the creative link: investigating the relationship between social network indices, creative...

Computers in Human Behavior xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Computers in Human Behavior

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

The creative link: Investigating the relationship between social networkindices, creative performance and flow in blended teams

0747-5632/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.chb.2013.12.003

⇑ Corresponding author.E-mail address: [email protected] (A. Gaggioli).

Please cite this article in press as: Gaggioli, A., et al. The creative link: Investigating the relationship between social network indices, creative perfoand flow in blended teams. Computers in Human Behavior (2013), http://dx.doi.org/10.1016/j.chb.2013.12.003

Andrea Gaggioli a,d,⇑, Elvis Mazzoni c, Luca Milani b, Giuseppe Riva a,d

a Department of Psychology, Catholic University of Milan, Italyb C.R.i.d.e.e., Department of Psychology, Catholic University of Milan, Italyc Department of Psychology, Alma Mater Studiorum, University of Bologna, Italyd Applied Technology for Neuro-Psychology Lab, Istituto Auxologico Italiano, Milan, Italy

a r t i c l e i n f o

Article history:Available online xxxx

a b s t r a c t

We present findings of an exploratory study, which investigated the relationship among the indices ofsocial network structure, flow, and creative performance in students collaborating in a blended setting.Thirty undergraduate students enrolled in a Media Psychology course were assigned to five groups taskedwith designing a new technology-based psychological application. Team members collaborated over atwelve-week period using two main modalities: face-to-face meeting sessions in the classroom (once aweek) and virtual meetings using a groupware tool. Social network indicators of group interaction andpresence indices were extracted from communication logs, whereas flow and product creativity wereassessed through survey measures. The findings showed that specific social network indices (in particularthose measuring decentralization and neighbor interaction) were positively related to flow experience.More broadly, the results indicated that selected social network indicators could offer useful insight intothe creative collaboration process. Theoretical and methodological implications of these results aredrawn.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

The integration of interactive social media, such as e-mail, chat,web conferencing, blogs, and Wikis, in instructional strategies isexpanding the array of creative teamwork tools that can be usedin the classrooms (Mortera-Gutierrez, 2006). In particular, blendedenvironment that allows students to meet occasionally face-to-face but otherwise use technology to connect to the universityand their peers has become an increasingly common delivery prac-tice in higher education (Mazzoni, 2014; Mazzoni & Iannone,2013). According to Graham (2006), this instructional approach be-comes so ubiquitous ‘‘that we will eventually drop the wordblended and just call it learning’’ (2006, p. 67). Although teams col-laborating in blended setting have received substantial attentionby scholars and educators over the last years, some of the issuesthat affect their effectiveness and performance have been scarcelyinvestigated, with creativity constituting one of these currently un-der-researched issues. Consistent with this need, the purpose ofthis research was to examine a conceptual and methodologicalframework called ‘‘Networked Flow’’ (Gaggioli, Milani, Mazzoni,& Riva, 2011; Gaggioli, Riva, Milani, & Mazzoni, 2013) with anaim to study creative collaboration in blended setting. Drawingon previous research on social creativity, the model argues that

the key to group creativity is the development of ‘‘collaborativezone of proximal development’’ in which actions of the individualsand those of the collective are in balance and a sense of social pres-ence is established. Further, the model suggests that if this condi-tion is achieved, the group has the opportunity to experimentgroup flow, an optimal experience that is able to produce a long-term change relevant to both the team and its individual members.At the methodological level, the Networked Flow framework iden-tifies Social Network Analysis (SNA) as a potentially useful ap-proach for investigating interaction dynamics that foster creativecollaboration. The first section of the paper presents the conceptsof the Networked Flow framework. Next, we describe preliminaryresults of a study in which longitudinal SNA and self-reported flowstates were used to explore the creative collaboration in fivegroups of students engaging in a blended environment. Finally,we discuss potential implications of the Networked Flow modelfor research and practice.

2. Conceptual background

2.1. Group creativity and flow

Creativity has been commonly referred to as the ability to cre-ate objects, artifacts, or thoughts, which may be defined and recog-nized as original, unexpected, high in quality and useful (Sternberg& Lubart, 1996). Thus, creativity can be described as both an

rmance

http://dx.doi.org/10.1016/j.chb.2013.12.003

mailto:[email protected]


http://www.sciencedirect.com/science/journal/07475632

http://www.elsevier.com/locate/comphumbeh


2 A. Gaggioli et al. / Computers in Human Behavior xxx (2013) xxx–xxx

outcome and a process in which individuals, groups or organiza-tions are engaged to produce creative outcomes, that is, noveland useful ideas. Traditionally, creativity has been mostly investi-gated from an individual perspective, i.e., by studying the psycho-logical features that characterize the creative person, such aspersonality traits, cognitive abilities and intellectual development(Sternberg & Lubart, 1999). More recently, however, there has beena shift in the focus from the individual to the social aspects of thecreative process (Amabile, 1983; Amabile, Conti, Coon, Lazenby, &Herron, 1996; Csikszentmihalyi, 1999; John-Steiner, 2000; Sawyer,2003; Sawyer, 2007). In line with this perspective, Sawyer has pro-posed a model of creative collaboration in which he argued that ateam performs at its best when it is able to achieve a state of‘‘group flow’’, an optimal collective experience defined as a ‘‘collec-tive state of mind’’ (p. 43). The concept of flow was originally intro-duced by Csikszentmihalyi (1975; 2000) who described it as anoptimal experience characterized by global positive affect, highconcentration and involvement, feeling of control, clear goals,and intrinsic motivation: in particular, a key feature of this experi-ence is the perception of high skills matched by equally high per-sonal resources (i.e. knowledge, abilities, proactive coping,positive engagement modes) to face them. Whereas Csikszentmih-alyi studied the link between flow and creativity at an individuallevel, Sawyer (a former Csikszentmihalyi’s student) extended theanalysis to group collaboration by considering two specific do-mains: jazz and theater improvisation (Sawyer, 2003). He used atechnique called ‘‘interaction analysis’’, which consists of an in-depth observation and classification of participants’ conversations,gestures, and body language. By examining the data collected overten years of observations of several performing groups, Sawyerconcluded that group flow requires members to develop a feelingof mutual trust and empathy, which culminates in a collectivemental state in which individual intentions harmonize with thoseof the group. Jazz music players often refer to this state as toachieving a ‘‘group mind’’ characterized by a profound emotionalresonance, which allows artists to be fully coordinated within theimprovisational flow. According to Sawyer, group flow ‘‘cannotbe reduced to psychological studies of the mental states or the sub-jective experiences of the individual members of the group’’ (2003,p. 46). In other words, group flow cannot be broken down into thework of individuals; rather, this phenomenon emerges from theinteractions occurring within a group and is able to positivelyinfluence overall performance. Furthermore, Sawyer suggestedthat the achievement of group flow involves a balance betweenthe extrinsic/intrinsic nature of the goal and pre-existing struc-tures shared by the team members (for example know-how,instructions, repertory of cultural symbols, set of tacit practices,etc.). An extrinsic goal, according to Sawyer, is characterized by aspecific and well-defined objective (i.e., how to fix a bug in soft-ware); therefore, it requires the achievement of more shared struc-tures. In contrast, an intrinsic goal is largely unknown andundefined (i.e., the task faced by an improvisation group in the-atre); therefore, it requires the achievement of structures that areless shared (2003, p. 167).

