Archival appraisal of research data

Introduction


There are some specific issues in the archival appraisal of research data:

  1. Most literature on archival appraisal does not adequately address research data.

  2. Appraisal decisions should proceed from a determination of specific projects or experiments in a laboratory that should be documented.

  3. Observational data is usually impossible to reproduce and, therefore, has a higher degree of evidential value than experimental data.

  4. Experimental data is generally less valuable for future scientific research because the data can often be recovered by running the experiment again.

  5. The way experimental data is gathered can indicate something important about research methods. Experimental data can reveal what choices were made in the scientific and technological process. 

  6. Exceptions might be made for experimental data if the answer to any of these questions is “yes”: 

    1. Was the experiment controversial? 

    2. Was the experiment time-consuming? 

    3. Would the experiment be difficult to reproduce? 

    4. Would the experiment be expensive to reproduce? 

Long-term preservation questions


When considering the long-term preservation of research data, consider the following questions:

  1. Has the creator/donor provided a rationale or argument for long-term preservation? Is this rationale explained in a research data management plan?

  2. If the donation contains sensitive data, does the consent form contain language that prohibits long-term preservation? (e.g., a promise to destroy after X years)?

  3. Does the data provide valuable evidence of research activity, and demonstrate potential ongoing social, scientific or historical value?

  4. Does the data concern under-documented or marginalized peoples? If so, have the relevant community stakeholders been consulted about the appropriate custodianship of the data?

  5. Is the donation unique (i.e., it is not comprised solely of third-party data and it has not been shared anywhere else)? If not, does the data still require long-term preservation? (e.g., data is also held by other repositories that do not provide adequate preservation support for the original data).

  6. Is the data the result of a model that would be difficult or impossible to recreate?

  7. Is the data available in preferred or accepted file formats for preservation and access? If not, is there a strong argument to develop new workflows to preserve the file formats submitted (e.g., increased adoption of this format within the research discipline)?

  8. Has the creator/donor assigned open licenses to the data? If not, will the data be open after a short period of time?

  9. Are the data files well documented? Does the donation include sufficient information to ensure that the data files will be correctly interpreted over time, including a README, clear description of methodology and variables, and/or links to scholarly publications?

Further reading


Haas, Joan K., Helen Willa Samuels, and Barbara Trippel Simmons. "Appraising the records of modern science and technology : a guide." Cambridge, MA: Massachusetts Institute of Technology (1985).

Jonathan Dorey, Grant Hurley, & Beth Knazook. (2022). Appraisal Guidance for the Preservation of Research Data. Zenodo. https://doi.org/10.5281/zenodo.5942236