Children and Youth Services Review 85 (2018) 143–150

Contents lists available at ScienceDirect

Children and Youth Services Review

journal homepage: www.elsevier.com/locate/childyouth

Analysing social networks for social work practice: A case study of the T
Facebook fan page of an online youth outreach project
⁎
Chitat Chan
Department of Applied Social Sciences, The Hong Kong Polytechnic University

A B S T R A C T

This study adopted social network analysis (SNA) to interpret a Facebook fan page used by a youth outreach
project in Hong Kong. The study aimed to: i) explore what types of open data can be captured from social media,
ii) explore in what ways SNA metrics can be used to generate meaningful indicators that help track relationship
changes, and iii) examine to what extent SNA findings agree with social workers' perceptions. These results
preliminarily suggest the feasibility and validity of SNA in social work practice research. This study adds to the
emerging knowledge base of how practitioners and researchers might use SNA to assess social networks, sup-
plement need assessments, track service users' behavioural changes, and evaluate services.

If one of the core missions of social work is to facilitate individuals 2017; Chan & Holosko, 2016; Leung et al., 2017; Oliver et al., 2015), or
to have more effective interactions with resources and social systems content analysis of communications on social media pages (Lee &
(Holosko, 2017; Moore, 1990), social media may likely offer new op- Suzanne Horsley, 2017; Oliver et al., 2015). Compared with other more
portunities for the profession (Chan & Holosko, 2017). Nonetheless, this traditional types of technology-supported intervention studies (e.g.,
potential is yet to be verified. online training) social media based studies have much less experi-
Social media refers to a set of online tools that support social in- mental designs (Chan, 2016). Qualitative case studies have offered
teraction between users, and cover a range of interactive Internet-based unique insights that have contributed to social work knowledge de-
applications such as YouTube, Facebook, Instagram, LinkedIn (Hansen, velopment, but the lack of empirical evidence has partly limited what
Schneiderman, & Smith, 2011; Kaplan & Haenlein, 2010). The term these studies can say.
social media is often used to contrast with more traditional media such There are some common hurdles in using social media data to
as television and books that deliver content to mass populations but do evaluate interventions. First, social media is a third-party platform,
not facilitate the creation or sharing of content by users (Hansen et al., consequently researchers as they do not have control of the interface
2011). design, the algorithm, and the data (Chan & Holosko, 2016). Algorithms
Research has been conducted on the use of social media in social are designed by people, and they only do what they are designed to do
work, outlining the increased use in direct practice (Best, Manktelow, & (Godinho et al., 2017), but social workers usually are not aware of that
Taylor, 2014; Chan, 2016; Goldkind, 2015; Valimaki, Athanasopoulou, and they may not be in a position to negotiate this with social media
Lahti, & Adams, 2016). There are healthcare interventions using social companies. Second, workers need to consider ways in which this data
media (Welch, Petkovic, Pardo, Rader, & Tugwell, 2016), but in social can be ethically collected and used. In some countries, such as the USA,
work, social media has been shown to be mainly used for service user interventions conducted via social media platforms (e.g., Skype) may
engagements (Chan & Holosko, 2016) or for sharing service information not be covered by insurance (Reamer, 2015), and this may affect
(Fitch, 2007). A 2016 scoping review reported that there are a limited practitioners' incentive to adopt social media in clinical interventions.
number of studies that posit social media as a medium to implement Third, social work does not have appropriate conceptual tools and re-
social work interventions, need assessments or evaluations, and most of search instruments for measuring social networks and interactions. In-
them reported using social media as an effective means to engage ser- tervention research in social work mainly adopts subjective measures
vice users (Chan, 2016). relying on self-reporting scales/instruments (Fraser, 2004; Holosko,
Among social work intervention studies adopting social media as a 2016), objective measures tracking behaviours, relationships or social
medium of intervention, most are qualitative studies exploring social networks are not as developed as these, and hence this has limited the
workers' or service users' experiences (Ballantyne, Lowe, & Beddoe, choices of research methods.

⁎
The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
E-mail address: [email protected].

https://doi.org/10.1016/j.childyouth.2017.12.021
Received 6 September 2017; Received in revised form 15 December 2017; Accepted 18 December 2017
Available online 19 December 2017
0190-7409/ © 2017 Elsevier Ltd. All rights reserved.
C. Chan Children and Youth Services Review 85 (2018) 143–150

