niches and networks: explaining network evolution through niche formation processes
TRANSCRIPT
Nf
Fa
b
c
d
a
ARR2AA
KSSL
1
tpicseig2G
tnG
UT
D(
0h
Research Policy 42 (2013) 613– 623
Contents lists available at SciVerse ScienceDirect
Research Policy
jou rn al h om epage: www.elsev ier .com/ locate / respol
iches and networks: Explaining network evolution through nicheormation processes
rans Hermansa,b,∗, Dirk van Apeldoornc,1, Marian Stuiverd,2, Kasper Kokc,3
Knowledge, Technology and Innovation, Wageningen University, P.O. Box 8130, 6700 EW Wageningen, The NetherlandsTelos, Brabant Centre for Sustainable Development, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The NetherlandsSoil Geography and Landscape, Wageningen University, P.O. Box 47, 6700 AA Wageningen, The NetherlandsAlterra, Wageningen University and Research Centre, Wageningen, The Netherlands
r t i c l e i n f o
rticle history:eceived 18 April 2011eceived in revised form4 September 2012ccepted 11 October 2012
a b s t r a c t
This paper uses the evolutionary perspective of Strategic Niche Management to investigate and explainthe network dynamics of a collaborative innovation network. Building upon the theories of socio-technical transitions, we link macro-level network dynamics to the micro-level niche processes of visionbuilding and experimentation. The paper describes a method to construct longitudinal two-mode affilia-
vailable online 30 November 2012
eywords:trategic Niche Managementocial Network Analysisongitudinal networks, Network evolution
tion networks and this method is illustrated with an analysis of the network properties of an agriculturalniche in the Netherlands over a period of 15 years. Results show how a successful niche grows moreconnected, even when it grows in size. We found three distinct phases during which the network com-position is more or less stable. Powerful actors are able to shape the composition of the network, eitherthrough providing the financial resources or through creating “legislative space” for the network to grow.
. Introduction
The fundamental question of how innovations can contributeo sustainable development is important for both researchers andractitioners. It calls for an understanding of how new technolog-
cal practices are developed and spread and how these processesan be managed effectively. Sustainable technologies go beyondimple technological fixes, but instead require a reordering of soci-tal structures and social change. The study of these large systemicnnovations has been taken up in the relatively new fields of Strate-ic Niche Management and Transition Management (Kemp et al.,001; Loorbach and Rotmans, 2006; Rip and Kemp, 1998; Schot andeels, 2008).
These transition theories hold an evolutionary perspective of
echnological development that focuses on the socio-technologicaliche as the place where new technologies emerge (Schot andeels, 2007). New and divergent technologies are allowed to∗ Corresponding author at: Knowledge, Technology and Innovation, Wageningenniversity, P.O. Box 8130, 6700 EW Wageningen, The Netherlands.el.: +31 317 48 4817.
E-mail addresses: [email protected] (F. Hermans),[email protected] (D. van Apeldoorn), [email protected]
M. Stuiver), [email protected] (K. Kok).1 Tel.: +31 317 4 87215; fax: +31 317 419000.2 Tel.: +31 317 4 81772.3 Tel.: +31 317 48 2422; fax: +31 317 419000.
048-7333/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.respol.2012.10.004
© 2012 Elsevier B.V. All rights reserved.
survive in small protected spaces where the mainstream pressurefrom the market or other regulatory forces is lower. Historical casestudies have shown how many successful innovations started outin a technological niche and how they gradually became moreimportant before they eventually took over the existing dominantsocio-technological regime (Geels, 2002, 2006; Geels and Schot,2007). The lessons from these historical case studies have inspiredpractitioners to purposefully create and manage socio-technicalniches that allow for experimentation in order to further promisingnovelties.
It is increasingly acknowledged that network structures play animportant role in explaining the potential of emerging technologiesto spread (Spielman et al., 2010; Van der Valk et al., 2011). An inter-esting approach to assess a niche is to look at its network. Caniëlsand Romijn (2008) were among the first to systematically investi-gate the network of a niche using Social Network Analysis (SNA)and more recently Lopolito et al. (2011) have used SNA to defineseveral development stages of a niche. This paper aims to take theseapproaches one step further by studying the characteristics of thenetwork of a niche as it evolves over time. The central questionsthis paper poses are: (1) how does the network of a socio-technicalniche evolve over time and (2) how can these changes in networkstructure be explained by the internal niche formation processes?