2.2. Collaborative zone of proximal development

Armstrong (2008) carried out a study to examine the conditionsthat foster (or hinder) the emergence of group flow in middleschool mathematics classroom setting. According to Armstrong,the occurrence of group flow indicates that the team is workingin a mutual zone of proximal development (Goos, Galbraith, &Renshaw, 2002; John-Steiner, 2000), which the author defined as‘‘an intellectual site where students are able first to negotiateshared meaning within their group (or part of their group)’’(p. 102). In particular, the author drew on the complex systems

Please cite this article in press as: Gaggioli, A., et al. The creative link: Investigaand flow in blended teams. Computers in Human Behavior (2013), http://dx.doi

model of mathematics classes developed by Davis and Simmt(2003) who identified five specific conditions that lead to theestablishment of such ‘‘space of joint action’’. These conditions in-clude (a) internal diversity, which is based on the various interestsand expertise present in the group. According to Davis and Simmt,this quality cannot be imposed ‘‘from the top down’’, that is, it can-not be assigned or legislated; instead, it must be assumed. The sec-ond condition is (b) redundancy. It refers to the creation of acommon ground shared by group members, which provides inter-nal coherence to the interaction. Such common ground does notonly involve shared vocabularies, symbol systems, and resources,but also a communion of experience, expectation and purpose (p.151). Davis and Simmt argued that redundancy plays two keyroles. First, it enables interaction among members and second, itallows members to compensate for others’ weak points and fail-ures. From this perspective, redundancy and internal diversity rep-resent two complementary sides. Whereas the first is moreoutward-oriented, enabling new opportunities for actions in re-sponse to change in context, the latter is more inward-oriented,enabling the co-acting of the agents. The third condition is (c)decentralized control. This feature refers to a situation in whichthe actions of a group and the decisions that it takes are sharedand distributed rather than managed by a single member. This con-dition is achieved when the knowledge does not reside within aparticular member of the group and the authority is not confinedto a specific person, argument or resource. Since decentralized con-trol fosters greater participation, it allows the group to fully exploitits internal diversity, which would otherwise remain silent. An-other condition concerns (d) organized randomness. According toDavis and Simmt, this is a critical aspect for the emergence of whatthey call a ‘‘collective learning system’’ (p. 163). It is achievedwhen the group is able to maintain the equilibrium between suffi-cient organization to guide members’ actions and to obtain suffi-cient randomness to allow for heterogeneous responses. Fromthis perspective, organized randomness can be seen as a structuralcondition that helps determine the balance between redundancyand diversity among members (p. 154). Finally, (e) neighbor inter-actions condition concerns the opportunity for group participantsto communicate and exchange ideas. In this process, the artifactsused to mediate such interaction play a critical role. Written mate-rials, such as notes, articles, and sketches not only facilitate thetransmission of ideas, but also serve as a record of emergent ones,acting as extra-somatic memories.

To understand the role that these five conditions play in theemergence of group flow, Armstrong observed the working pro-cesses of two small groups of students collaborating on a prob-lem-solving task. The sessions were recorded using videotapesand written transcripts. To identify the occurrence of group flow,Armstrong focused on specific physical and verbal behaviors,which would indicate a synchronization of actions and thoughts(i.e., physical closeness, echoing of gestures and phrases, the mir-roring of each other’s physical actions). The study found thatalthough both groups had the prerequisite structures to experiencegroup flow, only one group showed the characteristics of this opti-mal state. According to Armstrong, the absence of group flow thatemerged in the second group could be explained by a lower level ofdecentralization because one student took the lead and presented asolution, which other team members accepted passively. Conse-quently, some members of this group failed to establish a collabo-rative zone of proximal development in which they could developtheir ideas as a collective. Armstrong used these findings to drawimplications for practice. For example, according to this author, itis important to assign students to groups ‘‘where they feel a highlevel of comfort and trust so that all members feel safe to contrib-ute and develop a collective zone of proximal development’’(Armstrong, 2008, p. 114). Furthermore, the author stressed the

ting the relationship between social network indices, creative performance.org/10.1016/j.chb.2013.12.003


A. Gaggioli et al. / Computers in Human Behavior xxx (2013) xxx–xxx 3

importance of the level of student’s ability when working in agroup, with low-level and high-level students given equal opportu-nities to contribute to the discussion and find common level ofunderstanding.

2.3. Social presence

Previous research has identified social presence as an importantconstruct in distributed learning environments (Gunawardena,1995; Tu, 2002). Gunawardena and Zittle (1997) defined this con-cept as ‘‘the degree to which a person is perceived as ‘real’ in med-iated communication’’ (p. 2). Social presence has an importantinfluence on both participation and social interaction, which arethe key factors to enhance collaborative learning and knowledgeconstruction (Garrison & Vaughan, 2008). For example, a studyby Swan and Shih (2005) found that students perceiving high socialpresence projected themselves more into online discussions com-pared to students perceiving low social presence; further, studentsperceiving high social presence valued more the interaction withpeers. In the context of Networked Flow model (Gaggioli et al.,2013; Riva, Waterworth, Waterworth, & Mantovani, 2011), socialpresence is defined as the non-mediated perception of an enactingother (I can recognize his/her intentions) within an external world.According to this conceptualization, a subject is present within agroup if he is able to put his own intentions (presence) into prac-tice and understand the intentions of the other group members(social presence). This implies that not all groups are the same;hence, it is not enough to put together a group of people in orderfor them all to be ‘‘present’’. It is necessary to give the group thepossibility to express itself and understand the actions of eachindividual member. This becomes a critical requirement whenthe group breaks up and the members can use only mediated formsof communication. However, if this happened, the group maytransform itself and become a creative group characterized by anoptimal group experience, that is, Networked Flow. In this state,team members share the same intention (collective intention) toproduce a long-term change relevant both to the team and tothemselves.