Nevertheless, there are emerging studies which have partly over- Hong Kong (Wong, 2009). A closely related term often used in English
come these hurdles. For example, researchers have captured social speaking regions is NEET (Not in Education, Employment, or Training).
media data within the permitted parameters via the application pro- However, the meaning of hidden youth covers a wider scope, as some of
gramming interfaces (APIs) of social media platforms, and have used the hidden youth do have some short term trainings or employments.
such data to identify users' demographic data (Lewis, Kaufman, Those with less education and fewer economic resources are at a higher
Gonzalez, Wimmer, & Christakis, 2008), analyse group dynamics risk of being marginalized from traditional employment and main-
(Wölfer, Faber, & Hewstone, 2015), identify significant members in stream society.
social movements (Chan, 2013), and analyse social trends (Del Vicario, The Infinity Teens project has adopted an approach that enables
Zollo, Caldarelli, Scala, & Quattrociocchi, 2017). individuals to reach their potential, and facilitates effective interactions
A core technique behind these emerging social media studies is between individuals and local social systems. Under this service fra-
social network analysis (SNA), which is derived from the concept of mework, all cases are initially engaged via the Internet. In addition to
sociogram – a representation of social links. A sociogram consists of two using a project website, a range of social media platforms (e.g.,
primary building blocks: nodes and edges. Nodes (also called vertices, Facebook fan pages) are also used to identify and attract potential cli-
entities or items) can represent many things such as people, groups or ents. Practitioners promote their full worker identity on their social
even countries. Edges (also known as links, ties, connections or re- media account profiles. These workers might have their own personal
lationships) are the building blocks of networks. An edge connects two profiles, but that these working profiles reported in this study were
nodes together. Edges can represent many different types of relation- distinct from their personal profiles.
ships such as collaborations, kinship, friendship or citations. Based on Rather than from a particular physical catchment area, all service
these sociogram concepts, there are various metrics used to measure the users from all districts in Hong Kong are welcome. After engaging the
weighting or connectedness of individuals in a network, such as the clients online, practitioners assess the clients' needs and suggest a
number of in-coming connections, out-going connections, or the total variety of online and/or offline support services, which can include
number of people connected (Hansen et al., 2011). Further details social activities, training, job referrals and/or counselling.
about SNA metrics will be introduced in the Method section. As of December 31, 2016, the project had engaged 200 cases, all
In fact, social scientists have long proposed metrics aiming to cal- were directed from the Internet including the project website, apps,
culate network developments and individual weightings in social net- Facebook, Instagram and instant messaging apps. It should be noted
works (Freeman, 1977), but these metrics have not been widely applied that not all the fans on the fan page were cases. Cases were opened
until recently in social work or social sciences (Hansen et al., 2011). based on mutual agreements. The fan page users covered in this study
The development of various types of social media platforms has resulted were the online users “engaged” in their outreach activities. They did
in a large amount of structured data becoming available to the public not necessary become “cases”. The social workers on the project fre-
(Coulton, Goerge, Putnam-Hornstein, & de Haan, 2015). In conjunction quently use the project's Facebook fan page to promote their services, to
with the research tools currently available to collect, analyse, and vi- share relevant information, and to interact with Internet users. They
sualize social media data, new insights can be revealed (Hansen et al., might not work the same amount of time and in the same way. They
2011). If these new measurements can quantify and measure changes of posted weekly, but there was no maximum limit. In addition to posting
social network over time, bivariate and/or multivariate analyses can be or sharing relevant information and news, the social workers often use
applied, particularly useful for intervention studies. Facebook Live to discuss various issues and topics. On the project's
Against this background, we contend that SNA may possibly provide Facebook fan page, Internet users can access online self-assessment
new opportunities for social work intervention research. This study resources and find a way to seek online counselling services. Internet
aimed to depict the social network development of a Facebook fan page users can also make an appointment with social workers online. As of
used by a youth work service in Hong Kong, with special reference to June 10, 2017, there were 4497 Facebook users who liked the Infinity
the following questions: i) What types of data can we ethically get from Teens Facebook fan page and 4392 users who followed the fan page.
social media? ii) What analytical tools can we use to generate mean-
ingful indicators that help track changes? iii) To what extent the results 2. Method
of SNA metrics agree with social workers' perceptions? Validation is
both useful and essential, because it can tell us whether SNA findings 2.1. Overall study design
are trustworthy, therefore researchers and practitioners can use it to
triangulate findings from other data sources, and assess social media This study basically adopted an SNA approach, with special re-
networks without relying on network members' self-reporting. ference to the following analytical components. First, SNA software was
used to capture data allowed by social media platform. Social media
1. The case: the Facebook page of infinity teens offer different privacy settings, for example, there are private groups,
private communications, open group, public group, and public fan
This study analysed the Facebook fan page activities of the Infinity page, implying different levels of research subjects' consents. This study
Teens project (https://www.facebook.com/caritas.infinityteens/) op- only aimed to capture publicly available information supported by the
erated by Caritas Hong Kong, which is a local charitable organisation application programming interface (API) of Facebook. API is a set of
running various services in Hong Kong, such as social services, educa- protocols, and tools for building application software for communica-
tional services and medical services. tion between various software components. In this study, NodeXL was
The Infinity Teens project is one of three online youth service pro- used to import users' information supported by the API of Facebook.
jects operated by local NGOs with support from the Social Welfare Second, this study adopted social networker metrics supported by
Department (SWD) operating since 2011. These projects had been in- the SNA software, particularly the measure of Betweenness Centrality,
itially evaluated in 2015 and had received further support from the which measures the centrality of individual network members. The
Hong Kong Jockey Club Charities Trust in 2016 to run for a further study aimed to explore to what extent this measure can systematically
three years (SWD, 2011). These projects aim to serve young Internet indicate the centrality of network members over time.
users aged under 24 in Hong Kong, particularly those hidden youth who Third, a simple validation study was conducted with the objective to
may be at risk of socio-economic exclusion. The term hidden youth examine to what extent SNA findings reflect “reality”. There is thus far
means those young people who are socially-withdrawn, marginalized no known commonly accepted method of validating social media me-
and prone to emotional disturbances, which was a phenomenon ori- trics. Currently, there are research studies that examine to what extent
ginally recognized in Japan (called Hikikomori) and then spread to social media information corresponds to real life situations, for

