Twitter has generally been studied for its networking possibilities, its ability to easily transmit links to many people, and the way it removes boundaries between celebrities and the average person. But now, researchers believe they can study people’s tweets to ascertain telling information about human emotions.
As this article explains, the journal Science published a study by researchers at Cornell University who investigated 500 million tweets sent out by users in 84 countries over the last two years. They parsed out indications of positive and negative emotions—from enthusiasm and delight to distress and fear—as they were expressed at various times of day, days of the week, and times during the calendar year.
The findings confirmed many conventional notions of mood. Positive emotion was highest during the early morning, grew more negative during the workday, and improved again in the early evening. Moods were higher on weekends and when the sun was shining. The late evening hours, interestingly, featured high expressions of both positive and negative feelings, suggesting that the time is overall the most emotional one of the day.
The potential ramifications of Twitter accurately judging emotions would be far-reaching. As smartphones and tablets make it easier and easier for companies to reach consumers instantaneously, businesses might want to carefully allocate their messages based on the time of day. For example, attitude research suggests that people process less cognitive information when in a good mood—essentially, they don’t want to be faced with any reason for the good mood to dissipate—which might encourage companies to steer clear of promoting messages that involve the central route of processing at such times.
However, it must be strongly cautioned that this research was not particularly scientific. Aside from the admittedly subjective method of divining mood from tweets, the sample wasn’t remotely randomized. It slanted heavily towards young, Caucasian, wealthy people; furthermore, there was no demographic information for the tweets, making it much more difficult to separate the effects of gender, age, race, and SES from these results.
The linked article stated that these results might indeed be meaningful and generalizable because of “the fact that [the study] gives predictable answers.” Yet this is imperfect reasoning; the mere fact that the results meet our society’s ‘common sense’ expectations of emotions does not by itself make it legitimate. One of the consistent hallmarks of psychological research is proving common sense wrong; therefore, it should take more for such studies to gain critical acceptance.
Given these limitations, however, this marks an interesting step in the realm of using new-wave social media to glean emotional states. If such research proves legitimate, it will, of course, only be a matter of time before businesses try to find a competitive advantage from that.