A full data table carries every value a chart was built from, more reliably than any alternative
Aliases: data table · accompanying table
What it is
Among all alternatives to graphics, the complete data table is the most reliable: it carries every value the graphic was built from, so readers—whether using screen readers or simply needing precise figures—can extract information fully equivalent to the graphic, and often more (exact values, recomputability). Alt text carries only the conclusion level; the data table carries the complete data level—they complement rather than replace each other. The table simultaneously serves many non-disabled scenarios: writers citing exact numbers, skeptics recomputing aggregates, and low-vision users struggling with zoomed rendering.
Why it happens
The table's reliability comes from sharing the graphic's data source rather than sharing a "paraphrase": alt text is the maker's summary of the graphic (introducing summary error), while the table is the direct rendering of the raw numbers that generated the graphic (zero paraphrase). Screen-reader support for semantic tables (correct header structure) is mature and stable—every cell can be read and navigated cell by cell, the highest accessibility of any alternative technique. The engineering crux is semantic markup: a real table element with headers (th, scope attributes), not div-assembled "table-looking" layouts, which lose row-column structure in screen readers. The presentation trade-off is visibility: a table expanded by default occupies space and duplicates the graphic, so a collapsed expander ("view data table") is the common compromise—but on pages whose primary audience is screen-reader users, default-expanded is friendlier. The table also boosts the graphic's credibility in reverse: letting anyone recompute the graphic's aggregates from the table is the most direct form of "the data is verifiable."
Where it stops holding
Table size must be governed: million-row detail is neither possible nor appropriate as an alternative; the alternative table's granularity should match the graphic's aggregation level (the chart shows quarterly totals, the table gives quarterly totals), with finer access through export or query interfaces rather than on-page rendering. High-dimensional multi-series charts make tables wide (one column per series), which burdens screen readers' horizontal navigation—splitting by series (one long table per series) is more accessible than one wide table. The table also cannot replace all of the graphic's functions: pattern recognition (trends, shapes) requires the reader's own computation from tables, so the conclusion level (alt) and data level (table) each cover one segment—together, graphic, alt, and table form the complete accessibility stack.
Applying it
- Pair every informative chart with a semantic data table (real table element + header structure), collapsed by default but reachable.
- Generate the table and the graphic from the same data source so values stay consistent, updating in sync.
- Provide structured export (CSV) for users who need further processing.
- Verification: traverse the data table with a screen reader (or the system's speech function), confirming row-column structure and values read correctly; a table read out as "a string of numbers" means semantic markup is missing.
Related
- Same group: U11.01.1 Graphics need a text summary as equivalent content · U11.01.3 Interactive charts must support keyboard traversal
- Nearby: U11.01.1 Graphics need a text summary as equivalent content · U8.06.3 A drill-down path to raw granularity must be available
- Search terms:
data table·semantic table·accessible data