144
C. Chan Children and Youth Services Review 85 (2018) 143–150

example, whether social media users' close friends on social media are two consecutive commenters, d) comment author and reactor/
also close friends in real life (Bond et al., 2012) or whether social media commenter/sharer, e) post/comment author and user tagged.
users' online expressions correspond to what they say offline (Hampton
et al., 2014). Similarly, this study examined to what extent results from 2.3.2.3. Connected components. A connected component is a set of
SNA metrics agree with social workers' perceptions in selected domains. nodes that are connected to each other but not to the rest of the
These include: their assessment of fan page content popularity, their graph. The number and size of connected components in the graph can
assessment of overall social network development, and their assessment help reflect subgroup structures.
of network members' significance. Of course, this is a version of “rea-
lity” perceived by the social workers. Nonetheless, this is the first such 2.3.2.4. Single-node connected components. This refers to the number of
validation study in the social work literature. connected components that have only one vertex. This variable can
reflect isolated members in a network.
2.2. Data collection

2.3.2.5. Graph density. This is a ratio that compares the number of

2.2.1. Collecting social media data
edges in the graph with the maximum number of edges the graph would
The analysis measured the connectedness of fan page members in
have if all the nodes were connected to each other. Duplicate edges and
the Infinity Teens Facebook fan page (https://www.facebook.com/
self-loops are excluded. This variable can represent the extent to which
caritas.infinityteens/) in 2016. Data concerning fan page activities
network members are connected to each other.
were collected over four time points in 2016 in March, June, September
and December. Data collection and analysis were completed according
to the parameters set by Facebook. The data import, analysis, clustering 2.3.3. Network members' significance
and graphical presentation were conducted using NodeXL (http:// 2.3.3.1. Betweenness Centrality (BC). This is a quantified measure of the
nodexl.codeplex.com/) (Hansen et al., 2011), which is an open-source centrality of a member (node) in a particular network based on
extension of MS Excel that is increasingly used in social media studies calculating the paths passing through a node (Freeman, 1977; Hansen
(Chan, 2013; Doran, Doran, & Mazur, 2011; Hansen, 2011). The tech- et al., 2011). A path is measured as the distance from a member to
nical details of the software will not be illustrated here due to the another member in terms of the number of hops. The BC of a member in
nature and word limit of this article. a network is the total number of shortest paths from all members to all
others that pass through that member. The greater the BC of a member,
2.2.2. Collecting social workers' views the greater the number of other members relying on that member. For
There were five social workers on the Infinity Teens project who example, for calculating individual members' centrality in a
collaboratively managed the fan page. They were counted in the BC communication line having 5 nodes A → B → C → D → E, we need to
index, the solid squares in the revised socio-grams indicate workers' take every pair of nodes, and see which node(s) a path between the two
account. All of them were full time social workers who dedicated to this must pass through (see Table 1).
project. The study invited each of them to complete a short online In this example, the centrality value for each node will be as follows:
questionnaire, which covered some definite statements requiring them A = 0, B = 3, C = 4, D = 3, E = 0. Based on these numbers, we know
to indicate their levels of agreement using a 5 point Likert scale (From that node “C” is the most influential one because most paths between
1 = Strongly disagree to 5 = Strongly agree). The average ratings of other nodes have to pass through it. In a complex network, the BC of a
these statements were assumed to represent the views of the team. node (n) is given by the expression: ∑σst (n)/σst, where σst is the total
number of shortest paths from a node to another node and σst (n) is the
2.3. SNA Metrics number of those paths that pass through that node. This metric,
therefore, helps represent the importance (in terms of centrality) of
2.3.1. Fan page content popularity individual members in a network.
A simple test was used to identify whether posts about online ac-
tivities or posts about face-to-face activities were more popular. All 2.4. Social workers' perceptions
posts on the fan page within the study period were exported, the
number of likes received by each post were counted, and the top 10% of 2.4.1. Fan page content popularity
posts receiving the most likes were identified. These were then manu- Social workers indicated their levels of agreement with the state-
ally categorised into two broad categories: posts about face-to-face ment: Posts about face-to-face activities (e.g., news about face-to-face
activities (e.g., news about face-to-face gatherings, photos after activ- gatherings, photos after activities) are more often ‘liked’ by users than
ities) and posts about online activities (e.g., online voting, online posts about online activities (e.g., online voting, online counselling
counselling services). There could be different ways to categorize the services) (Q1). The objective was to examine to what extent social
posts (e.g., activity objectives, time), but this is one of the objective and workers' perceptions regarding the popularity of the fan page was in
commonly shared marker that can differentiate the nature of all posts. line with the statistics from SNA.
These two categories were adopted in order offer a convenient binary
concept to differentiate different types of posts. The total number of Table 1
“likes” of these two types of posts were identified and compared. An example of calculating the centrality of a node in a network.

