From my point of view, this week is all about me, me, ME! From your point of view, this week is all about YOU. We’re both right: ego networks are literally egocentric creations, starting from a single node looking out at the networked world. In lecture and reading, we’ll begin by considering ego alone, then coax the egoist into charting ties she or he may not see. Although ego networks are not complete networks, they are perhaps more accurate representations of the way the social world looks to us; as we go about living our lives, our range of vision is limited, not global.
Before you start to review this lecture, read Chapter 5 of Christina Prell’s text Social Network Analysis: History, Theory and Methodology. Also be sure to look over the Hansen et al. text on NodeXL, pp. 154-158. You have not read about the website Twitter before and are not expected to know about Twitter now (we’ll get to that later on in the semester). Regardless, the authors’ discussion of different levels of ego networks on pages 154-158 is a good way to nail down the concept.
Our lecture subjects this week are:
- Review: Ego Networks
- A Note on Density
- Why Study Ego Networks?
- Choice, Constraint and Homophily
- Ego Networks Online
Ego Networks: The Basics
A process generating an ego network, as you’ve learned in Prell Chapter 5, begins with a single individual at the center, named “ego.” and asks that individual to identify his, her (or its, in the case of a network of non-human entities) network contacts, named “alters.”
Ego networks spread out in levels. In a 1.0 level ego network, the process stops as described above, with a maximum network distance of 1 in the network. A sample 1.0 level ego network is shown below, spreading outward from a central ego (the black node) to a series of alters (gray nodes):
A “1.5 level ego network” is called that because its collection procedure stands in between a network distance of 1 and a network distance of 2. In a 1.5 level ego network we find out:
a) what alters ego is connected to,
b) and which of ego’s alters are connected to one another,
c) but we don’t ask if ego’s alters are connected to anybody else unless those people are already one of ego’s alters.
To clarify the somewhat convoluted meaning involved in part c), I’ve drawn a 1.5 level ego network below, working off of the 1.0 level ego network drawn above but now including new ties between alters. As before, ego is the black node in the middle.
In a 2.0 level ego network, we include all nodes up to a network distance of 2 from ego. This means that we find out and report on:
a) what alters ego is connected to,
b) and which of ego’s alters are connected to one another,
c) and now we do ask if ego’s alters are connected to anybody else, even if those people are not themselves ego’s alters.
Below is a 2.0 level ego network building off of the ego networks shown above. Again, ego is the black node and ego’s alters are shown in gray. But now some people who are not ego’s own alters (but are only the alters of alters) are included in the network, colored white:
The process of moving out from a 1.0 level to a 2.0 level ego network is one form of a more general process in network analysis called snowball sampling.
Have you ever tried to make a snowman in the the winter? You start with a tiny bit of snow and roll it. Every time you roll the snowball, more snow sticks to it. More and more snow sticks to the snowball every time you roll it because the snowball being rolled is bigger. Before you know it, your snowball is so big that you can hardly move it!
This is the idea behind snowball sampling:
Step 1. Select with a handful of individuals (or one, in the case of ego networks). This is your core.
Step 2. Ask those core individuals to identify what other individuals they’re tied to. This generates a set of individuals at a network distance of 1 from the core.
Step 3. Repeat Step 2, one step outward. Ask those individuals at network distance 1 to identify what other individuals they’re tied to. This generates a set of individuals at a network distance of 2 from the core.
Step 4. Continue to repeat until you have obtained a snowball sample of the size you’re interested in.
If you cut a rolled snowball in half, you’d notice a series of concentric rings, one for each time you rolled the ball in snow. Similarly, in a snowball sample you can imagine a set of concentric rings extending outward from each ego, getting larger and larger with each additional extension of network distance:
A Note on Density in Complete Networks vs. Density in Ego Networks
Let’s elaborate a bit on the difference between the density of a complete network, which you learned in Week 3, and the density of an ego network, which you read about last week. Recall that the density of a complete network refers to the number of actual ties in a network divided by the number of possible ties in the network. In Prell’s words, the density of an complete network “refers to the percentage of all possible ties in the [complete] network that are actually present.”
There’s a subtle but important contrast between complete network density and ego network density. As Christina Prell notes on p. 120 of her text, “The density of an ego network refers to the percentage of all possible ties in the ego network that are actually present, excluding the ego” [and ties to the ego]. That italicized difference is important.
The density of an ego-network actually has a few more wrinkles to resolve — wrinkles involving the level at which it’s observed. Consider a level 1.0 ego network: no ties between alters are measured, and so ego network density is impossible to calculate! Then consider a level 2.0 ego network: in this case, ties between alters are measured, but not ties between alters’ alters. Again, complete ego network density is impossible to calculate. Only for a level 1.5 ego network, in which alters’ alters aren’t included and all ties between alters are measured, can ego network density be accurately calculated.
In addition to this wrinkle, I should clarify what Prell means when she says “the density of an ego network refers to the percentage…”. If the density of an ego network is calculated as (number of actual ties)/(number of possible ties), as Prell indicates on p. 121, then the resulting number will always be between zero (when there are no actual ties) and one (when every possible tie is also an actual tie). In order to obtain a “percentage” (to which Prell refers on p. 120), you’d need to multiply by 100.
Why Study Ego Networks?
In the sections above, we’ve reviewed some basic techniques for gathering information on ego networks and describing their characteristics. But why would one want to bother doing so in the first place? What advantage is there in studying ego networks when we could just study entire, complete networks? This video considers those questions:
Choice, Constraint and Homophily
In Chapter 5, Christina Prell refers to homophily as “the social situation of actors preferring to have social relations with others who are similar to themselves.” This is not an entirely appropriate definition, as Prell’s ensuing commentary suggests:
“There are two main arguments regarding how homophily takes place. The first argument states that organizational settings determine that ties will form among similar actors…. The second argument states that actors are drawn to form ties with others who are similar to themselves.”
