To help you map and analyze your network’s connectivity as it evolves, we have developed examples of how we have used social network analysis (SNA) in practice, along with a survey template you can adapt to capture data for your network map.
Network connectivity reflects how well relationships are forming across a network. As the effectiveness of any network depends on the strength of the connections between participants, network connectivity is a critical means of assessing progress — especially in the early days of a network’s formation.
Social network analysis (SNA), also known as network mapping, is the most effective method currently available for visualizing and evaluating a network’s connectivity. SNA offers an empirical way to represent the patterns of connection and disconnection among participants at a given moment in time. This provides important clues about where clusters are forming, how information is flowing, and where to weave relationships.
SNA can be used to capture the degree to which participants have formed a relationship with one another, are communicating (sharing information, ideas, or data), are coordinating (connecting their efforts closely but maintaining separate resources and responsibilities), and collaborating (working in partnership, sharing resources, and making shared decisions). We aim to gather this information before the first network convening to get baseline data, and then again after major convenings, or at regular intervals thereafter (i.e., every 6 months) to assess change over time.
SNA results can support a network’s development in several ways, including:
Each of these applications are shown in the maps below. Following the examples are suggestions for getting started on conducting your own social network analysis.
SNA helps to assess the progress of a network during its formative stages. In particular, it can help leaders evaluate the effectiveness of network convenings in enabling participants to establish and deepen their relationships with one another. Seeing this type of progress, particularly before more tangible outcomes have been reached, can help to show progress and inspire further engagement among funders and participants alike.
For example, the pair of maps below shows the increasing connectivity of a 23-person network over 12 months. Over this time, the network’s density—the percentage of possible connections across the network that are actual connections—increased from 15% to 64%. This data was especially valuable in demonstrating the effectiveness of the network’s activities to its funders.
Social network analyses can also identify the key influencers in a network—the people others commonly turn to for information, guidance, or support. The map below shows a 325-person network featuring leaders from across a region. Each person in the network was asked to identify the people who have most significantly influenced their work. The size of the node correlates with the number of times a given person was listed as an influencer. The people listed most often are indicated with the largest nodes—these are the key influencers in the network. Meanwhile, the people on the periphery of the network are the least engaged, and may need some additional attention or support to find ways to plug in.
Network maps can reveal vulnerabilities and areas that need attention. The following map, which illustrates the connections between staff members of a large school district and its network of parent volunteers, revealed that the school district’s connections with an important minority community, Hmong Americans, was limited to single staff member. While this staff member created an essential bridge with the community, it was a particularly “narrow bridge”. If this person were to leave their job or become unavailable, the district would be at risk of losing many critical relationships.
Upon seeing the map, the school district mobilized to increase the diversity of connections between staff members and Hmong parents—to create what’s known as a “wide bridge.” Creating redundancies in networks through wide bridges allows resources and information to continue flowing even when certain individuals are unresponsive or unavailable.
SNA can also reveal how people are clustered together, and where there are opportunities to strategically weave those clusters together. For example, the map below shows a single network that is divided into two primary clusters, with a few isolated nodes. Visualizing the connections (or lack thereof) across this network presents a clear need for weaving—strategically connecting people with each other—in order to bridge clusters and engage disconnected individuals. In this network, deliberate weaving would go a long way towards creating a more connected network, allowing information and resources to flow more effectively.
The first step to conducting a social network analysis is collecting the data you need. This includes two components:
Individual data is gathered to incorporate identifying and contact information like name, title, affiliation, email address. Individual data may also include information relevant to an asset map, including areas of expertise, functional skills, connections with other stakeholder groups, and geographic focus. (Click here to learn how to develop an asset map).
Connectivity data is gathered to indicate how people are connected to one another. This information can take multiple forms, including how strong of a relationship people have, how much they are communicating, how often people are considered a source of guidance or support, and the degree to which people are collaborating with one another.
Asset maps consist of only individual data, and a social network analysis consists of only connectivity data. When combined together, however, we can create what’s known as a “social system map.”
Through an SNA survey, each network participant indicates their degree of connection with each other participant using a particular connectivity scale. You can offer multiple different connectivity scales (for example, one for relationships, and one for collaborations) but to keep things clear and simple, we like to offer a single connectivity scale whenever possible. Connectivity scales can be quite different depending on the network and what information is most relevant to you. Following is an example of a scale that combines multiple types of connection.
0. No connection: I have not met this person.
1. Connected: I have met this person but we are not in communication with one another.
2. Communicating: We regularly communicate with one another (for example, by talking to each other directly or by sharing information, ideas, or data)
3. Coordinating: We coordinate our individual efforts to advance shared goals (for example, by leveraging each other’s resources and expertise, providing peer assistance, or attending each other’s meetings) - but we maintain separate resources and responsibilities.
