Networks in Context

Networks in Their Surrounding Contexts

Dr. Bugra Caskurlu1

1 Computer Engineering Department

TOBB ETU University

February 4, 2021
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
However, each individual in a social network has a distinctive set of personal
characteristics. Similarities and compatibilities among two people’s characteristics can
strongly influence whether a link forms between them and hence affects the evolution of the
network.
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
However, each individual in a social network has a distinctive set of personal
characteristics. Similarities and compatibilities among two people’s characteristics can
strongly influence whether a link forms between them and hence affects the evolution of the
network.
The existing social connections of an individual affects the behaviors and activities she is
engaged with, and that in turn influence the links she form.
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
However, each individual in a social network has a distinctive set of personal
characteristics. Similarities and compatibilities among two people’s characteristics can
strongly influence whether a link forms between them and hence affects the evolution of the
network.
The existing social connections of an individual affects the behaviors and activities she is
engaged with, and that in turn influence the links she form.
These considerations suggest what we mean by a network’s surrounding contexts: factors
that exist outside the nodes and edges of a network, but which nonetheless affect how the
network’s structure evolves.
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
However, each individual in a social network has a distinctive set of personal
characteristics. Similarities and compatibilities among two people’s characteristics can
strongly influence whether a link forms between them and hence affects the evolution of the
network.
The existing social connections of an individual affects the behaviors and activities she is
engaged with, and that in turn influence the links she form.
These considerations suggest what we mean by a network’s surrounding contexts: factors
that exist outside the nodes and edges of a network, but which nonetheless affect how the
network’s structure evolves.
Today, we will consider how such effects operate, and what they imply about the structure
of social networks. We will see how the surrounding contexts affecting a network’s
formation can be viewed in network terms.
Networks in Context

Overview

What is next?
So far we studied some of the typical structures that characterize social networks, and
some of the typical processes that affect the formation of links in the network. Our
discussion of social networks was independent of the broader world in which it exists.
However, each individual in a social network has a distinctive set of personal
characteristics. Similarities and compatibilities among two people’s characteristics can
strongly influence whether a link forms between them and hence affects the evolution of the
network.
The existing social connections of an individual affects the behaviors and activities she is
engaged with, and that in turn influence the links she form.
These considerations suggest what we mean by a network’s surrounding contexts: factors
that exist outside the nodes and edges of a network, but which nonetheless affect how the
network’s structure evolves.
Today, we will consider how such effects operate, and what they imply about the structure
of social networks. We will see how the surrounding contexts affecting a network’s
formation can be viewed in network terms.
We will also see that several different processes of network formation can be described in a
common framework.
Networks in Context

Homophily

Definition and Historical Observations

Definition
A basic notion governing the structure of social networks is homophily: the principle that we
tend to be similar to our friends. Your friends are not a random sample of the underlying
population. They are mostly similar to you both in mutable and immutable characteristics.
Networks in Context

Homophily

Definition and Historical Observations

Definition
A basic notion governing the structure of social networks is homophily: the principle that we
tend to be similar to our friends. Your friends are not a random sample of the underlying
population. They are mostly similar to you both in mutable and immutable characteristics.
1 Immutable: Race, ethnicity, age, etc.
Networks in Context

Homophily

Definition and Historical Observations

Definition
A basic notion governing the structure of social networks is homophily: the principle that we
tend to be similar to our friends. Your friends are not a random sample of the underlying
population. They are mostly similar to you both in mutable and immutable characteristics.
1 Immutable: Race, ethnicity, age, etc.
2 Mutable: Places they live, their occupations, their levels of affluence, their interests, beliefs,
and opinions, etc.
Networks in Context

Homophily

Definition and Historical Observations

Definition
A basic notion governing the structure of social networks is homophily: the principle that we
tend to be similar to our friends. Your friends are not a random sample of the underlying
population. They are mostly similar to you both in mutable and immutable characteristics.
1 Immutable: Race, ethnicity, age, etc.
2 Mutable: Places they live, their occupations, their levels of affluence, their interests, beliefs,
and opinions, etc.

Figure: The underlying idea can be found in writings of Plato (”Similarity begets friendship”) and Aristotle
(”People love those who are like themselves”).
Networks in Context

Homophily

Observation in a Real Social Network

Figure: Homophily can divide a social network into densely connected, homogeneous parts that are weakly
connected to each other. In this social network from a town’s middle school and high school, two such
divisions in the network are apparent: one based on race (with students of different races drawn as differently
colored circles) and the other based on friendships in the middle and high schools, respectively.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
In the first case, a new link is added for reasons that are intrinsic to the network itself. In the
second case, the new link arises for a reason that makes sense only when we look at the
contextual factors beyond the network.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
In the first case, a new link is added for reasons that are intrinsic to the network itself. In the
second case, the new link arises for a reason that makes sense only when we look at the
contextual factors beyond the network.
There are strong interactions between intrinsic and contextual effects on the formation of
any single link; they both operate concurrently in the same network. For example, the
principle of triadic closure is supported by both intrinsic and contextual mechanisms.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
In the first case, a new link is added for reasons that are intrinsic to the network itself. In the
second case, the new link arises for a reason that makes sense only when we look at the
contextual factors beyond the network.
There are strong interactions between intrinsic and contextual effects on the formation of
any single link; they both operate concurrently in the same network. For example, the
principle of triadic closure is supported by both intrinsic and contextual mechanisms.
Motivating intrinsic mechanisms for triadic closure: When B and C have a common
friend A, then there are increased opportunities and sources of trust on which to base their
interactions, and A will also have incentives to facilitate their friendship.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
In the first case, a new link is added for reasons that are intrinsic to the network itself. In the
second case, the new link arises for a reason that makes sense only when we look at the
contextual factors beyond the network.
There are strong interactions between intrinsic and contextual effects on the formation of
any single link; they both operate concurrently in the same network. For example, the
principle of triadic closure is supported by both intrinsic and contextual mechanisms.
Motivating intrinsic mechanisms for triadic closure: When B and C have a common
friend A, then there are increased opportunities and sources of trust on which to base their
interactions, and A will also have incentives to facilitate their friendship.
Motivating the contextual mechanisms for the same: Since A-B and A-C friendships
already exist, due to the principle of homophily B and C are each likely to be similar to A,
and hence possibly similar to each other. Due to this similarity, there is an elevated chance
that B-C friendship will form; even if neither of them is aware that the other one knows A.
Networks in Context

