The Best Time to Catch a Data Problem Is Day One
A paradata check on day one of fieldwork in Uganda flagged three unusual patterns before they could spread across the dataset. None were automatically a problem, but all were worth understanding.
Data Quality Is Easier to Protect Than to Recover
Two inherited datasets, two problems no one caught at the point of collection. What slowing down to ask “does this actually make sense” revealed, and why how you respond matters as much as catching it.