Recent articles have emerged that criticize the importance of social media and the Internet in politics. This particular article claims that social media (broadly defined as the Internet) failed to predict the Iowa caucus outcome and is therefore irrelevant in the world of politics.
Is Social Media Revelant in Politics?
Graph 1.1 illustrates the voter age participation in the 2008 presidential election along with the age demographics of popular social networks.
Graph 1.1 Voter demographics in the 2008 presidential election
As displayed above, one can see the percentage of individuals in each age group that votes. One may also notice the correlation between each age group and the level of social media activity for that age group.
Is the Internet Relevant in Politics?
In their article, VentureBeat took their argument even further claiming that the internet was irrelevant. The internet, to begin with, is an organic, churning collection of this world’s activity delivered in a digital format. Although social media is an exponentially growing portion of the internet, social media encompasses just a portion of internet activity as a whole. Nearly 80 percent of the adult population uses the internet for everything from online shopping and entertainment, to research and academic pursuits. As such, when attempting to determine the buzz political candidates’ campaigns are creating by looking at internet activity, it’s helpful to collect a broader sample than merely data from social media outlets.
Let’s take a look at one of many internet traffic metrics we like to use: Wikipedia article views. We have found this internet tool to be a relatively accurate reflection of what’s occurring in the analog world. Here’s why: When a candidate is performing well and garnering new support, this creates buzz; people are likely to hear about such a candidate, and this moves many to attempt to learn more about the candidate using search engines. One of the top search engine results for each presidential candidate is her or his Wikipedia article. In Graph 1.2, we have gathered daily Wikipedia article hits for each candidate during the week leading up to the Iowa Caucus. Notice how the three candidates who saw the most Wikipedia article hits in the days prior to the Caucus were subsequently among the top three finishers in Iowa as shown in Graph 1.3.
Graph 1.2 – 2012 Republican Presidential Candidates Wikipedia article views: December 28 through January 3.
Graph 1.3 – 2012 Iowa Caucus Results
Another internet source that predict political performance is Intrade. Intrade is an online trading exchange website. Intrade users speculate the outcome of non-sporting events including the 2012 presidential election, and place cash on the table to back up those bets. Let’s take a look at Intrade predictions prior to the Iowa Caucus.
Graph 1.4 – Intrade Predictions on Iowa Caucus Results
Notice how Intrade predicted Mitt Romney to win the Iowa caucus three days prior to the event. Also notable is the fact that despite Rick Santorum only losing by a few votes on live television, Intrade traders still predicted Romney having significant higher probability of winning. Santorum’s price didn’t even exceed $30 during the displayed period of time. It is interesting that so many people were confident that Romney would win despite the race being so close. Perhaps individuals betting on Intrade were privy to knowledge most don’t have access to. Below is Romney’s Intrade price during the caucus.
Graph 1.5 – Mitt Romney’s Intrade Price
Correlation and Causality
Next time you’re at the beach, be careful not to purchase ice cream. According to a recent study, whenever ice cream sales rise, so do shark attacks! Well, actually, you’re probably fine purchasing that scope of your favorite desert. The above example displays the difference between correlation and causality. Are ice cream sales really causing sharks to be more aggressive, attacking more humans? No. What actually is happening is two independent events are converging when the summer causes a surge in people who want a cool treat, and a cool place to swim.
To appropriately measure causality, it’s necessary to analyze multiple variables. In the case of the shark attacks and the ice cream sales, one may look at sharks’ mating patterns, feeding grounds, etc. This is where PoliticIt’s It Score comes in. The It Score evaluates many variables in an attempt to provide a clearer picture of how well each candidate is performing on a national scale. It measures Internet activity including search engine activity and social media movement. Here’s a look at each candidate’s It Score prior to the Iowa Caucus.
Graph 1.6 – It Score vs. National Poll Results | December 19, 2011
Notice how it closely follows the national poll results. Since this point and up until the caucus, we began to see shifts in the position of each candidate on the It Score Scale. This change has reflected the performance of each candidate over time. Here are the It Scores of the current candidates following the Iowa Republican Caucus, posted January 9:
- Mitt Romney | 29
- Newt Gingrich | 20
- Rick Santorum | 15
- Ron Paul | 12
- Rick Perry | 4
Now that we have seen social media predictions during the recent weeks in the 2012 presidential race — specifically during the days leading up to the Iowa Caucus — let’s see what mainstream media predicted.
Graph 1.7 – Number of articles published by media outlets from Jan. 1 through Jan. 3
Media typically gives candidates that are perceived to be performing the best the most coverage. As seen above, Romney and Gingrich received the most media coverage while Rick Santorum — the man who basically tied for first place — received the least.
An individual by the name of Niall Cook was quoted in the VentureBeat article with the following statement: “Mainstream media is more a polished mirror of real life, and social media is still distorted and fragmented, like a fairground hall of mirrors.” Is there a way to get a more polished mirror of real life than what mainstream media provides?
Even with the best intentions, mainstream media makes up only a small percentage of the U.S. Is there a way to get a larger sample size? How is is possible to collect a sample from each voting person in population. It’s clearly a challenging task prior to an actual election. By measuring activity of voters on social media and through other web activity, we think we’ve found a way to get a gimps of what America is thinking using big data. That’s what the It Score is all about. You’ve seen how it leads national polls in Graph 1.6, it’s up to you to determine its significance.
Now that you have seen the data, you need to ask yourself if the internet plays a significant role in politics. Can mainstream media deliver a more accurate view of reality than social media and the internet as a whole? You decide.