research. Hastings Center Report 22,1: 24-29.

Dresser, R. 1999. Public Advocacy and Allocation of Federal Funds for Biomedical Research. Milbank Quarterly 77, 2: 257-82.

Easterbrook P, et al. 1991. Publication Bias in Clinical Research. Lancet 337: 867-72.

Enserink, M. 1996. Clinical Trials: Fraud and Ethics Charges Hit Stroke Drug Trial. Science 274:2004-5.

FDA. 1998. Financial Disclosure by Clinical Investigators. 21 Code of Federal Regulations 54.

Foltz, F. 1999. Five Arguments for Increasing Public Participation in Making Science Policy. Bulletin of Science, Technology, and Society 19,2: 117-27.

Frankel. M. 1999. Public access to data. Science 283: 1114.

Friedberg, M. et al. 1999. Evaluation of Conflict of Interest in New Drugs Used in Oncology. Journal of the American Medical Association 282:1453-7.

Fuchs, S. 1992. The Professional Quest for the Truth. Albany, NY: State University of New York Press.

Fuller, S. 1988. Social Epistemology. Bloomington: Indiana University Press.

Gibbs, W. 1996. The Price of Silence: Does Profit-minded Secrecy Retard Scientific

Progress? Scientific American 275, 5: 15-16.

Giere, R. 1991. Understanding Scientific Reasoning, 3^rd ed. Chicago: Holt, Rinehart and Winston.

Glick, J and Shamoo, A. 1993. A Call for the Development of "Good Research Practices" (GPR) Guidelines. Accountability in Research 2: 231-35.

Gould, S. 1981. The Mismeasure of Man. New York: WW Norton.

Gross, P. and Levitt, N. 1994. Higher Superstition. Baltimore: Johns Hopkins University Press.

Grunbaum, A. 1985. The Foundations of Psychoanalysis. Berkeley: University of California Press.

Haerlin, B. and Parr, D. 1999. How to Restore Public Trust in Science. Nature 400: 499.

Harding, S. 1986. The Science Question in Feminism. Ithica, NY: Cornell University Press.

Hess, D. 1999. Suppression, Bias, and Selection in Science: the Case of Cancer

Research. Accountability in Research 6, 4: 245-257.

Hilts, P. 1994a. Philip Morris Blocked ’83 Paper Showing Tobacco is Addictive, Panel Finds. New York Times (1 April1994):A1.

Hilts, P. 1994b. Two Scientists Testify Against Tobacco Firm. New York Times (29 April 1994): A1.

Howson, C. and Urbach, P. 1989. Scientific Reasoning. La Salle, Illinois: Open Court.

Hull, D. 1988. Science as a Process. Chicago: University of Chicago Press.

Huizenga, J. 1992. Cold Fusion. Rochester, NY: University of Rochester Press.

Jaffe, A. 1996. Trends and Patterns of Research and Development Expenditures in the United States. Proceedings of the National Academy of Sciences 93: 12658-12663.

Jasanoff, S. 1990. The Fifth Branch: Science Advisors as Policy Makers. Cambridge, MA: Harvard University Press.

Jasanoff, S. 1995. Science at the Bar. Cambridge, MA: Harvard University Press.

Joravsky, D. 1970. The Lysenko Affair. Cambridge, MA: Harvard University Press.

Kahneman, D.; Slovic, P.; Tversky, A. 1982. Judgment Under Uncertainty. Cambridge: Cambridge University Press.

Kennedy, D. 2000. Science and Secrecy. Science 289:722.

Kitcher, P. 1992. The Naturalists Return. The Philosophical Review 101, 1: 53-110.

Kitcher, P. 1993. The Advancement of Science. New York: Oxford University Press.

Kitcher, P. 1996. The Lives to Come. New York: Simon and Schuster.

Kosso, P. 1992. Reading the Book of Nature. Cambridge: Cambridge University Press.

Krimsky S, et al. 1996. Financial Interests of Authors in Scientific Journals: a Pilot Study of 14 Publications. Science and Engineering Ethics 4: 395-410.

Kuhn, T. 1962. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.

Kuhn, T. 1977. The Essential Tension. Chicago: University of Chicago Press.

