A User's Guide to Network Analysis in R (eBook)

Douglas Luke is Professor and Director of the Center for Public Health Systems Science at the George Warren Brown School of Social Work at Washington University in St. Louis. He is a leading researcher in the fields of health policy, and his work focuses on the evaluation, dissemination, and implementation of evidence-based public health policies. Dr. Luke has worked extensively with systems science methodologies, especially the analysis of social networks with regards to the implementation of public health policies. He is a member of the Institute for Public Health, a founding member of the Washington University Network of Dissemination and Implementation Researchers (WUNDIR), and serves on the Interagency Committee on Smoking and Health at the U.S. Department of Health and Human Services.

Preface 8
Contents 10
1 Introducing Network Analysis in R 14
1.1 What Are Networks? 14
1.2 What Is Network Analysis? 16
1.3 Five Good Reasons to Do Network Analysis in R 17
1.3.1 Scope of R 17
1.3.2 Free and Open Nature of R 18
1.3.3 Data and Project Management Capabilities of R 18
1.3.4 Breadth of Network Packages in R 19
1.3.5 Strength of Network Modeling in R 19
1.4 Scope of Book and Resources 19
1.4.1 Scope 19
1.4.2 Book Roadmap 20
1.4.3 Resources 21
Part I Network Analysis Fundamentals 22
2 The Network Analysis `Five-Number Summary' 23
2.1 Network Analysis in R: Where to Start 23
2.2 Preparation 23
2.3 Simple Visualization 24
2.4 Basic Description 24
2.4.1 Size 24
2.4.2 Density 26
2.4.3 Components 27
2.4.4 Diameter 27
2.5 Clustering Coefficient 28
3 Network Data Management in R 29
3.1 Network Data Concepts 29
3.1.1 Network Data Structures 29
3.1.1.1 Sociomatrices 30
3.1.1.2 Edge-Lists 31
3.1.2 Information Stored in Network Objects 32
3.2 Creating and Managing Network Objects in R 33
3.2.1 Creating a Network Object in statnet 33
3.2.2 Managing Node and Tie Attributes 36
3.2.2.1 Node Attributes 37
3.2.2.2 Tie Attributes 38
3.2.3 Creating a Network Object in igraph 40
3.2.4 Going Back and Forth Between statnet and igraph 42
3.3 Importing Network Data 42
3.4 Common Network Data Tasks 44
3.4.1 Filtering Networks Based on Vertex or Edge AttributeValues 44
3.4.1.1 Filtering Based on Node Values 44
3.4.1.2 Removing Isolates 46
3.4.1.3 Filtering Based on Edge Values 47
3.4.2 Transforming a Directed Network to a Non-directedNetwork 51
Part II Visualization 54
4 Basic Network Plotting and Layout 55
4.1 The Challenge of Network Visualization 55
4.2 The Aesthetics of Network Layouts 57
4.3 Basic Plotting Algorithms and Methods 59
4.3.1 Finer Control Over Network Layout 60
4.3.2 Network Graph Layouts Using igraph 62
5 Effective Network Graphic Design 64
5.1 Basic Principles 64
5.2 Design Elements 64
5.2.1 Node Color 65
5.2.2 Node Shape 69
5.2.3 Node Size 71
5.2.4 Node Label 75
5.2.5 Edge Width 77
5.2.6 Edge Color 78
5.2.7 Edge Type 79
5.2.8 Legends 80
6 Advanced Network Graphics 82
6.1 Interactive Network Graphics 82
6.1.1 Simple Interactive Networks in igraph 83
6.1.2 Publishing Web-Based Interactive Network Diagrams 83
6.1.3 Statnet Web: Interactive statnet with shiny 86
6.2 Specialized Network Diagrams 86
6.2.1 Arc Diagrams 87
6.2.2 Chord Diagrams 88
6.2.3 Heatmaps for Network Data 91
6.3 Creating Network Diagrams with Other R Packages 93
6.3.1 Network Diagrams with ggplot2 93
Part III Description and Analysis 97
7 Actor Prominence 98
7.1 Introduction 98
7.2 Centrality: Prominence for Undirected Networks 99
7.2.1 Three Common Measures of Centrality 100
7.2.1.1 Degree Centrality 100
7.2.1.2 Closeness Centrality 101
7.2.1.3 Betweenness Centrality 101
7.2.2 Centrality Measures in R 102
7.2.3 Centralization: Network Level Indices of Centrality 103
7.2.4 Reporting Centrality 104
7.3 Cutpoints and Bridges 108
8 Subgroups 112
8.1 Introduction 112
8.2 Social Cohesion 113
8.2.1 Cliques 114
8.2.2 k-Cores 117
8.3 Community Detection 122
8.3.1 Modularity 122
8.3.2 Community Detection Algorithms 125
9 Affiliation Networks 131
9.1 Defining Affiliation Networks 131
9.1.1 Affiliations as 2-Mode Networks 132
9.1.2 Bipartite Graphs 132
9.2 Affiliation Network Basics 133
9.2.1 Creating Affiliation Networks from Incidence Matrices 133
9.2.2 Creating Affiliation Networks from Edge Lists 135
9.2.3 Plotting Affiliation Networks 136
9.2.4 Projections 137
9.3 Example: Hollywood Actors as an Affiliation Network 139
9.3.1 Analysis of Entire Hollywood Affiliation Network 140
9.3.2 Analysis of the Actor and Movie Projections 145
Part IV Modeling 151
10 Random Network Models 152
10.1 The Role of Network Models 152
10.2 Models of Network Structure and Formation 153
10.2.1 Erd?s-Rényi Random Graph Model 153
10.2.2 Small-World Model 156
10.2.3 Scale-Free Models 159
10.3 Comparing Random Models to Empirical Networks 165
11 Statistical Network Models 168
11.1 Introduction 168
11.2 Building Exponential Random Graph Models 170
11.2.1 Building a Null Model 172
11.2.2 Including Node Attributes 174
11.2.3 Including Dyadic Predictors 176
11.2.4 Including Relational Terms (Network Predictors) 180
11.2.5 Including Local Structural Predictors (Dyad Dependency) 182
11.3 Examining Exponential Random Graph Models 184
11.3.1 Model Interpretation 184
11.3.2 Model Fit 185
11.3.3 Model Diagnostics 188
11.3.4 Simulating Networks Based on Fit Model 188
12 Dynamic Network Models 193
12.1 Introduction 193
12.1.1 Dynamic Networks 193
12.1.2 RSiena 195
12.2 Data Preparation 196
12.3 Model Specification and Estimation 202
12.3.1 Specification of Model Effects 202
12.3.2 Model Estimation 207
12.4 Model Exploration 207
12.4.1 Model Interpretation 207
12.4.2 Goodness-of-Fit 213
12.4.3 Model Simulations 216
13 Simulations 220
13.1 Simulations of Network Dynamics 220
13.1.1 Simulating Social Selection 221
13.1.1.1 Setting Up the Simulation 221
13.1.1.2 Creating an Update Function 222
13.1.1.3 Building a Simple Simulation of Social Selection 228
13.1.1.4 Interpreting the Results of the Simulation 230
13.1.2 Simulating Social Influence 231
13.1.2.1 Setting Up the Simulation 231
13.1.2.2 Creating an Update Function 232
13.1.2.3 Building the Simulation of Social Influence 234
13.1.2.4 Interpreting the Results of the Simulation 235
References 238