Homophily

Contrast Between Reasons of Link Formation

Intrinsic and Contextual Factors

Compare a friendship that forms because two people are introduced through a common
friend with a friendship that forms because two people attend the same school.
In the first case, a new link is added for reasons that are intrinsic to the network itself. In the
second case, the new link arises for a reason that makes sense only when we look at the
contextual factors beyond the network.
There are strong interactions between intrinsic and contextual effects on the formation of
any single link; they both operate concurrently in the same network. For example, the
principle of triadic closure is supported by both intrinsic and contextual mechanisms.
Motivating intrinsic mechanisms for triadic closure: When B and C have a common
friend A, then there are increased opportunities and sources of trust on which to base their
interactions, and A will also have incentives to facilitate their friendship.
Motivating the contextual mechanisms for the same: Since A-B and A-C friendships
already exist, due to the principle of homophily B and C are each likely to be similar to A,
and hence possibly similar to each other. Due to this similarity, there is an elevated chance
that B-C friendship will form; even if neither of them is aware that the other one knows A.
The point isn’t that any one basis for triadic closure is the ”correct” one. As more
factors that drive the formation of links in a social network are taken into account, it
becomes difficult to attribute any individual link to a single factor. And ultimately,
one expects most links to arise from a combination of several factors.
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
It would mean that the proportion of male and female friends that a person has looks like
the background male/female distribution in the full population.
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
It would mean that the proportion of male and female friends that a person has looks like
the background male/female distribution in the full population.
No-homophily: If we were to randomly assign each node a gender according to the
gender balance in the real network, then the number of cross-gender edges should not
change significantly relative to what is seen in the real network. That is, in a network with
no homophily, friendships are being formed as though random mixing were occurring
across the given characteristic.
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
It would mean that the proportion of male and female friends that a person has looks like
the background male/female distribution in the full population.
No-homophily: If we were to randomly assign each node a gender according to the
gender balance in the real network, then the number of cross-gender edges should not
change significantly relative to what is seen in the real network. That is, in a network with
no homophily, friendships are being formed as though random mixing were occurring
across the given characteristic.

Test Metric
Suppose we have a network in which a fraction p of all individuals are male, and a fraction
q are female. Consider a given edge in the network.
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
It would mean that the proportion of male and female friends that a person has looks like
the background male/female distribution in the full population.
No-homophily: If we were to randomly assign each node a gender according to the
gender balance in the real network, then the number of cross-gender edges should not
change significantly relative to what is seen in the real network. That is, in a network with
no homophily, friendships are being formed as though random mixing were occurring
across the given characteristic.

Test Metric
Suppose we have a network in which a fraction p of all individuals are male, and a fraction
q are female. Consider a given edge in the network.
If we independently assign each node the gender male with probability p and the gender
female with probability q, then both ends of the edge will be male with probability p2 , and
both ends will be female with probability q 2 . The edge is a cross-gender edge with
probability 2pq.
Networks in Context

Homophily

Measuring Homophily

Motivation of the Test Metric

Question: What would it mean for a network not to exhibit homophily by gender?
It would mean that the proportion of male and female friends that a person has looks like
the background male/female distribution in the full population.
No-homophily: If we were to randomly assign each node a gender according to the
gender balance in the real network, then the number of cross-gender edges should not
change significantly relative to what is seen in the real network. That is, in a network with
no homophily, friendships are being formed as though random mixing were occurring
across the given characteristic.

Test Metric
Suppose we have a network in which a fraction p of all individuals are male, and a fraction
q are female. Consider a given edge in the network.
If we independently assign each node the gender male with probability p and the gender
female with probability q, then both ends of the edge will be male with probability p2 , and
both ends will be female with probability q 2 . The edge is a cross-gender edge with
probability 2pq.
Homophily Test: If the fraction of cross-gender edges is significantly less than 2pq, then
there is evidence for homophily.
Networks in Context

Homophily

Homophily Test Example

Networks in Context

Homophily

Homophily Test Example

Measuring Homophily on a Social Network Example

In this example p = 2/3 and q = 1/3, and 5 out of 18 edges in the graph are cross-gender
edges. The expected number of cross-gender edges is 18 × 2 × p × q = 8. This example
shows some evidence of homophily since the number of cross-gender edges is
”significantly less than” its expectation.
Networks in Context

Homophily

Homophily Test Example

Measuring Homophily on a Social Network Example

In this example p = 2/3 and q = 1/3, and 5 out of 18 edges in the graph are cross-gender
edges. The expected number of cross-gender edges is 18 × 2 × p × q = 8. This example
shows some evidence of homophily since the number of cross-gender edges is
”significantly less than” its expectation.
Note that the ratio of cross-gender edges in a random assignment of genders will deviate
somewhat from its expected value of 2pq; therefore, to perform the test in practice one
needs a working definition of ”significantly less than”. Standard measures of statistical
significance can be used for this purpose.
Networks in Context

Homophily

Inverse Homophily and Nonbinary Characteristics

Definition
It is possible for a network to have a fraction of cross-gender edges that is significantly more than
2pq, in which case, we say that the network exhibits inverse homophily. The network of
romantic relationships typically exhibit inverse homophily.
Networks in Context

Homophily

Inverse Homophily and Nonbinary Characteristics

Definition
It is possible for a network to have a fraction of cross-gender edges that is significantly more than
2pq, in which case, we say that the network exhibits inverse homophily. The network of
romantic relationships typically exhibit inverse homophily.

Testing Homophily for Nonbinary Characteristics

We can extend the homophily test to any underlying characteristic (race, ethnicity, age,
native language, political orientation, and so forth).
Networks in Context

Homophily

Inverse Homophily and Nonbinary Characteristics

Definition
It is possible for a network to have a fraction of cross-gender edges that is significantly more than
2pq, in which case, we say that the network exhibits inverse homophily. The network of
romantic relationships typically exhibit inverse homophily.

Testing Homophily for Nonbinary Characteristics

We can extend the homophily test to any underlying characteristic (race, ethnicity, age,
native language, political orientation, and so forth).
When the characteristic can only take two possible values, then we can draw a direct
analogy to the case of two genders and use the same formula, 2pq.
Networks in Context

Homophily

Inverse Homophily and Nonbinary Characteristics

Definition
It is possible for a network to have a fraction of cross-gender edges that is significantly more than
2pq, in which case, we say that the network exhibits inverse homophily. The network of
romantic relationships typically exhibit inverse homophily.

