ABSTRACT
The survey sampler may know additional information about the population before sampling, such as the gender and major of every student at a university, the percentage of senior citizens in every county, or the dollar amount of every loan made by a bank. Stratified sampling uses this additional information about the population in the survey design. A stratified sample may use proportional allocation, in which every stratum has a sample size proportional to its population size, or disproportional allocation, in which some strata have larger sample sizes than their shares of the population. This chapter discusses how to use population information to obtain more accurate estimates from the sample—without spending more money on data collection.
