Skip to main content

Hey presidential candidates, welcome to the age of social media

Polarization, in politics, is the process by which public opinion becomes very divided and strays toward the two extremes, according to .  The country today is polarized.  Arguably more polarized than ever.  Evidence of this can be seen in many sources, one of which is visible at this link.  The graph in the previous image shows an accurate representation of how polarized the country is.  Other sources that emphasize how polarized the country is, with respect to the past, can be seen in the work of Keith Poole and Howard Rosenthal, two political scientists.  They plotted the evolution of the conservative versus liberal opinions in Congress amongst the two leading parties, available at this link.  All of this evidence shows how polarized the country actually is today; democrats are moving towards being ever more liberal and republicans moving towards being ever more conservative.

Now, when you look at it in terms of a network this is where it really becomes interesting.  You can represent a consensus among issues as a positive connection and a dissension on major issues as a negative connection.  Then, if you assume the graph of people in the country connected, at least in theory, you can start to draw out the graph.  One would assume the result is something like the graph showed on the first day of class by Adamic and Glance, 2005, see below, where political bloggers were separated into two separate groups; conservative and liberals.  The graph shows the sentiment of political bloggers (conservative as red, liberal as blue) before the 2004 elections.

This model could be extended to show the political sentiment of the country, where conservatives and liberals are separated by their dissenting viewpoints.  A similar graph to the one above would most likely be the result, assuming again that the graph is complete.  The phenomenon displayed would be analogous to the complete balanced graph where there are two groups of positive (similar viewpoints) separated by all negatives (dissimilar viewpoints).

However, one thing that isn’t captured in the representation above or the representation of the two distinct groups how dynamic political sentiment in the country as a whole is.  Even if the country seems to be very polarized, the sentiment of the country, at least based on social media data, mainly from Twitter, seems to indicate that the country swings back and forth between “feeling” liberal and “feeling” conservative very quickly.  The Twitter political index available at, seems to show this very well.  Over the course of the past five or so months the trends between support for Obama and Romney vary greatly, with one day Obama leading by 50 points and then next Romney up by 20 points, and so on.

Now, this brings up the interesting point of how credible are these Twitter numbers.  The country seems to switch viewpoints over night but the graphs don’t seem to change very much and the country seems to still remain polarized.  Is it because the Twitter numbers (and number from other social media outlets) are based on outspoken members of each respective community (democrats and republicans) and depending on what happened on a particular day members of one community are more outspoken than the other?  The Twitter index claims to analyze the 400 million tweets sent everyday and build up the index based on the semantics and sentiments expressed in those tweets about either of the candidates.  Are we to trust these calculations?  How are the numbers created beyond that vague description?

Each candidate spends a significant amount of time and money on social media, most recently with Obama doing a Q&A on reddit and Romney buying a Twitter trending topic (something that defeats the whole purpose of a trending topic).  Is it worth it?  It certainly doesn’t seem like it’s worth the effort and money both candidates put into it, since the polarized graph seems to stay relatively the same.

Now I’m not a government major so I’m sure there are tons of other factors that play into how successful a candidate is.  Additionally, the graphs shown above and the Twitter data are from slightly different times, but data does show that the country is as polarized as ever today as well.  But, purely from a network analysis point of view it seems as though all this hype about social media predicting the country’s political sentiment should be taken with a hefty grain of salt.



Hare, C. (Photographer). (2011). Political polarization in the u.s. house. [Web Photo]. Retrieved from

Twitter. (2012). The twitter political index. Retrieved from

voteview. (Photographer). (2011). 112th (2011-2012) congress. [Web Photo]. Retrieved from



One Response to “ Hey presidential candidates, welcome to the age of social media ”

Leave a Reply

Blogging Calendar

September 2012
« Aug   Oct »