Testing Homophily for Nonbinary Characteristics

We can extend the homophily test to any underlying characteristic (race, ethnicity, age,
native language, political orientation, and so forth).
When the characteristic can only take two possible values, then we can draw a direct
analogy to the case of two genders and use the same formula, 2pq.
When the characteristic can take on more than two possible values, we say that an edge is
heterogeneous if it connects two nodes that are different according to the characteristic in
question.
Networks in Context

Homophily

Inverse Homophily and Nonbinary Characteristics

Definition
It is possible for a network to have a fraction of cross-gender edges that is significantly more than
2pq, in which case, we say that the network exhibits inverse homophily. The network of
romantic relationships typically exhibit inverse homophily.

Testing Homophily for Nonbinary Characteristics

We can extend the homophily test to any underlying characteristic (race, ethnicity, age,
native language, political orientation, and so forth).
When the characteristic can only take two possible values, then we can draw a direct
analogy to the case of two genders and use the same formula, 2pq.
When the characteristic can take on more than two possible values, we say that an edge is
heterogeneous if it connects two nodes that are different according to the characteristic in
question.
We then ask how the number of heterogeneous edges compares to what we’d see if we
were to randomly assign values for the characteristic to all nodes in the network using the
proportions from the real data as probabilities.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
1 Selection: The tendency of people to form friendships with others who are like them. It
may operate at several different scales and with different levels of intentionality.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
1 Selection: The tendency of people to form friendships with others who are like them. It
may operate at several different scales and with different levels of intentionality.
Active choice: When people choosing friends who are most similar from among a clearly
delineated pool of potential contacts in a small group.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
1 Selection: The tendency of people to form friendships with others who are like them. It
may operate at several different scales and with different levels of intentionality.
Active choice: When people choosing friends who are most similar from among a clearly
delineated pool of potential contacts in a small group.
Implicit selection: When people live in neighborhoods, attend schools, or work for companies
that are relatively homogeneous compared to the population at large, the social environment
already favors opportunities to form friendships with others like oneself.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
1 Selection: The tendency of people to form friendships with others who are like them. It
may operate at several different scales and with different levels of intentionality.
Active choice: When people choosing friends who are most similar from among a clearly
delineated pool of potential contacts in a small group.
Implicit selection: When people live in neighborhoods, attend schools, or work for companies
that are relatively homogeneous compared to the population at large, the social environment
already favors opportunities to form friendships with others like oneself.
We refer to all these effects cumulatively as selection.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Introducing the Mechanisms Underlying Homophily

Defining Selection and Social Influence

The two mechanisms attributed to the principle of homophily are as follows:
1 Selection: The tendency of people to form friendships with others who are like them. It
may operate at several different scales and with different levels of intentionality.
Active choice: When people choosing friends who are most similar from among a clearly
delineated pool of potential contacts in a small group.
Implicit selection: When people live in neighborhoods, attend schools, or work for companies
that are relatively homogeneous compared to the population at large, the social environment
already favors opportunities to form friendships with others like oneself.
We refer to all these effects cumulatively as selection.
2 Social Influence: The process of people modifying their behaviors to bring them more
closely into alignment with the behaviors of their friends. Social influence can be viewed as
the reverse of selection: with selection, the individual characteristics drive the formation of
links, while with social influence, the existing links in the network serve to shape people’s
mutable characteristics.
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
2 Mutable: The selection and social influence processes operate concurrently. The feedback
effects between people’s individual characteristics and their links in the social network are
significantly complex.
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
2 Mutable: The selection and social influence processes operate concurrently. The feedback
effects between people’s individual characteristics and their links in the social network are
significantly complex.

Revealing the Complexity: Example and Question

We all observe that pairs of adolescent friends often have similar outcomes in terms of
scholastic achievement and delinquent behavior such as drug use. Why is that so?
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
2 Mutable: The selection and social influence processes operate concurrently. The feedback
effects between people’s individual characteristics and their links in the social network are
significantly complex.

Revealing the Complexity: Example and Question

We all observe that pairs of adolescent friends often have similar outcomes in terms of
scholastic achievement and delinquent behavior such as drug use. Why is that so?
1 Selection: Teenagers seek out social circles composed of people like them.
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
2 Mutable: The selection and social influence processes operate concurrently. The feedback
effects between people’s individual characteristics and their links in the social network are
significantly complex.

Revealing the Complexity: Example and Question

We all observe that pairs of adolescent friends often have similar outcomes in terms of
scholastic achievement and delinquent behavior such as drug use. Why is that so?
1 Selection: Teenagers seek out social circles composed of people like them.
2 Social Infleunce: Peer pressure causes them to conform to behavioral patterns within their social
circles.
Networks in Context

Homophily
Mechanisms Underlying Homophily

The Interplay of Selection and Social Influence

The Distinction on Mutable and Immutable Characteristics

The mechanisms leading to homophily for immutable and mutable characteristics exhibits some
differences:
1 Immutable: The order of events is clear: a person’s attributes are determined at birth, and
they play a role in how this person’s connections are formed over the course of his or her
life.
2 Mutable: The selection and social influence processes operate concurrently. The feedback
effects between people’s individual characteristics and their links in the social network are
significantly complex.

Revealing the Complexity: Example and Question

We all observe that pairs of adolescent friends often have similar outcomes in terms of
scholastic achievement and delinquent behavior such as drug use. Why is that so?
1 Selection: Teenagers seek out social circles composed of people like them.
2 Social Infleunce: Peer pressure causes them to conform to behavioral patterns within their social
circles.
Question: How can we understand how these two effects interact, and whether one is
more strongly at work then the other?
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time. This approach makes it possible to
see the behavioral changes that occur after changes in an individual’s network connections, as
opposed to the changes to the network that occur after an individual changes his or her behavior.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time. This approach makes it possible to
see the behavioral changes that occur after changes in an individual’s network connections, as
opposed to the changes to the network that occur after an individual changes his or her behavior.

Significance of Understanding the Mechanism

Question: Is there any use of quantifying the effect of different forces leading to homophily
other than just an intellectual curiosity?
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time. This approach makes it possible to
see the behavioral changes that occur after changes in an individual’s network connections, as
opposed to the changes to the network that occur after an individual changes his or her behavior.

Significance of Understanding the Mechanism

Question: Is there any use of quantifying the effect of different forces leading to homophily
other than just an intellectual curiosity?
It allows is to reason about the effect of possible interventions one might attempt in the
system. What will be the broad effect of a program that targets ”certain” high school
students and influences them to stop using drugs?
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time. This approach makes it possible to
see the behavioral changes that occur after changes in an individual’s network connections, as
opposed to the changes to the network that occur after an individual changes his or her behavior.

