Data loss

Our ability to generate complex, massive data has outstripped our ability to store, manage, and use it. The digital environment is straining practices refined on print materials. While the computing revolution has given us enormous capacity to generate, store, analyze, and visualize data, these systems and media are fragile. They do not have the stability or durability of paper. Communities of researchers, institutions, and funding agencies need to begin having conversations about prioritizing what data to preserve, curate, and share. The ultimate question here is: Will the data be accessible and usable when it is needed?

Responsible data management to improve research integrity

Data is a key piece of the scholarly record. As such, the way in which data is managed has an impact on the integrity of the scholarly record as well as the potential for data curation, sharing, and reuse or secondary analysis. This is recognized by the Office of Research Integrity,¹¹ the National Academies of Science,¹² federal funding agencies requiring data management plans,¹³ and initiatives like FORCE11.¹⁴ Kenneth Pimple describes data management as “the neglected, but essential, twin to the ‘scientific method.’”¹⁵ While careless data management can lead to gaps and errors in the scholarly record, effective data management strategies do exist for preventing inaccuracies. Documentation is one such strategy. Despite its importance in validating published results and preventing the types of errors that lead to retractions, researchers commonly acknowledge that research data are typically poorly documented. One cause may be the heavy administrative burden of conducting academic research.¹⁶

Variability in the research methods and processes used in different disciplines limits adoption of universal recommendations for managing data. However, some common focus areas are emerging from data management training programs:^{17, 18, 19} identifying and addressing ethical and legal obligations; providing detailed documentation that follows standard practices for the field; planning and execution of a data storage and backup plan; and archiving data for reuse and secondary analysis. Sharing data requires careful consideration and planning to protect sensitive information, while maximizing the availability of the data for verification, secondary analysis, and reuse. Responsible management of the data includes balancing ethical and legal obligations to the funding agency, institutions, researchers, and participants (if applicable). At minimum, the data underlying published findings should be available for verification purposes. There are many options beyond dark and open data, but many researchers are not aware of them.

Acknowledging responsible data management as foundational for research integrity is not sufficient. We need to value the processes and products of research equally by: 1) creating incentives for responsible management of data, 2) developing standards and practices for peer review that balance evaluation of methodological quality and research integrity with potential impact, and 3) carefully considering the resources necessary to responsibly manage and preserve newly created data for five-to-ten years after publication. The last is vital, given our current lack of capacity to store, manage, and use existing data. Efforts to develop better data management technologies and infrastructure have begun, but are not keeping pace with data creation.

Our role in supporting responsible data management

The academic community must improve the integrity of the scholarly record to regain credibility. The library has an important role to play in this by providing ongoing access to the scholarly record and teaching effective practices for ensuring its integrity. Despite our long-time role as stewards of the scholarly record, our involvement with research ethics education is often limited to teaching about plagiarism. Yet, our knowledge extends beyond this to include many of the principles for the responsible conduct of research,²⁰ excepting mentor and trainee responsibilities. As a profession, our deep engagement with scholarly communication and information management brings practical knowledge to ethics training that is often theoretical in nature. This is an opportunity to leverage our instructional skills to improve the data management practices of researchers and take a more active role in preventing misconduct throughout the research process. Providing education and support further upstream should result in more informed and carefully documented projects, leading to greater integrity and reduced data loss. Such instruction should focus on strategies such as data management planning, file organization and naming, storage and backup, metadata and documentation, and data registration and citation.

As practitioners, we can lead by example. Encouraging the editors of library and information science journals to implement review and data availability policies similar to those proposed by Frances S. Collins and Lawrence A. Tabak²¹ would increase the transparency and integrity of our own scholarly records. We can provide greater support for faculty in implementing tools for data deposit, registration, and citation into their process. We can support open science initiatives like study registration and replication studies by introducing these approaches into our outreach and instruction and by collaborating with researchers to implement strategies for transparency. Maintaining long-term access to data and contextual information can be aided by operating an institutional repository or participating in consortia like the Digital Preservation Network. As scholars, we can examine and measure the impact of these changes in practices and incentives on various research communities.

Although none of us will be involved with all of these activities, we should be cognizant of the importance of data management for research integrity and prepared to initiate these discussions with our faculty, students, and administration.