What is your talk about?
This seminar will illustrate various social network analysis (SNA) techniques and measures and their applications to research problems in education. These applications will be illustrated from our own research utilising a range of SNA techniques.
What are the key messages of your talk?
We will cover some of the ways in which network data can be collected and utilised with other research data to examine the relationships between network measures and other attributes of individuals and organisations, and how it can be linked to other approaches in multiple methods studies.
What are the implications for practice or research from your talk?
SNA is an approach that draws from theories of social capital to study the relational ties that exist between actors or institutions in a specific context. Such ties might include learning exchanges or advice-seeking interactions. SNA techniques allow researchers to incorporate the interdependence of participants within their research questions, whereas many traditional techniques assume our participants, and their responses to our questions, are independent of one another.
Social Network Analysis: applications for education research
1. Social Network Analysis:
applications for education
research
Dr Chris Downey and Dr Christian Bokhove
Southampton Education School Seminar Series
16th March 2017
The first 33 slides make up the main talk. The rest of the slides provide details
for each of the four projects. Slide 33 functions as a ‘table of contents.
3. Contents
• What is Social Network Analysis?
• Multilevel nature
• At the classroom level
– Dynamic SNA of classroom interactions
– Peer-status measures for social and learning
relationships
• At the institutional/system level
– Support networks of trainee teachers
– Teacher knowledge and resource exchange networks
4. Social Network Analysis
• Social network analysis (SNA) looks at social relationships
in terms of network theory, consisting of nodes,
representing actors within the network, and ties (or edges)
which represent relationships between the actors.
5. History
Originally the concept of ‘social networks’ has been studied
since the early 20th century to explore relationships between
members of social systems. In more recent years, social
network analysis has found applications in various academic
disciplines, as well as practical applications such as countering
money laundering and terrorism.
8. Freeman (2004)
Freeman (2004) reviewed the development of SNA from its earliest
beginnings until the late 1990s. He characterizes SNA as involving
four things
I. the intuition that links among social actors are important;
II. it is based on the collection and analysis of data that record
social relations that link actors;
III. it draws heavily on graphic imagery to reveal and display the
patterning of those links, and
IV. it develops mathematical and computational models to
describe and explain those patterns.
9. Fictional example
The application of SNA to classroom interaction is
demonstrated by the fictional network in figure 1 of one
teacher T01, and seven students S01 to S07, six nodes in total.
The nodes can have attributes, for example gender, which is
indicated by a colour (blue=female, pink=male).
12. Project 1
Dynamic SNA of classroom
interactions
Dr Christian Bokhove
Southampton Education School
13. Classroom observation
• Review classroom dialogue Howe and Abedin (2014)
– Quantitative vs Qualitative
• TIMSS (Trends in International Mathematics and Science Study) video
study (Hiebert et al., 1999)
– Video observations
– National patterns of teaching (Givvin, Hiebert, Jacobs,
Hollingsworth, & Gallimore, 2005)
• Lesson signatures
13
14. SNA for classroom interaction
• Case to use SNA for
classroom interaction
• Making it dynamic
– Classroom interaction
(Moody, McFarland,
& Bender-deMoll, 2005)
• Technological and methodological advances
– Observation apps
– Video recording easier
– Statistical techniques and packages to capture temporal aspects like
Gephi, ERGMs, Rsiena, Statnet, Relevent
15. This project
• Use dynamic social network analysis to describe classroom
interaction
• Data analysis and visualization software
– Gephi 0.8.2 beta
– R and Rstudio with the packages statnet (Handcock,
Hunter, Butts, Goodreau, & Morris, 2008) and ndtv
(Bender-deMoll, 2014)
18. Project 2
Peer-status measures for social
and learning relationships
Dr Chris Downey, Prof Daniel Muijs, Annie Brookman
Southampton Education School
EU Daphne III Project
Turning Obstacles into Opportunities – Early Interventions for
Developing Children's Bully Proofing Abilities
19. 19
Peer status
Establishes peer networks in a class (Coie and Dodge, 1982)
• Children make positive (‘Most Liked’) and negative (‘Least
Liked’) peer nominations of each of their peers in the class.
