2. Social Network Analysis
1. Description of the assignment and
choices made to fulfill requirements
2. Literature Review
3. Data Collection
4. Data Analysis - Sociograms
5. SLP Examples
6. Conclusion
3. Social Network Analysis
The Assignment:
- Explore an advanced research
method (Marketing area)
The Challenge:
- No student of Dr. Shackman had
ever studied this method before
4. Social Network Analysis
The Assignment
- Select an appropriate technique
- Explain its capabilities
- Explain its limitations
- Explain its data requirements
- Explain its applications
5. Social Network Analysis
Literature Review:
Scott (1988)
Trend Report
Social Network Analysis
Scott, J. (1988). Social Network Analysis. Sociology,
22(1), 109-127.
doi:10.1177/0038038588022001007.
9. Social Network Analysis
Literature Review:
Haythornthwaite (1996)
Social Network Analysis:
An Approach and Technique for
the Study of
Information Exchange
Haythornthwaite, C. (1996). Social network
analysis: An approach and technique for the study
of information exchange. Library & Information
Science Research, 18(4), 323-342.
10. Social Network Analysis
Literature Review:
Haythornthwaite (1996)
Three Attributes of SNA Relationships
- Content
- Direction
- Strength
11. Social Network Analysis
Literature Review:
Haythornthwaite (1996)
Two Types of Social Networks
- Egocentric; all of the connections
related to individual actors
- Whole Networks; all connections
within a single environment
12. Social Network Analysis
Literature Review:
Weber & Morrison (2004)
Network Analysis in Marketing
Webster, C.M. & Morrison, P.D. (2004). Network
Analysis in Marketing, Australasian Marketing
Journal (AMJ), 12(2), 2004, 8-18.
13. Social Network Analysis
Literature Review:
Weber & Morrison (2004)
SNA to Examine:
- Networks of relations
- Influence of these networks on
behavior
14. Social Network Analysis
Literature Review:
Weber & Morrison (2004)
General Avoidance of SNA for Marketing:
- Special data requirements
- Terminology used
- Cumbersome computer programs
(those initially used)
15. Social Network Analysis
Literature Review:
Contractor, Wasserman & Faust (2006)
Testing Multitheoretical, Multilevel
Hypotheses about Organizational Networks:
An Analytic Framework
and Empirical Example
Contractor, N. S., Wasserman, S., & Faust, K. (2006). Testing
Multitheoretical, Multilevel Hypotheses about
Organizational Networks: An Analytic Framework and
Empirical Example. Academy Of Management Review,
31(3), 681-703.
16. Social Network Analysis
Literature Review:
Contractor, Wasserman & Faust (2006)
Key Advantage of SNA:
- Ability to collect, collate, study
data at various levels of analysis
17. Social Network Analysis
Literature Review:
Contractor, Wasserman & Faust (2006)
Hypothesis Testing:
- Null hypothesis is that all ties are
independent with equal
probability
18. Social Network Analysis
SNA Capabilities
- Key relationships
- Visual representation
- Identify influencers
- Marketing research
19. Social Network Analysis
SNA Limitations
- Network size
- No hypothesis testing
- Complexity/Topography
limitations
21. Social Network Analysis
SNA Applications
- Uncover informal or hidden
networks
- Design effective marketing
campaigns
22. Social Network Analysis
Data Collection
- Using typical methods
- Recorded in a matrix
- Often collects data from the
entire network population rather
than a sample
23. Social Network Analysis
Data Analysis - Sociograms
- Visual representation of the
nodes, the edges (connections), and
the direction of the relationships.
29. Social Network Analysis
Conclusion
- SNA is primarily qualitative
- SNA is a useful tool for studying
relationships
- SNA results in sociogram
- SNA can provide valuable
insights for Marketing research
Hello. I’m Scott Gomer and I’m here to talk about an advanced research method known as Social Network Analysis or S-N-A.
Today’s presentation will follow this general outline. We’ll begin with details about the Mod four assignment and then move on to a literature review of S-N-A research developments. Following that, we look at the way data is collected and analyzed, provide examples from my SLP submission, and wrap up with concluding remarks.
