Social Network Analysis: Twitter Analysis of the White House
It has been often noted that the White House's Twitter feed has a high number of followers, but falls far behind many active entertainers. What those entertainers lack, however, is the political and historical significance of the White House and the Presidency.
While a graph of those following the White House on Twitter might not be very useful or interesting due to its popularity and the seeming ease with which one should be able to explain such a phenomenon, it might be very interesting to know who the White House is following on Twitter. In short, who does the President think is important? The graph and related metrics assembled here are based on the network of Twitter users that are followed by the White House.
(Have you ever used Deep Zoom Composer? Here is the link to zoom.it -- a tool to let you "zoom" into the image so that you can see the detail. )
Unsurprisingly, the White House has the highest degree of the 100 users polled in assembling this data. Behind the White House in the metric "degree" come several government agencies, groups, and cabinet members. This indicates that of the universe of users the White House follows, the most connected are all government entities. Further, this fact demonstrates on another level that Twitter is being used as a communication tool between government agencies, secretaries of various departments, and the White House. The metric "betweenness centrality" roughly follows the metric "degree" in terms of which vertices or nodes display which levels of connectedness, showing vertices to be as crucial to the network as they are connected.
In smaller graphs with fewer vertices, directional relationships are often important in seeing the flow of information as it moves from individual to individual across a network. In a data set with as many connections as the one at hand, such an understanding of the flow of information is difficult to get. However, the use of directional relationships as represented visually can serve a similar function on a larger scale. In this case, "edge opacity" was set to a moderately transparent level, and so denser, overlapping edges appear more opaque to the reader. Arrows indicating direction are so numerous that they are hard to distinguish, but their presence and layered opacity combine to present "halos" of color around vertices with a high "in-degree" metric and so indicate how much attention is paid to them by members of the network. This can indicate importance independent of that distinguishable by connectedness. For example, user "recoverydotgov" has a low "betweenness centrality" as compared to "irsnews", but has three times the "in-degree". This indicates that while "recoverydotgov" (a website that tracks government spending of tax dollars) does not play as major a bridging role between other users as does "irsnews" (the Internal Revenue Service), it certainly gets more attention!
Performing "clustering" in this particular case has yet to prove very useful in generating insights like those above, but may upon further investigation reveal special relationships between government entities that appear in this universe of Twitter users followed by the White House. While something may not necessarily jump out from the data by itself, the clustering algorithm indicates some set of relationships that bear further investigation. Unlike the party affiliations easily identified in Senate voting metrics, these are not so clearly relatable. This is, perhaps, the most important lesson one can learn regarding the power of metrics and analysis - patterns in large data sets can give users clues as to what undiscovered driving forces might be at play in apparently random or unorganized networks. In short, data can highlight future questions just as readily as it can highlight current answers.
What do you think? Start a discussion by leaving your comments! If you like what you have read, consider subscribing to our blog. It is easy! Just fill in your email address in the "Subscribe via email" box in the top right column.
--------
Did you like using Zoom.it to investigate the graphs? This tool is based upon Deep Zoom Composer. Deep Zoom Composer is one of the software productivity tools used in LIS 7410. If you would like more information on LIS 7410, click on the following button.
(function(){ var hsjs = document.createElement("script"); hsjs.type = "text/javascript"; hsjs.async = true; hsjs.src = "//cta-service.cms.hubspot.com/cta-service/loader.js?placement_guid=7e71ae97-9a81-492e-8662-0a2fa394d504"; (document.getElementsByTagName("head")[0]||document.getElementsByTagName("body")[0]).appendChild(hsjs); setTimeout(function() {document.getElementById("hs-cta-7e71ae97-9a81-492e-8662-0a2fa394d504").style.visibility="hidden"}, 1); setTimeout(function() {document.getElementById("hs-cta-7e71ae97-9a81-492e-8662-0a2fa394d504").style.visibility="visible"}, 2000); })();
Did you like this analysis of the White House Twitter Social Network Graph? Social Network Analysis is one of the tools used in LIS 7491. If you would like more information on LIS 7491, click on the following button.
(function(){ var hsjs = document.createElement("script"); hsjs.type = "text/javascript"; hsjs.async = true; hsjs.src = "//cta-service.cms.hubspot.com/cta-service/loader.js?placement_guid=73513bac-4fb9-440f-8247-d39c5a0f6a13"; (document.getElementsByTagName("head")[0]||document.getElementsByTagName("body")[0]).appendChild(hsjs); setTimeout(function() {document.getElementById("hs-cta-73513bac-4fb9-440f-8247-d39c5a0f6a13").style.visibility="hidden"}, 1); setTimeout(function() {document.getElementById("hs-cta-73513bac-4fb9-440f-8247-d39c5a0f6a13").style.visibility="visible"}, 2000); })();
--------
[1] Zoom.it
[2] "NodeXL is a free, open-source template for Microsoft® Excel® 2007 and 2010 that makes it easy to explore network graphs".