Significance of Understanding the Mechanism

Question: Is there any use of quantifying the effect of different forces leading to homophily
other than just an intellectual curiosity?
It allows is to reason about the effect of possible interventions one might attempt in the
system. What will be the broad effect of a program that targets ”certain” high school
students and influences them to stop using drugs?
1 If the observed homophily is based on some amount of social influence, such a program could
have a broad impact across the social network by causing the friends of these targeted students to
stop using drugs as well.
Networks in Context

Homophily
Mechanisms Underlying Homophily

Methodology and the Results

Longitudinal study
The interplay of selection and social influence mechanisms on homophily can be understood by
using longitudinal studies of a social network, in which both the social connections and the
behaviors within a group are tracked over a period of time. This approach makes it possible to
see the behavioral changes that occur after changes in an individual’s network connections, as
opposed to the changes to the network that occur after an individual changes his or her behavior.

Significance of Understanding the Mechanism

Question: Is there any use of quantifying the effect of different forces leading to homophily
other than just an intellectual curiosity?
It allows is to reason about the effect of possible interventions one might attempt in the
system. What will be the broad effect of a program that targets ”certain” high school
students and influences them to stop using drugs?
1 If the observed homophily is based on some amount of social influence, such a program could
have a broad impact across the social network by causing the friends of these targeted students to
stop using drugs as well.
2 If the observed homophily is arising almost entirely from selection effects, then the program may
not reduce drug use beyond the students it directly targets: as these students stop using drugs,
they change their social circles and form new friendships with students who don’t use drugs, but
the drug-using behavior of other students is not strongly affected.
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Question: Why this clustering is present?
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Question: Why this clustering is present?
1 Is it because of selection effects, in which people are choosing to form friendships with others of
similar obesity status?
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Question: Why this clustering is present?
1 Is it because of selection effects, in which people are choosing to form friendships with others of
similar obesity status?
2 Is it because of the confounding effects of homophily according to other characteristics, in which
the network structure indicates existing patterns of similarity in other dimensions that correlate
with obesity status?
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Question: Why this clustering is present?
1 Is it because of selection effects, in which people are choosing to form friendships with others of
similar obesity status?
2 Is it because of the confounding effects of homophily according to other characteristics, in which
the network structure indicates existing patterns of similarity in other dimensions that correlate
with obesity status?
3 Or is it because changes in the obesity status of a person’s friends was exerting a (presumably
behavioral) influence that affected his or her future obesity status?
Networks in Context

Homophily
Mechanisms Underlying Homophily

A Longitudinal Study Example

Work of Christakis and Fowler

Using longitudinal data, obesity status and social network structure of roughly 12, 000
people is tracked over a 32-year period.
Finding: Obese and nonobese people clustered in the network in a fashion consistent with
homophily, according to the measure we defined: people tend to be more similar in obesity
status to their network neighbors than in a version of the same network where obesity
status is assigned randomly.
Question: Why this clustering is present?
1 Is it because of selection effects, in which people are choosing to form friendships with others of
similar obesity status?
2 Is it because of the confounding effects of homophily according to other characteristics, in which
the network structure indicates existing patterns of similarity in other dimensions that correlate
with obesity status?
3 Or is it because changes in the obesity status of a person’s friends was exerting a (presumably
behavioral) influence that affected his or her future obesity status?
Result: Even accounting for the effects of the first two types, there is significant evidence
for an effect of the third type: obesity is a health condition displaying a form of social
influence, with changes in your friends’ obesity status in turn having a subsequent effect on
you. This evidence suggests that obesity can spread through the underlying social network
via social influence.
Networks in Context

Affiliation

Putting Context into Network

Affiliation Networks
Link formation may be based on contextual factors existing ”outside” the network, such as
similarities in characteristics of the nodes and behaviors (and activities) that the nodes
engage in.
Networks in Context

Affiliation

Putting Context into Network

Affiliation Networks
Link formation may be based on contextual factors existing ”outside” the network, such as
similarities in characteristics of the nodes and behaviors (and activities) that the nodes
engage in.
In fact, it is possible to put these outside contexts into the network itself, by creating a larger
network that contains both people and contexts as nodes. Though we can represent any
context in this way in principle, we will focus on how to represent the set of activities in
which a person takes part, and how these activities affect the formation of links.
Networks in Context

Affiliation

Putting Context into Network

Affiliation Networks
Link formation may be based on contextual factors existing ”outside” the network, such as
similarities in characteristics of the nodes and behaviors (and activities) that the nodes
engage in.
In fact, it is possible to put these outside contexts into the network itself, by creating a larger
network that contains both people and contexts as nodes. Though we can represent any
context in this way in principle, we will focus on how to represent the set of activities in
which a person takes part, and how these activities affect the formation of links.
Activity (Foci) Examples: Being part of a particular company, organization, or neigborhood;
frequenting a particular place; or pursuing a particular hobby or interest; etc.
Networks in Context

Affiliation

Putting Context into Network

Affiliation Networks
Link formation may be based on contextual factors existing ”outside” the network, such as
similarities in characteristics of the nodes and behaviors (and activities) that the nodes
engage in.
In fact, it is possible to put these outside contexts into the network itself, by creating a larger
network that contains both people and contexts as nodes. Though we can represent any
context in this way in principle, we will focus on how to represent the set of activities in
which a person takes part, and how these activities affect the formation of links.
Activity (Foci) Examples: Being part of a particular company, organization, or neigborhood;
frequenting a particular place; or pursuing a particular hobby or interest; etc.

