Unveiling the facade of blog posts to reveal subjective well-being : an exploratory study.

• Main Authors: Ou, Zhiqi., Tan, Justine Yi Ting., Tan, Jessie Zi Xi.
• Other Authors: School of Humanities and Social Sciences
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Social sciences::Economic theory
• Online Access: http://hdl.handle.net/10356/48860
• Institution: Nanyang Technological University

With the increased usage of social media over the past few years, online platforms like blogs and micro blogs may offer a valuable source of information with regards to behavioral economics and policy design. In this paper, we will carry out an exploratory study on the feasibility of harnessing these data from the social media to measure well-being. In comparison to the traditional method of conducting questionnaires to measure subjective well-being, using online platform enables us to attain revealed preferences instead of stated preferences. Hence, this paper serves as a bridge between the traditional methods and the usage of the social media as a new tool to assess subjective well-being. Our focus is on feelings recorded from blog posts in the 50 US states collected by We Feel Fine (www.wefeelfine.org). Due to the wide range of words that describes feelings, we have narrowed down the number of words that are rated as positive or negative based on the guidelines obtained from Affective Norms for English Words (ANEW) for our research. These feelings are aggregated and regressed against a range of objective indicators. They are also regressed against a subjective indicator of life-satisfaction obtained from the United States Behavioural Risk Factor Surveillance System (BRFSS). In this exploratory study to form a bridge between existing methods to gauge the well-being of people and the usage of social media as a data source, we find that there are substantial links between the subjective and objective well-being of people and the feelings expressed by people in blogs. Our two main findings demonstrate that eight out of 11 of the regressions involving the objective ranking as the main independent variable show the objective ranking having the correct expected sign at at least the 10% significance level. As for the subjective ranking, 11 out of 13 of the regressions show it having the correct expected sign at at least the 10% significance level. However, limitations exist for using the social media as a replacement for the traditional methods. First is the non-random nature of the sample available, whereby the characteristics of people who post on the social media may be different from those who do not. We also face the problem of “naive empiricism”, in which the same “feeling” is interpreted to be the same no matter how it is used in the sentence. The last limitation is that we are unable to obtain differences in the well-being of people with regards to specific policy changes. Our paper will form the foundation for future research to overcome these limitations and improve the usage of social media as a data source to measure subjective well-being. For the time being, social media could still be used as a supplement in gauging the well-being of people. A high level of well-being reflected in the social media could possibly be regarded as a necessary but not sufficient condition for a high level of well-being in the general population.