Paths of a pair of nodes Node (s) between the two ends of the path
2.3.2. Overall social network development
2.3.2.1. Total number of nodes. Nodes are service users, and the total A→B None
number of nodes in the graph is the total number of service users A→C B
A→D B, C
engaged via the fan page. A→E B, C, D
B→C None
2.3.2.2. Total number of edges. An edge connects two nodes together. B→D C
The total number of edges in the graph is the sum of all edges between B→E C, D
C→D None
all nodes. In this study, an edge is counted between two users whenever
C→E D
there is a relationship between: a) post author and reactor/commenter/ D→E None
sharer, b) users who reacted to/commented on/shared the same post, c)

145
C. Chan Children and Youth Services Review 85 (2018) 143–150

2.4.2. Overall social network development discs represent the 10 users (U1-U10) reported in the validation test. As
Social workers indicated their levels of agreement with several indicated in the four diagrams, the workers were in different subgroups
statements, including: Q3. The number of the project's Facebook fans in the first quarter (Jan to Mar 2016), and they gradually merged in a
has increased over the year; Q4. The number of relationships/connec- large core group (G1) over the year, and the biggest square, W1, was
tions between the project's Facebook fans has increased over the year; the official project account. The 10 users were in different groups in the
Q5. On the project's Facebook fan page, all your fans have been in- first quarter (Jan to Mar 2016), and U5, U6. U7, U8, U9, U10 were
creasingly connected with all other fans over the year, and; Q6. On the gradually absorbed into the central core group (G1).
project's Facebook fan page, subgroups have become more obvious over
the year, fans in the same subgroup are closely connected with each 3.2. Social workers' perceptions
other, but fans in different subgroups are just loosely linked or unlinked
with fans in other subgroups. All five social workers from the Infinity Teens project team (i.e.,
100% of the total number of social workers in the project) managing the
fan page responded to the online questionnaire, validating the results
2.4.3. Network members' significance (centrality)
computed by SNA. Average ratings of the levels of agreement indicated
The full definition of BC will confuse workers in the questionnaire,
by all the social workers (N = 5) were indicated beside each item (see
and therefore it was not used. Using the closest non-technical wordings,
Table 2). Key observations are as follows:
the meaning of BC is “widely connected”. In the questionnaire, social
First, social workers' perceptions regarding the tendency of page
workers were invited to indicate their level of agreement with the
content popularity was in line with the results from SNA. There were
statement: This Facebook fan page user [name] is widely connected
400 posts on the fan page. The analysis of the top 10% of posts liked by
with other Facebook fan page users (Q2) in regard to 10 randomly
users indicated that posts about face-to-face activities (e.g., news about
selected fan page users. Ten were selected in order to offer a straight-
face-to-face gatherings, photos after activities) were more often fa-
forward comparison showing to what extent workers' rating are in line
voured by users than posts about online activities (e.g., online voting,
with SNA findings. In other words, how many answers out of ten are in
online counselling services). This finding is in line with the social
line with SNA findings. Among these 10 randomly selected users, five
workers' rating (3.2/5.0), in which social workers tended to agree that
were among the top 10% having the highest BC, and five were among
posts about face-to-face activities are more often liked by users than
the lowest 10% having the lowest BC. The sequence of these 10 users
posts about online activities (see Table 2).
were randomly listed on the questionnaire. This analysis examined to
Second, social workers' perceptions of the overall development of
what extent users with high BC scores would be considered by workers
the fan page network was also in line with the SNA results. Both the
as widely connected with other users; and to what extent the users with
number of the Facebook fans (from 207 to 1458 users, i.e., an increase
low scores were considered by workers as not widely connected with
of over 600%) and the number of relationships/connections between
other users.
the Facebook fans increased over the year (from 17,247 to 134,230,
representing an increase of almost 700%). This is in line with the social
3. Results workers' subjective rating (4.2/5.0 and 3.8/5.0 respectively), which
indicated that social workers agreed that both the number of the
3.1. The SNA Findings Facebook fans and the number of relationships/connections did in-
crease over the year. In addition, when asked whether their fans have
SNA metrics including the total number of nodes, total number of been increasingly connected with all other fans over the year, social
edges, number of connected components, number of single-node con- workers tended to disagree (2.6/5.0). The SNA findings did show that
nected components, graph density, and BC at the four different time overall density of these connections decreased (from 0.08 to 0.04, re-
points were calculated (Table 2). These figures were visualised into four presenting a decrease of 50%), which indicated that as the group be-
sociogram graphics representing network structures in four different came bigger, the connections among members became irregular with
time points (Figs. 1, 2, 3, 4). The nodes are represented as discs, and the some only loosely connected. Further, the SNA findings showed that
size of a node is proportional to the significance of that node (in terms subgroups became more obvious, with larger ones becoming bigger and
of BC). The edges are represented as lines. Subgroups are identified smaller ones decreasing in size. Most fans in the network connected to
based on common connections. The solid squares represent workers' the same set of worker accounts, suggesting that there was a single
accounts (W1-W7). The circles represent user accounts, and the solid leadership centre. This is in line with the social workers' rating (3.6/
5.0) that they believe the connections among members became more
Table 2 concentrated, and subgroups were getting smaller over the year.
SNA findings at different time points. Third, social workers' assessments of the significance of individual
members based on their daily contact experiences were in line with the
Mar 2016 Jun 2016 Sep 2016 Dec 2016
SNA results. That is, all the selected users belonging to the top 10% BC
Total number of nodes 207 962 1324 1458 group were considered by social workers as widely connected with
Total number of edges 17,247 105,708 121,471 134,230 other users; and all the selected users belonging to the bottom 10% BC
Connected components 1 1 3 3
group were considered by social workers as not widely connected with
Single-node connected 0 0 1 1
components other users (see Table 3). A simple correlation test indicated the BC
Graph Density 0.08 0.07 0.04 0.04 scores of the 10 selected users and the social workers' average ratings
BC of User 1 381.46 13,865.89 43,898.41 46,591.65 were highly correlated (r = 0.93), though the sample size was not large
BC of User 2 NA 14,801.69 28,971.02 45,255.70 enough to claim statistical significance.
BC of User 3 NA 7852.20 35,578.74 36,561.89
BC of User 4 1101.05 20,510.41 23,128.39 29,264.92
BC of User 5 0.00 156.02 13,951.18 28,304.38 4. Discussion
BC of User 6 NA NA 0.00 0.00
BC of User 7 NA 0.00 0.00 0.00 This article particularly aims to illustrate how SNA can be used for
BC of User 8 NA 0.00 0.00 0.00
social work practice purposes. The findings basically showed that: i)
BC of User 9 NA 0.00 0.00 0.00
BC of User 10 NA 0.00 0.00 0.00 SNA software can capture publicly available information supported by
the API of social media, ii) BC and other SNA measure can system-
Notes: BC = Betweenness Centrality. atically show the change of network behaviours over time, iii) SNA