2.4. Social Network Analysis

In the previous paragraphs, we argued that Networked Flow isachieved when a collaborative zone of proximal development isestablished and participants experience high level of social pres-ence. When this happens, the group has the opportunity to expressits maximum creative potential. At a methodological level, we haveproposed Social Network Analysis (SNA) as a potentially useful ap-proach for investigating Networked Flow in mediated environ-ments (Gaggioli et al., 2011, 2013; Mazzoni, 2014). Byconsidering individuals as interdependent units as opposed toautonomous elements, SNA offers a promising methodology tostudy group dynamics as well as to investigate the role of the indi-viduals within these dynamics (Scott, 2000; Wasserman & Faust,1994). On the other hand, SNA has proven useful for gaining insightinto social network characteristics associated with creativity(Cattani & Ferriani, 2008; Gloor, 2006; Guimerà, Uzzi, Spiro, & Ama-ral, 2005). SNA focuses on various aspects of the relational structuresand the flow of information, which characterize a network of people,through two types of interpretation, graphs and structural indices(Mazzoni & Gaffuri, 2009; Wasserman & Faust, 1994). Graphs (orsociograms) plot the dots (individuals) and their social relationships(edges). Structural indices depict quantitatively the network of so-cial relations analyzed based on several characteristics (e.g., neigh-borhood, density, centrality, centralization, cohesion, and others).SNA is based on the flow of messages that individuals of a dyad,which are conceived and mutually dependent entities (i.e., each


message sent by X to Y is also a message received by Y from X), sendand receive within the network. This aspect is critical in online envi-ronments in which posting a message on a web forum, e.g., does notguarantee that all participants will read it. To address this issue,some authors (Manca, Delfino, & Mazzoni, 2009) have proposed acoding procedure to identify the receivers (or readers) in relationaldata collected within online environments. For each structural char-acteristic of a relational network, SNA provides two types of indices:individual indices (i.e., based on relations and exchanges character-izing each actor of the networks) and group indices (i.e., based onrelations and exchanges characterizing the network as a whole).Previous studies have used SNA to analyze the interactions in onlinelearning environment to understand the dynamics of the course(students’ involvement, students’ interactions with each other andwith teachers, their peripheral or central participation in discus-sions, etc.), knowledge construction and learning processes (e.g.,Aviv, Erlich, Ravid, & Geva, 2003; De Laat, Lally, Lipponen, & Simons,2007; Doran, Doran, & Mazur, 2011; Mazzoni & Gaffuri, 2009), andcollective creativity (e.g. Gaggioli et al., 2013; Mazzoni, 2014;Mazzoni, Gaffuri, & Gasperi, 2010). Most of these studies have usedrelational data collected in a specific moment, usually at the end ofthe process under observation. In some cases, interactions occurringover a given period were summarized by a single measurement,missing important insights on the evolution of the process. Further,in spite of the richness of group dynamics, which can be investigatedusing SNA, some authors (Gaggioli et al., 2013; Mazzoni, 2014;Mazzoni & Gaffuri, 2009) have expressed concerns regarding theuse of this approach with small groups in online environments dueto a possible ceiling effect that pushes indexes to their maximum le-vel. To address this limitation, some SNA scholars (Carrington, Scott,& Wasserman, 2005; Malin & Carley, 2007; Sijtsema et al., 2010)have suggested introducing the temporal dimension in the researchdesign, which allows researchers to explore the relational dynamicsof the groups longitudinally.

2.5. Objectives of the study

The goal of this explorative study was to evaluate the suitabilityof the Networked Flow model to examine the creative collabora-tion process of teams engaging in a blended environment. Themain novelty of the proposed methodology is that it used longitu-dinal Social Network Analysis to examine group interactions andsocial presence over time, and it related these to self-reported flowstates and creativity performance. In particular, we set the follow-ing research objectives:

� to explore the use of SNA indicators for investigating col-laborative zone of proximal development and social pres-ence in a blended environment;

� to analyze how the creative collaboration process evolvesover time by longitudinally monitoring the behavior ofSNA indicators;

� to explore the relationship among SNA indicators of ingroup interaction dynamics, flow state, and creativeperformance.

3. Method

3.1. Study setting and participants

This study took place at the Catholic University of Milan duringthe spring semester. It involved thirty participants (10 males and20 females, mean age = 24.00; SD = 0.48) who were all undergrad-uate students enrolled in a two-semester Media Psychology courseoffered through Psychology Department. The Media Psychologycourse met two class hours per week. Outside course hours,




students interacted virtually by using a web-based real-time col-laboration suite featuring chat, email, document sharing, andspreadsheet and presentation editing (Google Apps for Education).Participants were not compensated for their involvement in thestudy. Informed consent was obtained from all participants.

3.2. Research design

The research design (Fig. 1) consisted of a longitudinal collec-tion of social network data of student groups over 12 weeks of col-laboration, including a repeated-measure assessment of flowexperience at T1 (begin of the second week) and T2 (end of thetwelfth week) and a final assessment of the creative product at T2.

3.3. Materials and measures

3.3.1. CreativityFour judges, experts in Media Psychology, independently evalu-

ated the creativity of the project developed by each team using theCreative Product Semantic Scale (Besemer & O’Quin, 1989; CPSS).They used 55 items organized into subscales to measure threemain dimensions of creative products, each made up of underlyingfacets: novelty (the product is surprising, original), resolution (theproduct is logical, useful, valuable, and understandable), and elab-oration and synthesis (the product is organic, well-crafted, and ele-gant). Raters assess these dimensions using a semantic-differentialrating scale (e.g., surprising–unsurprising). A study carried out byBesemer (1998) indicated that the scale measures three dimen-sions and confirmed their ability to discriminate among products.The reliability of the three dimensions ranged from a = .69 toa = .87, with most coefficients being above .80. The CPSS has beentested extensively over one year, proving to be a valuable instru-ment for the analytical and objective evaluation of a wide rangeof creative products (Besemer & O’Quin, 1986; Besemer & O’Quin,1987, 1989; Besemer & O’Quin, 1993; O’Quin and Besemer, 2006).

