Research Guides
Before Someone Cites Your Comparison Table, Check What Each Cell Means
A comparison table can be easy to copy and hard to interpret. Before inviting readers to cite it, make sure they can identify the subject, measurement and limits without guessing from the surrounding article.
Start with a cell, not the design
Choose one ordinary value. Ask a colleague to explain it using only the table and its notes. For a fictional software comparison, “25” might mean projects, users, dollars or a monthly request allowance. Replace ambiguous labels with the actual unit and period, then repeat the exercise.
For each column, record its definition, unit, observation date and source. Keep an explicit “not verified” value separate from zero. A missing price is not a free plan; an unavailable feature description is not proof that a feature does not exist.
Preserve relationships in the published page
W3C explains that accessible data tables need structural relationships between header and data cells. HTML uses th for headers and td for data; more complicated layouts can need explicit associations. A caption helps identify the table's topic. A visually bold top row alone does not supply those relationships. W3C Tables Tutorial
Use this editorial handoff
- Table title and purpose
- Each column's definition and unit
- Relevant period or observation date
- Source URL for each row or claim
- Meaning of blank, zero and unavailable values
- Known differences that prevent direct comparison
- Owner and next review date
- Publication checks and unresolved questions
This is an editorial worksheet, not a claim of accessibility certification.
Test a realistic misunderstanding
Imagine a table comparing two fictional plans. Plan A permits 25 projects per account; Plan B permits 25 projects per user. A simple “Project limit: 25” row implies equivalence that the evidence does not support. Keep the unit in the cell or separate the plans into clearly labeled rows. If the terms cannot be reconciled, say so rather than computing a winner.
Verify the delivered version
Review the actual published table on a narrow screen and with keyboard navigation. If it scrolls, check that readers can discover and operate the scrolling area. Confirm that headers still match their values and source links remain usable. Inspect exported versions separately; a screenshot or document conversion is not proof that the original relationships survived.
Record what was tested and what remains unknown. A precise completion note is more useful than “table optimized”: “Units and observation dates added; row sources checked; narrow-screen reading tested; assistive-technology testing pending.” Clear evidence helps readers reuse a comparison responsibly, but does not guarantee citations, rankings or backlinks.