Types of Networks. Directed and non-directed relations; relational and affiliational networks; multiple networks; nested networks.

Size: px

Start display at page:

Download "Types of Networks. Directed and non-directed relations; relational and affiliational networks; multiple networks; nested networks."

Barbra Leonard
5 years ago
Views:

1 POL 279 Political Networks: Methods and Applications Course Website: Winter 2012 Zeev Maoz Wednesday 3:00-6:00 Office Hours: TTH-3:00-4:00 We live in a networked world. To some extent everybody in the planet is tied to everybody else through some indirect ties. Politics is also about networks. The structure of ties between people, organizations, groups, and nations is defined by certain principles. These ties also have an important impact on our ideas, behaviors, and their collective consequences. Social Networks Analysis (SNA) offers a rigorous, mathematically based approach to the study of relations among units organisms, individuals, groups, organizations, and nations. It is widely used in such disciplines as physics, biology, psychology, economics, sociology, and organizational behavior, to name just a few. In political science this approach had less impact although it is eminently suited for the study of relations among political actors. The purpose of this seminar is to introduce students to SNA and to the possibilities of employing this approach to study important processes in all fields of political science. Beyond introducing the basic concepts, techniques, and ideas of SNA, the seminar will cover applications of this approach in all empirical subfields of political science including American politics, comparative politics, and international relations. Students from all subfields are welcome to attend. Topics: Social Networks the Science of a Connected Age. A general introduction to SNA s philosophy; the historical evolution of this approach; how it was used to advance knowledge in various disciplines; some observations about the limitations of this approach. Types of Networks. Directed and non-directed relations; relational and affiliational networks; multiple networks; nested networks. The Basic Mathematics of Social Networks. Graphic and matrix representation of networks; Direct and indirect relations in social networks; basic network characteristics: density, transitivity, centrality and centralizations. Characteristics of Units: Egonets, centrality and prestige, dependence and interdependence, local transitivity. Characteristics of Subsets. Dyads, triads, cliques, communities, blocs and blockmodels, group characteristics analysis. Fit coefficients.

2 Political Networks Methods and Applications 2 Signed Networks and Structural Balance. Cognitive algebra, nodal balance, balanced cyclicality, group balance, network balance. Positional Analysis. Structural equivalence, role equivalence Network Composition and Compound Relations. Relations among networks and their implications. We will examine networks in international relations, in congress, and explore experimental study of networks in micro-political settings. We have three principal textbooks for this course. Stanley Wasserman and Katherine Faust Social Networks Analysis: Methods and Applications. New York: Cambridge University Press (second printing) Hereafter referred to as Wasserman and Faust (1997) Peter J. Carrington, John Scott, and Stanley Wasserman (eds.) Models and Methods in Social Networks Analysis. New York: Cambridge University Press hereafter referred to as Carrington et al. (2005). Zeev Maoz Networks of Nations: The Evolution, Structure, and Impact of International Networks, New York: Cambridge University Press hereafter referred to as Maoz 2011a. Recommended: Matthew O. Jackson Social and Economic Networks. Princeton: Princeton University Press. Note: This is a quite general but difficult book. It assumes a lot of math, but those of you who are into the topic, are welcome to peruse it. Software Packages: We will be using two software packages. UCINET is the most commonly used package in SNA. A free trial package can be downloaded for 60 days. The download website is After 60 days, you will need to get a licensed package. I am working on getting a budget so you can be reimbursed for buying it (it costs $40 for students). This is a pretty comprehensive package. Please be sure to download a package to your PC and bring your PCs to class so we can practice it. MaozNet. This program can be downloaded from my website Please be sure to follow the instructions on the Webpage before downloading it. Also, there is a bug with the online help file, so please download the help file separately. This program should also be on your PCs for the first session. Note: those using Macs should

