Research Guides
Keep Comparison Claims and Caveats Together in Reading Order
A research comparison looks clear in two columns: each company has a heading, a claim and a caveat. But when the page is read as a single sequence, the headings appear together, followed by both claims and then both caveats. The visual grouping no longer tells the reader which limitation belongs to which company.
This guide provides a reading-sequence review for published research. It does not claim that Colquhoun.AI performs automated accessibility certification.
1. Identify where order changes meaning
The W3C explanation of Meaningful Sequence says that when presentation order affects meaning, a correct reading sequence must be programmatically determinable. More than one order can be valid. Independent sections need not follow one prescribed arrangement, but their contents must not be interleaved in a way that changes meaning.
Start with relationships, not coordinates. Ask which heading identifies a passage, which qualification limits a claim, and which instruction must precede the next step. A left-hand position is not a sufficient description of those relationships.
2. Linearize a fictional comparison
Consider two fictional service summaries. North Studio offers a monthly plan, with a caveat that setup costs extra. South Studio offers an annual plan, with a caveat that the stated price assumes a yearly commitment.
A defensible reading sequence might be North's heading, plan and setup caveat, followed by South's heading, plan and annual-commitment caveat. Reversing the two complete summaries could also preserve their meaning. Reading both names, both plans and then two unlabeled caveats creates avoidable ambiguity.
The reviewer records the ambiguous sequence and the intended grouping. They do not merely request that the columns be swapped. A visual rearrangement alone may leave the underlying reading problem untouched.
3. Compare the visual and programmatic presentations
Inspect the rendered content structure and, where available, the accessibility tree. Follow the page using an appropriate screen reader to observe its actual reading sequence. Record the software and mode used so the finding can be reproduced.
A plain-text extraction can help reveal suspicious ordering, but it is not a substitute for every presentation used by assistive technology. Similarly, a screenshot demonstrates visual placement, not the order exposed to software. Label the evidence according to the method that produced it.
4. Repeat the check after layout changes
Review the same comparison when columns become a single stack on a narrow screen. Check that each caveat remains attached to the intended subject and that source links remain understandable in context.
Include expanded details if they form part of the evidence. A collapsed card may look correct while its newly revealed content appears in an unexpected position. Keep the tested state in the issue record rather than assuming the initial view represents every state.
5. Copy this reading-sequence card
- Content unit: comparison, procedure or other meaningful sequence.
- Required relationships: which heading, claim, caveat and source belong together.
- Visual presentation: layout and viewport observed.
- Programmatic evidence: structure or accessibility information inspected.
- Reading test: software, state and observed sequence.
- Mismatch: exact point where the meaning becomes unclear or changes.
- Proposed repair: grouping or order that preserves the content's meaning.
- Retest: affected layouts and interactive states checked.
Keep content changes separate from ordering repairs. If a caveat itself is inaccurate, that requires an editorial correction, not merely a different place in the sequence.
6. Close the issue at the right scope
A useful completion note states which content units and states were checked and whether their relationships remained clear. It does not certify an entire site after reviewing one card.
The goal is not to make every presentation look identical. It is to preserve the research's meaning when a reader encounters the evidence through a different presentation.