Missing data: an empty entry is not a measured zero

For everyday reading, start with a smaller distinction: a blank space does not automatically mean that someone observed a value of zero. We want to know what was recorded, what was not available and how the report treated the gap.

Three rows that should not become the same number

Consider an invented household inventory of three craft boxes. Box A contains zero glue sticks after somebody checks it. The entry for box B is blank because nobody opened it. Box C was removed from this week's inventory because it belongs to a different project.

Only the first row records a count of zero. The second is unknown. The third is outside the stated scope. If we replace all three with zeros, the rewritten sheet says more was checked than actually was.

This example is about record labels, not about collecting family behavior or health information. Nobody needs to start tracking a person's activities to practice noticing a missing entry.

Write the status beside the value

An optional inventory note could use:

• Recorded value: …
• Missing because, if known: …
• Not included in this inventory because: …
• Checked by agreement on this date: …
• Question left for the next update: …

Do not invent a reason for a blank. “Reason unknown” is different from “forgotten,” “refused” or “not needed.” If the sheet already uses a symbol such as a dash, look for its legend before assigning your own meaning.

Now return to the research report

When an article gives a total number of participants, you can ask whether that same number appears in every outcome's analysis. If a table has fewer observations, look for the authors' explanation instead of filling the difference with an assumed result.

A parent may want to understand why a summary gives different totals. A teen might notice that a skipped question was counted as an answer. A grandparent may be comparing two articles with different tables. An educator can help locate a table note without claiming to have resolved the analysis.

What this page leaves to specialists

We are not selecting a statistical method for replacing missing values, judging the trustworthiness of a particular study or calculating what its missing results would have been. Different assumptions would need explicit analysis; our craft-box example supplies none.

The task here is to keep an observed zero, an unknown value and an excluded item visibly distinct. When summarizing a report, preserve both its stated totals and its explanation of unavailable information, including any explanation you could not find.

Source

[1] Jonathan J. Deeks; Julian P. T. Higgins; Douglas G. Altman; Joanne E. McKenzie; Areti Angeliki Veroniki. Chapter 10: Analysing data and undertaking meta-analyses. Cochrane Handbook for Systematic Reviews of Interventions. Chapter citation says last updated November 2024, Handbook version 6.5; exact day not stated. Accessed September 6, 2026. Section 10.12.1 complete types-of-missing-data discussion, plus chapter citation. Supports the adjacent attributed claim. All illustrative scenes, numbers and optional note formats are original inventions, not source findings.

Related guides

• denominator
• outlier
• publication bias
• sample size