From Snapshot to Insight: Validating a Rapid Assessment Tool for Research Integrity Environments - We have developed and validated a novel, low-burden assessment tool, the Rapid Environmental Adjective Descriptor (READ), that provides reliable data about perceptions of research integrity environments in under two minutes. This project will test the validity of READ to produce research team-level results that predict established survey measures and will prototype the utility of READ as a diagnostic tool, with LLM-enabled reports based on participant responses. READ has potential to serve as an intuitive, low-burden early warning tool, aligning with ORI’s priority to support the development and evaluation of innovative tools and approaches to promote research integrity and to prevent misconduct by capturing a “snapshot” of research environments. Strong research environments are essential to fostering ethical research practices, yet institutions and research groups often lack timely, actionable information about the workplace cultures in which research occurs. Rather than focusing exclusively on investigating problems after they occur, rapid assessment could contribute to a broader prevention-oriented infrastructure that supports healthier research environments. This project, if successful, will provide teams and institutions with a practical mechanism for understanding research culture across settings, identifying both areas in need of support and environments that demonstrate strong, integrity-promoting practices. Using existing datasets, we have already established that the READ accurately predicts responses to longer working environment surveys validated in research settings, the Survey of Organizational Research Climates (SOURCE; Crain et al. 2013) and the Climate of Accountability, Respect, and Ethics Survey (CARES; Martinson, et al., 2025). Using AI tools and novel applications of existing analytical approaches, we will 1) seek to further validate the results using additional existing datasets; 2) extend the READ to predicting team science measures; 3) test whether accurate readings of group working environments are possible through experiments with three different methods of aggregating individual results; 4) mock-up a prototype diagnostic tool; and 5) perform usability tests. Final products will include an AI-supported automated process that takes READ results from individuals and groups, aggregates them into work-group level results, and produces an easily understood report. We will conduct a usability pilot with active science teams to understand issues to consider as we seek to deploy the READ for widespread use as an early warning and diagnostic tool. READ has the potential to serve as a safe way for those who experience research culture to share their experiences, including graduate students, postdoctoral scholars, faculty, and research staff. Research teams may benefit from periodic READ use for self-reflection and course correction. Institutional leaders and research administrators may benefit from earlier access to information about environmental strengths and vulnerabilities and tools that complement existing compliance systems. Rapid assessment could help leaders identify where additional resources, assessments, or support may be needed and where successful environments can provide models for broader institutional improvement. This project offers a low-risk, high-reward opportunity to advance research integrity assessment. By combining validated climate measures, psychometric evaluation, and a proven digital infrastructure, the project seeks to create a scalable, low-burden approach to monitoring research environments and supporting evidence-based institutional improvement.
