What information do we require of authors about replication packages?
What is the absolute minimum that data editors and readers interested in replication need to know? Provide justification.
- NB: bibliographic details of the article may not be know yet
- corresponding author (may be different from that on article)
- save your output
- where is your output
- output is file, widely readable (not .mat, not .dta)
- how do they map to those in the paper
- List of tables
- each table mapped to one or more output file in replication package
- List of figures
- each figure mapped to one or more output file in replication package
- metadata
Common metadata about each requirement
- unique citation
- including version number
- wave and date of accessinng for data
- how to access
- which package repository if not obvious
- which organization has the data
- license
- tested on
- entry point:
- single script to run (whatever language)
- which folder to start it from
- where to put missing inputs
- for each external requirement, give a mounting point
- run time
A structured format that is both machine- and human-readable.
- YAML format, saved as .txt?