3.3.2. FlowThe Italian version of the Flow State Scale (Jackson & Marsh,

1996) was utilized to assess flow experience. It comprises 36 itemsmeasured on 5-point Likert scales (ranging from 1 – strongly dis-agree to 5 – strongly agree). Flow is evaluated using 9 subscales:‘‘Challenge-Skill Balance’’ assesses the relationships between chal-lenges offered by the task and skills perceived in addressing them;‘‘Action-Awareness merging’’ measures the extent to which theactivity is done automatically, avoiding the perception of effort;‘‘Clear Goals’’ evaluates the extent to which the subject knows ex-actly what he/she is going to do; ‘‘Direct Feedback’’ assesseswhether the activity provides frequent and unambiguous feedbackto the subject; ‘‘Concentration’’ measures the level of attentionduring the execution of the activity; ‘‘Sense of Control’’ assessesthe extent to which the subject is able to modify the course of ac-tion; ‘‘Loss of Self Consciousness’’ refers to a sense of not being con-

Fig. 1. Experime


cerned with oneself while engaging in the activity; ‘‘TimeDistortion’’ assesses the level of distortion in time perception,whether speeded up or slowed down; and ‘‘Autotelic Experience’’evaluates the level of intrinsic motivation in doing the activity.Internal consistency of the 9 subscales of the Flow State Scale wereacceptable, with a mean a = .83 (Jackson & Marsh, 1996). Scores foreach subscale can be added to obtain a total flow score. The totalscore for each subscale ranged from 4 to 20 while the overall scoreranged from 36 to 180.

The questionnaire was administered at two time points, duringthe second week (T1) and at the end of the twelfth week of collab-oration (T2). Repeated assessment allowed evaluating whetherflow experience changed during the period of collaboration. Onthe accompanying instructions form, participants were asked to re-late the items of the questionnaire to their collaborationexperience.

3.3.3. Social Network AnalysisGroup interaction patterns were obtained by analyzing groups’

electronic communication logs by applying the coding procedureproposed by Manca et al. (2009) to generate the relational data(adjacency matrix) for SNA. The coding procedure adopted a con-tent analysis approach in combination with exogenous data (i.e.,properties of postings, such as, the name of the sender and area,date and time of dispatch, etc.) to identify senders and repliers/responders of the messages exchanged by team members. Thisprocedure allows determining whether a posting is addressed tothe whole group or to a specific participant (or sub-group of partic-ipants). Four social network indices were calculated: (i) Density;(ii) Centralization Degree; (iii) Centralization Betweenness; and(iv) Cliques’ Participation Index. Taken together, the first threemetrics are intended to measure ‘‘collaborative zone of proximaldevelopment,’’ as described by Armstrong (2008). In particular,Density is proposed to be an indicator of neighbor interactions,whereas Degree Centralization and Centralization Betweennessare proposed to be indicators of decentralized control. Finally,the Cliques’ Participation Index (CPI) is proposed to be a structuralmeasure of social presence. In the following, a more detaileddescription of each index is provided.

� Density: Density of a network (in this case a group) is defined asthe percentage (ranging from 0 to 1, or from 0% to 100%) ofaggregation of its members calculated based on the totality ofdirect contacts that each member has activated or received byothers (Scott, 2000; Wasserman & Faust, 1994). Since Densityis based on the interactions activated and received by eachmembers of a group, we propose that it can measure neighborinteractions among a networked set of individuals.� Group Centralization: Group Centralization (ranging from 0 to 1,

or from 0% to 100%) represents ‘‘the dependence of a networkon its ‘most important’ actors’’ (Mazzoni & Gaffuri, 2009, p.122). According to Wasserman and Faust (1994), it measures

ntal design.




the centrality of a variable or heterogeneity of the actor. It alsocan also be viewed as a measure of inequality of the individualactor values, as it (roughly) indicates the variability, dispersion,or spread’’. Regarding Degree and Betweenness Centralizationindices, the first simply indicates the extent to which singleindividuals are different from each other in terms of the quan-tity of links activated (Out-Degree Centralization index) andreceived (In-Degree Centralization index). The second deter-mines the centralization of the communicative structure basedon the individual participants’ mediating potential, since itmeasures the degree to which the group depends on the partic-ipants who act as mediators of interaction (Freeman, 1979;Mazzoni & Gaffuri, 2009; Wasserman & Faust, 1994). Takentogether, these two centralization indices act as a proxy ofdecentralized control.� Cliques Participation Index (CPI): This index measures the mean

involvement of group members in its cliques. The higher itsvalue, the more opportunities its members have to participatein different discussions (Gaggioli et al., 2013; Mazzoni, 2014).Cliques are defined as sub-graphs composed of at least threeadjacent completely connected nodes, i.e., each clique node isconnected to all other nodes of the same clique (Wasserman& Faust, 1994). Within any network, community, or group,although an individual may interact with a number of otherparticipants, he or she will preferentially interact with someindividuals rather than others. The clique index (the numberof cliques characterizing a group) can therefore indicate thepreferential interaction zones within which it is more likely thatindividuals will interact at a certain time (Gaggioli et al., 2013).As a study by Aviv et al. (2003) showed, the availability of a lar-ger number of cliques may provide group’s participants withmore opportunities to access different and varied opinionsabout the subjects discussed. The negotiation process that fol-lows this step could in turn enrich the number of arguments,eventually enhancing group’s productivity both quantitativelyand qualitatively. However, an issue with the clique index isthat it is affected by the number of participants in the groupand also by the number of participants in the cliques. To addressthis issue, Mazzoni (2014) introduced the Cliques ParticipationIndex (CPI), which is calculated by adding the participants whomake up the various cliques in a certain network, community,or group, and then dividing this number by the total numberof members of the main structure. This calculation considersthe main group dimensions and the participation of group par-ticipants in cliques. Defined in this way, the CPI can also beregarded as a structural indicator of the social presence thatcharacterizes a group. In fact, the CPI is an indicator of theextent to which a group enables its member to be involved indifferent cliques and benefit from the diverse discussions goingon within the group. The higher the CPI, the more group mem-bers participate in cliques, increasing the group’s internal cohe-sion, which is a key dimension of social presence, as identifiedin previous related research (Garrison & Vaughan, 2008; Sheaet al., 2010; Swan & Shih, 2005).

