Introduction

Let’s start by reviewing the basics, like a definition, why SNA is important, and the history of
the practice. If you want a quick intro to this methodology, download our Social Network
Analysis Brief.

Definition of Social Network Analysis (SNA)

Social Network Analysis, or SNA, is a research method used to visualize and analyze
relationships and connections between entities or individuals within a network. Imagine
mapping the relationships between different departments in a corporation. The outcome
would be a vivid picture of how each department interacts with others, allowing us to see
communication patterns, influential entities, and bottlenecks

The Importance of SNA

SNA is a powerful tool. It allows us to explore the underlying structure of an organization or
network, identifying the formal and informal relationships that drive the formal processes
and outcomes. This insight can enable better communication, facilitate change
management, and inspire more efficient collaboration.

This methodology also helps demonstrate the impact of relationship-building and systems
change efforts by documenting the changes in the quality and quantity of relationships
before and after the initiative. The maps and visualizations produced by SNA are an
engaging way to share your progress and impact with stakeholders, donors, and the
community at large.

Brief Historical Overview of SNA

The concept of SNA emerged in the 1930s within the field of sociology. Its roots, however,
trace back to graph theory in mathematics. It was not until the advent of computers and
digital data in the 1980s and 1990s that SNA became widely used, revealing new insights
about organizational dynamics, community structures, and social phenomena.

While it originated as an academic research tool, it is increasingly used to inform real-world

practice. Today, it is used in a broad variety of industries, fields, and sectors, including
business, web development, public health, foundations and philanthropy,
telecommunications, law enforcement, academia, and systems change initiatives, to name a
few.

Fundamentals of SNA
SNA is a broad topic, but these are some of the essential terms, concepts, and theories you
need to know to understand how it works.

Nodes and Edges

In SNA, nodes represent individuals or entities while edges symbolize the relationships
between them. For example, in an inter-organizational network, nodes might be companies,
and edges could represent communication, collaboration, or competition.
Network Types
Different types of networks serve different purposes. ‘Ego Networks’ focus on one node and
its direct connections, revealing its immediate network. ‘Whole Networks’, on the other
hand, capture a broader picture, encompassing an entire organization or system. Open
networks are loosely connected, with many opportunities to build new connections, ideal for
innovation and idea generation – while closed networks are densely interconnected, better
for refining ideas amongst a group who all know each other.

Network Properties
Properties such as density (the proportion of potential connections that are actual
connections), diameter (the longest distance between two nodes), and centrality (the
importance of a node within the network) allow us to understand the network’s structure
and function. Metrics also can measure relationship quality across the network, like our
validated trust and value scores.

Dyadic and Triadic Relationships

Dyadic relationships involve two nodes, like a partnership between two companies. Triadic
relationships, involving three nodes, are more complex but can offer richer insights. For
instance, it might show how a third company influences the relationship between two others,
or which members of your network are the best at building new relationships between their
peers.

Homophily and Heterophily

Homophily refers to the tendency of similar nodes to connect, while heterophily is the
opposite. In a business context, we might see homophily between companies in the same
industry and heterophily when seeking diversity in a supply chain. Many networks aim to be
diverse but get stuck talking to the same, similar partners. These network concepts underly
many strategies promoting network innovation to avoid group-think among likeminded
partners.

Network Topologies
Lastly, the layout or pattern of a network, its topology, can reveal much about its function.
For instance, a centralized topology, where one node is connected to all others, may indicate
a hierarchical organization, while a decentralized topology suggests a more collaborative
and flexible environment. This is also referred to as the structure of the network. Read more.

Theoretical Background of SNA

Many different theories have developed to explain how certain network properties, like their
topology, centrality, or type, lead to different outcomes. Here are several key theories
relevant to SNA.

Strength of Weak Ties Theory

This theory postulates that weak ties or connections often provide more novel information
and resources compared to strong ties. These “weak” relationships, which may seem less
important, can serve as important bridges between different clusters within a network. Read
more.

