Social-media-driven #fitspo is all the rage, so why not measure the nation's health using Twitter? A new study does just that in a brutal state-by-state comparison of calorie burn.
The Lexicocalorimeter creates a ratio of caloric input to output in geotagged locations. It does so by sifting through countless tweets, looking for mentions of food and physical activities. The former can be anything from "nacho cheese Doritos" to "tofu"; the latter ranges from "The Carlton Dance" to "bear hunting." The numbers used to calculate the calories found in certain foods—caloric input—and calories burned during physical activities—caloric output—are based on generalized values.
The study, led by researcher Peter Dodds and a team of scientists at the University of Vermont, provides an analysis of each state. As it turns out, New Jersey burns fewer calories than the average state; "getting my nails done" is the most prevalent activity on its low-intensity list and "running" appears less frequently than America's overall average. Colorado's input-to-output ratio was the most balanced of all the states (meaning they burned lots of the calories they inhaled), while Mississippi's was the least balanced.
Analyzing 50 million tweets from 2011 and 2012, the researchers found that the largest contributor to calories burned nationwide was "watching TV or movies" (calories burned!), and the food responsible for most calories consumed in each state was pizza. The only exceptions were in Wyoming, whose largest caloric contributor was cookies, and Mississippi, where ice cream was the calorie-bomb of choice.
Some results from the analysis come as no surprise—that a lot of people note eating lobster in Maine and Massachusetts, for example. Other results strike as more arbitrary, as with the relative popularity of talking on the phone, showering, and sitting in Delaware, Virginia, and Tennessee, respectively.
The researchers note that their study is limited;tracking Tweets about food and activity leaves out all kinds of other important details that make up an accurate picture of state health. That said, Dodds and his team believe that using the Lexicocalorimeter in tandem with other assessments can help a great deal in evaluating a population's overall well-being.
The Lexicocalorimeter is available for browsing. UVM has previously released a Hedonometer, which uses similar methods to measure relative happiness. Dodds said in a press release that his team also hopes to create an "Insomniameter"—and maybe a "Hangoverometer," too. But the team's main goal right now is to make the Lexicocalorimeter into a real-time assessment of public health.