3 Political Networks Methods and Applications 3 have a Windows emulation program in order to be able to run it. The same applies to UCINET. Datasets that can be used for your work. You will find a number of datasets on drive S in directory S:\graduate\pol \. These include the following datasets. 1. International Alliances. These are data on international alliance networks, spanning the period of , and measured annually. They had been collected by the ATOP Project (Leeds, 2005). There are three datasets in this subdirectory. a. A binary alliance variable with 1 denoting the presence of any type of alliance commitment between states at a given year, and zero if no alliance of any type exists between the states. b. A binary defense/offense variable, with 1 denoting the presence of a high-level alliance (either a defense pact or an offense pact) between the states for a given year, and zero otherwise. c. A continuous relative commitment variable denoting the strength of security commitment between the states for that given year. The coding of the relative commitment score is documented in Maoz (2011a: 41-42). 2. International Trade. Here there are two different trade datasets. a. The first was collected by Katherine Barbieri, Omar Keshk, and Brian Pollins (BKP in the file name), and covers the period of The data measure the share of the row s state (statea) GDP that have been exported to stateb at a given year. There is a continuous trade dataset and a binary trade dataset. The cutoff for binary trade is one thousandth of a GDP going from statea to stateb. b. The second was collected by Kristian Gledtisch for the period and expanded by John Oneal and Bruce Russett to cover the period Data are the same as the BKP data. The key difference is that the original BKP dataset has over fifty percent missing data. These data were assigned a score of zero in the files provided here. Gledtisch used a number of techniques to impute missing data, so that the proportion of missing data in that file is much lower (about 16%). Those were also assigned a score of zero in the file. 3. International Governmental Organizations (IGOs). These data list state membership in IGOs since Data were collected by Jon Pevehouse, Tim Nordstrom, and Kevin Wrenke. Data are set up in two types of file: the IGOaffil files are affiliation networks denoting whether a given state (row) is affiliated with one, two, or more IGOs. The igosimilarity dataset measures the relative IGO affiliation similarity of two states as explained by Maoz (2011a: 43-47). 4. European Party Systems. This is a dataset generated by Zeynep Somer-Topcu and Zeev Maoz. It covers the political systems of seventeen European states and Israel over the period of Data measure ideological similarity between pairs of political parties based on the similarity of their positions as stated in their political manifestos. 5. Senate Data. These data cover two types of behaviors of Senators from the 93 rd to the 110 th Congress. One dataset covers the extent to which two senators co-sponsored the same bills. The other examines the similarity of their roll-call votes.

4 Political Networks Methods and Applications 4 Feel free to peruse these data for your excursuses, papers, etc. If you confront any problems, or have any questions, please let me know. Course Requirements. Weekly exercises using social networks software and applying it to different political network data. The exercises account for 30% of the final grade. Research design paper: 4-6 double spaced paper outlining a research design applying social networks analysis to a political science topic of your choice. This accounts for 20% of the grade. Class presentation of the research design: 10% of the grade. Final research paper converting the research design to a full-fledged paper: 40% of the final grade. Course Schedule Week 1: What is Network Analysis? A general introduction to Network Analysis, including: The origins and development of the approach in various disciplines Relevance to Political Science Basic Concepts Types of Networks, Nodes, Vertices, Graphs, and Matrices Levels of Analysis: units, dyads, triads, groups cliques and blocks, structural aspects of entire networks Types of Network Data: Sociomatrices, Affiliation networks, Attribute data Network Representation: Graphs and Matrices Some limitations of Social Networks Research An introduction to network software: UCINET and MaozNet. Wasserman and Faust (1997: 1-66). Duncan J. Watts Six Degrees: The Science of a Connected Age. New York: W.W. Norton, pp Maoz 2011a. Chapter 1. Exercise 1: Graphic Presentation of a Political Network: An explanation of the data and the software package will be given in a separate instructions sheet, posted on the course s Website Week 2: Nodes and Social Networks