As you might recall, the assignment was for students to select an advanced research method related to their particular area of concentration. My area is Marketing and after a review of the background materials it seemed that Social Network Analysis, or S-N-A, was an interesting research technique to study. After all, Marketing is all about relationships and S-N-A provides a unique perspective into some of these relationships. When I proposed studying the S-N-A method to Dr. Shackman, he told me that no student had ever studied and reported on this technique in any of his previous classes. I felt that the reward of learning more about using this method far outweighed the challenge, so I dove right into the assignment.
The specific assignment requirements were to select an advanced research technique and then provide an overview of the technique’s features. As previously mentioned, the technique chosen for the assignment was Social Network Analysis, S-N-A. Following the literature review, I’ll be providing information about its capabilities, limitations, data requirements, and applications. The reason I chose S-N-A is because it looked very interesting and I thought it might be applied in many ways within the Marketing area. For example, S-N-A offers a way to examine connections behind consumer behavior that might uncover ways to better locate market influencers – those individuals and groups that generate additional sales through their networks of followers. S-N-A is also often used as a way to examine organizational communication; particularly the powerful, but often elusive, informal communication network. Let’s move on to a review of the relevant literature.
The seminal article on Social Network Analysis is a 1988 article in Sociology by Dr. John Scott, a British sociologist at the University of Plymouth on the far southwestern coast of England. This report, cited more than 5800 times since publication, describes the development of S-N-A up to that point in time. It traces S-N-A within the field of classical sociology and examines scientific and mathematical applications that aided in data collection, analysis, and presentation.
Scott described the methods and procedures employed within S-N-A as heavily dependent on graphical representations of group dynamics and connections. Individuals, groups, companies, industries, or even countries are represented by points or nodes that are connected by lines, arcs, or edges.
In order to depict connections with social networks, Scott presented the way mathematical graph theory has been adapted to represent social connections. From these graphical diagrams, it becomes possible to identify pockets of centrality among network “stars” – those points on the graph with the most connections – as well as network “isolates” – those points with only one or a few connections.
Here’s a visual representation from Scott’s article. This is a very simple diagram known as a sociogram. We will be looking at more complex sociogram models and examples later in this presentation.
Just eight years after Scott’s report, Caroline Haythornthwaite from the University of Illinois set about establishing clear principles and guidelines for practical application of Social Network Analysis; specifically, using S-N-A to track the exchange of information among individuals.
S-N-A enables researchers to study three primary attributes of relationships: the content that is exchanged between points or nodes within the network, the direction of those exchanges as either one-way or two-way, and the strength of the connection between nodes. For example, in this course, all of the students receive course content and feedback from Dr. Shackman via a strong connection that leads from each student directly to Dr. Shackman. However, although Dr. Shackman receives content through a mandatory connection with all students, he has relatively few, weak connections directly back to students seeking the same type of feedback from them.
Haythornthwaite explains two types of social networks – an egocentric social network that shows all of the connections related to individual actors and a whole network that tracks all relationships within a single environment such as a school, workplace, or geographic area.
One of the few studies I was able to locate that directly tied use of S-N-A to Marketing came from two researchers in Australia.
Webber and Morrison provided logical definitions of S-N-A techniques in a theoretical article positing the use of S-N-A to study the exchange of information within the context of a library or other information provider. They stated that S-N-A could be used to identify information needs, visualize information exposure, estimate information legitimization, display information routes, and uncover information opportunities.
The authors also gave three likely reasons that few marketing researchers use S-N-A . One reason is because of special data requirements such as collecting data for an entire network rather than just a representative sample. Another is because of the terminology – nodes, edges, sociograms – that is relatively unfamiliar territory for marketing researchers. Finally, many researchers were turned off by the clunky nature of the original computer programs that enabled the fast calculation of data matrices needed to provide the sociograms.
More recently, researchers have begun to experiment with highly complex analyses using Social Network Analysis in more quantitative ways. This study set up ten models to test eight hypotheses.
These complex hypotheses were made possible due to what the authors cite as the main advantage of using S-N-A, its ability to examine data at several levels of analysis.