3.4. Procedure

Five groups were created, each comprising 5–7 students.Students formed the groups themselves, without teacher’s interven-tion. Each group was tasked with designing a new technology-basedpsychological application as a part of the practical evaluation con-ducted at the end of the second semester. To ensure uniformity oflength, participants were provided with a presentation templatefor delivering the final project. Teams were given 12 weeks to com-plete their task. During this period, they collaborated in two mainmodalities: (a) face-to-face meeting sessions in the classroom (two


hour session once a week), and (b) virtually by using a web-basedreal-time collaboration suite (Google Apps for Education) featuringchat, email, document sharing, spreadsheet, and presentationediting.

4. Results

4.1. Creative product semantic scale

Scores obtained by each of the four judges were aggregated in asingle average score on each dimension of the CPSS. The followingtable (Table 1) presents the aggregated scores obtained by eachproject.

As the table shows, the project developed by group 3 obtainedthe highest score (38.09) in terms of overall creativity evaluation,whereas the product of group 4 obtained the lowest score(34.01). The remaining projects (1, 2 and 5) obtained almost equiv-alent scores (respectively: 37.78; 37.80, and 37.77). When lookingat the specific dimensions of the CPSS, the product of group 5 re-ceived the highest rating on the ‘‘Novelty’’ subscale, the projectdeveloped by group 3 obtained the highest rating on the ‘‘Resolu-tion’’ subscale, and the product of group 2 received the highest rat-ing on the ‘‘Elaboration and Synthesis’’ subscale. Finally, the projectof group 4 received lower scores on all dimensions compared toother group projects.

4.2. Flow State Questionnaire

Table 2 reports the mean group scores at T1 and T2 on the ninesubscales of the Flow State Questionnaire. Wilcoxon’s test was per-formed on each subscale to examine the changes over time. The in-crease in the score on ‘‘Action-Awareness Merging’’ as well as theincrease in the score on ‘‘Direct Feedback’’ was found to be statis-tically significant.

Table 3 shows the Flow State Scale’s global mean score for eachgroup at T1 and T2. Although not significant, an increase in globalflow was observed in all groups, with the exception of group 4,which showed a decrease from T1 to T2.

4.3. Social Network Analysis

Fig. 2 plots the evolution of the relational structure of teams,which allows comparing the interaction patterns among groups,whereas Figs. 3–6 illustrate the dynamics of selected SNA indica-tors (Density, In- and Out-Degree Centralization, BetweennessCentralization).

With reference to the concept of collaborative zone of proximaldevelopment, Fig. 3 shows that neighbor interactions (representedby Density) of groups 1, 3, and 5, varied remarkably over 11 weeks,whereas neighbor interactions of groups 2 and 4 remained rela-tively stable over time. Concerning decentralized control (repre-sented by the two centralization indices), all groups showed verylow In-Degree and Betweenness Centralization indices. Group 1,which showed high In-Degree Centralization during the first week,was the only exception. These data suggest that members in al-most all teams had an equal possibility to receive informationand played a similar role in mediating the exchange of information.Trends of Out-Degree Centralization presented a different picture,since all groups except one (group 2) showed relevant variations ofthis index over time. This indicates that even though each memberof each group had the same possibilities to receive information(low In-Degree Centralization), not all members of the groupsinitiated or continued a discussion actively.



Table 1Teams, active members in each team and average judges’ ratings for each dimension of the CPSS.

Group code Numerosity Novelty Resolution Elaboration and synthesis Overall evaluation

Group 1 5 6.30 16.38 12.85 37.78Group 2 7 6.42 16.26 12.90 37.80Group 3 6 6.14 16.80 12.80 38.09Group 4 6 5.08 14.61 12.70 34.01Group 5 7 6.64 15.94 12.75 37.77Mean 6.2 6.12 16.00 12.80 37.09Standard deviation 0.8 0.54 0.75 0.07 1.54

Table 2Mean scores and standard deviations of the flow state scale subscales at T1 and T2,with Wilcoxon’s test results.

Mean at T1 Mean at T2 Wilcoxon’s test sig.

Action-awareness merging 3.11 (0.23) 3.45 (0.17) .043Autotelic experience 2.89 (0.44) 3.05 (0.14) .345Challenge-skill balance 3.22 (0.31) 3.48 (0.13) .225Clear goals 3.06 (0.35) 3.29 (0.16) .144Concentration on task 2.85 (0.19) 3.04 (0.13) .225Loss of self-consciousness 2.86 (0.13) 2.94 (0.35) .686Paradox of control 3.06 (0.39) 3.21 (0.28) .225Transformation of time 2.63 (0.19) 2.76 (0.16) .345Unambiguous feedback 3.08 (0.25) 3.35 (0.21) .043

Table 3Mean scores and standard deviations of the flow state scale global score for eachgroup at T1 and T2.

Mean at T1 Mean at T2

Group 1 101.00 (13.56) 116.67 (4.16)Group 2 109.75 (9.91) 121.00 (14.93)Group 3 97.20 (10.47) 107.20 (24.84)Group 4 120.40 (13.05) 116.40 (12.86)Group 5 107.50 (12.29) 111.20 (15.61)


We can now turn our attention to the CPI as an indicator of so-cial presence. Fig. 7 plots the evolution of the CPI in the five groupsduring the 11 weeks.

Fig. 2. The evolution of the relational structures of


The first aspect to highlight in this figure is the considerablevariation in the CPI in groups 1, 3, and 4 compared to the relativestability of the CPI in groups 2 and 5. In particular, the CPI of group2 was very high in the first week during which each member wasinvolved in at least two cliques. In the remaining weeks, its valuewas lower (with only two peaks during the 3rd and the 7th week).Group 5 showed the highest CPI levels of all teams, with the excep-tion of the 7th and the 8th weeks during which the group membershad very poor interactions. These data suggest that compared toother teams, members of group 5 had the opportunity to partici-pate in more areas of exchange and discussion. In contrast, forexample, group 4 is characterized by a combination of the lowestneighbor interactions and highest centralized control, suggestingthat in this group, one or two members monopolized the discus-sion and the process of idea generation. In addition, group 4 dis-played the lowest CPI mean of all groups under study,considering both the first 6 weeks as well as the entire period ofcollaboration (Fig. 8).

