Replicating tasters on a single bottle does not replace independently sampling production batches when the claim concerns batch consistency.
A single-bottle study is useful when its question is specifically about that bottle. Independent batches are necessary only for broader production inference.
Sampling units and experimental units must match the intended inference. Taster replication estimates response variation conditional on the tested samples.
Sampling Before Generalization
Twenty tasters evaluating one bottle contribute information about assessor responses to that bottle. They do not provide twenty independent observations of production batches. A batch-consistency study must sample batches and preserve the sampling hierarchy in its analysis.
Academy Application
Contrast these designs in a sensory-method exercise. Ask learners to name the sampled units and the population their conclusion covers. No new whiskey trial has been performed.
Provenance
Meilgaard et al.,2016,printed pp.390–398.
Connection