Judgement Post-Stratification to Improve Inference of Link-Tracing Samples with Applications to Network Data

• Activity Number: 41
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Ran Wei* and Elizabeth Stasny and David Sivakoff
• Companies: The Ohio State University and The Ohio State University and The Ohio State University
• Keywords: Network Sampling ; Social Networks ; Hidden Population ; Link Tracing Sampling ; Web Crawling ; Improved Inference

Abstract

With the popularity of online social networks such as Facebook, Twitter and LinkedIn, the scale of network data has become enormous. Taking samples that are representative of the full network is one major concern. In particular, link-tracing sampling methods are effective for obtaining samples from hard-to-reach populations, however link-tracing methods often result in substantially biased samples.

Judgement Post-Stratification (JPS) is a data analysis method based on ideas similar to those in ranked set sampling. Besides a variance reduction role, as in traditional sampling schemes, stratification also helps reduce bias in a size-biased sampling scheme by down-weighting units that are more likely to be selected. Applications of JPS to improving estimation when using link-tracing sampling methods is discussed in the paper. Comparisons with traditional methods for compensating for size-biased samples, such as Horvitz-Thompson and Volz-Hackethorn Estimators, are given. JPS is demonstrated to provide a flexible framework, which incorporates various information and can be applied in broader cases. Analyses were conducted on both simulated data and real world social networks data.