Figure: An affiliation network is a bipartite graph that shows which individuals are affiliated with which groups
or activities. Here, Anna participates in both of the social foci on the right, while Daniel participates in only one.
Networks in Context

Affiliation

A Popular Affiliation Network

Figure: One type of affiliation network that has been widely studied involves the memberships of people on
corporate boards of directors. A very small portion of this network (as of mid-2009) is shown here. The
structural pattern of memberships can reveal subtleties in the interactions among both the board members and
the companies. You may want to search for news about the resignation of Arthur Levinson form the board of
directors of Google.
Networks in Context

Affiliation

Social-Affiliation Networks

Definition
In a social-affiliation network (as in an affiliation network), we have a node for each person
and each foci. But we have two distinct kinds of edges.
Networks in Context

Affiliation

Social-Affiliation Networks

Definition
In a social-affiliation network (as in an affiliation network), we have a node for each person
and each foci. But we have two distinct kinds of edges.
1 Edges of the social network: Each connects two people and indicates friendship.
Networks in Context

Affiliation

Social-Affiliation Networks

Definition
In a social-affiliation network (as in an affiliation network), we have a node for each person
and each foci. But we have two distinct kinds of edges.
1 Edges of the social network: Each connects two people and indicates friendship.
2 Edges of the affiliation network: Each connects a person to a foci and indicates the participation of
the person in the foci.
Networks in Context

Affiliation

Social-Affiliation Networks

Definition
In a social-affiliation network (as in an affiliation network), we have a node for each person
and each foci. But we have two distinct kinds of edges.
1 Edges of the social network: Each connects two people and indicates friendship.
2 Edges of the affiliation network: Each connects a person to a foci and indicates the participation of
the person in the foci.
We can observe the coevolution of social networks and affiliation networks as
closure processes in the social-affiliation networks.
Networks in Context

Affiliation

Social-Affiliation Networks

Definition
In a social-affiliation network (as in an affiliation network), we have a node for each person
and each foci. But we have two distinct kinds of edges.
1 Edges of the social network: Each connects two people and indicates friendship.
2 Edges of the affiliation network: Each connects a person to a foci and indicates the participation of
the person in the foci.
We can observe the coevolution of social networks and affiliation networks as
closure processes in the social-affiliation networks.

Figure: A social-affiliation network shows both the friendships between people and their affiliation with different
social foci.
Networks in Context

Affiliation

Closure Processes in a Social-Affiliation Network

Figure: The three closure types corresponding to the closing of a triangle in a social-affiliation network.
Networks in Context

Affiliation

Closure Processes in a Social-Affiliation Network

Figure: The three closure types corresponding to the closing of a triangle in a social-affiliation network.

Closure Processes
Suppose B and C have a common neighbor A. Consider the formation of the B-C edge:
Networks in Context

Affiliation

Closure Processes in a Social-Affiliation Network

Figure: The three closure types corresponding to the closing of a triangle in a social-affiliation network.

Closure Processes
Suppose B and C have a common neighbor A. Consider the formation of the B-C edge:
1 Triadic closure: A, B, and C each represent a person.
Networks in Context

Affiliation

Closure Processes in a Social-Affiliation Network

Figure: The three closure types corresponding to the closing of a triangle in a social-affiliation network.

Closure Processes
Suppose B and C have a common neighbor A. Consider the formation of the B-C edge:
1 Triadic closure: A, B, and C each represent a person.
2 Focal closure: B and C are persons, but A is a focus. Aspect of the principle of selection.
Networks in Context

Affiliation

Closure Processes in a Social-Affiliation Network

Figure: The three closure types corresponding to the closing of a triangle in a social-affiliation network.

Closure Processes
Suppose B and C have a common neighbor A. Consider the formation of the B-C edge:
1 Triadic closure: A, B, and C each represent a person.
2 Focal closure: B and C are persons, but A is a focus. Aspect of the principle of selection.
3 Membership closure: If A and B are persons, but C is a focus. Effect of social influence.
Networks in Context

Affiliation

An Example for Closure Processes at Work

Figure: In a social-affiliation network containing both people and foci, edges can form under the effect of
several different kinds of closure processes: two people with a friend in common, two people with a focus in
common, or a person joining a focus that a friend is already involved in.
Networks in Context

Affiliation

An Example for Closure Processes at Work

Figure: In a social-affiliation network containing both people and foci, edges can form under the effect of
several different kinds of closure processes: two people with a friend in common, two people with a focus in
common, or a person joining a focus that a friend is already involved in.

Discussion of the Example

Oversimplifying the mechanisms at work, we can summarize the evolution of the social-affiliation
network on the left to the one on the right as follows:
Networks in Context

Affiliation

An Example for Closure Processes at Work

Figure: In a social-affiliation network containing both people and foci, edges can form under the effect of
several different kinds of closure processes: two people with a friend in common, two people with a focus in
common, or a person joining a focus that a friend is already involved in.

Discussion of the Example

Oversimplifying the mechanisms at work, we can summarize the evolution of the social-affiliation
network on the left to the one on the right as follows:
1 Triadic Closure: Bob introduces Anna to Claire.
2 Focal Closure: Karate introduces Anna to Daniel.
3 Membership Closure: Anna introduces Bob to Karate.
Networks in Context

Tracking Link Formation in Online Data

Tracking Link Formation

Challenges and Caveats

We identified three different mechanisms that lead to the formation of links in social
networks. These mechanisms are good examples of social phenomena that are clearly at
work in small-group settings but have traditionally been very hard to measure quantitatively.
Networks in Context

Tracking Link Formation in Online Data

Tracking Link Formation

Challenges and Caveats

We identified three different mechanisms that lead to the formation of links in social
networks. These mechanisms are good examples of social phenomena that are clearly at
work in small-group settings but have traditionally been very hard to measure quantitatively.
Our strategy will be to try tracking these mechanisms as they operate in large populations,
where an accumulation of many small effects can produce something observable in the
aggregate.
Networks in Context

Tracking Link Formation in Online Data

Tracking Link Formation

Challenges and Caveats

We identified three different mechanisms that lead to the formation of links in social
networks. These mechanisms are good examples of social phenomena that are clearly at
work in small-group settings but have traditionally been very hard to measure quantitatively.
Our strategy will be to try tracking these mechanisms as they operate in large populations,
where an accumulation of many small effects can produce something observable in the
aggregate.
Challenge: Most of the forces responsible for link formation go largely unrecorded in
everyday life. We need a large, clearly delineated group of people (and social foci) and to
accurately quantify the relative contributions that these different mechanisms make to the
formation of real network links.
Networks in Context

Tracking Link Formation in Online Data

Tracking Link Formation

Challenges and Caveats

We identified three different mechanisms that lead to the formation of links in social
networks. These mechanisms are good examples of social phenomena that are clearly at
work in small-group settings but have traditionally been very hard to measure quantitatively.
Our strategy will be to try tracking these mechanisms as they operate in large populations,
where an accumulation of many small effects can produce something observable in the
aggregate.
Challenge: Most of the forces responsible for link formation go largely unrecorded in
everyday life. We need a large, clearly delineated group of people (and social foci) and to
accurately quantify the relative contributions that these different mechanisms make to the
formation of real network links.
Caveat: It is not clear how much one can extrapolate from digital interactions to interactions
that are not computer mediated, or even from one computer-mediated setting to another.
Networks in Context

Tracking Link Formation in Online Data

Some Questions About Triadic Closure

Figure: Pairs of people can have more than one friend (or more than one focus) in common. How does this
increase the likelihood that an edge forms between them?
Networks in Context

Tracking Link Formation in Online Data

Some Questions About Triadic Closure

Figure: Pairs of people can have more than one friend (or more than one focus) in common. How does this
increase the likelihood that an edge forms between them?

