"The best way to find out if you can trust somebody is to trust them."
— Ernest Hemingway
The days of solo researchers running their own labs with full knowledge and expertise about all aspects of their research methods, data collection and data analytic techniques are quickly receding into the past. Research methods and tools have become increasingly complicated, and answering today's complex research questions requires researchers to obtain sample sizes so large that it's impossible to complete a study at one site.
These changes mean that no one person, however talented, has the expertise and resources to do it all. This shift has given rise to team science in which investigators join together to answer a researchable question. With team science, trusting other members of the group is crucial and inescapable.
In team science, investigators team up based on mutual research interests. Individual members take responsibility for implementing a particular aspect of the research, such as a method, like functional magnetic resonance imaging (fMRI) or data analysis. Team members may be at a single site but typically are located at different sites, often distant from one another. Each site is generally responsible for its portion of the data collection. While team science is increasingly viewed as an approach that can speed the pace of discovery, enhance innovation and address questions that cannot be answered in smaller, more focused studies, it brings with it ethical complexities generally not present in the single-researcher model.
When I worked as a solo researcher, I was involved in all aspects of my studies. I had direct contact with my research participants and oversaw the collection of my research data. I knew my study's methodologies well and ensured their proper implementation. I personally analyzed my data and prepared reports for publication. I could easily attest to the integrity of the data and to the fact that my research participants gave their full, informed consent and were treated with respect. With team science, I no longer have this degree of direct involvement with so many aspects of a study. Given the current diversity and complexity of methods, even if I did, I would not be able to ensure that I provided quality work if I had to use methods outside the scope of my expertise.
I am currently involved with a research project that illustrates the ethical issues that come into play in team science. I am the principal investigator of a study examining the mechanisms of action of two treatments for individuals with borderline personality disorder. It is a complex, multi-method study that aims to understand whether our treatments improve responses to interpersonal stress or cognitive control of emotions (or both). To assess response to interpersonal stress, we obtain cortisol samples before and after a stressful experience in the laboratory. To determine changes in cognitive control of emotions, we employ a task in which patients' emotional and neural responses (using fMRI) to negative memories are evaluated under two conditions: 1. where they attempt to control their reaction; and 2. where they immerse themselves in the emotions evoked by the negative memories.
As the principal investigator, I am responsible for the integrity and veracity of all aspects of data collection, processing and analysis. Yet I am not an imaging researcher, and therefore am not competent to process and analyze data obtained from the fMRI portion of this project. Furthermore, I have neither the laboratory facilities nor the expertise to analyze cortisol data.
Similarly, my colleagues responsible for conducting and analyzing the fMRI results are not clinicians. Consequently, they rely on me to ensure that the patients they are testing do, in fact, have borderline personality disorder and that the treatments are delivered in a competent manner. I feel a great sense of responsibility for ensuring the soundness of all aspects of the research process — from design to consent to data accrual to analysis. But in reality, how can I fulfill that responsibility since I do not know how to process and analyze fMRI data? The very complex analyses of my outcome data are in the hands of a statistician who is trained in clinical trials data and who analyzes it on a daily basis. As principal investigator, how do I ensure that the data are carefully and correctly analyzed? For my cortisol analyses, how can I attest that the samples are analyzed according to prevailing laboratory standards of quality control?
Consider another instance of team collaboration in which I am just one of five principal investigators; the data are collected in several sites across the country. Although I know the investigators at each of the sites, I have never set foot in the institutions where the data are being collected. How do I know that the study participants are being informed adequately about the research? How do I know that the data are appropriately safeguarded to protect the confidentiality of participants? How do I know that the data are not fabricated or altered?
Does it all come down to trust — trust that your team colleagues will behave ethically, competently and loyally? Bennett and Gadlin (2012) identified three forms of trust: identity-based trust, in which parties effectively understand and appreciate each other's wants, desires and values; calculus-based trust, based on ongoing experience with individuals who prove to be reliable and honest; and competence-based trust, developed with the knowledge of the capabilities and skills of team members. Each of these forms of trust is important when research is conducted in the context of a team, particularly when team members are chosen. Is this sufficient? When we look to the not-too-distant past when much of psychological research was conducted by solo researchers, we see that scientists were not always trustworthy.
Past failures in trust gave rise to our current research regulations and the institutional review board assessment of research protocols. In response to the growth of team research, many guidebooks and toolkits are rapidly emerging on the science of team science. But this new material tends to focus on team management, conflict resolution and fairness with respect to budgeting and publications. We don't see, in the world of publications, how to ensure that our research partners are performing ethically, responsibly and competently. Perhaps we will ultimately decide to borrow from quality management for guidance, where external auditors review records and serve as "fact checkers." But this could be a cumbersome and costly process.
For now, we have to rely on our selection process by choosing team members whom we judge likely to conduct themselves with integrity and competence. But if we are to believe Ernest Hemingway — that the best way to find out if you can trust someone is to simply trust them — then we are likely destined for failure, with its high cost, in many ways, to the research community and the public. Witness, for example, the growing number of multi-authored paper retractions in which one author committed fraud unbeknownst to the other authors, with damage to the reputations of unwitting collaborators and their institutions. While trust is important, we need to develop additional means to protect the integrity of team science.
Barbara Stanley, PhD, is a research scientist in the division of molecular imaging and neuropathology at the New York State Psychiatric Institute and a professor of medical psychology in psychiatry at Columbia University College of Physicians and Surgeons. She is also a member and immediate past chair of APA's Committee on Human Research.