5 Political Networks Methods and Applications 5 The characteristics of individual positions, power, and prestige. Relationship between individual characteristics and network characteristics. Topics: Describing Egonets Attributes of Egonets Centrality and Prestige Symmetry issues of Centrality and Prestige Centrality and prestige in political networks Local transitivity Other characteristics of Egonets Wasserman and Faust (1997: ). Martin Everett and Stephen P. Borgatti. Extending Centrality. In Carrington et al., (2005: 57-76). Daniel Gomez, et al Centrality and Power in Social Networks: A Game Theoretic Approach. Mathematical Social Science, 46(1): Available on course website. Arthur S. Alderson and Jason Beckfield Power and Position in the World City System. American Journal of Sociology, 109(4): Available on course website. James H. Fowler Legislative Cosponsorship Networks in the U.S. House and Senate. Social Networks, 28(4): Maoz 2011a: Chapters 4 and 7. Zeev Maoz Network Power. Mimeographed, University of California, Davis. Exercise 2: Develop a set of hypotheses linking network centrality to some aspect of behavior of actors (or groups of actors) in the two networks which you analyzed in Exercise 2. Measure the centrality of actors (or groups) in these networks Measure nodal power in these networks Test the hypotheses. Report your findings and draw conclusions on the relationship between (a) centrality and power, (b) attributes and power, and (c) the effects of centrality, attributes, and network power on behavior in these networks. Week 3: Dyads and Triads in Social Networks Characteristics of and relations in dyads and triads as subgroups of social networks.

6 Political Networks Methods and Applications 6 Direct and Indirect links between dyads Dyadic Characteristics: structural equivalence, role equivalence Dyadic census Triadic relationships: transitivity Triadic censuses Triadic relationships: structural holes Introduction to balance theory Wasserman and Faust (1997: ). Patrick Doreian et al. Positional Analysis of Sociometric Data. In Carrington et al. (2005: 77-98). Zeev Maoz, Ranan D. Kuperman, Lesley G. Terris, and Ilan Talmud Structural Equivalence and International Conflict: A Social Networks Analysis. Journal of Conflict Resolution, 50(5): Zeev Maoz, Lesley G. Terris, Ranan D. Kuperman, and Ilan Talmud What is the Enemy of My Enemy: Causes and Consequences of Imbalanced International Relations, Journal of Politics, 69(1): Optional Reading: Wasserman and Faust (1997: ). (Difficult.) Exercise 3: One of three options. Exercise similar to Exercise 2, developing and testing hypotheses on dyadic relations and dyadic behavior/outcomes using measures of structural/role equivalence. (1) Measure the structural/role equivalence of dyads in your networks, (2) compare between the measures of structural and role equivalence in the networks, and (3) test the effect of structural/role equivalence on dyadic relations using an external measure of behavior/relations. For example, if you are using alliances as the network from which you are measuring structural/role equivalence, test the effects of equivalence on the probability of dyadic conflict. If you are using the Senate data, test the effect of equivalence on the probability of the two senators voting the same way in roll calls. Exercise similar to Exercise 2, developing and testing hypotheses on triadic relations and triadic behavior. Tasks similar to the process above. Exercise similar to Exercise 2, developing and testing hypotheses on the difference between direct and indirect dyadic relations, choosing appropriate measures of direct and indirect dyadic attributes. Tasks similar to the process above. Week 4: Partitioning networks into groups

7 Political Networks Methods and Applications 7 Cliques and N-Cliques Affiliations and overlapping Subgroups Blockmodeling methods: A comparative analysis Clusters Communities and modularity coefficient Relationships between grouping methods: Cliques, Blocks, Clusters, Communities Measures of Group Properties Goodness of Fit measures Groups as units of analysis Wasserman and Faust (1997: , ). Lucia Falzon Determining Groups from the Clique Structure in Large Social Networks. Social Networks, 22(2): Elizabeth Leicht and Mark E.J. Newman Community Structure in Directed Networks. Physical Review Letters, 100(118703): 1-4. Mason A. Porter et al Community Structure in the United States House of Representatives. Physica A. 386(1): Maoz 2011a. Chapters 4-5, 7, 9. Exercise 4: Use the alliance or congressional datasets to generate block partitions of the network. If you are using the alliance dataset, compare your blocks to the actual alliances that existed that year. If you are drawing blocks from the congressional datasets compare these with the Poole and Rosenthal common space score and the Fowler Closeness index. Discuss the extent to which actual closeness results match the blockmodel structure of the network. Week 5: Network Characteristics Density Transitivity Centralization Indices Interdependence Network Polarization

