
A Data Code Can Look Like a Number Without Being a Quantity
Distinguish identifiers from measurements before calculating with a column, using fictional specimen codes to show why numeric appearance is not enough.
From a download link to a useful question: approachable guides to understanding and working with data.

Distinguish identifiers from measurements before calculating with a column, using fictional specimen codes to show why numeric appearance is not enough.

Check units and conversion rules before combining measurements, using a small length example to show why matching column names are not enough.

Investigate apparently duplicated rows by checking identifiers, timestamps and the collection process before deleting records from a dataset.

Check whether row order encodes time or collection sequence before sorting a dataset, and retain an explicit field that preserves the original order.

Record conversions, filters and recoding decisions in a short transformation log so another reader can understand how a working dataset differs from its source.

Distinguish the number of recorded observations from the number of subjects in a dataset before interpreting repeated measurements as separate participants.

Before filling empty cells, distinguish an observed zero from an unrecorded, unavailable or inapplicable value.

Treat a shared pattern as a starting question by checking timing, alternative explanations and the design of the study.