NCRM International Visitor Exchange Scheme (IVES)

Designing New Methods and Data for Conducting Election Studies and Measuring Public Opinion (2017 -2018)

Professor Rachel Gibson (University of Manchester) (Rachel.Gibson@manchester.ac.uk) is visiting Professor Jonathon Nagler (New York University).

This research will examine the extent to which social media data, primarily Twitter and Facebook posts, can provide a new way of measuring public opinion. In particular we will assess to what extent they offer a possible replacement for survey data, or whether they are better viewed as an extension or enhancement of survey methods. To investigate this question the project is broken down into two main strands of activity. The first of these will seek to understand how we can increase the accuracy and value of social media data as a ‘stand alone’ source of public opinion. Social media platforms and particularly Twitter have attracted growing interest among social scientists in terms of the large volume of publicly accessible information that generate regarding individuals everyday lives and opinions. However, they clearly have numerous problems of bias and representation as well as practical challenges in terms of data storage, management and analysis.

This project will seek to address one of the key challenges that arises from Twitter data with regard to bias and representation – namely that of geolocation, i.e. the ability to assign a tweet and thus tweeter to a particular location. In order to render Twitter data representative of a wider population it is important first to be able to know that it is generated by members of the target group we are seeking to generalize to. The second project will look at the extent to which social media data can enhance survey data and the potential for linkage of the two forms. It will do so specifically in the context of the National Election Studies (NES) in the U.S. and the UK. These are the first national election studies to create this new hybrid type of data set and our goal will be to assess how each type of data can be used in conjunction to better understand political attitudes and drivers of individual electoral behaviour.