Forget GDP. Data Crunchers Measure Happy Tweets for Key Economic IndicatorBy
A group of Twitter-junkie data scientists launched a tool this week to measure the ups and downs of public contentment, an index they are calling the hedonometer that takes the pulse of the national mood and, the researchers hope, may even one day play a role in informing economic and public health policies.
Peter Dodds and Chris Danforth, mathematicians from the University of Vermont Complex Systems Center, worked with big-data specialists at Mitre Corp. to crunch 50 million Tweets per day; they aim to measure various other social media inputs such as Google Trend data in the future. The scientists say the measurements provide a deeper insight into public sentiment in real time. In introducing the hedonometer, the duo describe it as a “Dow Jones Index of Happiness.”
“Ultimately, we would like for government officials to use this tool in combination with more traditional indicators like GDP and consumer confidence when determining the emotional health of the population,” says Danforth. He adds that the index could help health officials gauge public adherence to new regulations, such as, say, New York City’s proposed ban on megasize soft drinks.
Here’s how it works: The hedonometer measures Twitter’s Gardenhose feed, a random daily sampling of roughly 50 million English-language tweets, generating usually about 100 million individual words. The entire word cache is then scored each day to determine a sliding mood scale. Drilling down to the word level, the scientists have scored 10,000 terms, weighting them for their positive or negative connotations. Words such as “happy,” for example, score high (8.3 out of 10) for positive sentiment, whereas words like “war” or “dead” score low. The more positive words that turn up in the cache, the higher the score. (You can see where ironic tweets might skew things.)
The scientists have been tracking public sentiment via Twitter since September 2008, the month Lehman Brothers went bust. The hedonometer assessment of Twitter chatter that day—Sept. 15, 2008—shows the term “bankruptcy” trending, sinking the overall mood of the index. The “saddest” day so far recorded was April 15, 2013, the Boston Marathon bombings. The most positive days consistently are public holidays, with Christmas regularly generating the highest score of the year.
Using similar methodology, the researchers prove Twitter can be tapped for assessing mood on a regional level. They have used the available geodata in tweets, for example, to determine America’s happiest town. It’s Napa, Calif., the heart of wine country. The least happy? Beaumont, Tex., the researchers found.
The scientists say they are building on ideas that are more than 100 years old. The 19th century political economist Francis Edgeworth coined the term “hedonometer” to describe a theoretical instrument to measure “the height of pleasure experienced by an individual.” And, more recently, the term “gross happiness indicator,” first uttered in the 1970s, is a belief that the true might of a country is not its wealth, but rather the collective contentment of its citizens. Many economists dismissed the concept of a GNH as too fruity and nearly impossible to quantify. But the idea has been revived in recent years, most recently with economist Joseph Stiglitz arguing that traditional benchmarks of economic success, namely GDP, are an inadequate measure of social progress, and that new measures are needed. Thanks to data-crunchers, those indicators may finally come to fruition. “I think our instrument will contribute to a dashboard of indicators,” says Danforth.