Questions
How much more likely is a link to form between two people in a social network if they
already have a friend in common?
Networks in Context

Tracking Link Formation in Online Data

Some Questions About Triadic Closure

Figure: Pairs of people can have more than one friend (or more than one focus) in common. How does this
increase the likelihood that an edge forms between them?

Questions
How much more likely is a link to form between two people in a social network if they
already have a friend in common?
How much more likely is an edge to form between two people if they have multiple friends
in common? In the figure, Anna and Esther have two friends in common, while Claire and
Daniel only have one friend in common. How much more likely is the formation of a link in
the first of these two cases?
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
1 Take two snapshots of the network at different times.
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
1 Take two snapshots of the network at different times.
2 For each k, identify all pairs of nodes who have exactly k friends in common in the first
snapshot, but who are not directly connected by an edge.
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
1 Take two snapshots of the network at different times.
2 For each k, identify all pairs of nodes who have exactly k friends in common in the first
snapshot, but who are not directly connected by an edge.
3 Define T (k ) to be the fraction of these pairs that have formed an edge by the time of the
second snapshot. This is an empirical estimate for the probability that a link will form
between two people with k friends in common.
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
1 Take two snapshots of the network at different times.
2 For each k, identify all pairs of nodes who have exactly k friends in common in the first
snapshot, but who are not directly connected by an edge.
3 Define T (k ) to be the fraction of these pairs that have formed an edge by the time of the
second snapshot. This is an empirical estimate for the probability that a link will form
between two people with k friends in common.
4 Plot T (k ) as a function of k to illustrate the effect of common friends on the formation of
links.
Networks in Context

Tracking Link Formation in Online Data

Addressing the Questions About Triadic Closure

A Methodology
We can address these questions empirically by using network data as follows:
1 Take two snapshots of the network at different times.
2 For each k, identify all pairs of nodes who have exactly k friends in common in the first
snapshot, but who are not directly connected by an edge.
3 Define T (k ) to be the fraction of these pairs that have formed an edge by the time of the
second snapshot. This is an empirical estimate for the probability that a link will form
between two people with k friends in common.
4 Plot T (k ) as a function of k to illustrate the effect of common friends on the formation of
links.
Note that T (0) is the rate at which link formation happens when it does not close a triangle,
while the values of T (k ) for larger k determine the rate at which link formation happens when it
does close a triangle. Thus, the comparison between T (0) and these other values addresses
our questions about the power of triadic closure.
Networks in Context

Tracking Link Formation in Online Data

An Experiment to Address Our Questions

Kossinets and Watts Experiment

They used a ”who-talks-to-whom” type of data set: the full history of e-mail communication
among roughly 22, 000 undergraduate and graduate students over a one-year period at a
large U.S. university.
Networks in Context

Tracking Link Formation in Online Data

An Experiment to Address Our Questions

Kossinets and Watts Experiment

They used a ”who-talks-to-whom” type of data set: the full history of e-mail communication
among roughly 22, 000 undergraduate and graduate students over a one-year period at a
large U.S. university.
They constructed a network that evolved over time, joining two people by a link at a given
instant if they had exchanged e-mail in each direction at some point in the past 60 days.
Networks in Context

Tracking Link Formation in Online Data

An Experiment to Address Our Questions

Kossinets and Watts Experiment

They used a ”who-talks-to-whom” type of data set: the full history of e-mail communication
among roughly 22, 000 undergraduate and graduate students over a one-year period at a
large U.S. university.
They constructed a network that evolved over time, joining two people by a link at a given
instant if they had exchanged e-mail in each direction at some point in the past 60 days.
They then determined an ”average” version of T (k ) by taking multiple pairs of snapshots:
they built a curve for T (k ) on each pair of snapshots using the aforementioned procedure,
and then they averaged all the curves they obtained.
Networks in Context

Tracking Link Formation in Online Data

An Experiment to Address Our Questions

Kossinets and Watts Experiment

They used a ”who-talks-to-whom” type of data set: the full history of e-mail communication
among roughly 22, 000 undergraduate and graduate students over a one-year period at a
large U.S. university.
They constructed a network that evolved over time, joining two people by a link at a given
instant if they had exchanged e-mail in each direction at some point in the past 60 days.
They then determined an ”average” version of T (k ) by taking multiple pairs of snapshots:
they built a curve for T (k ) on each pair of snapshots using the aforementioned procedure,
and then they averaged all the curves they obtained.
The observations in each snapshot were one day apart, so their computation gives the
average probability that two people form a link per day as a function of the number of
common friends they have.
Networks in Context

Tracking Link Formation in Online Data

Outcome of the Experiment

Figure: Quantifying the effects of triadic closure in an e-mail data set.

Comparing with a Baseline Model

Suppose that, each common friend that two people have gives them an independent
probability p of forming a link each day.
Networks in Context

Tracking Link Formation in Online Data

Outcome of the Experiment

Figure: Quantifying the effects of triadic closure in an e-mail data set.

Comparing with a Baseline Model

Suppose that, each common friend that two people have gives them an independent
probability p of forming a link each day. If they have k friends in common, the probability
that they form a link on any given day is Tb (k ) = 1 − (1 − p)k .
Networks in Context

Tracking Link Formation in Online Data

Outcome of the Experiment

Figure: Quantifying the effects of triadic closure in an e-mail data set.

Comparing with a Baseline Model

Suppose that, each common friend that two people have gives them an independent
probability p of forming a link each day. If they have k friends in common, the probability
that they form a link on any given day is Tb (k ) = 1 − (1 − p)k . Given the small effect of the
first common friend in the data, we also show a comparison to the curve 1 − (1 − p)k −1 .
Networks in Context

Tracking Link Formation in Online Data

Outcome of the Experiment

Figure: Quantifying the effects of triadic closure in an e-mail data set.