4. Collaborating: We work in partnership to advance shared goals (for example, by combining resources, making shared decisions, or creating something together).
Both individual and connectivity data can be collected in various ways, including physical (analog) or online surveys. The most effective way to capture this information is with sumApp, a survey tool specifically built for network mapping. The beauty of sumApp is that it integrates perfectly with two of our favorite mapping softwares, Kumu and Graph Commons, so that the data collected in sumApp is automatically updated in the resulting network maps. sumApp will also provide each participant with a unique survey link, which they can return to at any time to update their information. This makes it very easy to conduct an SNA at regular intervals as your network grows.
sumApp has multiple pricing tiers. The free version—Tier 1—might suffice for a basic SNA. If you want to add in additional individual information for an asset or social system map, however, you’ll need to upgrade to Tier 2. And if you’d like to capture multiple connectivity scales, you’ll need to sign up for Tier 3.
Once you’ve decided what type of information you’d like to collect, you can add the names and email addresses of all participants into sumApp, and then build your surveys within the software. Once you’re all set, you can copy the JSON link that sumApp provides and connect it to your map in Kumu or Graph Commons.
While Graph Commons is a great software, we use Kumu most often due to its visual aesthetic and responsive support team. Kumu is free for public projects, or you can host private password-protected projects with a monthly subscription.
Get started by creating a new project and choosing from one of Kumu’s templates. For an SNA featuring less than 200 participants, use the “Stakeholder template,” and for a map of more than 200 participants it’s probably a good idea to choose the “Big data template”.
Once your project is created, connect your JSON link from sumApp by clicking on the big green plus icon, selecting “Import”, and pasting the JSON link into the box titled “Link map to remote JSON”. Now all the data from your sumApp survey will be automatically integrated into this map!
The remaining steps are more art than science. Use Kumu’s editor in the right hand sidebar by clicking the settings icon. You can use their basic editor to color the nodes (representing survey respondents) and the connections to represent the information you collected. For example, start by making each level of your connection scale a different color. You can also use Kumu’s advanced editor if you’re comfortable using some code.
Follow this link for an example Kumu social system map, shown below, that you can help inform your own map design. You can even see and borrow the underlying code in the advanced editor!
The final step in conducting a social network analysis is to run SNA metrics to identify the most and least connected people (measured by degree or indegree), the top bridgers (measured by betweenness centrality), and overall network connectivity (measured by density).
In Kumu, you can run SNA metrics by clicking the cube icon in the lower-right. You can also add an SNA metrics dashboard that stays updated whenever you change the map’s view. You will see this dashboard at the bottom of the example map linked above.
Each of the most useful SNA metrics are described below.
Degree is the total number of connections a person has in the network. Those with a high degree might have more influence or access to information than others in the network. Indegree measures the total number of incoming connections, whereas outdegree measures the total number of outgoing connections.
Indegree is particularly useful when not everyone has completed a survey - measuring only incoming connections helps account for the fact that those who have completed the survey may have many outgoing connections while those who have not yet completed the survey will have none.
Density is the percentage of possible connections in the network that are actual connections. If the network density is 60%, that means that 60% of all the possible connections across the network are actual connections. Another way to think of this is the probability that any two people picked at random are actually connected with one another. This is a very helpful statistic that tells us how well-connected, or how dense, the network is. As the network evolves we hope this measure goes up.
Average Path Length is the average degree of separation between any two people in the network. Less connected networks have higher average path lengths, and more connected networks have lower average average path lengths. As a network evolves, we hope average path length decreases, as people become more connected with each other and have an easier time becoming connected with anyone else in the network. As you add new people into the network (or, at least, it into the survey), average path length will go back up.
Diameter is the highest degree of separation between any two people in the network. If the diameter of the network is five, that means it will take no more than five connections to get from any one person to any other person in the network. As the network evolves we hope that this number goes down.
Reciprocity is the likelihood that an expressed connection from one person to another is reciprocated back. When there is low reciprocity this tells us that people have very different understandings of their connections with one another, and there might be a problem or lack of clarity in how the questions were asked in the survey.
Betweenness Centrality is the extent to which a person is a connector between two other people who aren’t already connected with each other. People with high betweenness are bridgers in the network, as they provide essential connectors between clusters, communities, and organizations. People with high betweenness can also be seen as bottlenecks or gatekeepers between two groups. This is one of our favorite SNA metrics because of how it highlights the bridgers in networks who play a critical yet often hidden role.
Eigenvector Centrality is a measure that takes into account not only how many other people a given person is connected to, but also the connections among those people. People with high eigenvector centrality are connected to other well-connected people.
Closeness Centrality is a measure of how connected a person is with every other person in the network, on average. Those with the top closeness centrality ranking might have an easier time reaching anyone in the network than others, and their opinions may spread faster than others as well.