social network analysis as an evaluation tool 10 june 2015rick unicef, florence1

Download Social Network Analysis as an evaluation tool 10 June 2015Rick UNICEF, FLorence1

If you can't read please download the document

Upload: shannon-chapman

Post on 18-Jan-2018

219 views

Category:

Documents


0 download

DESCRIPTION

Agenda 1.What is a network [5] 2.My own perspectives on networks [4] 3.How to represent networks [8] 4.How to gather data on networks [12] 5.Making sense of complex networks [5] 6.Causal perspectives on networks 7.SNA software [10] 10 June 2015Rick UNICEF, FLorence3

TRANSCRIPT

Social Network Analysis as an evaluation tool 10 June 2015Rick UNICEF, FLorence1 The heart of all major discoveries in the physical sciences is the discovery of novel methods of representation Steve Toulmin (1953) The Philosophy of Science: An introduction. 10 June 2015Rick UNICEF, FLorence2 Agenda 1.What is a network [5] 2.My own perspectives on networks [4] 3.How to represent networks [8] 4.How to gather data on networks [12] 5.Making sense of complex networks [5] 6.Causal perspectives on networks 7.SNA software [10] 10 June 2015Rick UNICEF, FLorence3 1. What is a network? 10 June 2015Rick UNICEF, FLorence4 Any set of entities Connected by some kind of relationship Forming a larger structure Common purpose is not necessary Historical networks 5 Company Board networks 6 Metabolic networks 7 Where does SNA come from? Mathematics: Graph Theory, Euler, 1735 Sociology: Sociometry, Moreno, 1930s Computer programs: UCINET, 1970s> Conferences: INSNA, 1970s> Social Network apps, 2000>, LinkedIn, etc, 10 June 2015Rick UNICEF, FLorence8 2. Five Arguments For The Use Of A Network Perspective (2003) 1.Social Network Analysis is about social relationship, so are most development programs 2.There is a range of methods for describing networks 3.There is a range of theories about social and other networks 4.Network can be seen, and analysed, at many scales 5.Networks are not linear 10 June 2015Rick UNICEF, FLorence9 Nov 2005Rick Davies at NDN10 Background meta-theory (2005) Better representations of program designs will lead to better programs Because Better quality program designs => More evaluable designs => Improved M&E of those programs => Program improvements Assumptions (2010) 1.Evaluation involves comparisons. For example, between plans and achievements 2.Better descriptions will enable more informed comparisons and better evaluations 3.Projects described in terms of networks of actors will be more useful than those described as abstract and disembodied processes of change 4.SNA provides a flexible array of tools for describing networks of actors 11 Caveats: Its not just about theory (2015) SNA provides particular kinds of data not easily available elsewhere Examination of well presented data can prompt improvements in theory Ideally, there is an ongoing dialogue between theory and data 10 June 2015Rick UNICEF, FLorence12 3. Ways of representing networks Diagrams Matrices Mathematical measures Data formats readable by software 10 June 2015Rick UNICEF, FLorence13 14 A matrix view of 16 NGOs in Ghana In social network analysis, this is a one-mode network, showing Actors x Actors Cell values = frequency of references to each other, in their Progress Reports between Cells can represent many other aspects of their relationships Matrices can hold data on thousands of relationships Up to 120 here But it is hard to read. Measuring Effectiveness, Melbourne, Sept The same data, presented in diagram form.. Weak links Strong links Reciprocated links Well connected actors Unconnected and marginalised actors Groups of actors Mathematical measures 10 June 2015Rick UNICEF, FLorence16 Individuals positions within a network e.g. Degree centrality Closeness centrality Betweenness centrality The structure of a whole network e.g. Density Network centralisation Cliques 10 June 2015Rick UNICEF, FLorence17 Types of matrices (&diagrams) Adjacency matrices Actors x Actors e.g. Ghana NGOs Affiliation matrices Actors x Objectives e.g. Ghana PRSP Which can be converted to. Actors x Actors (by shared objectives) Objectives x Objectives (by shared actors) 10 June 2015Rick UNICEF, FLorence18 19 GPRS top priority governance objectives and networks of networks Modular matrices e.g. Actors x Objectives + Objectives x Activities + Activities x Projects 10 June 2015Rick UNICEF, FLorence20 4. Sources of network data and their uses Funding agreements Progress reports