Our analysis of these two questions provides both theoreti-cal and methodological contributions to the study of niches andtheir roles in socio-technical transitions. The theoretical contri-bution of this paper lies in its introduction of a perspective of
6 rch Po
ntl(tl
toKtitWnitr
2
wfatioos
tdtuanistsdfiaipn
frtcmnmstmtLpni2
14 F. Hermans et al. / Resea
etwork evolution in the study of niche developments. Studies onhe evolution of social networks show how changes in the macro-evel network structure can be explained by micro-level processesStokman and Doreian, 1997) and in this paper we will review howhe niche internal processes of convergence of expectations andearning and testing drives the network structure of the niche.
The methodological contribution of the paper lies in the applica-ion of Social Network Analysis on a dynamic network. Descriptionsf longitudinal networks are still relatively rare. So rare in fact thatnoben et al. (2006) speak of a “longitudinal gap” that exists in
he study of collaborative networks. In this paper we will apply annnovative method that helps in mapping the network characteris-ics of a network over time in a relatively straightforward manner.
e will illustrate this approach by investigating the changes in theetwork of a Dutch agricultural niche over a period of 15 years. The
mplications for Strategic Niche Management, the study of transi-ions in general and the possibilities this approach has for furtheresearch are presented in Sections 5 and 6.
. Niches and networks
A socio-technical niche can be defined as a protected spacehere promising new technologies are developed. As such a niche
orms the micro level of technological and social change wherectors are trying out new ideas in a series of dedicated experimen-al projects (Kemp et al., 1998; Raven et al., 2010). Raven (2005)dentified three internal processes that are important for the devel-pment of a niche: (1) the articulation and subsequent convergencef visions, (2) learning and experimentation and (3) the building ofocial networks.
The convergence of actors’ visions refers to the degree to whichheir strategies, expectations, beliefs and practices go in the sameirection. A shared vision between collaborating actors is impor-ant in order for the different actors to agree on the actions they willndertake (Beers et al., 2010). The actors in a niche are prepared toccept the initial low performance and higher costs of a new tech-ology and are willing to invest their time and resources to improve
t. Niche innovations are therefore often carried and developed bymall groups of pioneers: dedicated “outsiders” that are marginalo the existing networks of the socio-technical regime and do nothare some of the rules with respect to technical development (Vane Poel, 2000). When initial expectations of the innovation are con-rmed through positive results of projects and experiments, newctors and organisations are more likely to invest new resourcesn further developing the technology. This shared expectationrovides direction to the projects and experiments done in theiche.
Within a socio-technical niche, learning and experimentationunction therefore as a way to test the vision, and to gain expe-ience with a new practice or technology. In many SNM projectshere is often a strong focus on social learning and knowledgeo-creation. This form of organisational learning takes place inulti-disciplinary collaborative projects that create an opportu-
ity for people to interact, share their ideas and verify their ownental frameworks in discussion with others. During processes of
ocial learning, peoples’ perceptions change and they move fromypical first loop learning to second loop learning. Their individual
ental models are aligned into a shared group model enhancingrust between participants along the way (Argyris and Schön, 1978;eeuwis and Pyburn, 2002; Pahl-Wostl et al., 2007). Social learning
rocesses thus result in outputs, the practical plans, policies or tech-ical novelties that were produced, and some intangible outcomes:mproved relations and trust between actors (Burgess and Chilvers,006; Hermans et al., 2011).
licy 42 (2013) 613– 623
Finally there is the composition of the niche and its network.Complex innovations require different partners with differentresources and knowledge in order to perform different roles andtasks within the niche (Hermans et al., in press). Research showsthat a niche with a limited network in terms of diversity is likely tofail and that niches with broader networks provoke more second-order learning (Schot and Geels, 2008). Other network studies thatlook at the performance of individuals and corporations as a func-tion of their personal network characteristics show how certainnetwork characteristics can be advantageous for innovative per-formance, while other are not (Ahuja, 2000; Burt, 2005).