146
C. Chan Children and Youth Services Review 85 (2018) 143–150

Fig. 1. Network structure in March 2016.

findings can reflect network situations perceived by network co- interpretations. Therefore, this study adds to the emerging knowledge
ordinators (social workers who run the fan page). Major implications base of how practitioners and researchers might use SNA to support
for practice and research are as follows: longitudinal studies of groups/networks, to support cross-sectional
studies between different groups/networks, and compare different
4.1. Using SNA to interpret social networks working strategies.
It is worth noting that there are studies questioning to what extent
This analysis showed that SNA results largely agree with social findings from social media can accurately represent an individual's view
workers' ratings in their assessments of fan page content popularity, of the real situation. Fake information can be generated by humans or
overall social network development, and network members' sig- robots (Allcott & Gentzkow, 2017; Bessi & Ferrara, 2016), and some
nificance. These preliminarily results suggest the validity of SNA in researchers noted that people are less likely to express their true views
social work research, and imply that SNA can be used to estimate the on social media than in person (Hampton et al., 2014). However, many
situation of an online social network without relying on social workers' of these SNA limitations do not necessarily apply to social work

Fig. 2. Network structure in June 2016.

147
C. Chan Children and Youth Services Review 85 (2018) 143–150

Fig. 3. Network structure in September 2016.

research. While many social media studies predict social trends based 4.2. Using SNA to assess changes in service users' social networks
on information from anonymous individuals, social work researchers
and practitioners usually assess the behaviours of known individuals. As In the Introduction of this article, it is noted that the methodologies
such, social work researchers may not necessarily view SNA findings as of studies researching social media use in social work were relatively
facts reflecting the reality experienced by unknown users, but rather as limited compared to studies investigating other forms of information
a window revealing a particular version of the reality experienced by and communication technology (Chan, 2016). This is partly because of
service users, which may help triangulate findings from other data a lack of objective measures assessing behavioural changes (Chan,
sources. 2016). Social work interventions generally refer to activities purposely
designed to produce changes, and intervention research is the sys-
tematic study of purposive change strategies (Fraser & Galinsky, 2010;

Fig. 4. Network structure in December 2016.

148
C. Chan Children and Youth Services Review 85 (2018) 143–150

Table 3
Social workers ratings in relation to content popularity, network development, and individual members' significance.

Items Average rating (over In agreement with SNA

5) findings

1. Posts about face-to-face activities (e.g., news about face-to-face gatherings, photos after activities) are more often 3.2 YES
liked by users than posts about online activities (e.g., online voting, online counselling services)
2.1. User 1 is widely connected with other fan page users 4.2 YES
2.2. User 2 is widely connected with other fan page users 4.4 YES
2.3. User 3 is widely connected with other fan page users 3.2 YES
2.4. User 4 is widely connected with other fan page users 4 YES
2.5. User 5 is widely connected with other fan page users 3.2 YES
2.6. User 6 is widely connected with other fan page users 2.4 YES
2.7. User 7 is widely connected with other fan page users 2.6 YES
2.8. User 8 is widely connected with other fan page users 2.4 YES
2.9. User 9 is widely connected with other fan page users 2.4 YES
2.10. User 10 is widely connected with other fan page users 2.6 YES
3. The number of fans has increased over the year 4.2 YES
4. The number of relations/connections of fans has increased over the year 3.8 YES
5. ALL fans have been increasingly connected with ALL other fans over the year 2.6 YES
6. Subgroups have become more obvious over the year, fans in the same subgroup are closely connected with each 3.6 YES
other, but fans in different subgroups are just loosely linked or unlinked with fans in other subgroups

Notes: BC = Betweenness Centrality; Average ratings of the levels of agreement indicated by the N = 5 workers, using a 5 point Likert scale, in which 1 = Strongly Disagree, and
5 = Strongly Agree.

