Thursday, May 24, 2012

There Will Be Graphs; Part 1 of a Series

By Evan D. Robertson, Project Associate.

I’m about six feet tall. I have blue eyes. My hair: blonde. While these descriptions likely read as a personal ad, these data also share one very important thing in common. They all describe attributes particular to myself. Attribute data defines characteristics that belong to objects or, in this case, people. It is the type of data with which we are most familiar and serves as the cornerstone of our modern statistical society. Most surveys, interviews, and other data gathering techniques are essentially collecting attribute data in one form or another. For instance, the United States Census Bureau is engaged in the vital task of collecting and disseminating individual attribute data aggregated to some specified level of geography. Household income, educational attainment, age, and race/ethnicity are useful types of attribute data we see with regularity. In a growing world in which we have begun to realize the value of social networks, however, it is important to identify another class of data which, while used extensively in social science research over the last fifty years, has not reached such predominance in society’s consciousness.

Relational data defines the connections, ties, and attachments which relate one individual (or object, as you’ll see in my forthcoming post) to another. Without at least two agents, relational data doesn’t exist. Relational data hasn’t gained as widespread recognition for a variety of reasons but the most central: it is highly qualitative and difficult to gather. I have blue eyes; this is fairly easily observed and classified. But, how does one describe my connection with my friends? My coworkers? Close? How close? Do we measure our closeness by the number of emails exchanged? Phone call length? How much time we spend in the break room? There is, for now, a qualitative component of relational data that social scientists are still working to codify and develop a unified language to describe our social relations. Describing these relationships, however, is becoming an imperative in our increasingly networked world.

Social network analysis (SNA) is a tool that analyzes relational data. Social network analysis attempts to investigate the connections between individuals in a larger social web through the use of graphing techniques. SNA can be used in a variety of settings, depending, of course, on the needs of the user. For instance, if you are engaged in economic development marketing, then SNA would be a fairly useful tool to study your Twitter or other social media campaigns to identify gaps or groups which you feel that your organization isn’t as well connected to as it should be. In terms of analyzing an organization, SNA can help to identify potential barriers to information flow or general communication shenanigans within the economic development organization. And for those data geeks, so near and dear to my heart, social network analysis gives an outlet to visualize connections between various objects (yep, it doesn’t have to be people, simply objects that have interrelation…tune in on June 12th).

I’ll be the first to admit that there are a few short comings of social network analysis as an analytic tool. First off, depending on the size of the social network, I am not entirely convinced that social network analysis will show you something that you don’t already intuitively know. You can probably take a guess at which organizations your social media campaign isn’t penetrating. The data is also difficult to gather. While you can obtain information on your followers via and (see my colleagues past blog on these tools for more detail), it is more difficult to obtain the broader social structure you are attempting to influence. Matching your Twitter follower information to your broader social network, say the entire business community in your county, and understanding the quality of these connections is prohibitive from a time and cost perspective. The power of social network analysis is that it creates visually appealing, intuitive graphs. It is one thing to instinctively know the workings of your social network. It is another thing entirely too actually be able to see it as well as manipulate it (not used in the Orwellian sense of the word). Social network analysis gives the ability to organize data, to think about the connections between things, and to reorganize the system given new thoughts about their interrelations. It is an expedient tool to step away from our social world and gain a bird’s eye perspective.

So, I’m at 735 words which I've been informed is approaching the socially acceptable limit for blog length. On my next post, I’ll use social network analysis to order and reorder national occupational employment data published by the Bureau of Labor Statistics and see if we cannot glean any useful information. My next few blog posts are intended to be experimental, so if something doesn't exactly work out: you have my humble apologies. As promised, there will be graphs.