Structural Hole Theory

This theory posits that individuals who span the structural holes, or gaps, in a network—
acting as a bridge between different groups—hold a strategic advantage. They can control
and manipulate information and resources flowing between the groups, making their
position more influential. Read more

Small World Network Theory

This theory emphasizes the interconnectedness of nodes within a network. It suggests that
most nodes can be reached from any other node through a relatively short path of
connections. This property leads to the famous phenomenon of “six degrees of separation,”
indicating efficient information transfer and connectivity in a network.
Barabási–Albert (Scale-Free Network) Model
This model suggests that networks evolve over time through the process of preferential
attachment, where new nodes are more likely to connect to already well-connected nodes.
This results in “scale-free” networks, where a few nodes (“hubs”) have many connections
while the majority of nodes have few.

Data Collection and Preparation

Every network mapping begins by collecting and preparing data before it can be analyzed.
This data varies widely, but at a basic level, they must include data on nodes (the entities in
the network) and data on edges (the lines between nodes representing a relationship or
connection). Additional data on the attributes of the nodes or edges add more levels of
analysis and insight but are not strictly necessary.

Primary Methods for Collecting SNA Data

This can be as simple as conducting interviews or surveys within an organization. The more
complex the network, the more difficult it is to collect good primary data: If you have more
than 5-10 partners, interviews and surveys are hard to conduct by hand.

Network survey tools like PARTNER collect relational data by asking respondents who they
are connected to, and then asking them about aspects of their relationships to provide trust,
value, and network structure scores. This is impossible to do using most survey software like
Google Forms without hours of cleaning by hand.

Response rates are an important consideration if using surveys for data collection. Unlike a
typical survey where a small sample is representative, a network survey requires a high
response rate – 80% and above are considered the gold standard.

In an inter-organizational context where surveys are impossible, or you cannot achieve a

valid response rate, one might gather data through business reports, contracts, or publicly
available data on partnerships and affiliations. For example, you could visit an organization’s
website to note who they list as a partner – and do the same for others – to generate a basic
SNA map.

Secondary Sources of SNA Data

Secondary sources include data that was already collected but can be used again, often to
complement your use of primary data you collect yourself. This might include academic
databases, industry reports, or social media data. It’s important to ensure the accuracy and
reliability of these sources.

You can also conduct interviews or focus groups with network members to add a qualitative
perspective to your results. These mixed-method SNA projects provide a great deal more
depth to their network maps through their conversations with numerous network
representatives to explore deeper themes and perspectives.

Ethical Considerations in Data Collection

When collecting data, it’s crucial to ensure privacy, obtain necessary permissions, and
anonymize data where necessary. Respecting these ethical boundaries is critical for
maintaining trust and integrity in your work.
Consider also how your SNA results will be used. For example, network analysis can help
assess how isolated an individual is to target them for interventions. Still, it could also be
abused by insurance companies to charge these individuals a higher rate (loneliness
increases your risk of death).

Lastly, consider ways to involve the communities with stake in your SNA using approaches
like community-based participatory research. Bring in representatives from target
populations to help co-design your initiative or innovation as partners, rather than patients
or research subjects.

Preparing Data for Analysis

Data needs to be formatted correctly for analysis, often as adjacency matrices or edgelists.
Depending on the size and complexity of your network, this can be a complex process but is
crucial for meaningful analysis.

If you are new to SNA, you can start by laying out your data in tables. For example, the table
below shows a relational data set for a set of partners within a public health coalition. The
first column shows the survey respondent (Partner 1), the second shows who they reported
as a partner, the third shows their reported level of trust, and the fourth their reported level
of collaboration intensity. This is just one of many ways to lay out and organize network
data.

Depending on which analysis tool you choose, a varying degree of data preparation and
cleaning will be required. Usually, free tools require the most work, while software with
subscriptions do a lot of it for you.

Partner 1 Partner 2 Trust (1-4) Level of Collaboration

Mayor’s Office Local Hospital 3 Coordination
Public Health
Primary Care Clinic 4 Cooperation
Dept.
Mayor’s Office Public Health Dept. 2 Awareness
Network Analysis Methods & Techniques
There are many ways to analyze a network or set of entities using SNA. Here are some of
basic and advanced techniques, along with info on network visualization – a major
component and common output of SNA projects.

