People tweet happier on vacation

Social psychology is, i suppose, a science. It meets the criteria taught to eighth graders: there is a hypothesis and then experimental attempt to probe the hypothesis with experimental observations. True, the hypothesis part of this is more often than not omitted. The experiment is the thing is this science. I sometimes think the experiment is the only thing in this science. Dan Ariely, for example, continues to delight NPR reporters and New York Times reviewers with book after book of accounts of experiments involving such things as offering for sale candy bits to students at major universities and then musing endlessly about why one at one price rather than another at another price. Ariely is able to make all these experiments both sound well designed and pertinent to some important behavior or another and confirm some aspect of a very comfortable view of human nature. It never seems to bother anyone that impulse purchases by students involving literally pennies might not actually represent behavior of adults making prudent decisions with real hard-earned money. I guess the glib writing makes up for the insubstantial conclusions.

Seemingly silly experiments, with small data sets resulting in somewhat conventional interpretations of human behavior are just the sort of thing that newspapers (when there used to be newspapers) would describe wryly, and if government funds were involved could set the authors up for something like William Proxmire’s Golden Fleece Award, itself a silly publicity stunt usually involving a misinterpretation of the research and dedicated to wasting public attention on trivial expenditures. This blog has occasionally pointed out some studies of social psychology which produce something less than startling conclusions. But i won’t link them. Those posts were written to elicit cheap outrage. We have moved on from that.

But not all social science studies have devoted themselves to reflecting in lambent glow the lovable rationally irrational aspects of our soul.

Older experimenters in this general field produced darker views. Stanley Milgrim of Yale was able to persuade subjects to administer what they believed was fatal doses of current on those they believed were the real subject of the experiment. Those were the days! You are not likely to get funding for that sort of experiment nowadays. Of course you don’t much need it anymore either. You only have to look at our drone program to conclude that ordinarily well-meaning intelligent people can be persuaded that horrific and inhuman use of force is a routine policy option. Civilian statistics, by definition, never tell a tale.

Well, today’s study doesn’t involve any of those long dark nights of the soul; quite the contrary, nothing could be more joyous and light, it’s from the physics papers and involves the chief repository of the mental activity of modern man; twitter. It’s called: “Happiness and the Patterns of Life: A Study of Geolocated Tweets” and its written by Morgan R. Frank, Lewis Mitchell, Peter S. Dodds and Christopher M. Danforth and can the abstract and links to various formats of the full paper can be found here (or you can go directly to the PDF version if you hate making choices).

If you like you data in vast quantities, this is your paper. No Dan Ariely here. The authors crunched an amazing 37 million Tweets from 180,000 individuals during the year 2011. Maybe you yourself are involved in the study. Your tweet would have had to disclose your location, because that’s what they were analyzing: How happy you are depending on where you are. But don’t worry, these guys can track you down.

The paper begins like all social psychology experiments, by first humbly acknowledging that the authors stand on the shoulders of giants and they tell of past research and then, as required in these papers, they mouth the limitations of these previous efforts:

these studies previously had limited access to conversation content, rendering changes in expression as a function of movement invisible. In addition, they typically use the communication between a mobile phone and its nearest antenna tower to infer position, limiting the spatial resolution of the data to the geographical region serviced by each cellphone tower.

One almost feels sorry for these previous studies when you see what these guys did;

We use a collection of 37 million geolocated tweets to characterize the movement patterns of 180,000 individuals, taking advantage of several orders of magnitude of increased spatial accuracy relative to previous work. Employing the recently developed sentiment analysis instrument known as the hedonometer, we characterize changes in word usage as a function of movement, and find that expressed happiness increases logarithmically with distance from an individual’s average location.

If you like data analysis for its own sake, as I do, you will love this work. And, while the authors are no Edward R. Tufte, the paper has pleasing enough illustrations, and many different kinds: dot distribution maps, scatter plots, frequency charts and the like. Even their line graphs are tantalizing. And this is what they do. Crunching the data in a number of different ways, they posit where the tweeter lives and works and then analyze the happiness content of the tweets, depending on whether they are at home, work or away. And they conclude that the happiness content increases the farther one goes from work and home.

How do they determine that? Well, they assign happiness value to works in the tweet. Negative or downer words like “‘hate’, ‘damn’, ‘dont’, ‘mad’, ‘never’, ‘not’ and assorted profanity” give the tweet a negative score on the happiness scale. For example, “I hate my job” gives the tweet a negative value. I assume that the “never” in “I would never kick a puppy” is cancelled out by the positive “puppy” so if you are far away from home and tweeting that noble sentiment you are not considered to be sending out bad vibes away from work and home. Positive words like “‘great’, ‘new’,  ‘dinner’, ‘hahaha’, and ‘lunch’” are considered positive on the happiness scale.

With these rules in mind they correlate the happiness words with how far you tweet from home and work. And guess what? The farther you are from work and home, the more happy you tweet!

Now like all such papers there a large bit of “Yeah, so what’s new?” in it. In fact, it may all be in that category. After all, “beach” is one of the happiness words. And, I think it’s safe to say without need for crunching large quantities of data, that a person is more likely to tweet “beach” when he’s on vacation than when he is in an office cubicle. And when you start thinking about it, the whole concept is sort of self-evident. When are you large distances from hom? When you are on vacation. When are your likely to be in a better mood? Even when you are away on business, you likely have an expense account, your boss is not there (or is in a good mood because he/she has an expense account) and you get to stay in an Embassy Suite, which advertisements tell us make you happier. Yes there is some really elegant data analysis but what is the point?

So I was about to put this entire enterprise down to yet another exercise in pointless social psychology data analysis, until I remembered a fact that a sharp reader would have picked up ere this. The paper comes from the physics And a closer inspection shows the authors are not social psychologists, they are from the Computational Story Lab, Department of Mathematics and Statistics, Vermont Complex Systems Center, Vermont Advanced Computing Core of the University of Vermont. Not a seller of candies to university students among them. So why would these guys, who probably have much cooler Macs that I do, bother slumming in studies that belong in Current Directions in Psychological Science? So I look at who funded the study. (One of Bob Woodward’s wise fictional characters once said, “Follow the money.”) We find that the MITRE Corporation paid for the work. When you do some detective work, you find in Wikipedia, that the MITRE Corporation is a not-for-profit that distributes government funds supporting the Department of Defense, the Internal Revenue Service, the Department of Homeland Security (DHS), among other agencies.

Well, I guess that doesn’t help. It is a mystery to me why those organizations would get involved in the risible field of social psychology. It is even more perplexing why our government would pay for extensive computational work, designed to geolocate persons using Twitter and analyzing what they are thinking. And while it seems that the authors have shown that if anyone were interested in spending huge sums of money mining social media, that money ought to be directed to them, but, really, who would want to do that?


  1. April 30th, 2013

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s