The Annual Survey of Public Employment and Payroll (ASPEP) is conducted by the U.S. Census Bureau and provides employment statistics for state and local governments across the United States. The current sample has a probability proportional to size (pps) design with modified cutoff sampling, which results in complex variance estimation. In this paper, we examine an alternative single phase stratified design that implements a modified power allocation with pps methods for selecting certainties to determine if it results in improved ASPEP estimates. Our research uses a Monte Carlo simulation experiment with data from census years 2012 and 2017 to conduct the evaluation.