Basic Technique: Network Centrality

One of the most common ways to analyze a network is to look at the centrality of various nodes
to identify key players, information hubs, and gatekeepers across the network. There are three
types of centrality, each corresponding to a different aspect of connectivity and centrality.
Degree, Betweenness, and Closeness Centrality are measures of a node’s importance.

Degree Centrality

Can be used to identify the most connected actors in the network. These actors are considered
“popular” or “active” and they often have a strong influence within the network due to their
numerous direct connections. In a coalition or network, these nodes could be the organizations or
individuals that are most active in participating or the most engaged in the network activities.
They may be the ‘go-to’ people for information or resources and have a significant impact on
shaping the group’s agenda.

Betweenness Centrality

A useful for identifying the “brokers” or “gatekeepers” in the network. These actors have a
unique position where they connect different parts of the network, facilitating or controlling the
flow of information between others. In a coalition context, these could be the organizations or
individuals who have influence over how information, resources, or support flow within the
network, by virtue of their position between other key actors. These actors could play crucial
roles in collaboration, negotiation, and conflict resolution within the network.

Closeness Centrality

A measure of how quickly a node can reach every other node in the network via the shortest
paths. In a coalition, these nodes can disseminate information or exert influence quickly due to
their close proximity to all other nodes. These ‘efficient connectors’ are beneficial for the rapid
spread of information, resources, or innovations across the network. They could play a vital role
during times of rapid change or when swift collective action is required.
Advanced Techniques: Clusters and Equivalence
Clustering Coefficients

The Clustering Coefficient provides insights into the “cliquishness” or local cohesion of the
network around specific nodes. In a coalition or inter-organizational network, a high
clustering coefficient may indicate that a node’s connections are also directly connected to
each other, forming tight-knit groups or sub-communities within the larger network. These
groups often share common interests or objectives, and they might collaborate or share
resources more intensively. Understanding these clusters can be crucial for coalition
management as it can highlight potential subgroups that may need to be engaged
differently, or that might possess different levels of influence or commitment to the
coalition’s overarching goals.

Structural Equivalence

Structural Equivalence is used to identify nodes that have similar patterns of connections,
even if they do not share a direct link. In a coalition context, structurally equivalent
organizations or individuals often occupy similar roles or positions within the network, and
thus may have similar interests, influence, or responsibilities. They may be competing or
collaborating entities within the same sectors or areas of work. Understanding structural
equivalence can provide insights into the dynamics of the network, such as potential
redundancies, competition, or opportunities for collaboration. It can also reveal how changes
in one part of the network may impact other, structurally equivalent parts of the network.

Visualizing Networks
Network visualization is a key tool in Social Network Analysis (SNA) that allows researchers
and stakeholders to see the ‘big picture’ of the network structure, as well as discern
patterns and details that may not be immediately evident from numerical data. Here are
some key aspects and benefits of network visualization in the context of a coalition or inter-
organizational network:

Overview of Network Structure: Visualizations provide a snapshot of the entire network

structure, including nodes (individuals or organizations) and edges (relationships or
interactions). This helps to comprehend the overall size, density, and complexity of the
network. Seeing these relationships mapped out can often make the network’s structure
more tangible and easier to understand.

Identification of Key Actors: Centrality measures can be represented visually, making it

easier to identify key actors or organizations within the network. High degree nodes,
gatekeepers, and efficient connectors will stand out visually, which can assist in identifying
who holds influence or power within the network.

Detecting Subgroups and Communities: Visualization can also highlight clusters or

subgroups within the network. These might be based on shared interests, common goals, or
frequent interaction. Understanding these subgroups is crucial for coalition management
and strategic planning, as different groups might have unique needs, concerns, or levels of
engagement.

Identifying Outliers and Peripheral Nodes: Network visualizations can also help in
identifying outliers or peripheral nodes – those who are less engaged or connected within
the network. These actors might represent opportunities for further engagement or potential
risks for network cohesion.
Highlighting Network Dynamics: Visualizations can be used to show changes in the
network over time, such as the formation or dissolution of ties, the entry or exit of nodes, or
changes in nodes’ centrality. These dynamics can provide valuable insights into the
evolution of the coalition or network and the impact of various interventions or events.

