7 Exclusions, Biases and Conflicts of Interest
7.1 Why are exclusions, biases and conflicts of interest ethical concerns?
Exclusions
While in an ideal world every piece of data would be valid and relevant to a given experiment, that is simply not the case. Therefore, when we are conducting an experiment, we must carefully consider not only the data that we wish to use but also the data that we exclude. Ethically determining what data is excluded should occur at the beginning of an experiment. This is called exclusion criteria, which will determine the eligibility of samples, subjects or data (Neves & Amaral, 2020). When determining these criteria, they must also be justified. Unfortunately, sometimes scientists unethically exclude data because it is easier or reduces confounding variables. For example, physiology experiments sometimes exclude females to reduce the impact of hormonal fluctuation. Historically, this has resulted in an underrepresentation of females, especially in biomedical studies, causing females to have worse medical outcomes because their physiology is being assessed based on male traits (Wilson et al., 2020). Ethically speaking, it is not necessarily bad science to conduct an experiment in this way, in fact it is sometimes necessary if, for instance, the experiment is studying anatomy that one sex lacks (such as in the case of prostate cancer research). However, scientists should justify the exclusion and take care not to generalize their findings to an entire species if they have excluded half of the population (Lew et al., 2022). Not every possible problem can be anticipated and therefore, pre-established exclusion criteria may not be optimal (Neves & Amaral, 2020). However, by establishing these criteria in advance, the risk of confirmation bias is reduced.
Bias
Confirmation bias is the human tendency to process information based on preexisting beliefs, expectations, or desires (Casad & Luebering, 2024). Researchers are understandably invested in their work, and with the pressure to publish, can lead them to preferentially select for a specific outcome, at times ignoring conflicting data. This can shape how we collect, analyze and interpret data (Guiney et al., 2020; Casad & Luebering, 2024). As a result, confirmation bias can cause scientists to draw incorrect conclusions which taint our basis of knowledge and contributes to the crisis of trust that has developed between science and the public (Flier, 2022). It is vital for scientists to do their utmost to mitigate confirmation bias through transparent, honest research with established standards to support or reject hypotheses (Neves & Amaral, 2020). All scientists should be aware that they have biases, both conscious and unconscious, and carefully consider if their opinions and conclusions are based on evidence or their own expectations (Guiney et al., 2020; Flier, 2022).
Confirmation bias is not the only type of bias that scientists must contend with. Unintentional bias is perhaps the most challenging to manage but with consideration and ethically-minded thinking, it is possible to reduce the impact of bias. Bias can begin at the outset of an experiment when choosing a topic of inquiry where a researcher may choose a topic based on funding availability (or some other incentive) rather than the value of the research itself. It can also affect how scientists design their experiments, the methods they use to collect and analyze their data and even how they read and present their results. Unfortunately, confirmation bias can cause scientists to behave unethically as they unintentionally bias their data in their favour (Kalichman et al., 2016). Even researchers with good intentions can be biased without being aware that they are contributing to unethical behaviour. Additionally, despite efforts in recent years to improve diversity and accessibility, biases surrounding gender, ethnicity and disability within lab culture persists (Batty & Reilly, 2022; Eaton et al., 2020). In 2021, 35% of individuals working in STEM identified as women and 24% of individuals were Hispanic, black or of indigenous descent. In both instances, these individuals had lower median earnings for the same roles compared to white men (National Center for Science and Engineering Statistics, 2023). Although these statistics are much improved compared to 2011, it is important to be aware that these biases continue to have an impact on lab culture and the individuals who are welcomed into STEM spaces.
Conflicts of Interest
Biases within science can arise due to conflicts of interest. A conflict of interest (COI) is when an individual is placed in a position of actual or perceived conflict between their research or work and other interests. These interests may be personal, financial or related to the institution in which they work (Bell et al., 2022). It is an unfortunate reality that perceived impropriety can have consequences that are just as damaging as those that result from intentional misconduct (Kalichman et al., 2016). In 2024, it is easy to get immediate access to new information. New content is being created and shared constantly through social media but unlike news media in the past which required (arguably) some evidence to support reported information, social media has no such burden of proof. As a result, we are now in the age of perception where what is perceived to be true can matter as much or more than information backed by evidence (Kalichman et al., 2016). Moreover, these perceptions can be widely shared more rapidly than ever before, and the internet never forgets. To safeguard our reputations as scientists in this era of scrutiny, we must do our utmost to uphold our integrity and that of our research.
