Tech Watch: Nate Silver, Facebook, and Why Numbers Don?t Lie
It's all just a matter of numbers. Can we predict the election before voters take to the polls?
The election is over, the polls have closed, and Barack Obama has been re-elected as President of the United States. Pundits on CNN, MSNBC, and Fox News all touted the election as being "too close to call"…but to those of us looking at the numbers, the 2012 election was simply a matter of, well, numbers.
Nate Silver, the statistician and creator of the blog FiveThirtyEight, argued that the election was not as close as many pundits would have Americans think. Based on his logic, it was the local state polling data that mattered more than the national polls. By creating algorithms to account for the validity of poll numbers and the change in data from one week to another along with other factors, he was able to confidently project that President Obama would win re-election from day one (even before Mitt Romney became the Republican Nominee).
Despite the pundit's protestations on the subject of Silver's quantitative approach to political science, I happened to agree with the foundation of his logic. Despite the poor economy, a terrible debate performance by Obama, and a general apathy from the Democratic Party, the numbers were still in the incumbent's favor. Silver looked at the state-by-state data, as well as information about TV ad spending to project who would win the electoral college.
However, I would argue that Silver left out one key element to his game of numbers: digital media. While poll numbers and TV ads are – and will always be – critical to any election cycle, the trend in recent elections points toward a greater focus online as well as offline. Specifically with this most recent election, both presidential campaigns had entire teams dedicated to the candidate's digital presence. By taking a greater look at these numbers, you can easily project who is going to win an election based on digital media.
For example, a Washington Post article recently reported that the Obama campaign budgeted over twice as much for Google Adwords (Google Paid Search Results) and over three times as much for Digital Display Ads (banner ads, ads on YouTube, etc). Beyond this, Obama had a more robust and effective digital strategy on social media as well. The Barack Obama Facebook page had 32,720,395 "likes" and over 2 million people interacting with the page consistently as of Election Day. Compare that to Mitt Romney's Facebook page, which had only about a third of the total "likes" as President Obama (12,113,130) and the same amount of people interacting with the page. Looking at Twitter, Obama crushes Romney in terms of followers (22,640,472 to 1,775,085), and as of last night one of Obama's tweets became the most popular tweet of all time.
However, this trend towards projecting the winner of an election based on digital media is not limited to national races. Locally, we had a number of candidates that took to social media. In the PA State House District 3 race between Ryan Bizzarro and Jason Owen, the winning candidate (Ryan Bizzarro) had over four times as many "likes" on his Facebook page and over three times as many people interacting with the page on a regular basis. In the State Senate District 49 race between Janet Anderson and Sean Wiley, a similar trend emerges. Unfortunately, I can't give specific numbers on the differences between the two campaigns because Janet Anderson's Facebook page is already deleted.
In general, Silver's method works. He's predicted essentially every race in the 2008, 2010, and 2012 elections accurately. Moving forward, I hope he includes some aspects of social media into his analysis and trending. The effective implementation of a digital strategy has been proven time and time again to work well for political candidates, and the results of those efforts must be quantified and taken into account for any realistic measure of a campaign's health.
Now, who's ready for 2016??
Michael Haas can be contacted at Epic@ErieReader.com.