Software and Tools for SNA

SNA software helps you collect, clean, analyze, and visualize network data to simplify the
process of of analyzing social networks. Some tools are free with limited functionality and
support, while others require a subscription but are easier to use and come with support.
Here are some popular s tools used across many application

Introduction to Popular SNA Tools

Tools like UCINet, Gephi, and Pajek are popular for SNA. They offer a variety of functions for
analyzing and visualizing networks, accommodating users of varying skill levels. Here are
ten tools for use in different contexts and applications.

1. UCINet: A comprehensive software package for the analysis of social network data
as well as other 1-mode and 2-mode data.
2. NetDraw: A tool usually used in tandem with UCINet to visualize networks.
3. Gephi: An open-source network analysis and visualization software package written
in Java.
4. NodeXL: A free and open-source network analysis and visualization software
package for Microsoft Excel.
5. Kumu: A powerful visualization platform for mapping systems and better
understanding relationships.
6. Pajek: Software for analysis and visualization of large networks, it’s particularly good
for handling large network datasets.
7. SocNetV (Social Networks Visualizer): A user-friendly, free and open-source tool.
8. Cytoscape: A bioinformatics software platform for visualizing molecular interaction
networks.
9. Graph-tool: An efficient Python module for manipulation and statistical analysis of
graphs.
10. Polinode: Tools for network analysis, both for analyzing your own network data and
for collecting new network data.

Choosing the Right Tool for Your Analysis:

The right tool depends on your needs. For beginners, a user-friendly interface might be a
priority, while experienced analysts may prefer more advanced functions. The size and
complexity of your network, as well as your budget, are also important considerations.

PARTNER CPRM: A Community Partner Relationship

Management System for Network Mapping
 The problem with most
social network analysis tools is the lack of specialization. They require a lot of customization
and integration to complete specialized tasks and analyses – the kind that provides the most
useful insight and value. A host of new SNA tools and software are developing that
incorporate relationship mapping into their operation for a specific niche or need, reducing
the time spent cleaning data and greatly increasing its value.

For example, we created PARTNER CPRM, a Community Partner Relationship Management

System, to replace the CRMs used by most organizations to manage their relationships with
their network of strategic partners. Incorporating data collecting, analysis, and visualization
features alongside CRM tools like contact management and email tracking, the result is a
powerful and easy-to-use network mapping tool.

Learn more about PARTNER CPRM

SNA Case Studies

Looking for a real-world example of a social network analysis project? Here are three
examples from recent projects here at Visible Network Labs.

Case Study 1: Leveraging SNA for Program Evaluation

SNA is increasingly becoming a vital tool for program evaluation across various sectors
including public health, psychology, early childhood, education, and philanthropy. Its
potency is particularly pronounced in initiatives centered around network-building.

Take for instance the Networks for School Improvement Portfolio by the Gates Foundation.
The Foundation employed PARTNER, an SNA tool, to assess the growth and development of
their educator communities over time. The SNA revealed robust networks that offer valuable
benefits to members by fostering information exchange and relationship development. By
repeating the SNA process at different stages, they could verify their ongoing success and
evaluate the effectiveness of their actions and adjustments.

Read the Complete Case Study Here

Case Study 2: Empowering Coalition-building

In the realm of policy change, building a coalition of partners who share a common goal can
be pivotal in overturning the status quo. SNA serves as a strategic tool for developing a
coalition structure and optimizing pre-existing relationships among the members.

The Fix CRUS Coalition in Colorado, formulated in response to the closure of five major
peaks to public access, is a prime example of this. With the aim of strengthening state
liability protections for landowners, the coalition employed PARTNER to evaluate their
network and identify key players. Their future plans involve mapping connections to
important legislators as their bill progresses through the state legislature. Additionally, their
network maps and reports will prove instrumental in acquiring grants and funding.

Case Study 3: Boosting Employee Engagement

In the private sector, businesses are increasingly harnessing SNA to optimize their employee
networks, both formal and informal, with the goal of enhancing engagement, productivity,
and morale.

