Sampling Method

A sampling method is a rule for selecting a subset of pages, screens, or components from a larger product to be evaluated for accessibility. Because full manual evaluation of every page on a real-world website, app, or PDF library is usually infeasible, every practical audit depends on some sampling method — and the method itself strongly shapes the resulting accessibility score. Common families include ad hoc selection (home page, contact page, one representative page per section, as recommended by W3C/WAI), uniform random sampling, random-walk methods that follow links from a seed page, and stratified sampling that first clusters pages by their error profile and then samples from each cluster. Empirical work shows that ad hoc selection — the most commonly used method in practice — can introduce substantial inaccuracy compared with stratified or error-profile-based methods.