– Which children do you most like to play with in your
class?
– Which children do you find it hardest to play with in
your class?
• Children nominate up to 3 children in each category but
need not nominate at all
• Results are processed using some statistical analysis and
can be used to produce a social map of the class known as a
‘sociogram’
• http://www.sussex.ac.uk/Users/robinb/socio.html
20. 20
Peer-nomination form
Name: ________________________ School:_______________________
People you like
to play with
1.
2.
3.
People you find it
hard to play with
1.
2.
3.
People you think you
work well with
1.
2.
3.
People you find it
hard to work with
1.
2.
3.
23. Project 3
Support networks of
trainee teachers
Dr Christian Bokhove and Dr Chris Downey
Southampton Education School
24. Context
• Teacher training in UK
• PGCE
– Provider Led (PL)
– School Direct (SD)
– NQT
• Secondary Maths and Science
– cohort size (~35)
– Uni context
– longevity of course
27. Conclusions
• Views on support (SUPPORT), network intentionality
(NETWORK) and peer trust (TRUST) were quite trait-like
and did not change much.
• Self-perceived self-efficacy (DEVELOPMENT) increased
significantly over the four waves.
• Trainees did not develop significantly less or external ties,
but they did lose internal ties and subsequently an
increased EI-index . These changes, however, did only set in
after wave 2.
28. Project 4
Teacher knowledge and resource
exchange networks in schools
Dr Chris Downey
Southampton Education School
29. 29
Background
Case studies of two schools.
• judged to be outstanding by Ofsted
• also Lead Schools in a Teaching School Alliance
Cross sectional survey of all teaching staff.
Collected bounded whole networks of teaching staff.
During the last month, with who have you …
• exchanged teaching resources?
• developed your own teaching and learning?
• exchanged data about your students?
• evaluated the data about your students?
31. 31
DHT – “You think of something like MFL.
They are physically contained in one area,
one corner of a rectangle of our school and
also, by the nature of accessing their
course...”
AHT - “It’s also about other roles those
people have as well.
Secondary school –
teaching resource exchange
32. 32
HT – “You’re Mr Data really.”
AHT – “Too much
dependency on
one person”
Secondary school –
data collaboration
33. What now?
• Demo Gephi – software for SNA
• More details on one of the projects?
– Dynamic SNA of classroom interactions
– Peer-status measures for social and learning
relationships
– Support networks of trainee teachers
– Teacher knowledge and resource exchange networks
33
34. Exploring classroom
interaction with dynamic
social network analysis
Dr. Christian Bokhove
University of Southampton
SUNBELT XXXV
26th June 2015
35. Rationale
• Dynamic model (Creemers & Kyriakides, 2008)
– Multilevel: students in classrooms in schools
– Classroom interaction
• Social networks
– Actors and interactions
– Multidisciplinary (Freeman, 2004)
36. Classroom observation
• Review classroom dialogue Howe and Abedin (2014)
– Quantitative vs Qualitative
• TIMSS (Trends in International Mathematics and Science Study) video
study (Hiebert et al., 1999)
– Video observations
– National patterns of teaching (Givvin, Hiebert, Jacobs,
Hollingsworth, & Gallimore, 2005)
• Lesson signatures
36
37. SNA for classroom interaction
• Case to use SNA for
classroom interaction
• Making it dynamic
– Classroom interaction
(Moody, McFarland,
& Bender-deMoll, 2005)
• Technological and methodological advances
– Observation apps
– Video recording easier
– Statistical techniques and packages to capture temporal aspects like
Gephi, ERGMs, Rsiena, Statnet, Relevent
38. This project
• Use dynamic social network analysis to describe classroom
interaction
• Data analysis and visualization software
– Gephi 0.8.2 beta
– R and Rstudio with the packages statnet (Handcock,