Consider the case of Acuity Insurance. In response to a transition to a Hybrid-model amid

the COVID-19 pandemic, the company started using PARTNER to gather network data from
their employees. Their aim was to maintain their organizational culture and keep employee
engagement intact despite the model change. Their ongoing SNA will reveal the level of
connectedness within their team, identify employees who are over-networked (and hence at
risk of burnout), and pinpoint those who are under-networked and could be missing crucial
information or opportunities.

Read More About the Project Here

Challenges and Future Directions in Network

Analysis
Like all fields and practices, social network analysis faces certain limitations. Practitioners
are constantly innovating to find better ways to conduct projects. Here are some barriers in
the field and current trends and predictions about the future of SNA.

The Limitations of SNA

SNA is a powerful tool, but it’s not without limitations. It can be time-consuming and
complex, particularly with larger networks. Response rates are important to ensure
accuracy, which makes data collection more difficult and time-consuming. SNA also requires
quality, validated data, and the interpretation of results can be subjective. Software that
helps to address these problems requires a significant investment, but the results are often
worth it.

Lastly, SNA is a skill that takes time and effort to learn. If you do not have someone in-house
with network analysis skills, you may need to hire someone to carry out the analysis or
spend time training an employee to build the capacity internally.

How to Read Sociograms / Sociomatrixes

The most basic way to approach a sociogram is as a data visualization. A data visualization consists of
the data, the analyst and his / her questions, the mapping of the data to various visual attributes, and
some level of interactivity. A surface reading of the visual will reveal some information, but applying
various statistical analysis and rendering tools to the data will reveal even more.

Micro and Macro-Level Analysis: Node-link diagrams may be understood to be evaluated both at the
micro and macro levels. The micro level involves particular ego nodes, so the individual point-of-view
(POV) at that egocentric level. At this level, one can look at the node trajectories. One can also look at
various features of individual nodes. The macro levels involve the nature of the social network (a
sociocentric view) its breadth and depth; its types of nodes (its social network composition); its various
connections / connectivity; its path lengths; the density of ties; its clustering; how it evolves; and what
moves through the social network. In between are subnetworks or sub-cliques that may be pulled out for
analysis. "Islands" and ego neighborhoods may also be focal points for research, depending on what it is
that the data analyst may want to understand. A social network may be broken out into"partitions" that
make it easier to analyze.

Layering (Complex) Information: Additional information may be layered or superimposed over a social
network. There are visualizations that enable multi-dimensional viewing of a social network. Each of the
layers of information have to be input and rendered for the various visualizations.

It helps to apply various statistical tools and algorithms to the data in order to output different
visualizations that may be revealing. A transposition of data may show the opposite of a particular factor,
to reveal further information. Or there can be data turned into binary information (like a dummy variable)
to draw out further information.

Underlying all sociograms are various statistical techniques used in research. Understanding these
techniques is critical in terms of using the software accurately and then representing the data analysis
results.

Node (Ego) Attributes: The attributes of nodes are critical as well. The attribute may be described as an
identity or role with related self-interests. Or a social network may actually be populated with real-world
personalities and remote psychological readings on each based on their past patterns of behaviors, public
statements, and public personas. The nature of nodes affects their choices and behaviors in a strategic
context. Further, nodes may play multiple overlapping roles in a social network. The visual depiction
should be relevant, and it should not be overwhelmed by complexity...but it also should not over-simplify
the reality. Further, without textual labels of the nodes and more nuanced measures and indicators of
relationships of the links, a diagram by itself will not have deeper analytical value.

The Criticality of Context / the Social Ecology: Further, a node-link diagram cannot be understood in
isolation. It has to be understood in the context of the field. Looking at a node-link diagram without the
background is not deeply informative. Said another way, social networks expressed as node-link
diagrams (or other diagrams) are not fully stand-alone. They are one channel of information that must be
used with other channels for actual meaning. In the same way that data must be triangulated with other
sources, a social network diagram offers some information that may be combined with other data--such
as geographic information systems (GIS) or spatialized data, survey or self-reported data, demographic
information, press reports, and published analyses.

