Basic guidelines for authors
- Clearly link the analysis with the data provided. If it is not clear how the data and the analysis are related — as can occur when working from multiple spreadsheet worksheets — then the analysis is extremely difficult to check.
- When possible, try to use nonproprietary software. Proprietary software — particularly software that is not in common use — can be difficult for other researchers to obtain.
- Provide data cleaning instructions or code. It should be clear how the data set you analyzed arose from the complete data set.
- No “mystery meat”. All files should be accompanied by enough explanation that another person can understand the contents of the files, and how these are related to what is discussed in the paper.
- Use reliable hosting. Do not use personal cloud storage providers, or personal web hosts, to host your data. These are prone to be lost as you clean or reorganize your own files, or change website hosts. Use a third party designed for this purpose, such as the Open Science framework or a data journal/repository.
Helpful resources
- Scripting languages, such as R and python are ideal for reproducible analyses. Both are free.
- The RStudio interface for R offers an easy way to create documents including analysis code and graphics. See, for instance, Rmarkdown.
- Making your code citable: A guide to posting your code to github and assigning a DOI number, enabling it to be cited.
- See the resources under Making your data public for information about where you can store your analyses. Repositories will typically accept both data and analysis code together.
Supporting the spread of open research practices
I highly recommend adding Jupyter notebooks as a helpful resource for sharing analyses https://jupyter.org/