Hunter, Butts, Goodreau, & Morris, 2008) and ndtv
(Bender-deMoll, 2014)
40. Data analyses
• Three data analyses approaches
– A: transcripts of TIMSS used ‘as is’ because low effort
with existing transcripts Gephi
– B: TIMSS videos re-observed
to get more detail Gephi, Rstudio (statnet and ndtv)
– C: Five observation of maths lessons in a secondary
school in the south of the United Kingdom Using
Lesson App, Gephi (incl. animations)
42. Two TIMSS lessons: US1 and HK1
• US1
– USA 8th grade
– Maths: graphing linear equations
– 44m, 36 students, mainly self work and private interaction
• HK1
– Hong Kong SAR 8th grade
– Maths: square numbers and roots
– 34m, 40 students, whole class first then exercises
42
43. Results – analysis B
HK1 US1
Nodes 44 35
Edges 51 95
Average degree 1.159 2.714
Average weighted degree 3.273 21.4129Duration of
interaction
45. Results – analysis C
Lesson R1 Lesson R4
Topic
Proportions Area of triangles
Year
Year 10 Year 7
Visualisation
Nodes (*)
16 25
Edges (**)
33 75
Degree
The size of the nodes indicates the total degree
Average
degree
2.062 3.0
Av.clust.coeff.
0.334 0.322
45
48. What might it tell us?
• Teacher student interaction
– Whole class, directionality
• Student interactions
– Groups and cliques
• Individual behaviour
– Help seeking
– Disturbances
– Central students
• Perhaps, patterns over classes, schools, countries (analogue TIMSS
video study)
49. Conclusions and discussion
• Proof of concept to capture classroom interaction technology useful
• SNA methods
• Longitudinal and temporal data can be modelled
• Challenges and limitations
– Quality of data (protocols)
– Capturing (all) interactions (and whole class?)
– Nature of the interactions
– Logistics and ethical with regard to video
– Complex character of analysis methods
– Interpretation
50. Future work
Use more advanced models
Mainly in R
• Temporal ERGM
• Rsiena
• R packages relevent (Butts, 2015) and observR (Marcum & Butts, 2015)
Aggregate data (multilevel modelling)
• Multiple lessons into a teacher or class profile
• Multiple classes/teachers into a school
• Multiple schools into countries
50
51. Question
• This was an example on classroom interaction. Can you
think of other examples in education. What do the nodes
denote? What do the ties denote?
52. Peer-status measures for social
and learning relationships
Dr Chris Downey, Prof Daniel Muijs, Annie Brookman
Southampton Education School
EU Daphne III Project
Turning Obstacles into Opportunities – Early Interventions for
Developing Children's Bully Proofing Abilities
53. 53
Data from teachers
Child Behaviour Scale (Ladd & Profilet, 1996)
•a measure of children’s aggressive, withdrawn,
and prosocial behaviors consisiting of 17
statements
•teachers respond with 1 = doesn’t apply, 2 =
applies sometimes, 3 = certainly applies
•two scales:
(i) aggressive with peers and (ii) prosocial with
peers
54. 54
Data from teachers
Example statements:
•Tends to react to classmates’ distress by teasing
them or making things worse
•Seems concerned when classmates are distressed
•Taunts and teases classmates
•Threatens classmates
•Is kind toward classmates
•Listens to classmates
•Compromises in conflicts with classmates
55. 55
Peer status
Establishes peer networks in a class (Coie and Dodge, 1982)
• Children make positive (‘Most Liked’) and negative (‘Least
Liked’) peer nominations of each of their peers in the class.
– Which children do you most like to play with in your
class?
– Which children do you find it hardest to play with in
your class?
• Children nominate up to 3 children in each category but
need not nominate at all
• Results are processed using some statistical analysis and
can be used to produce a social map of the class known as a
‘sociogram’
• http://www.sussex.ac.uk/Users/robinb/socio.html
57. 57
Peer-nomination form
Name: ________________________ School:_______________________
People you like
to play with
1.
2.
