I promise I’m not the type to bash scientists for what they do. Unless you’ve clearly misrepresented your data with underhanded intentions, I can respect your scientific endeavor. Ninety-nine percent of the time. But this research I just read about made me literally laugh out loud, so I may sound a bit facetious at times in this post.
Read on to find out how scientists analyzed millions of Twitter accounts and drew conclusions that left me thinking, “Tell me something I don’t know…?”
These two scientists from Cornell wanted to learn more about how people’s moods change over the course of a day, and across seasons. This type of information has been studied plenty in the past, but, to the authors’ credit, analyzing Tweets removes the risk of people inaccurately recalling their past emotions, as with surveys, and there’s no issue of response bias where people try to give answers they think sound good. Twitter can tell you how people are feeling in real time, with minimal pretense.
So these researchers looked at 509 million Tweets made between February 2008 and January 2010, representing 2.4 million Twitter users. The Tweets were analyzed by something called the Linguistic Inquiry and Word Count (LIWC), a lexicon that basically matches words to various behavioral and psychological concepts. For example, the word “cried” would be matched to five categories: sadness, negative emotion, overall affect, verb, and past tense verb. Scores are assigned to categories so that the words in a Tweet can basically be translated into a numerical measure of “positive affect” or “negative affect.”
Would probably be interpreted as a sad Tweet.
Don’t get me wrong, these guys definitely had a nice experimental setup, and they were careful about the caveats of their study: they don’t know the gender, race, occupation, or anything really, of the people they’re analyzing — other than that they’re Twitter users. Their results shouldn’t be applied to everyone globally. All they could say was what’s known about Twitter users in general: they’re 51% White, 24% Black, and 17% Hispanic, and generally richer and more educated than the population at-large.
Despite all this, what ultimately amuses me is the conclusions that they make. Below, I’ll highlight some of the main findings of this work:
1. “[Positive mood] levels were generally higher on Saturday and Sunday….” How curious! Even more curious, the peak in happy morning moods on the weekends occurred 2 hours later than on weekdays, so the authors made a bit of a logical stretch and suggested that “people awaken later on weekends.”
2. “[Positive mood] decreased midmorning (at the start of the work day) and increased in the evening (at the end of the work day).” This isn’t my area of study, but I’d hazard a guess that work isn’t most people’s “happy place.” But, again to credit the authors, they point out that the overall ups and downs in mood throughout the day are similar on weekdays and weekends, just shifted in time. Since most people don’t go to work on the weekends, the authors point to “sleep and the biological clock as important determinants of affect.” Sleep? Affecting my mood? The concept is a little unorthodox, but I’ll allow it.
A likely candidate for some sleep-induced "emotional refreshment."
3. “…People may be emotionally “refreshed” by sleep.” I just don’t know about this one…aren’t the happiest people usually the ones who pull all-nighters and make it through the day on energy drinks and high anxiety? Oh well, I can be open-minded.
4. “Most people are most active in the afternoon and evening….” Odd. I hear so many voices on the street outside my window at, say, 3AM. I’m actually not joking about this one….
5. “…Baseline [positive mood]…is higher when change in daylength is positive (as the summer solstice approaches) than when it is negative (as the winter solstice approaches).” Well, I can only speak for myself here, but I feel like the existence of seasonal affective disorder kind of beat these guys to the punch. Always nice when your data make sense, though.
To end on a nice note, this research is definitely worth a look. The methodology was cool, and they did get into the fact that differences in nationality and culture can manifest as differences in mood patterns across time. And in all seriousness, the idea of scientifically analyzing Twitter on a large scale for the purposes of learning about human behavior and psyche is an immensely cool concept. Maybe in the future this sort of analysis can be done on even more complex psychological parameters, if such can be gleaned from 140-character Tweets. For doing something that’s never quite been done before, I give these researchers props.
Speaking of Twitter…hit up our Facebook/Twitter/RSS (way over on the right-hand side of your screen)! Your support makes our nerdy world go ’round 
Scott A. Golder, & Michael W. Macy (2011). Diurnal and Seasonal Mood Vary with Work, Sleep, and Daylength Across Diverse Cultures Science, 333 (6051), 1878-1881 DOI: 21960633
I liked this study a lot. Some of the conclusions are fairly obvious but the finding that day length and circadian cycles have a small but noticeable effect on mood around the world is interesting, it might have been culture-specific.
I actually went into the article expecting responses to day light and circadian rhythms to be deeply rooted in human biology and moods, regardless of culture, but it was nice to see that born out in the data as well. I think this is a great proof of principle study!