In order to enable quantitative hypothesis testing, the authors established the null hypothesis as a state where all ties in the network are independent of one another with an equal probability of a connection between them. Using this new and advanced technique, the authors were able to find support for three of their eight hypotheses.
Returning momentarily to the Mod four assignment, here’s what was learned about Social Network Analysis capabilities. S-N-A is a useful method for locating key relationships within networks that may not match formal organizational charts. Provides visual representation of network activity. And S-N-A is often the only way to truly identify which network nodes are most influential so that they can be contacted to help spread information or other valuable content. Also, as stated by Webster and Morrison, S-N-A can be applied to a wide range of marketing research including word-of-mouth communication, relationship marketing, information acquisition, and adoption of new products and services.
First and foremost, S-N-A is limited by the size of the network being studied. Since data is in the form of a matrix, relationships expand exponentially so that even studying a class of 30 students results in the analysis of 870 connections (that is a matrix of 30 x 29). Move up to a group of 100 and the network connections climb to 9900. Another limitation of using S-N-A is the complexity required for hypothesis testing or other quantitative research. S-N-A is by nature primarily a qualitative research method. S-N-A is also hindered by its limited ability to accurately represent the content, direction, and strength of connections for certain types of complex network arrangements and topographies.
The data requirements for S-N-A can be as simple as using a 0 or 1 to indicate a relationship between points in the network. This can be represented with the same set of actors on both axes, known as a one-mode network, or as a matrix that depicts different actors on the axes, known as a two-mode network.
This advanced research method lends itself nicely to the tasks of uncovering otherwise invisible networks and toward effective designs to support more effective marketing campaigns.
Data for Social Network Analysis can be gathered using any of the typical measurement instruments including surveys, interviews, or observation. Data are normally recorded in some sort of matrix that fits the data entry requirements of the software program used to analyze the data. One thing that is markedly different is the requirement to collect data from the entire population within the network so as to accurately show all of the connections at once.
Sociograms are the contribution from mathematical graph theory that help show relationships within Social Network Analysis. Let’s look at an example.
Here’s the sociogram presented by Webster & Morrison of the connections between 27 pharmaceutical companies in Australia. Notice how certain labs stand out as “stars” with several connections to and from other laboratories while many labs are just “isolates” with no incoming connections.
Here’s another example from my SLP assignment. Using what I found to be one of the more popular software programs for analyzing Social Network Analysis, a program called UCINET, I located a dataset from the UCINET website. In this case, the dataset represents communication relationships for a small sawmill that employees both English and Spanish speaking workers. The letter in front of each node is either an E for English speaker or H for Hispanic, Spanish speaker. As you can see, a Spanish speaking employee named Juan is the “star” of the network as he has the most connections. Those connections include the owner, mill manager, and several of the Spanish speaking employees. In all, Juan has connections to 13 of the 26 other workers in the mill. This indicates that Juan is most likely a very effective bilingual worker who is well respected by both upper management and workers at the lower levels of the organization. He is probably the one most everyone turns to whenever some sort of translation is needed within the communication chain.
Here’s another example from my SLP. As mentioned previously, one of the major limitations of S-N-A is its difficulty to handle large datasets. The sociogram depicted here shows all of the connections between the 500 largest US airports. So, which patterns do you see here?
The UCINET program offers a few filters that can adjust the view to make it a little easier to interpret the results. This view incorporates the Gower theoretical model to bring the relationships a little closer into focus. Of course, researchers can also reduce the dataset as demonstrated in the next slide.
Here is a spread out view of the connections between just the top 100 US airports. You can now clearly see the clusters surrounding the top 20 or so major airline hubs.
Social network analysis is a longstanding, but little understood, research technique that may apply to your own area of study. The method emerged from the field of sociology, but can apply to many types of business research as well. Think of S-N-A primarily as a qualitative tool to gather more information about the relationships of interest. Learn how to analyze data in a visual way through the sociograms that represent the relationships under review. And consider using S-N-A to address a wide variety of research questions in the area of Marketing and Market research.
Thank you for watching this presentation. Should you have any questions or like to discuss this technique further, please feel free to contact me via the email address on screen. Thanks again.