introducing network analysis as a research technique part

Introducing Network Analysisas a Research Technique

Part Two - Workshop

by

Graham Durant-Law

BSc, MHA, MKM, Grad Dip Def, Grad Dip Mngt, Grad Cert Hlth Fin, psc.

Copyright © 2009 Knowledge Matters™

Workshop Administration

AblutionsFire escapeWorkshop organised as four sessions

– Session 1: Introducing UCINET– Session 2: Visualising Networks with NetDraw– Session 3: Structuring Data– Session 4: Helper Applications - Pajek

Break between each sessionEnd time - 16:30Do ask questionsDo challengeDo participate

Copyright © 2009 Knowledge Matters™ 2

In seeking wisdom, the first step is silence, the second is listening, the third remembering, the fourth practicing,

the fifth – teaching others.

— Solomon ibn Gabirol

In seeking wisdom, the first step is silence, the second is listening, the third remembering, the fourth practicing,

the fifth – teaching others.

— Solomon ibn Gabirol

Session One – Introducing UCINET

‘Each of us is part of a large cluster, the worldwide social net, from which no one is left out. We do not know everyone on this globe, but it is guaranteed that there is a path between any two of us in this web of people’.

Albert-Laszlo Barabasi, Physicist, 2002


Copyright © 2008 Graham Durant-Law 4

UCINET has a default folder which can be changed here.

The current path.

UCINET Opening Screen

Drop down menus.

Helper Applications.

Data manipulation icons.

Drop down menus.

First Points to Note

UCINET assumes you know what you are doing and what the measures mean!UCINET of itself does not provide visualisations.UCINET creates many log files. Ensure you create a Session Directory.Ensure you name log files logically.There are several ways to get data into the tool.


Exercise One - Sampson Monastery

Open the Sampson Monastery file (sampson.##h) using the “spreadsheet” icon.

– Note there are multiple sheets– Note the structure of the sheets– Note the scoring system

Open the Sampson Monastery file (sampson.##h) using the “D display dataset” icon.

– Note the format of data this time

Import the Sampson Monastery file (samplike.txt) DATA-IMPORT-DL , or VNA, or RAW


Second Points to Note

UCINET is the quantitative engine for network analysisSome of the most common measures are centrality, density, and distance.

– Centrality: the extent to which a network is organised around one or more central nodes.

– Density: the percentage of connections that exist out of the total possible that could exist.

– Distance: degrees of separation or the diameter of a network


Exercise Two - Sampson Monastery

Import the Sampson Monastery file (samlike.txt) DATA-IMPORT-DLCalculate the out-degree and in-degree NETWORK-CENTRALITY-DEGREE

– Who is most popular?– Who is most active?– Can we really draw these conclusions?

Calculate node and edge betweeness centrality using the Freeman algorithm.

– Note the different outputs– What does this mean?


Exercise Three - Sampson Monastery

Import the Sampson Monastery file (samlike.txt) DATA-IMPORT-DLCalculate the density of the network. NETWORK-COHESION-DENSITY

– What is the report telling us?– Is the network well connected?– What is the ideal density?

Calculate network distances NETWORK-COHESION-DISTANCE

– Note the output– What is the report telling us?


Session Two – Visualising Networks with NetDraw

‘Simplicity is the key to effective scientific inquiry.’

Stanley Milgram, Sociologist, 1973


Copyright © 2008 Graham Durant-Law 11

NetDraw Opening Screen

Drop down menus.

Lightening spring layout icons.

Relationship filtering and attributing.

Relationship and node tab

Points to Note

NetDraw is not dependant on UCINET.NetDraw is the visualisation engine for network analysis.Note some measures can be applied from within the tool.NetDraw is very easy to use and misuse.Colour, size, and shapes are easy to apply.Several layouts are possible, and layouts can be filtered.Visualisations can be misleading.


Exercise Four - Sampson Monastery

Open the Sampson Monastery file (sampson.##h)

– Click the lightening icon to see the network.

– Experiment with several different layouts, using the LAYOUT menu.

– Colour the nodes.– Turn all the “like” relationships on.– Turn some of the nodes off.– Size the nodes by betweeness centrality.– What are the limitations of this dataset?– Is visualisation sufficient to explain the

network?


Exercise Five – Drug Dealers Network

Open the Drug Dealers network file (drugnet.##h)

– Click the lightening icon to see the network.

– Note the isolates.– What is the structure of this network?– What are the limitations of these views?

Open the Drug Dealers network attributes file (drugattr.##h)

– Turn the labels and arrows off.– Colour the nodes by ethnicity.– Shape the nodes by gender.– What does colouring and shaping the

nodes reveal?– Progressively remove the isolates and

pendants. Which nodes are of interest and why?

– Turn the arrows on. Now which nodes are of interest and why?


Exercise Six – Davis Women’s Events

Open the Davis Women’s Events file (davis.##h).

– Note Netdraw interprets it as a two-mode (bipartite) graph. Why?

Open the Davis Women’s Events file (davis.##h) in UCINET using the “D display dataset” icon.

