Twitter map predicted outcome of the US election
A Twitter map created by Dr Mark Graham from the Oxford Internet Institute with US colleagues predicted the re-election of Barack Obama before the actual results came in.
The team collected about 30 million geocoded tweets between 1 Oct and 1 Nov and pulled out all references to Obama and his Republican contender Mitt Romney. They found that if the election was decided purely on Twitter mentions, then Democrat President Obama would be re-elected. They were right – even if the real contest was not as tight as the Twitter map had suggested.
The Twitter map showed a total of 132,771 tweets mentioning Obama and 120,637 mentioning Romney, giving Obama 52.4% of the total mentions (and Romney 47.6%). In fact, by Wednesday morning, President Barack Obama was well past the winning post of 270 electoral college votes, having swept 303 compared with only 206 for his Republican challenger Mitt Romney.
According to the percentage of tweets registering his name, the only states that Romney would win would be Maine, Massachusetts, New Mexico, Oregon, Pennsylvania, Utah, and Vermont, as well as the District of Columbia. What did the Twitter map tell us? Dr Graham said: ‘I doubted that they would accurately predict that Obama would win in Texas or that Romney would win in Massachusetts. But the overall outcome was accurate. They also reveal that many internet users in California, Texas, and much of the country prefer talking about Obama to talking about Romney.’
There is a surprising similarity between mentions of the candidates on Twitter and the number of votes cast on the day of the US election.
Dr Mark Graham
The data can be further examined using a sliding scale in order to better see how close the margin of victory is in each state, and Twitter mentions of Romney and his running mate Paul Ryan can be compared. However, Dr Graham concedes the map takes a very ‘broad brush’ approach as it does not differentiate between tweets that were supportive from those that were critical of the candidates. ‘We would need to employ sentiment analysis or manually read a large number of the election-related tweets in order to figure out whether we are seeing messages of support or more critical posts,’ he said.
Some of the results seem to be interesting reflections of social and political characteristics of particular places. For example, the tweet map shows that Romney appeared to capture more of the public imagination in Utah (perhaps due to the state’s large Mormon population) and Massachusetts (the state where he once served as governor).
Obama said in his election speech: ‘We will continue our journey forward’, a sentiment that Dr Graham has taken on board. ‘When aggregated across all states there is a surprising similarity between mentions of the candidates on Twitter and the number of votes cast on the day of the US election,’ he said. But he added there is still work to do: ‘The Twitter results for each state may not ultimately tell us much about the Electoral College results.’