Tables on the web accessible? unfortunately not!
Waqar Haider, Yeliz Yesilada · 2020 · Proceedings of the 17th International Web for All Conference (W4A) · doi:10.1145/3371300.3383349
Summary
This paper presents the first large-scale systematic study specifically focused on the accessibility of HTML tables on the web. The authors crawled approximately 30,000 web pages sourced from Google, Google Tables, and the Alexa top 10,000 sites, ultimately analyzing over 16,000 table elements extracted from 925 pages. They developed a custom Python-based tool to evaluate tables against WCAG 2.1 Success Criterion 1.3.1 (Info and Relationships, Level A), which requires that information and structure conveyed visually must also be programmatically determinable. The study classified tables into six types: layout tables, row header tables, column header tables, single row-column tables, and mixed tables. The evaluation checked for the presence of key accessibility techniques specified by WCAG: captions (H39), id and headers attributes (H43), proper use of th and td elements (H51), scope attributes (H63), and summary attributes (H73). The researchers also examined the use of structural elements thead, tbody, and tfoot.
Key findings
The results paint a bleak picture of table accessibility on the web. Nearly half (49.28%) of all table elements were used for layout purposes rather than presenting data — a practice that creates confusion for screen reader users. Among the roughly 8,162 genuine data tables: only 4.71% included a caption element; only 3.12% had a summary attribute; 0% used id and headers attributes to associate data cells with headers, despite over 80% being complex mixed tables that require these associations; 100% had scope attributes (suggesting automated template usage rather than intentional accessibility); only 9.39% used thead; 39.67% used tbody; and just 0.17% used tfoot. The near-total absence of captions, summaries, and id/headers associations means screen reader users encounter tables with virtually no programmatic structure to help them understand and navigate the data.
Relevance
This study provides empirical evidence of a widespread and fundamental accessibility failure that affects virtually every website. Tables are among the most complex HTML structures for assistive technology to interpret, yet the data shows that even the most basic Level A WCAG requirements are almost universally ignored. For web developers and accessibility practitioners, the findings are a clear call to action: use CSS for layout instead of tables, always include caption elements on data tables, use th with scope attributes for headers, and employ id/headers for complex tables. The study also highlights a gap in automated testing tools — existing tools evaluate entire pages but lack specialized table analysis capabilities, suggesting an opportunity for AI and machine learning approaches to automatically detect table types and remediate accessibility issues. The near-zero compliance rates found across popular websites demonstrate that accessibility awareness campaigns and developer education around table markup remain critically needed.
Tags: web accessibility · HTML tables · automated testing · WCAG compliance · screen readers · semantic HTML
Standards referenced: WCAG 2.1