This document discusses the criteria for selecting a sampling procedure. There are two costs involved in sampling: the cost of data collection and the cost of incorrect inference from the data. Incorrect inferences can arise from either systematic bias or sampling error. Systematic bias results from errors in the sampling procedure and cannot be reduced by increasing sample size, while sampling error can be reduced by a larger sample. Sources of systematic bias include an inappropriate sampling frame, defective measuring devices, non-response bias, the indeterminacy principle, and natural bias in self-reported data.