An ROC Approach to Estimating Interpersonal Networks

• Activity Number: 498
• Type: Contributed
• Date/Time: Wednesday, August 3, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Deniz Yenigun* and Gunes Ertan and Michael Siciliano
• Companies: Istanbul Bilgi University and Koc University and University of Illinois at Chicago
• Keywords: Network sampling ; cognitive social structures ; network estimation

Abstract

This paper focuses on estimating interpersonal networks from a random sample of individuals that belong to the network. A design known as Cognitive Social Structures (CSS) asks individuals to report on their own ties as well as their perceptions of the networks that surround them. Based on data collected using CSS design, first section of the paper dissects cognitive errors as errors of commission and omission, and displays the relationship between the two. Second section of the paper uses a Reciever Operating Characteristic (ROC) based approach to estimate networks from a small sample of randomly selected individuals, while controlling for both types of errors. Results from a numerical study are provided to illustrate the performance of the suggested method.