goodthingswithdata.com

4  Design choices and constraints: What data were used?

4.1 Describe data sources in anticipation of analysis and interpretation

Principle: Clearly describe data sources (whether commercial or noncommercial), sizes, and structures, including information about choices and constraints. Include information that will influence analysis and interpretation, such as selection criteria and biases, quality, completeness, missingness, inclusions, exclusions, and losses.

Observations: Reports consistently describe data in sufficient detail. Reports from web-based surveys include mm6924e1 (Czeisler, Tynan, et al. 2020), mm6932a1 (Czeisler, Lane, et al. 2020), and mm6949a2 (Gilbert et al. 2020). Reports from electronic health records include mm6802a1 (García et al. 2019), mm6911a5 (Schieber et al. 2020), mm7114e1 (Block et al. 2022), and mm7121e1 (Bull-Otterson et al. 2022).

Recommendations:

  1. Current practice is strong. MMWR or CDC could develop guidance for authors on how to relate characteristics of data to analysis and interpretation, especially for emerging and nontraditional data sources and types, such as web-based surveys and other convenience samples; administrative, claims, and electronic health records; fusion of multiple data sources; and wastewater and other ecological measures.
  2. Reports do not typically need to indicate data management technology, such as REDCap, dplyr (R package), or pandas (Python library), unless data management practice or technology substantively influences analysis, interpretation, or potential reproducibility.