Rothman & Thomas, 1994). In general, intervention research should Newton, 2014). There are also well-known commercial examples that
have some sort of pretest-posttest design, and behavioural change use social networking methods to meet various types of needs and de-
should be assessed over time (Holosko, 2016). mands, such as online dating sites for specific sexual orientations (e.g.,
This study showed that it is possible to evaluate network behaviours MyLadyboyDate), transportation services (e.g., Uber), or short-term
over time. SNA provides quantified metrics that help track individuals' accommodation services (e.g., Airbnb). Similarly, social networking
behaviours in a network, such as the number of individual nodes, the methods may contribute to social work, as the very nature of social
number of edges associated with individuals, and BC. The analysis in- work focuses on both enabling individuals to better reach their poten-
dicated that researchers did successfully track the results of these me- tial and on facilitating more effective interactions with local social
trics over time (See Table 2), and these metric results are in line with systems (Holosko, 2017; Moore, 1990), social networking can provide a
practitioners' perceptions. range of resources for service users to experiment new roles and op-
As such, online behaviours can be quantified and tracked, and they portunities. For example, in the Infinity Teens project researched in this
can be compared with other quantified variables in experimental de- study, practitioners engaged participants online, they assessed their
signs. Therefore, bivariate analysis and multivariate analysis become needs and suggested a variety of online and/or offline support services,
possible, and it is possible to examine to what extent service users' self- such as counselling, social activities, and job referrals.
reporting outcomes at different time points correlate with various on- In some way, the insights from this study are nothing new. Social
line behaviours and/or network events. work has long been advocating such an idea. Based on a person-in-
environment principle, social work scholars see that the purpose of
social work is to improve the quality of transactions among people and
4.3. Concluding remarks: social networking for social work practice
their physical and social environments (Mattaini & Huffman-
Gottschling, 2012). Social work has long emphasized the role of social
This study has some limitations. First, the case studied here is an
network systems in assessments and interventions. Such system per-
online youth outreach project, which is not (yet) a practice method
spectives are almost universally accepted in social work (at least on
widely adopted in social work, and therefore the generalisability of the
paper), as they provide a framework for exploring and intervening in
findings is limited. Second, SNA applied to one platform may not work
such networks in their complexity (Mattaini & Huffman-Gottschling,
equally well in another platform, since different social media platforms
2012). This orientation is also important in making social work dif-
support a different set of networking capabilities. Third, this study
ferent from other intervention-oriented disciplines, which either focus
merely examined to what extent SNA agreed with social workers' per-
on social functioning (e.g., counselling, clinical psychology, psychiatry)
ceptions, but not service users' perceptions. Notwithstanding these
or social change (e.g., social activism, social revolution).
limitations, this study has explored analysis methods that have not
Unfortunately, such system perspectives have been criticised for
previously been addressed in social work literature, and suggests it has
lacking substance. In the 90s, Wakefield (1996) dealt a devastating
potential for further research in the field.
blow to the credibility of systems theory, claiming that systems theory
It is worth noting that this study analysed a social work project that
is clinically useless. Recently, Thyer, Dulmus, and Sowers (2012) also
purposefully used a Facebook fan page for practice method purposes,
stated that social workers' disastrous embrace of the vague principles of
and not merely for general NGO marketing purposes. As claimed by the
systems theory has “side tracked the profession” for years, because this
service agency, social workers participating in the project used the
provides a vague theoretical rationale for just about anything one
project's Facebook fan page to enable individuals to reach their po-
chooses to do in practice (p. 465).
tential, and to facilitate effective interactions between individuals and
Nevertheless, recent advances in social media technologies and SNA
local social systems (Chan & Holosko, 2017). Although this online
invite us to reimagine what we can do with social network systems in
practice method is in its infancy, it suggests that social networking is an
social work practice. Social networking technologies have already been
intervention component, rather than an unimportant fringe component,
used to help people solve real life problems for almost two decades, and
in practice. In social services, there are emerging practices which work
data analytics technologies have recently become more advanced and
with transactions, relationships, networking, and resource matching,
user-friendly. SNA software, such as NodeXL as used in this study, can
rather than clinical therapeutic interventions (Chan & Ho, 2017; Chan
support researchers when tracking and evaluating social network
& Holosko, 2016; Holosko, 2017; Martin, Upham, & Budd, 2015;

149
C. Chan Children and Youth Services Review 85 (2018) 143–150

behaviours, which provide empirical support to systems which focus on 10.1177/1049731503262150.