3.
People you find it
hard to play with
1.
2.
3.
People you think you
work well with
1.
2.
3.
People you find it
hard to work with
1.
2.
3.
65. 65
References
• Coie, J.D. and Dodge, K.A. (1982) Continuities and Changes in
Children's Social Status: A Five-Year Longitudinal Study, Merrill-
Palmer Quarterly, 29(3), 261-282.
• Ofsted (2007) Developing social, emotional and behavioural skills in
secondary schools: A five term longitudinal evaluation of the Secondary
National Strategy pilot, (London, Office for Standards in Education).
66. Mapping Changes in Support: A
Longitudinal Analysis of
Networks of Preservice
Mathematics and Science
Teachers
Christopher Downey
Christian Bokhove
Social Side of Teacher Education Symposium
AERA Annual Meeting – Washington, DC 8-12 April 2016
67. Context
• Teacher training in UK
• PGCE
– Provider Led (PL)
– School Direct (SD)
– NQT
• Secondary Maths and Science
– cohort size (~35)
– Uni context
– longevity of course
68. Role of networks
Liou, Forbes, Hsiao, Moolenaar & Daly (2013)
•Pre-service elementary school teachers - mathematics
– Trust and self‐efficacy are positively associated with
pre‐service teacher’s outcome performance on a
mathematics teaching assessment.
– The social network position of a pre‐service teacher is
also related to performance.
•Importance of support relationships as a buffer/resilience in
a pressured environment
Liou, Y. , Forbes, C. A., Hsuao, J. , Moolenaar, N. and Daly, A. J. , (2013) "Investing in Potential: Exploring Preservice Teachers’
Social Capital and Outcomes" Paper presented at the annual meeting of the UCEA Annual Convention, Hyatt Regency,
Indianapolis, IN Online <PDF>. 2014-12-10 from http://citation.allacademic.com/meta/p674423_index.html
69. Data
• General
– Basic demographic (sex, age)
– Programme of Study (subject, mode)
• Related factors
– Peer trust
– Self perception of development as teachers
– Views on support
– Network intentionality
• Peer-network (bounded whole networks for Ma & Sci)
• Wider network (external actors from different categories)
70. Approach
• Longitudinal - 4 ‘waves’ of data collection (every 2 months)
– PL and SD differences in programme structure
• Directed network question: “During the last month, to
whom have you turned for support?”
• Both instrumental and affective aspects of support
• Online questionnaire instrument
– shared instruments (San Diego & Barcelona)
71. Research question
RQ1: Are certain network characteristics (such as network
homophily, network intentionality, peer trust and views on
support) significantly associated with the growth in perceived
self-efficacy of these pre-service teachers?
RQ2: How do the support networks of trainee teachers vary
between Provider Led (PL) and School Direct (SD)
programmes?
73. Network intentionality
• 22 questions
• 5 point Likert scale
• Example question
– I attempt to connect to people who are prominent or central in the
course/at school
– I periodically evaluate the nature of my connections and networks
within the course/at school
78. Observations repeated ANOVA
• TRUST, NETWORK, SUPPORT constant
• DEVELOPMENT increased: F(1.900, 77.925) = 21.032,
p<0.001
• E not significantly different over waves: F(2.351,
119.884)=.908, p=.419
• I and EI were different over waves but not from wave 1 to
wave 2: F(2.521, 128.578)=22.238, p<.001 and F(2.389,
119.467)=17.589, p<.001
78
83. Conclusions
• Views on support (SUPPORT), network intentionality
(NETWORK) and peer trust (TRUST) were quite trait-like
and did not change much.
• Self-perceived self-efficacy (DEVELOPMENT) increased
significantly over the four waves.
• Trainees did not develop significantly less or external ties,
but they did lose internal ties and subsequently an
increased EI-index . These changes, however, did only set in
after wave 2.
84. Differences PL and SD
• Self-perceived efficacy as represented by DEVELOPMENT
between the two groups SD and PL differed: SD starts out
higher, but PL increases more from wave 1 to 4.
