Signal in the network: Measuring public opinion without polls

The project objective is to develop a method for measuring public opinion using social media data. Presently the ability to derive externally valid inferences from social media data is impeded by issues such as sample bias and data structure. By applying recent innovations in machine learning to account for such issues, this project aims to develop a robust, reliable and inexpensive means by which to continuously gauge public opinion with respect to a given topic.
This project benefits the partner organization by helping advance its mission of promoting democratic participation. The benefit to the intern is the ability to work with the partner organization’s unique data which are central to the development of the proposed method.

Faculty Supervisor:

Paul McNicholas

Student:

Corentin Vande Kerckhove

Partner:

Vox Pop Labs Inc

Discipline:

Statistics / Actuarial sciences

Sector:

Professional, scientific and technical services

University:

McMaster University