Comparing with a Baseline Model

Suppose that, each common friend that two people have gives them an independent
probability p of forming a link each day. If they have k friends in common, the probability
that they form a link on any given day is Tb (k ) = 1 − (1 − p)k . Given the small effect of the
first common friend in the data, we also show a comparison to the curve 1 − (1 − p)k −1 .
Both the real curve and the baseline curve are close to linear, and hence qualitatively
similar; but the fact that the real curve turns upward while the baseline curve turns slightly
downward indicates that the assumption of independent effects from common friends is too
simple to be fully supported by the data.
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Kossinets and Watts supplemented their university e-mail data set with information about
the class schedules for each student. In this way, each class became a focus, and two
students shared a focus if they had taken a class together.
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Kossinets and Watts supplemented their university e-mail data set with information about
the class schedules for each student. In this way, each class became a focus, and two
students shared a focus if they had taken a class together.
They could then compute the probability of focal closure by direct analogy with their
computation for triadic closure, determining the probability of link formation per day as a
function of the number of shared foci.
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Kossinets and Watts supplemented their university e-mail data set with information about
the class schedules for each student. In this way, each class became a focus, and two
students shared a focus if they had taken a class together.
They could then compute the probability of focal closure by direct analogy with their
computation for triadic closure, determining the probability of link formation per day as a
function of the number of shared foci.
Results: A single shared class turns out to have roughly the same absolute effect on link
formation as a single shared friend, but after this point the curve for focal closure behaves
quite differently from the curve for triadic closure: it turns downward and appears to
approximately level off rather than turning slightly upward.
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Kossinets and Watts supplemented their university e-mail data set with information about
the class schedules for each student. In this way, each class became a focus, and two
students shared a focus if they had taken a class together.
They could then compute the probability of focal closure by direct analogy with their
computation for triadic closure, determining the probability of link formation per day as a
function of the number of shared foci.
Results: A single shared class turns out to have roughly the same absolute effect on link
formation as a single shared friend, but after this point the curve for focal closure behaves
quite differently from the curve for triadic closure: it turns downward and appears to
approximately level off rather than turning slightly upward.
Thus, subsequent shared classes after the first produce a ”diminishing returns” effect.
Networks in Context

Tracking Link Formation in Online Data

Similar Question for Focal Closure

Question and Experiment

Question: What is the probability that two people form a link as a function of the number of
foci they are jointly affiliated with?
Kossinets and Watts supplemented their university e-mail data set with information about
the class schedules for each student. In this way, each class became a focus, and two
students shared a focus if they had taken a class together.
They could then compute the probability of focal closure by direct analogy with their
computation for triadic closure, determining the probability of link formation per day as a
function of the number of shared foci.
Results: A single shared class turns out to have roughly the same absolute effect on link
formation as a single shared friend, but after this point the curve for focal closure behaves
quite differently from the curve for triadic closure: it turns downward and appears to
approximately level off rather than turning slightly upward.
Thus, subsequent shared classes after the first produce a ”diminishing returns” effect.
Comparing to the same kind of baseline, in which the probability of link formation with k
shared classes is 1 − (1 − p)k , we see that the real curve turns downward more
significantly than this independent model.
Networks in Context

Tracking Link Formation in Online Data

Plot of Results of Focal Closure Experiment

Figure: Quantifying the effects of focal closure in an e-mail data set. Again, the curve determined from the data
is shown as the solid black line, while the dotted curve provides a comparison to a simple baseline.
Networks in Context

Tracking Link Formation in Online Data

Membership Closure Experiment

Figure: Quantifying the effects of membership closure in a large online data set: The plot shows the probability
of joining a LiveJournal community as a function of the number of friends who are already members.

Experiment
The figure is based on the blogging site LiveJournal, where friendships are designated by
users in their profiles and where foci correspond to membership in user-defined
communities.
The plot shows the probability of joining a community as a function of the number of friends
who have already done so.
Networks in Context

Tracking Link Formation in Online Data

Membership Closure Experiment on Wikipedia Data

Figure: Probability of editing a Wikipedia article as a function of the number of friends who’ve already done so.

Experiment
The social-affiliation network contains a node for each Wikipedia editor who maintains a
user account and user talk page on the system.
An edge joins two such editors if one editor has written on the user talk page of the other.
Each Wikipedia article defines a focus and an editor is associated with the respective focus
if she has edited the article.
Thus, the plot shows the probability that a person edits a Wikipedia article as a function of
the number of prior editors of that article with whom she has communicated.
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Measuring Effects of Mechanisms Leading to Homophily

Defining the Experimental Setup

Question: How do similarities in behavior between two Wikipedia editors relate to their
pattern of social interaction over time?
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Measuring Effects of Mechanisms Leading to Homophily

Defining the Experimental Setup

Question: How do similarities in behavior between two Wikipedia editors relate to their
pattern of social interaction over time?
The social network consists of all Wikipedia editors who maintain talk pages, and there is
an edge connecting two editors if they have communicated, with one writing on the talk
page of the other.
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Measuring Effects of Mechanisms Leading to Homophily

Defining the Experimental Setup

Question: How do similarities in behavior between two Wikipedia editors relate to their
pattern of social interaction over time?
The social network consists of all Wikipedia editors who maintain talk pages, and there is
an edge connecting two editors if they have communicated, with one writing on the talk
page of the other.
An editor’s behavior corresponds to the set of articles she has edited. Let the similarity of
two editors A and B be the neighborhood overlap of them in the affiliation network, i.e., the
following ratio:

number of articles edited by both A and B

number of articles edited by at least one of A and B
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Measuring Effects of Mechanisms Leading to Homophily

Defining the Experimental Setup

Question: How do similarities in behavior between two Wikipedia editors relate to their
pattern of social interaction over time?
The social network consists of all Wikipedia editors who maintain talk pages, and there is
an edge connecting two editors if they have communicated, with one writing on the talk
page of the other.
An editor’s behavior corresponds to the set of articles she has edited. Let the similarity of
two editors A and B be the neighborhood overlap of them in the affiliation network, i.e., the
following ratio:

number of articles edited by both A and B

number of articles edited by at least one of A and B

Observation: Pairs of Wikipedia editors who have communicated are significantly more
similar in their behavior than pairs of Wikipedia editors who have not communicated, so we
have a case where homophily is clearly present.
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Measuring Effects of Mechanisms Leading to Homophily