– Note the format of data. – Which measures are appropriate for

this type of network? Why?


Session Three – Structuring Data

‘Science is built with facts, as a house is with stones. But a collection of facts is no more a science than a heap of stones is a house’.

Henri Poincare, Mathematician, 1901.


Session Three – Structuring Data

How are data collected and structured?What is a one-mode network?How are data imported into UCINET and NetDraw?A short friendship exercise.What is a two-mode network?A two-mode network exercise.


How are data collected and structured? The Generic Process

Determine the unit of analysis. This is arguably the most important step, as it determines how data is collected and which tools and analysis techniques should be employed. Determine the questions. The questions depend on the unit of analysis, and what you want to discover. Collect the data. Typically the questions are answered using a survey. The survey can be done in person, on paper, or be web-enabled. Where appropriate data collection can also be done using data-mining techniques. For example intra-departmental e-mail traffic could be mined. In the case of a policy relationship mapping exercise the documents are parsed for key words, headings and other relevant attributes. Import the data into a visualisation tool. Typically data is entered into an Microsoft EXCEL workbook or database, and then imported into a visualisation tool.


How are data collected and structured? Sample Questions

Type Example QuestionsOrganisational Interface Mapping

Please identify up to 10 people who work in external departments and who are important to you in your professional network. These can be people who provide you with information to do your work, help you think about complex problems posed by your work, or provide developmental advice or personal support helpful in your day-to-day working life. These may or may not be people you communicate with on a regular basis and must come from an organisation external to yours.

Project Interface Mapping

Please identify the people in other project teams that you rely on to provide information for your project. For each person you have identified please assign a score based on the amount of contact you have with them. 1 is the most amount of contact. 10 is the least amount of contact. Each score should be different.

Information Flow Mapping

Please identify the people in your department you have passed documents or e-mails to in the last month. These may or may not be people you communicate with on a regular basis, but they must be part of your department.

Collaboration Mapping

Please identify the people who are important to you in your professional network. These can be people who provide you with information to do your work, help you think about complex problems posed by your work, or provide developmental advice or personal support helpful in your day-to-day working life. These may or may not be people you communicate with on a regular basis and must come from within your organisation.

Social Capital Mapping

In your workplace who do you go to for information that helps you solve problems or capitalise on opportunities?


How are data collected and structured? A Mind Map


What is a one-mode network?


What is a one-mode network?

John Thomas Anna James Peter Mary Michael David Anthony Bobby

John 0 0 0 0 0 0 0 1 0 0

Thomas 0 0 0 0 0 0 1 2 0 0

Anna 0 0 0 0 0 0 0 0 0 0

James 0 0 0 0 0 0 0 0 0 0

Peter 0 0 0 0 0 0 0 0 0 0

Mary 0 0 0 0 2 0 0 0 0 1

Michael 0 0 0 2 0 0 0 0 0 0

David 0 2 0 0 0 0 0 0 0 0

Anthony 0 0 0 0 0 0 0 0 0 0

Bobby 0 0 0 0 4 1 0 0 0 0


How are data imported into UCINET and NetDraw?


Exercise Seven - Friendship Exercise

Using any method you want create your friendship network.– list between 10 and 20 friends– assign categorical attributes

such as age and gender– assign value attributes such as

strength of friendship.Create the same dataset in a different format – for example a text file instead of an Excel file.Calculate centrality, distance and density measures.Visualise the network.


What is a two-mode network?


What is a two-mode network?

ID Item 1 Item 2 Item 3 Item 4 Item 5 Item 6 Item 7 Item 8 Item 9 Item 10

CY10034 0 0 0 1 1 0 0 0 0 1

CY10039 1 0 0 1 0 0 0 0 0 0

CY10044 0 0 0 1 1 0 0 0 0 0

CY10045 0 0 1 0 0 0 0 0 0 0

CY10047 0 0 1 0 0 0 0 0 0 0

CY10054 1 0 0 1 0 0 0 0 0 0

CY10055 0 1 0 0 1 0 1 0 0 0

CY10057 0 1 0 0 0 0 0 0 0 0

CY10059 0 1 0 0 0 0 0 0 0 0


Exercise Eight – Two Mode Network

Using any method you want create a two-mode network of friends and where they work or study.Create the same dataset in a different format – for example a text file instead of an Excel file.Visualise the network.


Session Four – Helper Applications

‘A wonderful harmony is created when we join together the seemingly unconnected’.

Heraclitus, Greek Philosopher, circa 500 BC.


Exercise Nine – Helper Applications Pajek

Open NetDraw and load the Drug Dealers network file (drugnet.##h) and attributes file (drugattr.##h)– Save the file as a Pajek.net file.– Save the file Pajek.clu file for

ethnicity.– Save the file Pajek.vec for

gender.Open Pajek. Load the filesView the graph:– using a circular layout. – using the partition file and a

spring layout.


Workshop Concluding Remarks

‘The most incomprehensible thing about the world is that it is at all comprehensible’.

Albert Einstein, Physicist, 1921.