understanding networks of transactional relationships (rather than ob- Fraser, M. W., & Galinsky, M. J. (2010). Steps in intervention research: Designing and
developing social programs. Research on Social Work Practice, 20, 459–466. http://dx.
jects per se) and self-organizations in those networks (Mattaini & doi.org/10.1177/1049731509358424.
Huffman-Gottschling, 2012). Freeman, L. C. (1977). A set of measures of centrality based on betweenness. Sociometry,
Taken together, we see that advances in social media technologies 35. http://dx.doi.org/10.2307/3033543.
Godinho, P., Godinho, P., Moutinho, L., Moutinho, L., Pagani, M., & Pagani, M. (2017). A
and SNA have unquestionably brought fresh opportunities for social memetic algorithm for maximizing earned attention in social media. Journal of
work practice. Of course, it is questionable whether offline social net- Modelling in Management, 12, 364–385. http://dx.doi.org/10.1108/JM2-10-2015-
works can also be tracked and analysed equally well. Nonetheless, it 0078.
Goldkind, L. (2015). Social media and social service: Are nonprofits plugged in to the
may be important for the social work field to notice that however much digital age? Human Service Organizations Management Leadership & Governance, 39,
social networking technologies matter today, they may be going to 380–396. http://dx.doi.org/10.1080/23303131.2015.1053585.
matter even more in the future. Driven by our growing connectedness, Hampton, K., Rainie, L., Lu, W., Dwyer, M., Shin, I., & Purcell, K. (2014). Social media
and the ‘spiral of silence’. Retrieved from http://www.pewinternet.org/2014/08/26/
social network data will penetrate further into our daily lives, and it
social-media-and-the-spiral-of-silence/.
may become more openly available. That is, tomorrow, even more than Hansen, D. L. (2011). Exploring social media relationships. On the Horizon, 19, 43–51.
today, social network analysis may be an important skill for any social http://dx.doi.org/10.1108/10748121111107726.
worker who cares about an issue, who wants to build a reputation, and Hansen, D. L., Schneiderman, B., & Smith, M. A. (2011). Analyzing social media networks
with NodeXL: Insights from a connected world. Burlington, Mass.: Morgan Kaufmann.
who wants to connect with others around the world who share a vision. Holosko, M. J. (2016). Introduction to the special issue on intervention research. Research
Further research and discussion are indeed necessary. on Social Work Practice, 26, 5–7. http://dx.doi.org/10.1177/1049731515581015.
Holosko, M. J. (2017). Social work case management: Case studies from the frontlines.
Thousand Oaks: SAGE.
Acknowledgements Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and
opportunities of social media. Business Horizons, 53, 59–68. http://dx.doi.org/10.
The study was derived from part of a larger evaluation study of the 1016/j.bushor.2009.09.003.
Lee, A. R., & Suzanne Horsley, J. (2017). The role of social media on positive youth
Cyber Youth Outreach Project supported by The Hong Kong Jockey development: An analysis of 4-H Facebook page and 4-H'ers' positive development.
Club Charities Trust and three universities in Hong Kong. Children and Youth Services Review, 77, 127–138. http://dx.doi.org/10.1016/j.
childyouth.2017.04.014.
Leung, Z. C., Wong, S. S., Lit, S.-w., Chan, C., Cheung, F., & Wong, P.-I. (2017). Cyber
References youth work in Hong Kong: Specific and yet the same. International Social Work, 60(5),
1286–1300 Retrieved from: https://doi.org/10.1177/0020872815603784.
Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Lewis, K., Kaufman, J., Gonzalez, M., Wimmer, A., & Christakis, N. (2008). Tastes, ties,
Journal of Economic Perspectives, 31, 211–236. http://dx.doi.org/10.1257/jep.31.2. and time: A new social network dataset using Facebook.com. Social Networks, 30,
211. 330–342. http://dx.doi.org/10.1016/j.socnet.2008.07.002.
Ballantyne, N., Lowe, S., & Beddoe, L. (2017). To post or not to post? Perceptions of the Martin, C. J., Upham, P., & Budd, L. (2015). Commercial orientation in grassroots social
use of a closed Facebook group as a networked public space. Journal of Technology in innovation: Insights from the sharing economy. Ecological Economics, 118, 240–251.
Human Services, 35, 20–37. http://dx.doi.org/10.1080/15228835.2017.1277903. http://dx.doi.org/10.1016/j.ecolecon.2015.08.001.
Bessi, A., & Ferrara, E. (2016). Social bots distort the 2016 US Presidential election online Mattaini, M. A., & Huffman-Gottschling, K. (2012). Ecosystems theory. In B. A. Thyer, K.
discussion. http://dx.doi.org/10.5210/fm.v21i11.7090. M. Sowers, & C. N. Dulmus (Eds.). Human behavior in the social environment theories for
Best, P., Manktelow, R., & Taylor, B. (2014). Online communication, social media and social work practice (pp. 297–325). Hoboken, N.J: Wiley.
adolescent wellbeing: A systematic narrative review. Children and Youth Services Moore, S. T. (1990). A social work practice model of case management: The case man-