The Representation of Entitles and Relationships: Node-link diagrams are really about entities and the
relationships between those entities. There is the assumption of dynamism, stochastic factors, and node-
level self-interests. The social context, though, may shape power realities.
Network Centrality: Spatiality sometimes matters in a sociogram, such as in some which may place
nodes in a core, semi-periphery, and periphery. Other times, spatiality is only a tool used for the
expression of certain nodes and links. There are tools within various social networking software
visualisation packages that enable clearer depictions (with less visual clutter). Other times, the users of
the software can manually move the nodes to locations where the relationships are more visible (and the
links follow automatically). Visual coherence is critical. Further, it helps to have an aesthetic sensibility
regarding the diagram.

The Need for Accurate Data Sets (on the Back End): It is not possible to truly reverse engineer a data
set and a data array from a sociogram. It helps to have the original data from which the sociogram was
created. In this light, it's critical to originate your own data sets and ensure that the information is high
quality; otherwise, it'll be much harder to analyze the information from that data.

Random Patterns: Not every pattern found is meaningful. Some apparent patterns may be due to
random error alone. (The social network visualization and the interpretation of that visualization has to
make sense with other known data, to a degree.)

A Human-Created Universe (of Sorts): One philosophical point at the heart of social network research
is that humans do reify certain realities. They co-create certain realities in the world. In this context,
perception and human decision-making matter.

Some Assumptions of Sociometry

Costs and Transactional Relationships: There are costs to create ties and linkages to others and to
maintain them. While it is costly to maintain relationships in terms of social capital / time / resources /
attention and other elements (and one has to assume a transactional element in all human relationships),
there are further costs and risks to breaking ties, too, particularly if one's acquaintances are linked to a
particular node that one is considering breaking off from. These ties entail "constraints." They limit some
freedom of action; they limit the ability to de-link (temporarily or permanently) from a network or a portion
of a network.

All social alliances are strategic. They are created for particular ends. Without shared interests, alliances
tend to break off. In game theory, this is referred to as the continuous Prisoner's Dilemma, in which a
player chooses constantly whether to cooperate with or defect from another. (With the prospect of a
continuing relationship, the optimal way is for both to cooperate.)

Social Network Types: Heterogeneous networks are more beneficial than homogeneous ones.
Heterogeneous ones include much more diversity and many more links, enabling many more
connections. However, networks are about those who are included and those who are excluded. There
are spoken and unspoken rules for joining and maintaining membership in a social network.
What Moves through Social Networks: All sorts of things move through social networks--information,
resources, habits, diseases, attitudes, culture, and other elements. Some are positive, and some are
negative. Understanding social networks enables ways to engage the larger world to troubleshoot issues
and to multiply positive effects and to cut off negative ones (to a degree). There is the understanding that
some rewiring of social networks may be possible.

Types of Sociograms

Sociograms (sociomatrixes) consist of entities and relationships depicted on a 2-D grid space / graph.

Different types of information are better aligned to be expressed in certain visual formats. Certain types of
data can only be expressed coherently in certain formats.

a dendrogram (a tree diagram, usually showing taxonomic relationships)

Social Network Data

In subject area: Computer Science
Social network data refers to information about individuals within a community that is critical for various
applications such as criminology, terrorism, and public health. It involves sensitive information that, when
shared across organizations, can help in integrating social networks for analysis and mining purposes.
Social network data are important for discovering knowledge about a community, which is
critical in criminology, terrorism, public health, and many other applications. At the same
time, there is a great deal of private information about individuals in a social network,
which makes it sensitive when social network data are shared across organizations.
Without sharing social network data, each organization may only have part of a large global
social network. For example, each local law enforcement unit may have a criminal social
network. Without integrating the social networks of multiple law enforcement units, each
unit may not be able to identify the relationship between suspects or groups precisely. In
this chapter, we review the literature on privacy-preserving data publishing and sharing.
We define the problems of social network integration, analysis, and mining. We define
the τ-tolerance privacy leakage to ensure that a specified tolerance of privacy leakage must
be satisfied. We develop the subgraph generalization technique for sharing insensitive
information, which can be integrated for social network analysis and mining of the global
social network. Our preliminary work has shown promising results of the proposed
technique but there are more avenues to explore to enhance its performance.