Conflicts of interest are an ethical concern because they pose a threat to a researcher’s ability to be objective and they call into question the integrity of the research (Kalichman et al., 2016). Conflicts of interest can impact many aspects of research. It can introduce bias into the experiment, cause the results to be questioned and scrutinized, and it can damage the reputation of those involved. Additionally, it can undermine the integrity of peer review as reviewers themselves can be biased. A reviewer may accept a paper outlining a poorly conducted experiment because they agree with the results. Or they might reject the paper because they are biased against the author, results, methodology or topic (Measey, 2023). Moreover, conflicts of interest contribute to the erosion of public trust in science as the public questions the integrity of work when researchers have underlying incentives (Bell et al., 2022). To combat this issue, scientific journals, academic institutions and funding groups require a conflict of interest statement from all authors and contributors when manuscripts are submitted for review (Kalichman et al., 2016; Measey, 2023).
7.2 Example: Glyphosate
Some industries such as the tobacco and oil industries have had longstanding campaigns to corrupt science and introduce contrasting scientific research to sow doubt about legitimate results that negatively impact their interests. The pesticide industry employs similar tactics. In 2015, the World Health Organization (WHO) classified the herbicide Glyphosate made by Monsanto as a probable carcinogen. Glyphosate is one of the most common herbicides used globally, and is the primary herbicide in Canada (Shochat & Fournier, 2019; Malkan, 2024). Monsanto (now owned by Bayer) has been sued multiple times by farmers and workers who developed non-Hodgkin Lymphoma after using RoundUp, the main ingredient of which is Glyphosate (Shochat & Fournier, 2019; Malkan, 2024). In response to the WHO’s classification, Monsanto hired 15 Canadian researchers who were supposedly unbiased parties to do their own studies of the side effects of Glyphosate. They published 5 papers that showed that it could not be proven to be carcinogenic. It later was revealed that 12 of the authors had been consultants for Monsanto and two authors admitted that they had worked for Monsanto directly in previous years. None of these blatant conflicts of interest were disclosed in the papers (Marcus, 2018). During a court case, it was proven that Monsanto employees reviewed and even edited some of the papers prior to publishing, engaging in unethical ghostwriting. Ghostwriting is a practice they have used in the past to dictate the narrative of Glyphosate while hiding their involvement (Malkan, 2024; Marcus, 2018). Although the journal has asked that the authors clarify their conflicts of interest, as of 2024 the papers have not been retracted, nor are they likely to be. Based on a review of these papers and other studies, many of which were not peer reviewed, Health Canada reapproved Glyphosate for use in Canada until 2032 (Shochat & Fournier, 2019). Monsanto has used the same practices in the EU and the US to ensure that their product remains in use (Malkan, 2024).
7.3 Practice Questions
- Data banks and exclusions
You are doing an experiment examining the DNA sequences of different primates to try to determine if a gene is conserved within populations. You must use a database to source your data points but not every sample contains the same information.
2. Excluding data and conflicting results
An experiment that you are performing uses several controls that indicate if the experiment was successful or not. The controls indicate that one set of data was unsuccessful and therefore there must have been an error.
As you continue the experiment, you find that you are getting conflicting results. Some of your data sets show significance and some are proving to be non-significant. The non-significant data does not support your hypothesis and out of 10 data sets, 7 were significant.
3. Conflicts of Interest
You are working on your PhD research investigating the effects of green tea on cellular metabolism. Your hypothesis is that certain types of tea are beneficial over time. Your supervisor has published papers recently that show that specific strains of herbal teas increase the rate of cellular replication and reduce water content. There has been some pushback about the results, with others in the field questioning the methodology. Your supervisor has been putting pressure on you to finish your results and get them published to provide supporting research to bolster their conclusions. Although you tell your supervisor that your preliminary results do not support their conclusions, they believe that your final results will. They further say that PhD committees are less likely to view a thesis positively if it lacks interesting results. In conversation with other students, you learn that your supervisor hopes to use both their results and yours to draw investors with the goal of making a health supplement.
