Mina Zheng
Research Mentor(s): Josh Pasek
Mentor Department: Center for Political Studies
Authors:
Session: Session 3 (11:00am – 11:50am)
Presentation Type: Poster 36
Abstract
This project examines the set of approaches used in contemporary political polling and assesses how decisions made for different election surveys shaped the accuracy of the results. To assess this, we scraped data from multiple polling sites to compile a complete collection of methodological data that would allow us to make sense not only of the methodological choices being made, but also how those shaped predictions of voting behaviors across racial, age, and education categories, among other groups. To understand how results collected from polling firms differed from the actual result from the election, we focused on classifying polling information across two main attributes: use of weighting variables and sampling method .We compare manual coding of these categories across coders, with AI, and with election results. Accuracy rates vary across the different categories. This research provides insight into the value of Open AI for identifying the evolving landscape of political polling, specifically when determining the strengths and weaknesses of sampling and surveying techniques.
