Fast Moving Sampling Designs in Temporal Networks

TL;DR

This paper analyzes idealized sampling designs in temporal networks, focusing on link tracing and probabilistic procedures to understand network dynamics and inform intervention strategies.

Contribution

It introduces simplified models of network sampling designs to better understand their properties and implications in dynamic network environments.

Findings

Designs use link tracing and probabilistic procedures

Provides inference methods for time-changing networks

Enhances understanding of natural and human network behaviors

Abstract

In a study related to this one I set up a temporal network simulation environment for evaluating network intervention strategies. A network intervention strategy consists of a sampling design to select nodes in the network. An intervention is applied to nodes in the sample for the purpose of changing the wider network in some desired way. The network intervention strategies can represent natural agents such as viruses that spread in the network, programs to prevent or reduce the virus spread, and the agency of individual nodes, such as people, in forming and dissolving the links that create, maintain or change the network. The present paper examines idealized versions of the sampling designs used to that study. The purpose is to better understand the natural and human network designs in real situations and to provide a simple inference of design-based properties that in turn measure properties of the time-changing network. The designs use link tracing and sometimes other probabilistic procedures to add units to the sample and have an ongoing attrition process by which units are removed from the sample.