Estimating sample size in conceptual property norms by standardizing coverage.

Conceptual properties norming (CPN) studies are a fundamental tool in cognitive science, where participants list properties (e.g., features, attributes) they associate with several concepts, generating rich semantic profiles used to study mental representation. Traditionally, CPNs standardize the number of participants per concept-a method we argue is flawed, as it ignores differences in semantic richness (that is, how many distinct properties can be listed for each concept). Instead, we propose standardizing coverage, which refers to the proportion of all possible properties for a concept that are identified in a study (e.g., if 70 out of 100 possible properties are listed, coverage is 0.70). We develop a mathematical framework for estimating the sample size needed to achieve a target coverage level, based on the average number of properties participants list. Models were built and validated using empirical data from both concrete and abstract concepts. Applying our framework revealed that sample sizes required for robust coverage vary widely across concepts, underscoring the limitations of fixed-participant approaches. To facilitate adoption by the research community, we provide pre-calculated look-up tables and an interactive HTML coverage calculator, allowing researchers to easily determine sample sizes or estimate achieved coverage. By focusing on coverage rather than participant count, our method ensures more accurate and comparable semantic profiles across diverse concepts. We recommend a coverage target between 0.70 and 0.80 for robust analysis and propose a practical two-stage sampling strategy. This approach significantly enhances the reliability, comparability, and replicability of CPN data.