Measuring Social Well Being in The Big Data Era: Asking or Listening?

TL;DR

This paper introduces a novel method to measure social well-being by analyzing social media conversations using supervised sentiment analysis, providing a spontaneous and scalable alternative to traditional self-report surveys.

Contribution

It proposes the Social Well Being Index (SWBI) derived from social media data using the innovative iSA sentiment analysis technique, enabling large-scale, unbiased welfare perception measurement.

Findings

SWBI effectively captures social well-being from Twitter data.

The method scales across languages and large datasets.

SWBI provides insights into welfare trends in Italy from 2012-2015.

Abstract

The literature on well being measurement seems to suggest that "asking" for a self-evaluation is the only way to estimate a complete and reliable measure of well being. At the same time "not asking" is the only way to avoid biased evaluations due to self-reporting. Here we propose a method for estimating the welfare perception of a community simply "listening" to the conversations on Social Network Sites. The Social Well Being Index (SWBI) and its components are proposed through to an innovative technique of supervised sentiment analysis called iSA which scales to any language and big data. As main methodological advantages, this approach can estimate several aspects of social well being directly from self-declared perceptions, instead of approximating it through objective (but partial) quantitative variables like GDP; moreover self-perceptions of welfare are spontaneous and not obtained as answers to explicit questions that are proved to bias the result. As an application we evaluate the SWBI in Italy through the period 2012-2015 through the analysis of more than 143 millions of tweets.