Toward an enriched understanding of factors influencing Filipino behavior during elections through the analysis of Twitter data

Social networking sites (SNS) have become increasingly popular means for self-expression, communication, and influence, particularly in relation to political activity. In cases involving elections, for example, studies have been conducted to explore how these sites have been mobilized as a tool to disseminate opinion, encourage others to vote, publicize one's voting decisions, and even predict election outcomes. In this study, we explore how data from the SNS site Twitter can be used to enrich current understandings of the nature of voters in a Philippine setting. We mobilized a combination of manual and computer-based natural language-processing (NLP) techniques to analyze Twitter data in Metro Manila concerning the 2013 Philippine elections for a two-week period. Specifically, we combined language modeling using n-gram and manual discourse analysis to generate over 30 themes from Twitter users' comments on the 2013 Philippine elections. The themes cohered around a number of different narratives that shed light on voters' behavior, specifically why they vote, how they vote, and who they vote for, thus deepening our understanding of the different interlocking factors that shape election and voting processes.