Based on the three niche internal processes of Raven, Lopolitoet al. (2011) derived a taxonomy of the potential stages a niche canfind itself in, see Table 1. A linear development process is defined inwhich first a shared vision has to be present, the right actors are tobe involved and finally the experimentation and learning can start.
It is clear that the internal niche processes are closely linkedto each other and form an iterative cycle of activities in the niche(Loorbach and Rotmans, 2006). Through testing and experimenta-tion the vision will be adapted in a continuous process: promisesand practices in a niche develop simultaneously (Stuiver andWiskerke, 2004). The results of successful experimental projectswill make it easier to enrol new actors and expand the network.Negative results, or results that are below the initial expectations,do they opposite: they reduce the faith in the new technology lead-ing to a shrinking network and less resources made available forfurther testing (Geels and Raven, 2006).
Our first goal is to move from the rather static, linear descrip-tion of development stages portrayed above to a more dynamicapproach that takes into account the changes in the niche networkover time. This means that we will look at the network structure ofa niche and we will explain the changes in the structural character-istics of the network by referring to the two underlying processes ofvision convergence (shared purpose) and learning and experimen-tation. Following Lopolito et al. we formulate our first proposition:
Proposition 1. Technological niches have different developmentphases in which the purpose of the actors involved, their learningand experimentation define the network properties of the niche.
According to Head (2008), the character of cooperation withinnetworks change over time with the establishment of trust. Inthe early stages of the collaborative network, its projects oftencan be characterised as forms of cooperation in which the workis task-focused, generally short term and participants maintaintheir organisational identities as they strive to obtain the goals andobjectives of their own organisation. As trust between participantsdevelops, successful co-operations may lead to more complex andambitious projects being organised that require more coordinationamong the network participant and the installation of a centralcoordinating organisation. Joint planning or the implementationof an agreed joint working programme for the medium term canbe established. The network stabilizes and a central coordinatingorganisation is created that can take the form of a special platformor a consortium that coordinates interactions in the network andstimulate its further expansion. Since technological niches are notyet ready to function as a market niche, the coordinating role withinthese kind of networks is often reserved for the government (Raven,2005).
Proposition 2. The network structure of a niche becomesincreasingly centralised as trust builds up between actors and
organisations and they move from cooperation to more coordinatedforms of collaboration.However, there is also a competing force at work. As the net-work of the niche grows, more and more people will be involved
F. Hermans et al. / Research Policy 42 (2013) 613– 623 615
Table 1Taxonomy of niche development stages (adapted from Lopolito et al., 2011).
Stage I Stage II Stage III Stage IV
Convergence of expectations Absent Present Present Present
attCtto(lm
Pw
cnilt2faAfitblTfibdhaAtft
3
3W
pItahMVd
lia
Networking with relevant actors AbsentLearning and experimentation Absent
Status Absence
nd they are less likely to know all the other actors involved inhe network. A growing network can easily suffer from a decline ofrust between the people involved. This line of reasoning followsoleman (1988) and Granovetter (1985, 1992) who have arguedhat closed networks facilitate the effective enforcement of sanc-ions as all the actors are connected and therefore know eachther’s actions: a denser network will therefore induce more trustBuskens, 1998). After the network has reached a certain size actorsose the overview of the whole network and the trust between its
embers is likely to go down.
roposition 3. A growing network will become less cohesivehich will lead to a loss of trust between partners.