This distinction is crucial to understand and therefore is worthy of some further discussion. The homophily principle declares that the probability of a social tie between two individuals becomes smaller the more different those two individuals are from one another in some socially-salient characteristic such as age, education, income, occupation, religion, racial/ethnic group or geographic location. Put slightly more succinctly, homophily is the extent to which social ties between similar people occur more often than chance alone would predict. The qualifier of more often than chance is vital, because chance alone has the potential to explain the emergence of a great deal of similarity in our relations.
Miller McPherson and Lynn Smith-Lovin (1987) identify two kinds of explanations for the existence of homophily in social relations. The theory of choice homophily supposes that people associate disproportionately with similar others because human beings prefer (for rational or irrational reasons) similar others. If choice homophily holds true, then if people enter into a room of similar and dissimilar strangers, they will seek out those who are like themselves and avoid those who are unlike them.
Induced homophily, on the other hand, can occur because we form social ties with the people we encounter, and because the people we encounter resemble us. No psychological preference is required for induced homophily to occur. Induced homophily theorists begin with the observation that the usual consideration of the “chance” of forming a tie with someone of a particular social category is the proportion of people of that category in the general population. The 2010 U.S. Census found that people in the self-reported racial category “Black or African American” make up 13.6% of the United States population. Is it then fair to say by chance alone any person should have a friendship network that is 13.6% Black or African American?
The problem with this simple formulation of the “chance” of a tie occurring is that we rarely enter a room, a church, a neighborhood, a community, a school, a business, or any other environment that is not in some way more homogeneous than the general population. In other words, the places and circumstances in which we meet people are already segregated (by law, through bias, or through unintentional processes that sort people). The same 2010 Census data, for instance, show that Maine is the whitest state in the nation, with a population that is 94.6% white. Is the makeup of people encountered by someone living in Maine liable to be 13.6% Black or African American? Within a geographic region, there are other factors that make cross-category contact still less likely. A recent study by Christopher Schietle and Kevin Dougherty (2010), for instance, finds support for Martin Luther King, Jr.’s contention that “11:00 on Sunday morning” is “the most segregated hour of Christian America”: in the typical American congregation, the most-common racial group accounts for 91 percent of the congregation’s members.
If our states are segregated, and if within states our communities are segregated, and if within communities our neighborhoods and workplaces and churches and schools and clubs and sports teams are segregated, and if the chance of our encountering and forming a social tie with someone is more likely in these contexts as argued by Ronald Breiger and Scott Feld, then by chance alone within these contexts we are likely to form and maintain ties with people highly similar to ourselves. Proponents of induced homophily argue that the extent of the role played by choice in the matter is highly overrated because people are inclined to attribute their behavior to choices and because people fail to appreciate how segregated their organizational environments are.
In a 1987 classic study of 10 communities in Nebraska, J. Miller McPherson and Lynn Smith-Lovin found that the average difference between pairs of people in the population of these communities was greater than the average difference between pairs of people in organizations within these communities — but the average difference between people in actual reported social ties was smaller still:
In case you’re feeling a little confused, the following video talks about the difference between choice homophily and induced homophily in another way… using the example of bears — teddy bears, to be exact:
To try to consolidate the ideas of heterophily and homophily, choice and induced constraint, I’d like you to apply them to your own life. In the Padlet below, I’d like you to post a reflection on an ego network that has surrounded you in the past or that is surrounding you now. What kind of relation was that ego network based upon? Pick some demographic characteristic (like age, sex, race, or religion) and describe the share of your alters in that ego network who were or are similar to you. Is that ego network characterized by heterophily or homophily? How did the ties in the ego network form? As a result of choice or induced by circumstances? (As always, if you can’t manage to get the Padlet to work for you in your browser, please send your response to me in an e-mail message.)
Ego Networks Online
An increasingly popular use of ego network analysis is to help people understand their place in online social networks. The social networking site Linkedin helps people maintain a professional profile online and connect with influential people in their field — if you don’t have a Linkedin account already, you should get one soon so that you’re ready to present yourself to employers as you prepare for graduation. Linkedin maintains a mapping service that allows users to create Level 1.5 ego networks of their professional connections to others. Here’s a copy of mine:
As the ego network map shows, most of my LinkedIn contacts are academics within the University of Maine system where I work (coded blue), and those academics are strongly connected to one another, forming a single connected component. A much smaller set of Linkedin connections are made to social media professionals (coded orange) and people who make a living describing the meaning of communications in the business sector (coded green).
You may or may not have a Linkedin account now, but if you are a typical American you almost certainly have a Facebook account. Did you know that there is an app in Facebook you can use to generate a Level 1.5 ego network with you at the center and your Facebook friends around you? The app is called NameGenWeb, and this video shows you how to use it:
Breiger, Ronald. 1974. “The Duality of Persons and Groups.” Social Forces 53(2): 181-190.
Feld, Scott L. 1981. “The Focused Organization of Social Ties.” American Journal of Sociology 86(5): 1015-1035.
McPherson, J. Miller and Lynn Smith-Lovin. 1987. “Homophily in voluntary organizations: Status distance and the composition of face-to-face groups.” American Sociological Review 52(3): 370-379.
Scheitle, Christopher and Kevin Dougherty. 2010. “Race, Diversity, and Membership Duration in Religious Congregations.” Sociological Inquiry 80(3): 405–423.