Researchers from Northeastern University and Harvard College embarked on an interesting endeavor: they fed about 300 million tweets to a system using ANEW (Affective Norms for English Words, from the University of Florida), which assigns a positive or negative emotional value to words. They then mapped the results onto a tweet density preserving US map, which makes land masses larger relative to the amount of tweets its population generates.
A fair criticism about the method is that the word-by-word ANEW rating is a pretty rudimentary model when one wants to capture the mood of a complex sentence, even limited to 140 characters. For instance, a tweet like “I like the dark color of this painting” may be misinterpreted as negative because of the word "dark." But the researchers argue that this kind of bias is uniformly distributed over the country, like a background noise, which may affect the basic scale but not the relative positivity of mood. Regardless of the study’s imperfections, it results in a pretty cool video showing the American mood evolving over two days.
From this animation, the West Coast is clearly happier than the East Coast. Within one day, the best mood occurs in the morning and evening, with a low between noon and 3PM. Not surprisingly, the weekend shows the highest level of positive mood over the week.
10 Comments - Add comment