Methods that focus on obtaining large numbers of subjects in a trial, or large numbers of systems that share the property being studied, in order to generalise results. Such an approach is typically impractical for studying properties of generative artefacts (things that make other things) because of the vast number of trials required. Alternatives include employing smaller amounts of rich data or very targeted evaluation combined with some form of justification (reasoned argument) as part of a holistic validation.
Used on page 133