Panel on Scientific Responsibility and the Conduct of Research. 1992. Responsible Science. Washington: National Academy Press.

LaFollette M. 1992. Stealing into Print. Berkeley: University of California Press.

Lakatos, I. 1978. Philosophical Papers, J. Worrall and G. Curie (eds.). Cambridge: Cambridge University Press.

Latour, B. and Woolgar, S. 1986. Laboratory Life, 2^nd ed. Princeton: Princeton University Press.

Laudan, L. 1984. Science and Values. Berkeley: University of California Press.

Laudan, L. 1990. Science and Relativism. Chicago: University of Chicago Press.

Longino H. 1990. Science as Social Knowledge. Princeton, NJ: Princeton University

Press.

Meadows, J. 1992. The Great Scientists. New York: Oxford University Press.

Merton, R. 1973. The Sociology of Science. Chicago: University of Chicago Press.

Misakian A and Bero L. 1998. Publication Bias and Research on Passive Smoking:

Comparison of Published and Unpublished Studies. Journal of the American

Medical Association 280: 250-53.

Munthe, C. and Welin, S. 1996. The Morality of Scientific Openness. Science and Engineering Ethics 2: 411-28.

NAS. 1994. On Being a Scientist. Washington: National Academy Press.

Nelkin, D. 1995. Selling Science, revised ed. New York: WH Freeman.

ORI. 1999. Federal Policy on Research Misconduct. Federal Register 64, 198: 55722-55725.

Peirce, C. 1955. Philosophical Writings of Peirce. New York: Dover.

Popper, K. 1959. The Logic of Scientific Discovery. London: Routledge.

Popper, K. 1972. Objective Knowledge. Oxford: Clarendon Press.

Raloff, J. 1994. The Great Nicotine Debate: Are Cigarette Recipes Cooked to Keep Smokers Hooked? Science News 145, 20: 314-317.

Resnik, D. 1998a. The Ethics of Science: an Introduction. New York and London: Routledge.

Resnik D. 1998b Conflicts of Interest in Science. Perspectives on Science 6, 4: 381-408.

Resnik D. 1998c Industry-sponsored Research: Secrecy Versus Corporate Responsibility. Business and Society Review 99,1: 31-35.

Resnik D. 1999. Privatized Biomedical Research, Public Fears, and the Hazards of Regulation: Lessons from Stem Cell Research. Health Care Analysis 7: 273-87.

Resnik, D. 2000. Statistics, Ethics, and Research: an Agenda for Education and Reform. Accountability in Research 8: 163-188.

Shamoo, A. (ed.). 1989. Principles of Research Data Audit. New York: Gordon and Breach: 111-120.

Shrader-Frechette, K. 1994. The Ethics of Scientific Research. Lanham, MD: Rowman and Littlefield.

Sokal, A. 1996. Transgressing the Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity. Social Text 46/47: 217-52.

Solomon, M. 1994. A More Social Epistemology. In F. Schmitt (ed.) Socializing Epistemology. Lanham, MD: Rowman and Littlefield, pp. 218-233.

Swazey, J. et al 1993.Ethical Problems in Academic Research. American Scientist 81: 542-53.

Tauer, C. 1999. Testing Drugs in Pediatric Populations: the FDA Mandate. Accountability in Research 7: 37-58.

Thompson, D. and Gutman, A. 1996. Deliberative Democracy. Cambridge, MA: Harvard University Press.

Wade, N. 1994. The Erosion of the Academic Ethos: the Case of Biology. In: Bowie, N. University-Business Partnerships: an Assessment. Lanham, MD: Rowman and Littlefield, pp. 143-158.

Wadman, M. 1996. Drug Company Suppressed Publication of Research. Nature 381: 4.

Weiner, T. 1994. Inquiry Finds "Star Wars" Plan Tried to Exaggerate Test Results. New York Times (22 July 1994): A1.

Whitbeck, C. 1998. Ethics in Engineering Practice and Research. Cambridge: Cambridge University Press.

Ziman, J. 1996. Is Science Losing Its Objectivity? Nature 382: 751-754.

Zolla-Parker, S. 1994. The Professor, the University, and Industry. Scientific American 270, 3:120.