Defining the Experimental Setup

Question: How do similarities in behavior between two Wikipedia editors relate to their
pattern of social interaction over time?
The social network consists of all Wikipedia editors who maintain talk pages, and there is
an edge connecting two editors if they have communicated, with one writing on the talk
page of the other.
An editor’s behavior corresponds to the set of articles she has edited. Let the similarity of
two editors A and B be the neighborhood overlap of them in the affiliation network, i.e., the
following ratio:

number of articles edited by both A and B

number of articles edited by at least one of A and B

Observation: Pairs of Wikipedia editors who have communicated are significantly more
similar in their behavior than pairs of Wikipedia editors who have not communicated, so we
have a case where homophily is clearly present.
Specific Question: Does the homophily arise because editors are forming connections
with those who have edited the same articles (selection), or is it because editors are led to
the articles of those they talk to (social influence)?
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Experiment

Details of the Experiment

For each pair of editors A and B who have ever communicated, record their similarity over
time, where ”time” in this case moves in discrete units, advancing by one ”tick” whenever
either A or B performs an action (editing an article or communicating with another editor).
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Experiment

Details of the Experiment

For each pair of editors A and B who have ever communicated, record their similarity over
time, where ”time” in this case moves in discrete units, advancing by one ”tick” whenever
either A or B performs an action (editing an article or communicating with another editor).
Declare time 0 for the pair A-B to be the point at which they first communicated. This
results in many curves showing similarity as a function of time (one for each pair of editors
who ever communicated).
Networks in Context

Tracking Link Formation in Online Data

Quantifying the Interplay Between Selection and Social Influence

Experiment

Details of the Experiment

For each pair of editors A and B who have ever communicated, record their similarity over
time, where ”time” in this case moves in discrete units, advancing by one ”tick” whenever
either A or B performs an action (editing an article or communicating with another editor).
Declare time 0 for the pair A-B to be the point at which they first communicated. This
results in many curves showing similarity as a function of time (one for each pair of editors
who ever communicated).
Averaging all these curves yields the plot below, which shows the average level of similarity
relative to the time of first interaction, over all pairs of editors who have ever interacted.
Networks in Context

A Spatial Model of Segregation

Spatial Segragation

Chicago Example
One of the most readily perceived effects of homophily is in the formation of ethnically and
racially homogeneous neighborhoods in cities such as the one in the figure.

Figure: The tendency of people to live in racially homogeneous neighborhoods produces spatial patterns of
segregation that are apparent both in everyday life and when superimposed on a map. Blocks drawn lighter in
the map have populations with the lowest percentages of African-Americans, while blocks drawn darker in the
map have populations with the highest percentages of African-Americans.
Networks in Context

A Spatial Model of Segregation

A Simple Model

Aim and Spatial Model

The Schelling Model is a famous model that shows how global patterns of spatial
segregation can arise from the effect of homophily operating at a local level, even when no
one explicitly wants a segregated outcome.
Networks in Context

A Spatial Model of Segregation

A Simple Model

Aim and Spatial Model

The Schelling Model is a famous model that shows how global patterns of spatial
segregation can arise from the effect of homophily operating at a local level, even when no
one explicitly wants a segregated outcome.
We assume a population agents, each of which is of type X or O that represent a binary
immutable characteristic.
Networks in Context

A Spatial Model of Segregation

A Simple Model

Aim and Spatial Model

The Schelling Model is a famous model that shows how global patterns of spatial
segregation can arise from the effect of homophily operating at a local level, even when no
one explicitly wants a segregated outcome.
We assume a population agents, each of which is of type X or O that represent a binary
immutable characteristic.
The agents reside in the cells of a grid. Some cells of the grid contain agents while others
are unpopulated. A cell’s neighbors are the cells that touch it, including diagonal contact.

Figure: Agents care about whether they have at least t neighbors of the same type.
Networks in Context

A Spatial Model of Segregation

Shelling Model

Details
The fundamental constraint driving the model is that each agent wants to have at least t
other agents of its own type as neighbors.
Networks in Context

A Spatial Model of Segregation

Shelling Model

Details
The fundamental constraint driving the model is that each agent wants to have at least t
other agents of its own type as neighbors.
An agent fails to have t neighbors of the same type is called unsatisfied with its current
location.
Networks in Context

A Spatial Model of Segregation

Shelling Model

Details
The fundamental constraint driving the model is that each agent wants to have at least t
other agents of its own type as neighbors.
An agent fails to have t neighbors of the same type is called unsatisfied with its current
location.

Dynamics
Agents move in a sequence of rounds: in each round, we consider the unsatisfied agents in
some order; for each one in turn, we move it to an unoccupied cell where it will be satisfied.
Networks in Context

A Spatial Model of Segregation

Shelling Model

Details
The fundamental constraint driving the model is that each agent wants to have at least t
other agents of its own type as neighbors.
An agent fails to have t neighbors of the same type is called unsatisfied with its current
location.

Dynamics
Agents move in a sequence of rounds: in each round, we consider the unsatisfied agents in
some order; for each one in turn, we move it to an unoccupied cell where it will be satisfied.
After this, the round of movement has come to an end, representing a fixed time period
during which unsatisfied agents have changed where they live. These new locations may
cause different agents to be unsatisfied, which leads to a new round of movement.
Networks in Context

A Spatial Model of Segregation

Shelling Model

Details
The fundamental constraint driving the model is that each agent wants to have at least t
other agents of its own type as neighbors.
An agent fails to have t neighbors of the same type is called unsatisfied with its current
location.

Dynamics
Agents move in a sequence of rounds: in each round, we consider the unsatisfied agents in
some order; for each one in turn, we move it to an unoccupied cell where it will be satisfied.
After this, the round of movement has come to an end, representing a fixed time period
during which unsatisfied agents have changed where they live. These new locations may
cause different agents to be unsatisfied, which leads to a new round of movement.
There are numerous variations in the specific details of how the movement of agents within
a round is handled but all of them produces similar outcomes.
Networks in Context

A Spatial Model of Segregation

Simulations with Threshold 3

Figure: Two runs of a simulation of the Schelling model with a threshold t = 3, on a 150 × 150 grid with 10, 000
agents of each type. In each image, there is a dot for each cell of the grid; black dots indicate cells that are
empty, and the red and blue dots corresponds to agents with different types.
Networks in Context

A Spatial Model of Segregation

Simulation with Threshold 4

Figure: Four intermediate points in a simulation of the Schelling model with a threshold t = 4, on a 150 × 150
grid with 10, 000 agents of each type. As the rounds of movement progress, large homogeneous regions on
the grid grow at the expense of smaller, narrower regions.