Data-Intensive Science and Research Integrity

Disclosing and Managing Non-Financial Conflicts of Interest in Scientific Publications

In the last decade, there has been increased recognition of the importance of disclosing and managing non-financial conflicts of interests to safeguard the objectivity, integrity, and trustworthiness of scientific research. While funding agencies and academic institutions have had policies for addressing non-financial interests in grant peer review and research oversight since the 1990s, scientific journals have been only recently begun to develop such policies. An impediment to the formulation of effective journal policies is that non-financial interests can be difficult to recognize and define. Journals can overcome this problem by providing guidance concerning the types of non-financial interests that should be disclosed, including direct research interests, direct professional interests, expert testimony, involvement in litigation, holding a leadership position in a non-governmental organization, providing technical or scientific advice to a non-governmental organization, and personal or professional relationships. The guidance should apply to authors, editors, and reviewers.

Examining paleobotanical databases: Revisiting trends in angiosperm folivory and unlocking the paleoecological promise of propensity score matching and specification curve analysis

Paleobotany is at a crossroads. Long-term trends in the fossil record of plants, encompassing their interactions with herbivores and with the environment, are of the utmost relevance for predicting global change as pCO2 continues to rise. Large data compilations with the potential to elucidate those trends are increasingly easy to assemble and access. However, in contrast to modern ecology and unlike various other paleontological disciplines, paleobotany has a limited history of “big data” meta-analyses. Debates about how much data are needed to address particular questions, and about how to control for potential confounding variables, have not examined paleobotanical data. Here I demonstrate the importance of analytical best practices by applying them to a recent meta-analysis of fossil angiosperms. Two notable analytical methods discussed here are propensity score matching and specification curve analysis. The former has been used in the biomedical and behavioral sciences for decades; the latter is a more recent method of examining relationships between, and inherent biases among, models. Propensity score matching allows one to account for potential confounding variables in observational studies, and more fundamentally, provides a way to quantify whether it is possible to account for them. Specification curve analysis provides the opportunity to examine patterns across a variety of schemes for partitioning data—for example, whether fossil assemblages are binned temporally by stage, epoch, or period. To my knowledge, neither of these methods has been used previously in paleontology, however, their use permits more robust analysis of paleoecological datasets. In the example provided here, propensity score matching is used to separate latitudinal trends from differences in age, climate, and plant community composition. Specification curve analysis is used to examine the robustness of apparent latitudinal trends to the schema used for assigning fossil assemblages to latitudinal bins. These analytical methods have the potential to further unlock the promise of the plant fossil record for elucidating long-term ecological and evolutionary change.

Perceptions of plagiarism among PhDs across the sciences, engineering, humanities, and arts: Results from a national survey in Brazil

Plagiarism allegations are not rare in the history of science, and credit for prior work was and continues to be a source of disputes, involving notions of priority of discovery and of plagiarism. However, consensus over what constitutes plagiarism among scientists from different fields cannot be taken for granted. We conducted a national survey exploring perceptions of plagiarism among PhD holders registered in the database of the Brazilian National Council for Scientific and Technological Development (CNPq). This survey was sent to 143,405 PhD holders across the fields, including biologists, physicists, mathematicians, and engineers as well as linguists, philosophers, and anthropologists, with a 20% response rate. The results suggest that core principles about plagiarism are shared among this multidisciplinary community, thus corroborating Robert K. Merton's observations that concerns over plagiarism and priority disputes are not field specific. This study offers insight into the way plagiarism is perceived in the research community and sheds light on the problem in the context of international collaborative research networks. The data focus on a particular research system in Latin America, but, given the cultural similarities that bind most Latin-American nations, these results may be relevant to other PhD populations in the region and should provide an opportunity for comparison with studies from other emerging, non-Anglophone regions.

Auditing data resolves systemic errors in databases and confirms mycorrhizal trait consistency for most genera and families of flowering plants

Nearly 150 years of research has accumulated large amounts of data on mycorrhizal association types in plants. However, this important resource includes unreliable allocated traits for some species. An audit of six commonly used data sources revealed a high degree of consistency in the mycorrhizal status of most species, genera and families of vascular plants, but there were some records that contradict the majority of other data (~ 10% of data overall). Careful analysis of contradictory records using rigorous definitions of association types revealed that the majority were diagnosis errors, which often stem from references predating modern knowledge of mycorrhiza types. Other errors are linked to inadequate microscopic examinations of roots or plants with complex root anatomy, such as phi thickenings or beaded roots. Errors consistently occurred at much lower frequencies than correct records but have accumulated in uncorrected databases. This results in less accurate knowledge about dominant plants in some ecosystems because they were sampled more often. Errors have also propagated from one database to another over decades when data were amalgamated without checking their suitability. Due to these errors, it is often incorrect to designate plants reported to have inconsistent mycorrhizas as "facultatively mycorrhizal". Updated protocols for resolving conflicting mycorrhizal data are provided here. These are based on standard morphological definitions of association types, which are the foundations of mycorrhizal science. This analysis also identifies the need for adequate training and mentoring of researchers to maintain the quality of mycorrhizal research.

Is it time to revise the definition of research misconduct?

U.S. federal policy defines research misconduct as fabrication of data, falsification of data, or plagiarism (FFP). In recent years, some have argued or suggested that the definition of research misconduct should also include sexual harassment, sabotage, deceptive use of statistics, and failure to disclose a significant conflict of interest (COI). While the arguments for revising the definition of misconduct used by federal agencies to include misbehaviors other than FFP are not convincing at this point in time, the arguments for revising definitions used by other organizations, such as professional societies, universities, or journals, may be. Since these other organizations play an important role in promoting integrity in science and deterring unethical behavior, they may consider adopting definitions of misconduct that extend beyond FFP. Debates about the definition of research misconduct are a normal and healthy part of broader discussions about integrity in science and how best to promote it. These debates should continue even if the federal definition of misconduct remains unchanged.