Romney in a Landslide: Using Facebook Data for Predictive Analysis

Election coverage is filled with plenty of subjectivity, so here’s a little more:

Based on Facebook data, Mitt Romney is set to win in a landslide over Barack Obama, claiming 391 of the 538 votes from the Electoral College. That’s not exactly matching what the media or polls are saying … so how did I get there?

The Inspiration

Facebook’s EdgeRank Algorithm says that a Like is worth less than a Comment because a Like is a more passive form of engagement, while taking the time to type something out is seen as more active engagement.

I believe that the same can be said for people who Like a Facebook Page versus those who actively join groups or type something out in the interests section of their personal Page.

Following this logic, if we use Facebook Advertising to determine the relative size of an audience in the network based not solely on likes but on interests and groups as well, we should be able to identify the number of people actively engaged on behalf of a candidate.

It stands to reason that those who care enough to actively engage on Facebook are more likely to actively engage by voting.

Additional Details

Total “Popular Votes” counted through Facebook: 25,527,910

“Popular Votes” For Obama: 12,250,120

“Popular Votes” For Romney: 13,277,790

My Electoral College Results

“Votes” For Romney: 391

“Votes” For Obama: 147

Here is a comparison of my data when examined alongside three Electoral College Maps, CNN, Rasmussen and Karl Rove.

So how do these results stand up to “traditional” thinking?

  • In 2008, young voters went against election trends by showing up and voting (see: Young voters feel less engaged this year). The Facebook numbers I’ve pulled this time around seem to indicate a return to the traditional model with older voters driving the outcome of the election. While my projections show Obama carrying the popular vote in the 18-20, 21-24 and 25-34 age categories, they do not outnumber voters aged 35 and up who favor Romney.
  • The South votes Republican and the Northeast votes Democrat.
  • California is weird. Nothing new here, but perhaps the biggest anomaly in the numbers shows Romney carrying California because of a large number of 45-64 year olds “voting” for him. In addition to California, Delaware, Hawaii, Illinois, Maine, Maryland, Massachusetts, New Jersey, Rhode Island and Washington project for Romney.

The Process

I used the Facebook Ad Manager to gather data on the number of people on the Facebook network that have either Obama, or Romney listed in their Timeline using the Precise Targeting tool.

I segmented the search information by Keyword and broke results out by State and Age Range. To be as accurate as possible, I also used the Exact Age Match feature.

Precise Targeting looks at a person’s interests, activities, education, job titles, pages they like and groups they belong to, and Exact Age Match eliminates overlap between age groups. For example if you search 25-34 year olds without Exact Age Match, the targeting engine may include 24 and 35 year olds in the audience. I used these features with the goal of creating a comparison to how we would typically look at any other brand’s audience when taking a high level pass at their potential audience for a Facebook Ad buy.

I also used the hashtag feature in front of the candidate name to collect all possible interest categories, and get the broadest base of support.

The age ranges used were based on Voter Data as measured by the US Census, 18-20, 21-24, 25-34, 35-44, 45- 64 and over 65.

I then used an Electoral College calculator to allocate votes in the Electoral College based on the number of “active voters” for each candidate.


The data set is limited and only shows what it shows based on the Facebook definition of Precise Targeting and Exact Age matching. There are a lot of outside factors that influence the data. For instance, younger audiences may be engaging more actively on other platforms, like Tumblr or Twitter.

User activity by age range is a big takeaway. People in the older demographic groups are actively using Facebook when they encounter something that fits their needs, or interests. While a younger person may update their Status or Like a Page, an older person seems more likely to join a Group or update their Profile information to reflect their voting behaviors.

The 45 and up categories generate more targetable activity than the younger demos and are ripe for smart marketing.

For a closer look at the data pulled, I’ve made it available here as a spreadsheet:


JOSH JORDAN is the Founder and President of Make Me Social. Josh has spent the majority of his career blending his passion for people, technology and community development to create real relationships for brands and their message. Josh and his wife Jennifer live in St. Augustine, FL where they volunteer their time and energy to support the local arts and children’s charities and spend endless hours keeping their 19 month old son, George, entertained.



Filed under Josh Jordan, Social Media

9 responses to “Romney in a Landslide: Using Facebook Data for Predictive Analysis

  1. David Quigley

    I love it !!!

  2. Pat

    Another example of how traditional tools are being challanged by social media.

  3. I have to say, I completely disagree. Although, how you came about your opinions may be correct, California and Illinois most definitely would not go Republican. I would bet on that.

    • Hey Ryan, thanks for reading the post and responding. The data was a little surprising and we’re planning to compare actual numbers and see if we can determine anything else from it. The California anomaly, for example. While Obama is projected to carry the state using traditional models, are the Facebook numbers accurate in showing a large group of 44+ supporters for Romney?

      Stay tuned.

      – Josh

  4. Tim Mann

    Maryland for Romney? I don’t see that as a likely outcome.

    • Hi Tim, thanks for reading the post and responding.

      The data we pulled (again limited in scope) showed the following:

      18-20 21-24 25-34 35-44 45-64 65+ TOTAL
      18200 22340 37820 35700 57680 14760 186500

      18-20 21-24 25-34 35-44 45-64 65+ TOTAL
      12740 17480 34280 35180 69480 19400 188560

      The pattern is pretty typical of what you’d expect between the two candidates, with the younger demos showing better for Obama than Romney. But again we see in these numbers that the older demos are more “active” at least as it relates to the Facebook Ad manager data set.

      If you didn’t get a chance, you can click on this link to see how each state and age group pulled:

      It’ll be fun to see how things report post election and what conclusions we can draw.

      – Josh

  5. Pingback: Marketing Lessons from the Election « Redefining Outsourcing

  6. Josh, thank you for this very interesting and hopeful article. We will see how it goes next week. As a couple of other posters commented, it will be surprising to see if your predictions on several of the states play out. Like you said . . . let’s all stay tuned.

  7. Pingback: This Country Is Condemnced | Middle-Age Cranky

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 )

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s