Create a data validation protocol to minimize human error and bias

[theecrit]

Description

Manually pulling data from PDF into structured spreadsheets introduces the potential for human error and bias. We can reduce this potential by maintaining a process for pulling, reviewing, and publishing data.

Proposed solution

• Start with another brigade's process or other frameworks and adapt. The only alternative I can think of off-hand is to start from scratch, which is a larger lift.

Things to consider:

• Pulling data, reviewing data, and publishing data each introduce particular challenges. Are there other stages that should be addressed?
• Some ideas:
  • Use forms validation, conditional formatting, etc during data entry to help catch errors as they occur.
  • Require each handler of data to complete a short checklist that triggers a "gate unlock" to the next step on completion (eg you can't skip a step when handing off).

To close this issue, the following conditions should be met

Check off as you go to track progress. This list can be changed and/or broken out into individual issues during the discussion phase if needed.

• Team discussion on the issue, with key decisions documented in the comments (even if just a summary of asynch or live conversation).
• Draft protocol created and documented.
• Draft protocol shared with brigade and publicly via OO channels for comment.
• Protocol revised and published publicly (at minimum, with project documentation).