4.4. Correlations between flow and social network indices

A Pearson’s correlation was conducted to evaluate the rela-tionship between selected SNA indices and Flow State subscalesat T1 and T2. For the reason of brevity, only correlations thatwere significant at the 95% level of confidence or higher are re-ported. Density was positively correlated with ‘‘Challenge-SkillBalance’’ at both T1 (r = 0.88) and T2 (r = 0.95), ‘‘Action

the groups across the 11 weeks of observation.



Fig. 3. The evolution of the Density of the groups along the eleven weeks.

Fig. 4. The evolution of the In-Degree Centralization of the groups along the eleven weeks.

Fig. 5. The evolution of the Out-Degree Centralization of the groups along the eleven weeks.


Awareness Merging’’ at T1 (r = 0.87), ‘‘Clear Goals’’ at T1(r = 0.90), and ‘‘Concentration’’ at T1 (r = 0.92). Degree of Central-ization correlated positively with ‘‘Action Awareness Merging’’ atboth T1 (r = 0.86) and T2 (r = 0.93), ‘‘Concentration’’ at T1(r = 0.94), and ‘‘Paradox of Control’’ at T2 (r = 0.93). A positivecorrelation was also found between the number of links and dif-ferent flow dimensions, namely, ‘‘Direct Feedback’’ at both T1(r = 0.86) and T2 (r = 0.95), ‘‘Concentration’’ at T1 (r = 0.90), and‘‘Time Distortion’’ at T1 (r = 0.93). Finally, mean weight corre-lated positively with ‘‘Autotelic Experience’’ at T1 (r = 0.90), ‘‘Par-adox of Control’’ at T1 (r = 0.87), ‘‘Time Distortion’’ at T1(r = 0.92), ‘‘Clear Goals’’ at T1 (r = 0.96), and ‘‘Action AwarenessMerging’’ at T2 (r = 0.94).


5. Discussion

The first objective of this explorative study was to evaluate thesuitability of SNA for investigating collaborative zone of proximaldevelopment and social presence in a blended setting. This trans-lates into the question of whether the selected SNA indicators(Density, In- and Out-Degree Centralization, Betweenness Central-ization, CPI) can provide a consistent representation of these con-cepts, i.e., by allowing the identification of meaningfuldifferences between groups. To better examine this issue, we fo-cused on comparing SNA profiles of group 5 and group 4. Althoughthese two groups did not show considerable differences over thefirst eight weeks of collaboration, differences became more



Fig. 6. The evolution of the Betweenness Centralization of the groups along the eleven weeks.

Fig. 7. The evolution of the CPI in all groups across the eleve weeks of observation.

Fig. 8. The CPI mean by considering the first 6 weeks and all the 11 weeks.


pronounced in the last three weeks. In particular, from week 9 toweek 11, group 5 obtained higher values on Density and lower val-ues on Centralization, which indicate, respectively, high neighborinteractions and high decentralized control, the two key conditionsfor the emergence of a collaborative zone of proximal develop-ment. The opposite pattern (low neighbor interactions and lowdecentralized control) was observed in group 4, suggesting thatmembers of this group had less opportunities to participate inthe discussion. When looking at SNA measurement of social pres-ence (CPI index), group 5 had the highest mean level of social pres-ence, whereas group 4 had the lowest mean level of socialpresence. In addition, when considering global flow experience,members in group 5 increased their global flow, whereas group 4reported a decrease in global flow from T1 to T2. Interestingly, dif-ferences observed in the structural dynamics and flow experienceof these two groups are reflected in the evaluation of creative prod-


ucts. Whereas the project developed by group 5 obtained the high-est evaluation on the CPSS dimension of ‘‘Novelty’’, group 4obtained the lowest scores on all dimensions of the CPSS. Taken to-gether, these findings provide a coherent description of the SNAdifferences observed among groups. Moreover, they confirm thatSNA indicators of density and centralization are highly informativemeasures of team creativity (Choi, Lee, & Seo, 2013).

The second focus of this research was to explore the relation-ship between social network dynamics, presence, and flow state.The underlying theoretical assumption of the Networked Flowmodel is that when the conditions of a collaborative zone of prox-imal development are present, the group has a greater chance tohave an optimal experience and increase its creative potential.The analysis of the relationship between SNA indicators and FlowState Scale revealed that Density and Degree of Centralization(which, in our terms, are two key indicators of collaborative zoneof proximal development) correlated positively with several FlowState Scale dimensions, but mostly at T1. In general, this observa-tion seemed to support the previously proposed idea that theestablishment of a ‘‘zone of joint action’’ can be associated withgroup flow (Armstrong, 2008; Gaggioli et al., 2013). Moreover,the observation that most correlations were found at T1 could beindicative of the fact that the emergence of an optimal group expe-rience can start in the early stage of the collaboration process.However, while the positive relationship between Density andFlow Subscales, such as ‘‘Challenge-Skill Balance’’, ‘‘Action-Aware-ness Merging’’, ‘‘Clear Goals’’, and ‘‘Concentration’’, is in line withthe model’s assumption that neighbor interactions can facilitatethe emergence of group flow, one could have expected that higherlevels of Centralization (which indicate the emergence of a morehierarchical structure) have a negative correlation with flow sub-scales. For instance, in the previously mentioned study carried




out by Armstrong (2008), the absence of group flow in the secondgroup was explained by a higher level of centralization due to thefact that one student took the lead and presented the group with asolution, which other team members passively accepted through-out the discussion. In contrast, our results showed that Centraliza-tion correlated positively with two flow dimensions, namely‘‘Action Awareness Merging’’ at both T1 and T2, and with ‘‘DirectFeedback’’ at T1. A possible interpretation of these findings mightbe that Flow State Scale has been administered at the very begin-ning and at the very end of collaboration when a limited numberof members with specific know-how could have directed the ef-forts of the team to take care of specific tasks, e.g., organizing theteam agenda, preparing the final presentation, or emailing the re-sults to the teacher. Since the Action Awareness Merging subscalemeasures the extent to which the activity is performed automati-cally – minimizing the perception of effort – it could be that thedelegative decisional style characterizing the initial and finalphases of the process has contributed to reduction of the percep-tion of effort among members. This might also explain the positivecorrelation between Centralization and ‘‘Direct Feedback’’ at T1:the intervention of one or more coordinators at the beginning ofthe project could have provided participants with frequent andunambiguous feedbacks on how they were doing the task.

