Because datasets can be queried in myriad ways, it is important to detail what decisions you make in your analysis and why you have made them.
What to do:
- While investigating your data and drawing your conclusions keep notes about what software you are using and why you are approaching the analysis the way you are. For example, what variables are you collapsing and how are you determining your weights?
Why do it:
- Because data without descriptions of the collection and analysis process are of little use. There can be no informed replication or reproduction without documentation of your decisions.
How to do it:
- Document your process
- Document known data issues
- Document file versions to track transformations
Things to consider:
- Who will have access to this data for analysis and what will they need to know?