I read an article in the New York Times Magazine the other day about the analysis of content from social media websites (such as Twitter and Facebook) for predictions of everything from box office success, to changes in the Dow Jones Industrial Average.
By examining Tweets, and classifying them as falling into six mood categories (happiness, kindness, alertness, sureness, vitality, and calmness), Johan Bollen and his research team at Indiana University were able to predict changes in the Dow Jones Industrial Average. Pretty intriguing stuff right? But how credible can this really be?
Don’t get me wrong, I’m not trying to argue the correlation found between the two. What I’m questioning is the credibility of the source of information. Having majored in Psychology in college, I know a thing or two about reliable sources and data, and at no point in time did any of my professors mention Tweets.
There are more than a handful of things that sound fishy about this, but let’s just focus on the simple fact of classifying Tweets into emotional categories. For example, let’s decode and classify this Tweet:
Sometimes I think I’m the only person in the world who hates musicals, but don’t worry guys, I hate them enough for all of us. 6:42 PM Dec 15th via HTC Peep
I would classify this Tweet as: alert, sure, vital, and calm. Did you come up with the same classifications? Maybe; but probably not. That’s because interpreting someone’s emotions is about as subjective and biased as it gets. (By the way, that’s my own Tweet, and as the author, I don’t even know that those emotional categorizations are correct/appropriate).
Additionally, if people are Tweeting happy Tweets because they’ve got some extra dough to spend (because let’s be honest, that brightens anyone’s day) aren’t these Tweets telling us information we would have known anyway from the strength of the market?
The details of the study weren’t divulged in the blurb from NYTM, so perhaps these things were somehow accounted for, but lets just say I’m not quite ready to throw the T-Distribution charts out the window.
Posted by Courtney