Policy Volume: AC
Responsible Executive: Senior Data Librarian
Responsible Office: Cornell Center for Social Sciences
Revised: 2014-04-03, 2020-11-05
REASON FOR POLICY
To ensure that data versioning criteria are consistently applied to changes in data files and data documentation in the CISER Data & Reproduction Archive.
For reasons detailed below a significant percentage of data studies are altered within the first year of being published. These changes may be initiated by the data producer, the CISER Data & Reproduction Archive, or data user. CCSS Research Support has adopted a set of criteria which distinguish between significant and minor changes to a data study to determine whether a new dataset version is merited.
These guidelines ensure that data versioning criteria are consistently applied to changes in data files and data documentation (including, but not limited to, correction for error, amendments, additional variables, changes in access conditions, format changes) for inclusion in the CISER Data & Reproduction Archive. This will often involve working closely with the data producer.
Significant changes are those that will have a high impact on the use or interpretation of the data, whereby minor changes are those that will have a low impact in relation to interpretation or use for research purposes. CCSS assigns a new version to a data study with significant changes such as: addition of new variables; revision of incorrect data; revision of miscoded data; formatting changes; substantial documentation changes; changes in access conditions; withdrawal of data elements or documentation files. Minor changes such as small changes in variable labels, spelling corrections in metadata, minor changes in documentation will be made to relevant content and recorded in ancillary and accompanying documentation, but no new version assigned. Any and all changes are recorded in an internal notes field in our database.
Where possible CCSS will clearly label and make available earlier versions of data and documentation through the data catalog. Version record numbers are captured in metadata held in CCSS relational databases. CCSS retains the right to withdraw an older version of a data study where significant change may be misrepresentative or a copy is held in another major data archive (such as ICPSR).
In some cases data studies are issued as completely new study editions (this may be as part of a series) if there are changes to data/variables, major changes to documentation, new waves have been added to a series, or there have been methodological changes.
Policy Review Process: CCSS will review these policies every three years in conjunction with the CoreTrustSeal certification process or any future certification process.