Missing Data: Drop, Impute, or Flag

A blank cell isn't one problem — it's three, and they need different responses:

• Missing at random: a sensor dropped packets — the gaps have no pattern. The safest case.
• Missing for a reason: high earners skip the income question; patients who feel fine skip follow-up tests. The blank itself carries information — it correlates with what you're studying.
• Missing because of the answer: a churned customer simply stops appearing. The absence is the outcome.

Before touching anything, ask why it's blank. Fill in a systematic gap carelessly and you can erase the very pattern that explains your result.