HP 350L Lecture Notes - Lecture 5: Stratified Sampling, Quota Sampling, Normative Social Influence
Document Summary
Population is divided into homogenous subpopulations (called strata or clusters). Randomly draw the same proportion of people from each stratum so that it resembles population size. A large scale variant of stratified random sampling. Divides the population into geographic unties (zip codes) Within each start, geographic units are randomly selected. Requirements: all geographic areas have seem chance of selection; and units within the area are chosen impartially. Convenience sampling: easy to obtain, generally because the whole sample exists in one location. Quota sampling: survey of quota of people who have certain characteristics. Similar to stratified sampling except that there is no random selection. Snowball sampling: ask participant to ask a friend to participate. Responednt-driven sampling: diverse seeds are selected and each reform debris of network into study. One difficulty in obtaining random samples is the refusal of some individuals to participate in the survey. Nonresponse bias is the error due to non participation or non response.