STAT 210 Lecture Notes - Lecture 2: Stratified Sampling
Document Summary
To perform any statistical analysis/inference, data is needed. Select a sample of the population and only measure/contact subjects in sample is known as a sampling procedure. Experimental designs have a treatment imposed on experimental units/subjects and a response is observed. Then compare the effect the treatment had on the response. Sampling designs has a population researches want statements about, so a representative sample of the population is used and the data from the sample helps make inferences about the population. Sample should be as representative of the population as possible. Some subjects and/or outcomes are favored over others. Selection bias: one or more types of subjects are systematically excluded from the sample. Undercoverage: inference cannot me made to population, but only to a subset of the population. Nonresponse bias: individual randomly chosen cannot be contacted or fails to respond. Response bias: respondent gives inaccurate info or the interviewer influences subject to respond a certain way.