• <add some more points from last page paper>
84
85. Selected references
Bender-deMoll, S. 2014. ndtv: Network Dynamic Temporal Visualizations. R package version 0.5.1. [Software].
Available from http://CRAN.R-project.org/package=ndtv
Butts, C.T. (2015). relevent: Relational Event Models. R package version 1.0-4, URL http:
//CRAN.R-project.org/package=relevent.
Creemers, B. P. M., & Kyriakides, L. (2008). The dynamics of educational effectiveness: A contribution to policy,
practice and theory in contemporary schools. London: Routledge
Freeman, L. (2004). The development of Social Network Analysis: A Study in the Sociology of Science. Empirical
Press.
Gephi Consortium. (2014). Gephi (Version 0.8.2 beta) [Software]. Available from https://gephi.github.io/
Givvin, K.B., Hiebert, J., Jacobs, J.K., Hollingsworth, H., & Gallimore, R. (2005). Are there national patterns of
teaching? Evidence from the TIMSS 1999 Video Study. Comparative Education Review, 49(3), 311-343.
Handcock, M.S., D. Hunter, C. Butts, S. Goodreau, P. Krivitsky, S. Bender-deMoll, and M. Morris. 2014. Statnet:
Software Tools for the Statistical Analysis of Network Data. The Statnet Project. http://www.statnet.org. R package
version 2014.2.0.
Hiebert, J., Gallimore, R., Garnier, H., Givvin, K. B., Hollingsworth, H., Jacobs, J., Chui, A. M., Wearne, D., Smith, M.,
Kersting, N., Manaster, A., Tseng, E., Etterbeek, W., Manaster, C., Gonzales, P., & Stigler, J. (2003). Teaching
Mathematics in Seven Countries: Results from the TIMSS 1999 Video Study, NCES (2003-013), U.S. Department of
Education. Washington, DC: National Center for Education Statistics.
Howe, C., & Abedin, M. (2013). Classroom dialogue: A systematic review across four decades of research, Cambridge
Journal of Education, 43(3), 325-356.
Marcum, C.S., & Butts, C.T. (2015). Constructing and Modifying Sequence Statistics for relevent Using informR in R.
Journal of Statistical Software, 64(5).
Moody, J., McFarland, D.A., & Bender-deMoll, S. (2005). Dynamic network visualization: Methods for meaning with
longitudinal network movies. American Journal of Sociology, 110, 1206-1241.
86. Utilising social network
approaches to determine the
roles of teachers within key
resource-sharing networks
in schools
Dr Chris Downey
Associate Professor in Education
Southampton Education School
ICSEI 2016 Glasgow 8th July 2016
87. 87
Background
Case studies of two schools.
• judged to be outstanding by Ofsted
• also Lead Schools in a Teaching School Alliance
Cross sectional survey of all teaching staff.
Collected bounded whole networks of teaching staff.
During the last month, with who have you …
• exchanged teaching resources?
• developed your own teaching and learning?
• exchanged data about your students?
• evaluated the data about your students?
88. 88
Network graphs - key
• each square/shape is a
teacher
• Each nomination is
represented by an arrow
(tie)
• A reciprocated nomination
is represented by a double-
headed arrow
• The size of each square
indicates how sought after
the teacher is for the
resource in question (in-
degree).
All classroom teachers and leaders in each
school were asked to nominate those
colleagues with whom they had engaged
over the previous month in four areas of
practice; two related to learning and
teaching and two related to data use.
92. 92
Primary school –
learning & teaching collaboration
“Someone told me it was because I am
approachable. I think it’s also because
I’m a classroom teacher”
Why do people come
to you?
“I was an Advanced Skills Teacher and now I’m a Specialist Leader of
Education. Next year I will be given a role out of the classroom.
I go to visit other schools and when I see new ideas I bring them back
here and share them.
“Sometimes I come back and kiss the ground and realise how grateful I
am to be working in a school like this.”
