Automated misconduct detection: Uses and effects of digital tools for investigating research misconduct
Principal Investigator: Dr. Felicitas Heßelmann
Researcher: Rocío Fonseca, Jacqueline Sachse
This project is concerned with digital tools for detecting and investigating possible instances of scientific misconduct, such as plagiarism scanners, tools for statistical analysis or tools for detecting image manipulation. Looking at the development, application and potential effects of those tools, its focus is on the respective definitions, criteria and normative understandings of problematic behavior that are inscribed in the tools and hence influence procedures and decisions about possible misconduct.
The project aims at developing an overview over the current state of software use by university investigative committees, ombudspersons, and journal editors, in order to gauge the potential effects and shifts the software causes in the work of those actors. Here, the focus lies on understanding how these tools influence the definitions and concepts of scientific misconduct and good scientific practice that are at stake in misconduct investigations. As such, the project also contributes to current research on the uses and effects of automated tools for decision-making in high-stakes situations, such as court cases or procedures for resource allocation.
Empirically, the project starts out with a survey analyzing the current state of uses of automated tools by ombudspersons and journal editors. We will then conduct qualitative case studies addressing the development of selected tools for plagiarism detection, detection of statistical irregularities as well as the detection of image manipulation. The case studies will combine key informant expert interviews, document analysis, socio-technical walk-throughs as well as observations in order to reconstruct the relevant concepts of research integrity inscribed in the tools. To investigate potential effects of automated detection tools, we will conduct observations of two editorial offices where such tools are used. Subsequently, we will compare and contrast those results with observations and interviews investigating manual detection procedures. In addition, the potential uses of automated tools in university procedures will be explored in an intervention with the ombudspersons of the Berlin University Alliance taking place over the course of the project.
Supported by
