arXiv subjects

Steve Thompson

Publications and source records attributed to Steve Thompson.

6 recordsLinked to original sources

arXiv2020

Estimation for network snowball sampling: Preventing pandemics

Snowball designs are the most natural of the network sampling designs. They have many desirable properties for sampling hidden and hard-to reach populations. They have been under-used in recent years because simple design-based estimators and confidence intervals have not been available for them. The needed estimation methods are supplied in this paper. Snowball sampling methods and accurate estimators with them are needed for sampling of the people exposed to the animals from which new coronavirus outbreaks originate, and to sample the animal populations to which they are exposed. Accurate estimates are needed to evaluate the effectiveness of interventions to reduce the risk to the people exposed to the animals. In this way the frequencies of major outbreaks and pandemics can be reduced. Snowball designs are needed in studies of sexual and opioid networks through which HIV can spread explosively, so that prevention intervention methods can be developed, accurately assessed, and effectively distributed.

stat.ME↗

arXiv2020

New estimates for network sampling

Network sampling is used around the world for surveys of vulnerable, hard-to-reach populations including people at risk for HIV, opioid misuse, and emerging epidemics. The sampling methods include tracing social links to add new people to the sample. Current estimates from these surveys are inaccurate, with large biases and mean squared errors and unreliable confidence intervals. New estimators are introduced here which eliminate almost all of the bias, have much lower mean squared error, and enable confidence intervals with good properties. The improvement is attained by avoiding unrealistic assumptions about the population network and the design, instead using the topology of the sample network data together with the sampling design actually used. In simulations using the real network of an at-risk population, the new estimates eliminate almost all the bias and have mean squared-errors that are 2 to 92 times lower than those of current estimators. The new estimators are effective with a wide variety of network designs including those with strongly restricted branching such as Respondent-Driven Sampling and freely branching designs such as Snowball Sampling.

stat.ME↗

arXiv2019

Design-adherent estimators for network surveys

Network surveys of key populations at risk for HIV are an essential part of the effort to understand how the epidemic spreads and how it can be prevented. Estimation of population values from the sample data has been probematical, however, because the link-tracing of the network surveys includes different people in the sample with unequal probabilities, and these inclusion probabilities have to be estimated accurately to avoid large biases in survey estimates. A new approach to estimation is introduced here, based on resampling the sample network many times using a design that adheres to main features of the design used in the field. These features include network link tracing, branching, and without-replacement sampling. The frequency that a person is included in the resamples is used to estimate the inclusion probability for each person in the original sample, and these estimates of inclusion probabilities are used in an unequal-probability estimator. In simulations using a population of drug users, sex workers, and their partners for which the actual values of population characteristics are known, the design-adherent estimation approach increases the accuracy of estimates of population quantities, largely by eliminating most of the biases.

stat.AP↗

arXiv2018

Simple estimators for network sampling

A new estimation method is presented for network sampling designs, including Respondent Driven Sampling (RDS) and Snowball (SB) sampling. These types of link-tracing designs are essential for studies of hidden populations, such as people at risk for HIV. The simple idea behind the new method is to run a fast-sampling process on the sample network data to estimate the inclusion probabilities of the actual survey, and incorporate those in unequal probability estimators of population means and proportions. Improved versions of the usual RDS and SB designs are also proposed, termed RDS+ and SB+, to obtain information on more of the within-sample links. In simulations using the network from the Colorado Springs study on the heterosexual spread of HIV, the new estimators produce in most cases lower bias and lower mean square than current methods. For the variables having the largest mean square errors with current estimators, the improvement with the new estimator is dramatic. The estimates are improved even more with the enhanced design versions. For estimating the population mean degree, the efficiency gains using he new method are 29 for RDS, 54 for RDS+, 26 for SB and 80 for SB+. This means for example, with the ordinary RDS design, the mean square error with the new estimator, same data, is 1/29 that of currently used estimators. The new method is computationally intricate but is fast and scales up well. The new estimation method can be used to re-analyze existing network survey data. For new network sampling studies, it is recommended to use the improved designs as well as the new estimators.

stat.ME↗

arXiv2014

Estimating the size and distribution of networked populations with snowball sampling

A new strategy is introduced for estimating population size and networked population characteristics. Sample selection is based on a multi-wave snowball sampling design. A generalized stochastic block model is posited for the population's network graph. Inference is based on a Bayesian data augmentation procedure. Applications are provided to an empirical and simulated populations. The results demonstrate that statistically efficient estimates of the size and distribution of the population can be achieved.

stat.ME↗

arXiv2012

Estimating Population Size with Link-Tracing Sampling

We present a new design and inference method for estimating population size of a hidden population best reached through a link-tracing design. The strategy involves the Rao-Blackwell Theorem applied to a sufficient statistic markedly different from the usual one that arises in sampling from a finite population. An empirical application is described. The result demonstrates that the strategy can efficiently incorporate adaptively selected members of the sample into the inference procedure.

stat.ME↗