Av deg 4.6 Andy – outdegree 8 (3rd); indegree 10(1st =); betweeness 2nd
P4C - “That’s how it works. You find something and research it and 9
times out of 10 she will say yes”
“I have asked for two half days to keep
me grounded in the classroom”
How will that work
in the new role?
93. 93
Primary school –
data collaboration
“There are two key
people I go to, and we
all go to, in making
sense of the data”
“They are the people
that have the know-how
to make sense of the
data”
“Even in this school we have our core people who are
familiar with the data and after that it falls off”
95. 95
DHT – “You think of something like MFL.
They are physically contained in one area,
one corner of a rectangle of our school and
also, by the nature of accessing their
course...”
AHT - “It’s also about other roles those
people have as well.
Secondary school –
teaching resource exchange
96. 96
DHT - “The large red block, blue block and the
grey block we would hope are people one our T&L
steering group…
That’s what we would want it to be”
Secondary school –
teaching resource exchange
97. 97
AHT – “We’d expect [teachers A, B & C] to be there.
Someone like [teacher D] would be increasingly in the middle], more over
last year.”
DHT - “ And also [teacher E]
Secondary school –
learning & teaching collaboration
98. 98
HT–“Now that is really encouraging... Very encouraging. That has been totally
intentional.
DHT–“The Lead Teacher idea started… because we wanted to spread good
practice more widely than by just having ASTs. We grew them didn’t we?”
HT–“They were identified and promoted through as lead professionals.”
AHT-“These are people who we have identified as exemplary teachers that
also have a certain skill.”
Someone like myself would be well recognised as very, very good teachers but
don’t necessarily have that transferable skill…They have trainability. They
provide good quality training resources. They are also very accessible by the
nature of the people they are and we have grown them because they are the
people who we would hope would act as hubs.”
Secondary school –
learning & teaching collaboration
99. 99
AHT – “This doesn’t surprise me… [science have] an internal data system
they put out there. ‘We have our own system, we set up our [data type] in a
certain way’.
This makes sense to me. Maths share their own data, ‘Because our data
doesn’t make sense to other people’.
Secondary school –
data exchange
100. 100
AHT – “The fact that this cluster sits together it kind of aligns well with that
doesn’t it? I’ve been in a 2.5 hour meeting where we’ve discussed 20 odd kids
moving in the literacy groups poring through assessments and sheets.”
“We’ve made an ueber faculty in a very loose sense. They’ve become cousins
in a funny old way. We’ve almost forced a link with English.”
Secondary school –
data exchange
101. 101
AHT – “I don’t think they should interplay for a funny old reason, but most
people who are slightly detached from teaching and learning think they
should interplay.
“That pedagogical discussion, why does it need to have any form of data
as a basis to it? Is it an art or a science, and it’s an art really isn't it? You
kind of feel your way around pedagogy don’t you? ...The data signposts
something.”
“If you look at our more able groups, we’ve never discussed the data of
children once, and I think that’s a healthy thing… The networks should
relate but I wouldn’t want to see them overlap.”
Secondary school –
data collaboration
102. 102
HT – “You’re Mr Data really.”
AHT – “Too much
dependency on
one person”
Secondary school –
data collaboration
104. AHT – “The red that that is English is right in the centre of the diagram. So
pedagogically speaking that’s a hub in itself. We set up literacy as a key
dimension here. This suggests that our literacy programme is at the centre
of all that we do. That’s why we get interplay between these other
subjects…If you had done this 3 years ago they would actually have been
separate hubs…A couple of years ago you would probably have seen 6 or
7 people interplaying…
This is the extraordinary bit for me. It looks to me like an intended
consequence of something we chose to do”
HT – “For me the point is we set out to do something and it looks as
though we might be achieving it to some degree… I really am encouraged
through…I was always taught that structure follows strategy. That’s why
we are where we are”
Me - “This is your staff telling you what the structure for teaching and
learning is in your school. If it matches your intention I think that’s really
quite something.”
104
Global thoughts on networks…