Label Percentage Denominators in Tables and Charts

Define the population behind each percentage

A percentage is a count divided by a specified denominator, multiplied by 100. The arithmetic can be correct while the table is misleading if readers assume a different denominator. State whether percentages use all participants, eligible participants, observed responses, row totals or column totals.

Cochrane’s interpretation guidance emphasizes clear interpretation of quantitative results. For descriptive tables, a practical safeguard is to preserve numerator and denominator alongside the percentage wherever space and confidentiality permit.

Audit the table structure

Identify whether categories are mutually exclusive and exhaustive. A single-choice category distribution may sum to 100%, apart from rounding, when every participant is represented. A multiple-response question can exceed 100% because one person contributes to several categories.

Handle missing and not-applicable values explicitly. Excluding them changes the denominator and may make a percentage look larger without any change in the numerator. Use footnotes to explain the rule, but avoid forcing readers to reconstruct it from scattered notes.

Compare hypothetical denominators

Suppose 100 participants are enrolled, 80 are eligible for a questionnaire item and 60 provide a usable answer. Twenty report symptom X. The percentage is 20% of all enrolled participants, 25% of those eligible and 33.33% of those answering. These are different summaries, not conflicting calculations.

If the item asks participants to select all applicable symptoms, 20 may also select symptom Y. Adding symptom percentages does not recover the number of unique symptomatic people. A table should say multiple responses allowed and define whether its denominator is respondents or total selections.

Verify labels against calculations

For each displayed percentage, trace the numerator and denominator to the actual filtered data. Check that a column headed n (%) does not mix several denominator rules without explanation. Reconcile totals before rounding, and retain exact counts in the analysis output.

When denominators vary by outcome or time, show them in the relevant row rather than relying on one overall sample-size label. If disclosure rules prevent exact small counts, explain the suppression convention without inventing a precise percentage. The finished display should allow readers to identify who could contribute, who actually contributed and what comparison the percentage supports.

Source verification: 6 November 2026. Examples are hypothetical.