Sampling Revisited

This page addresses advanced topics on sampling that includes chapters 4 and 5 of Fowler.

Nonresponse Bias

Definition of nonresponse bias: the extent that nonresponse makes the actual sample estimate systematically different than that of the population.

Three reasons for non-response

  • Cannot access the survey
  • Chooses not to respond
  • Cannot respond to the answers

Estimating bias

  • Consider response rate
  • Compare first phase estimates with second phase
  • Analytical methods?

Reducing bias

  • Facilitate survey response process
  • Increase incentives
  • Follow up on non-responses
  • Adjusting for bias

More sampling issues

  • Personal interviewing vs. self-adminstered surveys
  • Distributing and promoting the survey
  • Examples of incentives

Discussion question

How do these issues of sampling and nonresponse bias apply to your proposed survey?