Finally, the observation that global flow scores generally in-creased from T1 to T2 might suggest that the degree of involve-ment and enjoyment of participants in group activities tends togrow with time, as the group members become more familiar witheach other as well as with groupware tools. Previous research hasshown that knowledge sharing with Web 2.0 tools can be associ-ated with flow and further employee creativity at the individuallevel (Yan, Davison, & Mo, 2013).

Concerning the relationship between the CPI and flow, no corre-lation was found. Therefore, these findings do not support the exis-tence of a relation between this social network measure of socialpresence and optimal experience. Since the CPI is essentially ameasure of the degree of involvement of group members in differ-ent cliques, this result might also suggest that the degree to whichmembers participate in different discussions does not play a role infacilitating (or hindering) group flow.

5.1. Limitations of this study

This study had a number of limitations. First, the small size ofgroups could have made SNA indices more prone to the ceiling ef-fect. Second, although longitudinal SNA indicated meaningful dif-ferences between groups, particular differences found may bespecific to the context of the research. Furthermore, while SNAindicators were explored in the context of online interactions,the assessment of flow concerned both offline and face-to-faceinteractions. Thus, the extent to which a SNA model of group inter-action resulting from online data can be considered a reliable mod-el of overall group dynamic is questionable. Finally, while SNAindicators were monitored for several weeks, flow was assessedonly at two time points. A longitudinal assessment of flow shouldbe the focus of future research.

6. Conclusions

The major aim of this study was to evaluate the suitability of anovel theoretical and methodological framework – NetworkedFlow – for investigating creative collaboration in a blended envi-ronment. To accomplish this goal, the key components of model,including the collaborative zone of proximal development, socialpresence, and group flow were examined. Although preliminary,the findings of the current study suggests that combining qualita-


tive evaluation of participants’ experience and SNA is a potentiallyuseful approach for investigating new forms of interaction emerg-ing in a blended setting.

The results of the research reported in this paper have potentialimplications for practice. The research on online learning contextshas emphasized the importance of fostering social presence andoptimal experience among learners to promote creativity andknowledge co-construction (Shea et al., 2010). If these preliminaryfindings are confirmed in further studies, the proposed SNA met-rics could provide researchers and practitioners with a more effi-cient and cost-effective approach to assess creative collaborationsvia online tools. Further, a goal of future research is to evaluatethe potential of this methodology not only in the classroom, butalso in organizations, where the promotion of networked creativityis becoming a central issue to improve a company’s innovation po-tential. Therefore, it is our hope that the present research willmotivate other researchers interested in this relationship to im-prove the methodology and overcome the limitations of our study.

References

Amabile, T. M. (1983). Social psychology of creativity: A componentialconceptualization. Journal of Personality and Social Psychology, 45, 357–377.

Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996). Assesing thework environment for creativity. Academy of Management Journal, 39,1154–1184.

Armstrong, A. C. (2008). The fragility of group flow: The experiences of two smallgroups in a middle school mathematics classroom. Journal of MathematicalBehavior, 27, 101–115.

Aviv, R., Erlich, Z., Ravid, G., & Geva, A. (2003). Network analysis of knowledgeconstruction in asynchronous learning networks. Journal of AsynchronousLearning Networks, 7(3), 1–23.

Besemer, S. P., & O’Quin, K. (1986). Analysis of creative products: Refinement andtest of a judging instrument. Journal of Creative Behavior, 20(2), 115–126.

Besemer, S. P., & O’Quin, K. (1987). Creative product analysis: Testing a model bydeveloping a judging instrument. In S. G. Isaksen (Ed.), Frontiers of creativityresearch: Beyond the basics (pp. 341–357). Buffalo, NY: Bearly Ltd..

Besemer, S. P., & O’Quin, K. (1989). The development, reliability and validity of therevised creative product semantic scale. Creativity Research Journal, 2, 268–279.

Besemer, S. P., & O’Quin, K. (1993). Assessing creative products: Progress andpotentials. In S. G. Isaksen (Ed.), Nurturing and developing creativity: Theemergence of a discipline (pp. 331–349). Norwood, New Jersey: Ablex PublishingCorp..

Besemer, S. P. (1998). Creative product analysis matrix: Testing the model structureand a comparison among products – Three novel chairs. Creativity ResearchJournal, 11, 333–346.

Carrington, P. J., Scott, J., & Wasserman, S. (Eds.). (2005). Models and methods in socialnetwork analysis. Cambridge University Press.

Cattani, G., & Ferriani, S. (2008). A core/periphery perspective on individual creativeperformance: Social networks and cinematic achievements in the hollywoodfilm industry. Organization Science, 19(6), 824–844.

Choi, D. Y., Lee, K. C., & Seo, Y. W. (2013). A longitudinal analysis of team creativityevolution patterns based on heterogeneity and network structure: An approachwith agent-based modeling. In K. C. Lee (Ed.), Digital creativity: Individuals,groups, and organizations (pp. 115–128). New York, USA: Springer-Verlag.

Csikszentmihalyi, M. (1999). Implications of a systems perspective for the study ofcreativity. In R. J. Sternberg (Ed.), Handbook of creativity (pp. 313–338).Cambridge, UK: Cambridge University Press.

Csikszentmihalyi, M. (1975/2000). Beyond boredom and anxiety. San Francisco:Jossey-Bass.

Davis, B., & Simmt, E. (2003). Understanding learning systems: Mathematicseducation and complexity science. Journal for Research in MathematicsEducation, 34(2), 137–167.

De Laat, M., Lally, V., Lipponen, L., & Simons, R. J. (2007). Investigating patterns ofinteraction in networked learning and computer-supported collaborativelearning: A role for Social Network Analysis. International Journal of Computer-Supported Collaborative Learning, 2(1), 87–103.

Doran, P. R., Doran, C., & Mazur, A. (2011). Social network analysis as a method foranalyzing interaction in collaborative online learning environments. Journal ofSystemics, Cybernetics and Informatics, 9(7).

