Exercise

Comparing survey weights of different designs

Remember that an observation's survey weight tells us how many population units that observation represents. The weights are constructed based on the sampling design. Let's compare the weights for the three samples of the api dataset. For example, in simple random sampling, each unit has an equal chance of being sampled, so each observation gets an equal survey weight. Whereas, for stratified and cluster sampling, units have an unequal chance of being sampled and that is reflected in the survey weight.

Instructions 1/3

undefined XP
  • 1

    Using ggplot2, construct a histogram of the survey weights, pw, for the simple random sample, apisrs.

  • 2

    Now construct a histogram of the survey weights, pw, for the stratified sample, apistrat.

  • 3

    Lastly, construct a histogram of the survey weights, pw, for the cluster sample, apiclus2. Compare the histograms!