Hello!

I am trying to draw a random representative sample of cities from the census. I have three stratification variables that I have listed below guiding my stratification process. My plan is to categorize the first two stratification variables (Labor force participation and unemployment rate) at 2 quartiles above and 2 quartiles below the median. The variable city population is categorized into four groups (1-9,999; 10,000-24,999; 25,000-99,999 and 100,000 and above). In my sample, I would like to over sample from the 1-10,000 population group and include all cities in the 100,000 group.

My total population is 29,000 cities and I would like to sample a total of 2000 cities. My question is:
  1. How do I create strata from the overall sampling frame by expanding the grid of all possible combinations of the three categorized stratification variables?
  2. How do I draw a representative sample of 2000 cities from the strata?
Stratification Variables:
  • Labor force participation, (percentage)
  • Unemployment rate (percentage)
  • City population (count)
Thank you for helping.
-Seba