Combining cluster sampling and link-tracing sampling to estimate the size of a hidden population: Asymptotic properties of the estimators

Abstract

Félix-Medina and Thompson proposed a variant of link-tracing sampling to estimate the size of a hidden population such as drug users or sexual workers. In their variant a sampling frame of sites where the members of the population tend to gather is constructed. The frame is not assumed to cover the whole population, but only a portion of it. A simple random sample of sites is selected; the people in the sampled sites are identified and are asked to name other members of the population, who are added to the sample. Those authors proposed maximum likelihood estimators (MLEs) of the population size that derived from a multinomial model for the numbers of people found in the sampled sites and a model that considers that the probability that a person is named by any element in a particular sampled site (link-probability) does not depend on the named person, that is, that the probabilities are homogeneous. Later, Félix-Medina et al. proposed unconditional and conditional MLEs of the population size, which derived from a model that takes into account the heterogeneity of the link-probabilities. In this work we consider this sampling design and set conditions for a general model for the link-probabilities that guarantees the consistency and asymptotic normality of the estimators of the population size and of the estimators of the parameters of the model for the link-probabilities. We showed that the unconditional and conditional MLEs of the population size are consistent, that they have different asymptotic normal distributions, and that the unconditional ones are more efficient than the conditional ones.