FDSC 300 Lecture Notes - Lecture 2: Sampling Error, Cuvette, Plans
Document Summary
Sample take representative of entire population -> test -> make inference about population cheaper/faster/practical than testing whole population. But: chance of error: consumer risk (accept faulty product, producer risk (reject good product) Bigger sample -> more accurate (but more 16871$, time consuming) Error may occure at each step in analysis -> magnified. Sampling plan: what size sample, how to retrieve sample. Purpose -> quality or safety, accept/reject, measure levels, determine consistency. Population -> lot of variability or uniform, how it is distributed, distinct units?\ Attribute sampling: accept/reject (yes/no outcome) determine if product has characteristic: sample is 10x smaller than pop (ex: test for botulinum) Variance sampling: estimate amount of substance in product (scale) Manual vs continuous: continuous does not req personnel, less bias, is on-line. Population: all objects under testing (all oranges from company) Lot: quantity of same product that is to be tested (oranges in a warehouse)