Freeman, L. C. (1979). Centrality in social networks: Conceptual clarification. SocialNetworks, 1, 215–239.

Gaggioli, A., Milani, L., Mazzoni, E., & Riva, G. (2011). Networked flow: A frameworkfor understanding the dynamics of creative collaboration in educational andtraining settings. The Open Education Journal, 4, 107–115.

Gaggioli, A., Riva, G., Milani, L., & Mazzoni, E. (2013). Networked flow – Towards anunderstanding of creative networks. Dordrecht: Springer.

Garrison, R. D., & Vaughan, N. D. (2008). Blended learning in higher education:Framework, principles, and guidelines. San Francisco, CA: Jossey-Bass.


http://refhub.elsevier.com/S0747-5632(13)00445-7/h0005




























































Gloor, P. (2006). Swarm creativity: Competitive advantage through collaborativeinnovation networks. Oxford: Oxford University Press.

Goos, M., Galbraith, P., & Renshaw, P. D. (2002). Socially mediated metacognition:Creating collaborative zones of proximal development in small group problemsolving. Educational Studies in Mathematics, 49, 193–223.

Graham, C. R. (2006). Blended learning systems: Definition, current trends, andfuture directions. In C. J. Bonk & C. R. Graham (Eds.), Handbook of blendedlearning: Global perspectives, local designs (pp. 3–21). San Francisco, CA: PfeifferPublishing.

Guimerà, R., Uzzi, B., Spiro, J., & Amaral, L. (2005). Team assembly mechanismsdetermine collaboration network structure and team performance. Science, 308,697–702.

Gunawardena, C. (1995). Social presence theory and implications for interactionand collaborative learning in computer conferences. International Journal ofEducational Telecommunications, 1(2/3), 147–166.

Gunawardena, C., & Zittle, F. (1997). Social presence as a predictor of satisfactionwithin a computer mediated conferencing environment. American Journal ofDistance Education, 11(3), 8–26.

Jackson, S. A., & Marsh, H. W. (1996). Development and validation of a scale tomeasure optimal experience: The Flow State Scale. Journal of Sport and ExercisePsychology, 18, 17–35.

John-Steiner, V. (2000). Creative collaboration. New York: Oxford University Press.Malin, B., & Carley, K. (2007). A longitudinal social network analysis of the editorial

boards of medical informatics and bioinformatics journals. Journal of theAmerican Medical Informatics Association, 14(3), 340–348.

Manca, S., Delfino, M., & Mazzoni, E. (2009). Coding procedures to analyseinteraction patterns in educational web forums. Journal of Computer AssistedLearning, 25(2), 189–200.

Mazzoni, E. (2014). The Cliques Participation Index (CPI) as an indicator of creativityin online collaborative groups. Journal of Cognitive Education and Psychology,13(1).

Mazzoni, E., & Gaffuri, P. (2009). Monitoring activity in e-learning: A quantitativemodel based on web tracking and social network analysis. In A. A. Juan, T.Daradoumis, F. Xhafa, S. Caballe, & J. Faulin (Eds.), Monitoring and assessment inonline collaborative environments: emergent computational technologies for E-learning support (pp. 111–130). IGI Global.

Mazzoni, E., & Iannone, M. (2013). From high school to university: Impact of socialnetworking sites on social capital in the transitions of emerging adults. BritishJournal of Educational Technology. http://dx.doi.org/10.1111/bjet.12026.

Mazzoni, E., Gaffuri, P. & Gasperi, M. (2010), Individual versus collaborative learningin digital environments: the effects on the comprehension of scientific texts in


first year university students. In L. Dirckinck-Holmfeld, V. Hodgson, C. Jones, M.De Laat, D. Mcconnell, & T. Ryberg (Eds.), Proceedings of the seventh internationalconference on networked learning 2010. A research-based conference on networkedlearning in higher education and lifelong learning. Aalborg, Denmark. 3rd & 4thMay 2010 (pp. 293–300).

Mortera-Gutierrez, F. (2006). Faculty best practices using blended learning in e-learning and face-to-faceinstruction. International Journal on E-Learning, 5(3),313–337.

O’Quin, K., & Besemer, S. P. (2006). Using the creative product semantis scale as ametric for results-oriented business. Creativity and Innovation Management,15(1), 34–44.

Riva, G., Waterworth, J. A., Waterworth, E. L., & Mantovani, F. (2011). From intentionto action: The role of presence. New Ideas in Psychology, 29(1), 24–37.

Sawyer, K. R. (2003). Group creativity: Music, theater, collaboration. New Jersey: LEA.Sawyer, K. R. (2007). Group genius: The creative power of collaboration. New York:

Basic Books.Scott, J. (2000). Social network analysis: A handbook (2nd ed.). London: Sage.Shea, P., Hayes, S., Vickers, J., Uzuner, S., Gozza-Cohen, M., Mehta, R., et al. (2010). A

re-examination of the community of inquiry framework: Social network andquantitative content analysis. The Internet and Higher Education, 13(1–2), 10–21.

Sijtsema, J. J., Ojanen, T., Veenstra, R., Lindenberg, S., Hawley, P. H., & Little, T. D.(2010). Forms and functions of aggression in adolescent friendship selectionand influence: A longitudinal social network analysis. Social Development, 19(3),515–534.

Sternberg, R. J., & Lubart, T. I. (1996). Investing in creativity. American Psychologist,51(7), 677–688.

Sternberg, R. J., & Lubart, T. I. (1999). The concept of creativity: Prospects andparadigms. In R. J. Sternberg (Ed.), Handbook of creativity (pp. 3–15). New York:Cambridge University Press.

Swan, K., & Shih, L. F. (2005). On the nature and development of social presence inonline course discussions. Journal of Asynchronous Learning Networks, 9(3),115–136.

Tu, C. (2002). The measurement of social presence in an online learningenvironment. International Journal on E-Learning, 1(2), 34.

Wasserman, S., & Faust, K. (1994). Social network analysis. Methods and applications.Cambridge University Press.

Yan, Y. L., Davison, R. M., & Mo, C. Y. (2013). Employee creativity formation: Theroles of knowledge seeking, knowledge contributing and flow experience inWeb 2.0 virtual communities. Computers in Human Behavior, 29(5), 1923–1932.






































http://dx.doi.org/10.1111/bjet.12026




































the creative link: investigating the relationship between social network indices, creative...

Documents