Workshop participant lists Workshop planning exercises Workshop drawing exercises Card sorting exercises Online SNA surveys Household survey data WWW sources 21 October 2006 Rick Davies22 PETRRA funding relationships Visualising data from Progress Reports 23 July 2007UK Social Network Conference24 M4P project, Hanoi Nodes = workshops Links = movement of participants from one workshop to another Thicker link = more participants moving Node colour = type of workshop (by topic ) July 2007UK Social Network Conference25 MSC Training Workshop, Accra Nodes = workshop participants Links = intend to talk to each other after workshop, but had not worked together before. (=Old minus new links) Thick lines = reciprocated intentions Output to Purpose relationships 26 Sept 2005Rick Davies at EES SS27 Objectives x programs (Vietnam) Rick Davies, Uppsala, April 2008, Part 3 28 Network model based on discussion 29 Categorisation of districts in Indonesia Online survey of NGOs & Issue coalitions Key: Blue squares = coalitions; Red circles = NGO respondents; Purple triangles = NGO respondents who are also listed as coalitions; Grey lines = NGO membership in a coalition. Sensemaker self-signified stories 10 June 2015Rick UNICEF, FLorence31 Websites linked by hypertext 10 June 2015Rick UNICEF, FLorence32 5. Making sense of complex networks Aggregating Grouping actors into higher level categories Filtering by link, type, strength and other attributes Consider filtered networks as flooded landscapes By node type, degree and other attributes Clustering Using layout algorithms and cluster identification procedures 33 October 2006 Rick Davies34 PETRRA funding relationships October 2006 Rick Davies35 PETRRA funding relationships simplified 36 G-RAP Baseline survey circa 2006 Meeting participants 37 Orange = participating organisations Blue, red and green = meetings Links = meetings attended Meetings can be filtered out in temporal sequence 6. Causal perspectives on networks Actors attributes + their network position + wider network structure => outcomes Multiple pathways to the same outcome Impact mediated via x other actors Incomplete knowledge of network structrure 10 June 2015Rick UNICEF, FLorence38 Aspects of networks Links may have direction, or not Connections may or may not be reciprocated In-degree and out-degree may differ E.g. NGO and donor relationship Links may be singular or multiplex Distance is in links (degrees), not cm/inches Connected links create pathways Structure is important, so layout is important, to minimise cross-overs 39 10 June 2015Rick UNICEF, FLorence40 Pathways through networks A way of telling a simplified story, that Can be identified by one person, as they see it Or generated bottom up, by asking all to prioritise their relationships 41 11th June 2009Rick Davies at BOND42 Which can be captured in tabular form as a Social Framework, a Logical Framework re-designed as if people and their relationships mattered 9. Software for network visualisation and analysis Criteria Free User friendly manual On-screen network drawing is easy Imports and exports multiple formats Ability to display attributes by colour/size/shape Continuing to be developed Multiple layout options Multiple network measures 43 Who has used what software? Hands up exercise 44 NodeXL, an Excel Template Free, manual exists, continuing development You can give nodes and links as many different attributes as desired Including objective descriptions, results data And these can be visualised in the colour/size/shape of nodes and links 45 46 yED, diagramming software Free Wide range of ways of visualising differences in nodes and links Copes with Excel files that vary in their layout Diagrams can be uploaded as interactive web pages 47 48 UCINET & NetDraw The most widely used SNA package Lots of data manipulation and network analysis tools Lots of layout and filtering options Continually being developed Steep learning curve Courses available 49 50 Visualyzer Very easy on screen drawing User friendly manual Attributes of nodes and links can be made visible by mouse-over, clicks and coding Expensive Not under continuing development 51 52 The attributes of the actors can be made visible by cursor rollover+shift Sept 2005Rick Davies at EES SS53 A common problem? 10 June 2015Rick UNICEF, FLorence54 which can be dealt with by filtering By kinds of actors Types Centrality measures By kinds of relationships Types Strengths 10 June 2015Rick UNICEF, FLorence55