Definition: Implications: Analysis:

Transcription

1 Analyzing Two mode Network Data Definition: One mode networks detail the relationship between one type of entity, for example between people. In contrast, two mode networks are composed of two types of entities. These entities could be people and teams or people and organizations. Two mode networks summarize the association between one entity and another, for example, the teams that individuals are members of. For this reason, two mode networks are often called affiliation networks. They can also be called bipartite networks. Implications: Gathering data on two mode networks, such as people and teams, can be a quick way of understanding the web of relationships in an organization. You can use this data as a proxy for employee employee relationships. However, caution must be taken when using two mode data in large teams or teams that evolve over time because not every person on a given team may know one another. In addition, individuals have ties that are outside their teams. Data comprising employee affiliations with teams can be useful in itself. It can highlight which teams (and the individuals in the teams) are central in the network and hence influential. It can also highlight which teams and individuals are playing broker roles within an organization. Analysis: Two mode data can be visualized and analyzed in several ways. It can be visualized so that ties between an individual and teams make up the structure of the network. UCINET also has some specific analysis routines for two mode data. Two mode data can be made into square matrices called bipartite networks, which allows for all the analysis routines in UCINET to be used. Finally two mode networks can be reduced to one mode of teams teams or employees employees. Each of these options is detailed in the following sections.

2 Visualizing Two mode Networks Visualizing two mode data in Netdraw In this analysis, we are going to visually examine our two mode network. The Excel spreadsheet below shows individuals who are members of teams. Notice the names of the respondents are in the first column and the team names are along the first row. Step 1. Load the network into UCINET the same way that you would any network. Name the file Teams. In Netdraw, File > Open > UCINET dataset > 2 Mode network. Then select the Teams file and press OK. In the network diagram, the teams are the blue squares and the individuals are the red circles. There is obvious clustering by team, with Team 6 connecting the two subgroups together.

3 Analyzing Two mode Networks Analyzing two mode data in UCINET Step 1. Network > 2 Mode networks > 2 Mode Centrality. Step 2. Select the Teams network and press OK. This routine produces normalized scores based upon the maximum possible value. The people with scores of have the most ties. The most central team is team 5, with Note that Team 6 (the one on the center of the network diagram) has the highest betweenness score.

4 Analyzing Two mode Networks Analyzing two mode data in UCINET We cannot run all the analytical measures in UCINET on two mode data because the matrix is not square. To create a square matrix, we need to run the bipartite function. This function adds the names of the teams to the rows and the names of the people to the columns. Step 1. Transform > Graph Theoretic > Bipartite Step 2. Select the Teams network and press OK. Step 3. Once you have constructed your bipartite dataset, you can analyze it the same way as other networks. The interpretation of the findings, however, needs to take into account that the data has two modes. Analysis 1. Degree centrality. Network > Centrality and Power > Degree. In the input network box type Teams Bip or click and select the Teams Bip network. Analysis 2. Brokerage (betweenness). Network > Centrality and Power > Freeman Betweenness > Node Betweenness. Degree Centrality Betweenness

5 Creating One mode Networks from Two mode Networks Creating one mode networks from two mode data Sometimes it may be useful to transform two mode data into one mode data. For example in our teams network, you can create a one mode network of ties between individuals where the ties represent being part of the same team. Step 1. Data > Affiliations (2 mode to 1 mode) Step 2. Select the Teams network, then under mode select the rows button, then press OK. In the new network, each value represents the number of teams each pair of people works on together. For most of the analysis in UCINET, you would need to dichotomize the data. You can also visualize the data in Netdraw in the usual way. In the diagram on the right, the width of the ties indicates the number of shared teams each pair of people are on using the properties > lines > size > tie strength option.

6 Bibliography Methodological papers and books: Borgatti, S. P., & Everett, M. G Network analysis of 2 mode data. Social Networks, 19(3), Borgatti, S. P., Everett, M. G., & Johnson, J. C Analyzing Social Networks. Los Angeles, CA: Sage. Borgatti, S. P., & Halgin, D. S Analyzing affiliation networks. The SAGE Handbook of Social Network Analysis, Hanneman, R. A., & Riddle, M Introduction to social network methods. University of California, Riverside. Published in digital form at Wasserman, S., & Faust, K Social Network Analysis: Methods and Applications. Cambridge, United Kingdom: Cambridge University Press. Empirical and conceptual papers: Breiger, R The duality of persons and groups. Social Forces 53: Cattani, G., Ferriani, S., Negro, G., & Perretti, F The structure of consensus: Network ties, legitimation, and exit rates of US feature film producer organizations. Administrative Science Quarterly, 53(1), Davis, A., B. B. Gardner and M. R. Gardner Deep South: A Social Anthropological Study of Caste and Class. Chicago: University of Chicago Press. Hoppe, B., & Reinelt, C Social network analysis and the evaluation of leadership networks. Leadership Quarterly, 21(4), Robins, G., & Alexander, M Small worlds among interlocking directors: Network structure and distance in bipartite graphs. Computational & Mathematical Organization Theory, 10(1), Snijders, T. A., Lomi, A., & Torló, V. J A model for the multiplex dynamics of two mode and one mode networks, with an application to employment preference, friendship, and advice. Social Networks, 35(2), Andrew Parker, PhD, is a visiting professor at the University of Kentucky. He has conducted network analysis in over 100 multinational organizations and government agencies. He was a Senior Consultant at IBM s Institute for Knowledge Management, a research fellow at the Network Roundtable at the University of Virginia as well as an advisor to the Knowledge and Innovation Network at Warwick Business School. His research has appeared in Science, Organization Studies, Journal of Applied Psychology, Journal of Applied Behavioral Science, Social Networks, Management Communication Quarterly, Sloan Management Review, Organizational Dynamics and California Management Review. He is also the co author of The Hidden Power of Social Networks. He received his PhD from Stanford University.