Model
Digital Document
Publisher
Florida Atlantic University
Description
The Hedonometer analyzes Twitter data using human evaluations of happiness to give a happiness score for a given day. The goal of this study was to be able to predict the Hedonometer’s happiness score for a given day using the Linguistic Inquiry and Word Count (LIWC). Using a sample of over 15 million Tweets gathered from Archive.org’s Twitter Stream Grab, the positive emotion dictionary of the LIWC was able to predict happiness on an independent sample, R2 = .57, p < .001. When adding seven additional LIWC dictionaries and using lasso regression, predictive power improved, R2 = .85, p < .001. This reveals that different language analysis metrics may also be able to reveal positivity and happiness within the population of Twitter.
Member of