8 Political Networks Methods and Applications 8 Wasserman and Faust (1997: 101-3, ; 165; ). Zeev Maoz The Effects of Strategic and Economic Interdependence on International Conflict across Levels of Analysis. American Journal of Political Science, 53(1): Zeev Maoz Network Polarization. Mimeo. University of California, Davis. To be distributed. Maoz 2011a, pp Note: Research Design Paper is due Week 6: Relations among Multiple Networks Comparing networks and determining relationships among multiple networks. Network Composition and compound relations Relational Algebras Network roles and positions in composite networks Comparing different networks with the same units Comparing similar networks with different units Comparing different networks with different units Complexity in multiple networks Wasserman and Faust (1997: ). Zeev Maoz Network Polarization, Network Interdependence, and International Conflict, Journal of Peace Research, 43(4): Maoz 2011a. Chapter 11. Exercise 6: Details will be provided on a special instructions sheet on the course s website. Week 7: Signed Networks and Structural Balance Analyzing networks with positive and negative relations. What are signed networks? Cognitive algebra and the analysis of signed networks Relationships between transitivity and balance in signed networks Estimating imbalances at the nodal, dyadic, group, and network level Estimating imbalances across different networks

9 Political Networks Methods and Applications 9 Wasserman and Faust (1997: ). Zeev Maoz National Choices and International Processes. Cambridge: Cambridge University Press, pp Week 8: Network Dynamics Measuring change in networks over time including change in the size of the network and the composition of the units. Measuring and predicting change through network attributes Measuring and predicting change in subgroups Measuring and predicting change in units attributes and egonets Measuring the effect of change in one network on change in other networks Tom A. B. Snijders Models for Longitudinal Network Data. In Carrington et al. (2005: ). Mark Huisman and Tom A. B. Snijders Statistical Analysis of Longitudinal Network Data with Changing Composition. Sociological Methods and Research, 32(2): Katherine Faust and John Skvoretz Comparing Networks across Space and Time, Size and Species. Sociological Methodology, 8(2): Zeev Maoz From Cooperation to Conflict: An Agent-Based Model of Security Network Formation. Mimeographed, University of California, Davis. To be distributed. Zeev Maoz Preferential Attachment, Homophily, and the Structure of International Networks, Conflict Management and Peace Science (forthcoming). Will be distributed. Week 9: Random Graphs and Network Formation Models: A Basic Introduction The Random Graph Approach to Social Networks Analysis Uncovering Latent Dependencies through Exponential Random Graphs Students Presentations of Draft Paper. Students must circulate drafts among all class participants by Monday noon. (This includes both students who are scheduled to present and those who are scheduled to present the following week.) Stanley Wasserman and Gary Robbins An Introduction to Random Graphs, Dependent Graphs and p*. In Carrington et al. (2005: )

10 Political Networks Methods and Applications 10 Garry Robbins and Phillipa Patterson Interdependencies and Social Processes: Dependence Graphs and Generalized Dependence Structures. In Carrington et al. (2005: ). Peter D. Hoff and Michael D. Ward Modeling Dependencies in International Relations Networks. Political Analysis 12 (2): Skyler J. Cranmer and Bruce A. Desmarais Inferential Network Analysis with Exponential Random Graph Models. Political Analysis, 19(1): Skyler J. Cranmer, Bruce A. Desmarais, and Elizabeth J. Menninga Complex Dependencies in the Alliance Network. Conflict Management and Peace Science (forthcoming). Article will be distributed. Students Presentations of Research Design Readings Week 10: Social Networks Analysis in Political Science Promise and Limitations An open discussion of the possibilities and problems entailed in using social networks analysis to study political processes. I will try to bring in several faculty members from Poli Sci and Sociology for this discussion. No Readings. Note Final Paper Due.

Blockmodels/Positional Analysis Implementation and Application. By Yulia Tyshchuk Tracey Dilacsio

Blockmodels/Positional Analysis Implementation and Application By Yulia Tyshchuk Tracey Dilacsio Articles O Wasserman and Faust Chapter 12 O O Bearman, Peter S. and Kevin D. Everett (1993). The Structure