Defining Two-mode Networks

tnet » Two-mode Networks » Defining Two-mode Networks
 Network with two types of nodes
Networks are representations of systems in which the elements (or nodes) are connected by ties (Wasserman and Faust, 1994).
Most networks are defined as one-mode networks with one set of nodes that are similar to each other. However, several networks
are in fact two-mode networks (also known as affiliation or bipartite networks; Borgatti and Everett, 1997; Latapy et al., 2008). These
networks are a particular kind, with two different sets of nodes, and ties existing only between nodes belonging to different sets. A
distinction is often made between the two node sets based on which set is considered more responsible for tie creation (primary or
top node set) than the other (secondary or bottom node set).
One of the first two-mode datasets to be analysed was the Davis’ Southern Women dataset (Davis et al., 1941), which recorded the
attendance of a group of women (primary node set) to a series of events (secondary node set). A woman would be linked to an
event if she attended it. Another category of two-mode networks that has become popular in recent years is scientific collaboration
networks (Newman, 2001). The two sets of nodes are scientists and papers, and a scientist is linked to a paper if she or he is listed
as an author. As scientists generally decide whether or not they would like to work on a paper, they are often assumed to be the
primary nodes. However, it is not always obvious which node set is the primary one, and in these cases, the research question
guides the choice. For example, in the case of interlocking directorates where the two node sets are directors and corporate boards,
and ties represent affiliation of directors with boards, it is not clear whether directors or boards are the primary node set (e.g.,
Levine, 1979; Mizruchi, 1996; Seierstad and Opsahl, 2011). This is likely to be due to tie formation being a mutual process where
the directors must (1) be invited to join the board, and (2) accept the invitation.
Analyse as one-mode networks?
Two-mode networks are rarely analysed without transforming them. This is because most network measures are solely defined for
one-mode networks, and only a few of them have been redefined for two-mode networks (Borgatti and Everett, 1997; Latapy et al.,
2008). Transforming a two-mode network to a one-mode network is often done using a method known as projection. This method
operates by selecting one of the two node sets (often the primary node set) and linking nodes from that set if they were connected to
at least one common node in the other set. Although the two-mode structure is discarded in this process, it is possible to define tie
weights based on it. Specifically, the tie weights are often defined as the number of common nodes. This method was extended by
Newman (2001) who argued that tie weights among authors in scientific collaboration networks should be discounted if the authors
collaborated on papers with many others. For more information, see the page on projection.
The projection of two-mode networks creates a number of issues. First, each tie in a prototypical one-mode network is assumed to
be created separately; however, this is not the case in projected two-mode networks. For example, while a standard phone call
creates a communication tie from one person to another, a director forms ties with all the other directors on a board when she or he
joins that board. This has direct implications for frameworks that utilize random networks to detect a baseline level (e.g., Opsahl et
al., 2008) and when comparing measures observed in a network with those found in corresponding random networks. This is due to
the fact that ties in classical random networks are assumed to be independent of each other (Erdos and Rényi, 1959). Although this
is neither the case in prototypical one-mode nor projected two-mode networks, the random networks are less comparable to
projected two-mode networks than to prototypical one-mode networks as multiple ties can be created due to a single event in these
networks. Second, depending on the degree distribution of the non-projected node set, a projected two-mode network tends to have
more and larger fully-connected cliques than prototypical one-mode networks (Wasserman and Faust, 1994). These are produced
when three or more nodes are connected to a common node in the two-mode network (e.g., all the directors on a single board are
connected and form a fully connected clique). This feature impacts a number of network measures, especially those based on
triangles including the structural holes measures (Burt, 1992, 2005) and the clustering coefficients (for a review, see Opsahl and
Panzarasa, 2009). To exemplify the cliques, and the many triangles, produced when projecting a two-mode network, the figure
below shows the main component of the interpersonal network among Norwegian directors.
The network structure among directors (circles) who form part of the largest group of interconnected directors. Two directors are connected if they are
members of the same board. The solid circles refer to women, whereas the hollow circles refer to men (Seierstad and Opsahl, 2011).