Some Cautions

Mathematical approaches to network analysis tend to treat the data as ‘deterministic’. That is, measurements are viewed as an accurate reflection of the ‘real’ or ‘final’ or ‘equilibrium’ state of the network.

Observations are usually regarded as the population of interest rather than a sample of some larger population of possible observations.

You must understand your organisation, the data, the resultant network and the assumptions you are making!

Hanneman, R & Riddle, M 2005, Introduction to social network methods,http://faculty.ucr.edu/~hanneman/


http://faculty.ucr.edu/~hanneman/

Some Issues

Must clearly define the ‘unit of analysis’ – that is what are nodes, what are ties, and what are attributes.Must define the population, and then cover the whole population to get meaningful network statistics.Requires specialist software.Usual limitations of survey techniques. Few research exemplars.


Ethical Considerations

Ethics in network analysis is a vexed question.Non-response does not immediately guarantee omission from the study.It is very easy for data, and the visualisations to be used in unintended ways and/or to be misinterpreted. Must read “Toward ethical guidelines for network research in organizations ” by Professor Steve Borgatti.


http://www.analytictech.com/borgatti/papers/ethics2005.pdf

http://www.analytictech.com/borgatti/papers/ethics2005.pdf

Presentation Considerations

To remove ambiguity always include the question on the slide.Remove node labels unless they are necessary for understanding.Try to space the diagram to minimise link crossovers.Where appropriate include arrowheads to show direction.Do use colour, size and shapes.Do filter data to show only what is necessary for understanding.Always remember there are multiple ways to present data, and hence multiple interpretations.


Web Resources

Connections - The official journal of INSNA. It has full text articles back to 1977. http://www.insna.org/indexConnect.htmlInternational Network for Social Network Analysis http://www.insna.org/Hanneman, R & Riddle, M 2005, Introduction to social network methods, http://faculty.ucr.edu/~hanneman/Vancouver Network Analysis Team http://www.sfu.ca/~richards/Ronald Burt's homepage http://gsb.uchicago.edu/fac/ronald.burtVisible Path – the web pages and blog of Stanley Wassermann http://www.centralityjournal.com/Social Networks Laboratory The University of Melbourne. http://www.psych.unimelb.edu.au/research/labs/UHSN_lab.htmlVirtual Observatory for the Study of Online Networks - Australian National University http://voson.anu.edu.au/


http://www.insna.org/indexConnect.html

http://www.insna.org/


http://www.sfu.ca/~richards/

http://gsb.uchicago.edu/fac/ronald.burt

http://www.centralityjournal.com/

http://www.psych.unimelb.edu.au/research/labs/UHSN_lab.html

http://voson.anu.edu.au/

Essential Reading

Barabasi, AL 2003, Linked. How everything is connected to everything else and what it means for business, science, and everyday life, Plume, New York.

Breiger, R, Carley, K & Pattison, P (eds) 2001, Dynamic social network modeling and analysis, The National Academies Press, Washington.

Buchanan, M 2002, Nexus. Small worlds and the groundbreaking theory of networks, W.W. Norton & Company, New York.

Cross, R & Parker, A 2004, The hidden power of social networks, Harvard Business School Publishing, Boston.

Durlan, MM & Fredericks, KA (eds) 2006, Social network analysis in program evaluation, Wiley, Minnesota.

Scott, J 2005, Social network analysis: a handbook, 2 edn, Sage Publications, London.

Surowiecki, J 2004, The wisdom of crowds. Why the many are smarter than the few, Abacus, London.

Watts, DJ 2003, Six degrees: the science of a connected age, W.W. Norton & Company, New York.


More Advanced Reading

Carrington, P, Scott, J & Wassermann, S (eds) 2005, Models and methods in social network analysis, Cambridge University Press, Cambridge.

Cross, R, Parker, A & Sasson, L (eds) 2003, Networks in the knowledge economy, Oxford University Press, New York.

de Nooy, W, Mrvar, A & Batagelj 2005, Exploratory social network analysis with PAJEK, Cambridge University Press, New York.

Freeman, LC 2004, The development of social network analysis. A study in the sociology of science, Empirical Press, Vancouver.

Hanneman, R & Riddle, M 2005, Introduction to social network methods, http://faculty.ucr.edu/~hanneman/

Hill, R & Dunbar, R 2002, ‘Social Network Size in Humans’ in Human Nature, Vol. 14, No. 1, pp. 53-72.

Kilduff, M & Tsai, W 2005, Social networks and organisations, Sage Publications, London.

Wassermann, S & Faust, K 1999, Social network analysis, Cambridge University Press, Cambridge.



Workshop Questions?

‘The real questions refuse to be placated… They are the questions asked most frequently and answered most inadequately, the ones that reveal their true natures slowly, reluctantly, most often against your will’.

Ingrid Bengis, Author, 1973


Contact Me


Phone +61 408 975 795Website http://www.durantlaw.infoBlog http://www.durantlaw.info/blog Email mailto:[email protected]

http://www.durantlaw.info/

http://www.durantlaw.info/

http://www.durantlaw.info/blog

mailto:[email protected]

introducing network analysis as a research technique part

Documents