Inverse Bernoulli Sampling and Applications
Two-stage sampling, Pareto Sampling, Sequential Poisson Sampling, Systematic Sampling, COVID-19.
The absence of a listing frame that identify and provides access to the elements of a target-population is one of the most recurrent adversities faced by sampling surveys. When sample frames are available not seldom, they need to be updated to be used in practice. When the elements of a target-population are grouped in clusters, the challenge very often rely on the non-existence or the outdating of existing listing frame of elements within clusters. In this Thesis the Inverse Bernoulli Sampling design is presented, its statistical properties discussed and its potential use in the second stage of two-stage sampling designs, to select a sample at the same time an updating screening process is carried out, is investigated. The performance of two-stage designs combining Pareto Sampling or Sequential Poisson sampling in the first stage, with Inverse Bernoulli Sampling or Systematic Sampling in the second stage, is studied by a computational Monte Carlo experiment using data from the serological Survey Sample Continuar Cuidando, carried out in the Brazilian state of Paraiba, to monitor the COVID-19 epidemics.