These are notes from using the STARS framework to guide the sharing of Stata do-files with analysis for an article currently being submit to journals:
Comparative responsiveness of preference-based health-related quality of life, social care and wellbeing, condition-specific and generic measures in the context of multiple sclerosis
Elizabeth Goodwin, Amy Heather, Colin Green, Nia Morrish, Jenny Freeman, Kate Boddy, Sarah Thomas, Jeremy Chataway, Rod Middleton, Annie Hawton
Liz extracted do-files from MS Register - presently just those from analysis (and not pre-processing)
STARS framework reflections
Uncertainty around what to include when working with personal person-level data. Is including just analysis scripts sufficient? Or is it ideal to also include pre-processing? In which case, along with dummy data?
This is a different context to what STARS was designed for for several reasons:
- Not FOSS - this was analysis in Stata which requires payment to use
- Not discrete event simulation - this was descriptive analysis and regression
STARS framework reflections
Applicability and other considerations when non-FOSS. Here, there are things we can't do, like have free-to-use remote instances of this.
Also, when non-DES. In this case, arguably not particularly relevant to have a web application.
Made a skeleton README with categories based on STARS (e.g. ORCID, citation). Also making notes about archiving, acknolwedgements (for funding and dataset).