U.S. Mood Measured Through Twitter, 2006-2009
A group of researchers have published a simple word analysis of 300 million tweets (you know, those short, 140 character-maximum status updates from individuals) from Twitter and discovered something amazing — people are happier on the weekends, and before and after work. Yes, that’s right — people are happier when they are not working!
It took three researchers from Northeastern University and two from Harvard Medical School to arrive at these stunning conclusions.
Now, since researchers didn’t actually look at 300 million tweets individually, the mood of each tweet was inferred using the ANEW word list — Affective Norms for English Words — a word-rating system that gives normative emotional ratings for English language words. These kinds of analyses are indirect and rough measurements — they can only note very large trends because they are not necessarily reliable.
Click through to watch the video and read more.
For instance, such an analysis can’t take into account things like sarcasm, double-negatives, or opposite meanings in human language (which may be especially prevalent in informal social networking services like Twitter). “I hate this sunny day ” for instance can be scored as being negative by the researchers, when in fact it’s likely a positive tweet. Since the researchers didn’t say, we have no way to know how many of these kinds of tweets might have been misclassified.
Another limitation of the data (and it’s interesting to note that since this wasn’t published in a peer-reviewed journal, the researchers don’t mention the limitations of their data or methodology) is that it also infers location only from those users who’ve enabled Twitter to post their geographic location. So are people in the southwest U.S. tweeting more? Or are they simply more open to turning this feature on in Twitter? It seems like an obvious alternative explanation.
It was enough to create a cool video that I’m certain will make the rounds on the Internet (a video that also shows that we’re not apparently very happy while we’re sleeping). The graphs also show that the West Coast’s “mood” is approximately 3 hours behind the East Coast’s mood — an amazing correlation to the actual time difference between the two coasts of our nation.
Sorry, but when I see data like this that state the obvious, I just can’t help but scratch my head and wonder why five researchers spent dozens (hundreds?) of hours to analyze 300 million tweets. Did they honestly think they were going to find something astounding in such “data”? There are so many other areas of mental health and real epidemiological work that deserve research attention, it just seems like a waste of some obviously-smart, forward-thinking researchers.
Read the full report on their website: Pulse of the Nation: U.S. Mood Throughout the Day inferred from Twitter
Grohol, J. (2010). U.S. Mood Measured Through Twitter, 2006-2009. Psych Central. Retrieved on May 29, 2015, from http://psychcentral.com/blog/archives/2010/07/23/u-s-mood-measured-through-twitter-2006-2009/