Tweeting Alone? A Computational Analysis of Twitter Users' Social Capital and their Expressed Well-Being

Much scholarly research has been devoted to the relation of social capital and happiness (for a review see Helliwell, 2001). This study builds on such previous research by examining whether social capital can be used to predict the expressed well-being of Twitter users. Apart from its focus on Twitter users, the contributions of this study are twofold. First, it distinguishes between bonding, bridging and overall social capital, a differentiation that has been neglected in most previous studies examining the relationship between social capital and happiness although most theoretical discussions of social capital make this distinction. Second, I take the perspective of the rather new discipline of Computational Social Science, which aims to study societal processes through the analysis of large volumes of data. I used a computational framework for the determination of social capital developed by Smith, Giraud-Carrier, Ventura et al. (2011) to determine the social capital of Twitter users and I calculated their expressed well-being based on a sentiment analysis of their tweets. To my knowledge, the framework for the determination of social capital has not been applied to Twitter before to this extent and I expanded on it in several ways to make it applicable to this study. Since research suggests that the positivity of tweets is indicative of their author's happiness (Kramer, 2010; Bollen et al., 2011) and much research shows that social capital is related to happiness (for a review see Helliwell, 2001), I expected to find that social capital can be used to predict the expressed well-being of Twitter users. I analyzed the social capital and expressed well-being of 214 Twitter users that were determined randomly according to specific criteria which made them suitable for the analysis. In order to determine their social capital, I collected a dataset consisting of 43,670,346 user profiles and 79,579,346 tweets. Contrary to the expectations, neither bonding, bridging nor overall social capital were found to be significant predictors of Twitter users' expressed well-being.