Measuring the Ideology of Crowds: A Validation Test using Twitter Data

• In Event: Information, Media, Technology & Mass Protest in Comparative Perspective

Abstract

What are the political preferences of protesting crowds? To answer this question, existing research has relied either on retrospective nationally representative surveys or in-protest surveys of participants. The former is commonly limited by a small positive n; the latter is beset by problems of selection and non-response. Both techniques pose prohibitive financial constraints on the researcher. In this article, we propose a method for estimating the distribution of ideological preferences in a crowd using Twitter data. The method relies on combining ideal point and GIS techniques to estimate the ideological position of individually-geolocated protesters along a protest route. We demonstrate the validity of this method using a sample of >7m tweets from the 2017 Women’s March in the USA and compare against an in-protest survey of participants at the same protest. We find that the method provides comparable estimates of the distribution of ideological preferences as well as crowd demographics.

Authors

• Christopher Barrie

• Arun Frey, University of Oxford