Review, 41, 27–36. http://dx.doi.org/10.1016/j.childyouth.2014.03.001. agement grid. Social Work, 35, 444. http://dx.doi.org/10.1093/sw/35.5.444.
Bond, R. M., Fariss, C. J., Jones, J. J., Kramer, A. D., Marlow, C., Settle, J. E., & Fowler, J. Newton, C. (2014). TaskRabbit is blowing up its business model and becoming Uber for
H. (2012). A 61-million-person experiment in social influence and political mobili- everything. Retrieved from https://www.theverge.com/2014/6/17/5816254/
zation. Nature, 489, 295–298. http://dx.doi.org/10.1038/nature11421. taskrabbit-blows-up-its-auction-house-to-offer-services-on-demand.
Chan, C. (2013). Young activists the anti-patriotic education movement in postcolonial Oliver, D. P., Washington, K., Wittenberg-Lyles, E., Gage, A., Mooney, M., & Demiris, G.
Hong Kong: Some insights from Twitter. Journal of Citizenship, Social and Economics (2015). Lessons learned from a secret Facebook support group. Health & Social Work,
Education, 12, 148–162. http://dx.doi.org/10.2304/csee.2013.12.3.148. 40, 125–133. http://dx.doi.org/10.1093/hsw/hlv007.
Chan, C. (2016). A scoping review of social media use in social work practice. Journal of Reamer, F. G. (2015). Clinical social work in a digital environment: Ethical and risk-
Evidence-informed Social Work, 13, 263–276. http://dx.doi.org/10.1080/23761407. management challenges. Clinical Social Work Journal, 43, 120–132. http://dx.doi.
2015.1052908. org/10.1007/s10615-014-0495-0.
Chan, C., & Ho, C. Y. S. (2017). Youth-at risk of socio-economic exclusion, information Rothman, J., & Thomas, E. J. (1994). Intervention research: Design and development for the
and communication technology, and case management practice. In M. J. Holosko human service. New York: Routledge.
(Ed.). Case studies from the frontlines: Social work case management (pp. 223–233). SWD (2011). Invitation to operate pilot cyber youth outreaching projects service specifications.
Thousand Oaks, CA: Sage Publications. Hong Kong: Social Welfare Department (SWD).
Chan, C., & Holosko, M. J. (2016). An overview of the use of Mechanical Turk in beha- Thyer, B. A., Dulmus, C. N., & Sowers, K. M. (2012). Human behavior in the social en-
vioral sciences: Implications for social work. Research on Social Work Practice, 26, vironment theories for social work practice. Hoboken, N.J.: Wiley.
441–448. http://dx.doi.org/10.1177/1049731515594024. Valimaki, M., Athanasopoulou, C., Lahti, M., & Adams, C. E. (2016). Effectiveness of
Chan, C., & Holosko, M. J. (2017). The utilization of social media for youth outreach social media interventions for people with schizophrenia: A systematic review and
engagement: A case study. Qualitative Social Work, 16, 680–697. http://dx.doi.org/ meta-analysis. Journal of Medical Internet Research, 18. http://dx.doi.org/10.2196/
10.1177/1473325016638917. jmir.5385.
Coulton, C. J., Goerge, R., Putnam-Hornstein, E., & de Haan, B. (2015). Harnessing big Wakefield, J. C. (1996). Does social work need the eco-systems perspective? Part 1. Is the
data for social good: A grand challenge for social work (Grand Challenges for Social perspective clinically useful? Social Service Review, 70, 1–32. http://dx.doi.org/10.
Work Initiative Working Paper 14). Retrieved from http://aaswsw.org/grand- 1086/604163.
challenges-initiative/12-challenges/. Welch, V., Petkovic, J., Pardo, J. P., Rader, T., & Tugwell, P. (2016). Interactive social
Del Vicario, M., Zollo, F., Caldarelli, G., Scala, A., & Quattrociocchi, W. (2017). Mapping media interventions to promote health equity: An overview of reviews. Health
social dynamics on Facebook: The Brexit debate. Social Networks, 50, 6–16. http://dx. Promotion and Chronic Disease Prevention in Canada-Research Policy and Practice. 36.
doi.org/10.1016/j.socnet.2017.02.002. Health Promotion and Chronic Disease Prevention in Canada-Research Policy and Practice
Doran, P. R., Doran, C., & Mazur, A. (2011). Social network analysis as a method for (pp. 63–75). . http://dx.doi.org/10.24095/hpcdp.36.4.01.
analyzing interaction in collaborative online learning environments. Journal of Wong, V. (2009). Youth locked in time and space? Defining features of social withdrawal
Systemics, Cybernetics & Informatics, 9, 10–16. and practice implications. Journal of Social Work Practice, 23, 337–352.
Fitch, D. (2007). Wherefore wikis? Journal of Technology in Human Services, 25, 79–85. Wölfer, R., Faber, N. S., & Hewstone, M. (2015). Social network analysis in the science of
http://dx.doi.org/10.1300/J017v25n04_05. groups: Cross-sectional and longitudinal applications for studying intra- and inter-
Fraser, M. W. (2004). Intervention research in social work: Recent advances and con- group behavior. Group Dynamics: Theory, Research, and Practice, 19, 45–61. http://dx.
tinuing challenges. Research on Social Work Practice, 14, 210–222. http://dx.doi.org/ doi.org/10.1037/gdn0000021.

150