It is important to note that the potential of a niche to be suc-essful depends not only on the internal characteristics of theiche, but also on its relationship with the incumbent technolog-
cal regime and the broader environment, the socio-technologicalandscape both the niche and the regime are embedded in, concep-ualised in the multi-level perspective of transitions (Geels, 2002,005; Geels and Schot, 2007). The multi-level perspective (or MLPor short) brings the elements of learning, bottom-up innovationsnd processes of social change in a single research framework.lthough the multi-level perspective has become a very popular
ramework to study transformative innovations within society, its not completely undisputed. For example, the analytical distinc-ions between the different levels of the MLP sometimes seem toe somewhat arbitrary. The differences in structuration of different
evels are of a gradual nature in which one level blends into the next.he core concepts of niches, regimes and landscape therefore dif-er from study to study, leading to a wide range of definitions beingn use (Markard and Truffer, 2008; Raven et al., 2010). Boundariesetween niches and regimes sometimes become blurred or evenisappear altogether (Elzen et al., 2008; Smith, 2006, 2007). Thisas led some authors to call for more methodological rigour in thepplication of the MLP (Genus and Coles, 2008; Smith et al., 2010).n additional goal of this paper is therefore to investigate whether
he application of Social Network Analysis can provide such a moreormal approach to link the relationships of the actors in the nicheo the actors outside the niche.
. Method
.1. Case: the environmental cooperatives of the Northern Frisianoodlands
The case of the Northern Frisian Woodlands (NFW) is an exam-le of a socio-technical niche operating in the agricultural sector.
n the niche two related ideas were championed that challengedhe prevailing Dutch agricultural regime regarding dairy farmingnd manure application. The establishment and history of the NFWas been described by various authors in terms of Strategic Nicheanagement, see for instance: Roep et al. (2003), Wiskerke and
an Der Ploeg (2004) and Stuiver (2008). We will draw on theseescriptions for our historical overview described below.
The Northern Frisian Woodlands is an area of about 60,000 haocated in the north of the Netherlands dominated by dairy farm-ng. It consists of small-scale, closed landscapes on high sandy soils,lternated by relatively open areas on lower peat-clay soils. The
Absent Present PresentAbsent Absent PresentEmbryonic Proto-niche Full
small-scale landscapes are formed by hedges and belts of aldertrees surrounding the plots of land, resulting in a unique mosaicof parcels (Van Apeldoorn et al., 2011). In the 1990s, national reg-ulations were drafted that imposed stringent measures to reducethe environmental impact of agricultural activities. These nationalregulations were tailored to the existing socio-technical regimein dairy farming: prescribing the use of large manure injectors topump the sludge directly into the soil of the grassland thus reduc-ing direct ammonia emissions. The new law did not allow anyother form of manure application anymore and forced all farmers towork with these manure injectors. However, these manure injec-tors consisted of very large and heavy machinery that conflictedwith local field conditions and threatened the operations of localdairy farms within their small-scale landscape. As a response to thisthreat, regional environmental farmer cooperatives were estab-lished with the aim to move towards viable and environmentalfriendly agro-systems attuned to the local landscape. VEL (Verenig-ing Eastermars Lansdouwe, landscape association of Eastermar) andVanla (Vereniging Agrarisch Natuur en Landschapsbeheer Achtkar-spelen, Agrarian Nature and Landscape Association of Achtkarspelen),were the first two environmental farmers cooperatives in theNetherlands (Renting and Van Der Ploeg, 2001).
After its foundation in 1992, a subsidy of the Ministry of Hous-ing, Spatial Planning and the Environment (VROM in Dutch) createdthe financial room for VEL to work out its ideas for landscapemanagement and nutrient reduction into a consistent pilot plan.Based on this plan, VEL and Vanla joined forces with three othernew Dutch environmental cooperatives and successfully lobbiedthe Ministry of Agriculture to let them implement their vision andexplore and develop their own means of combating nutrient losseson their farms. In 1996 the Minister, overruling his own civil ser-vants, granted the environmental cooperatives a formal exemptionfor the national manure regulations and allowed them to startexperimenting with locally developed alternative measures. Withthis formal exemption the environmental cooperatives got theirniche status as a protected space (Van der Ploeg et al., 2004). Theexemption from the national legislation for this region has beenextended ever since in order for the cooperatives to develop newknowledge and experiment with other forms of nutrient manage-ment in close cooperation with scientists (Eshuis and Stuiver, 2005;Stuiver et al., 2003; Stuiver and Wiskerke, 2004).
In 1998, VEL and Vanla and three other regional environmentalcooperatives merged into a new regional environmental coopera-tive, The Northern Frisian Woodlands (NFW). At almost the sametime two large research projects commenced. The first project is theNutrient Management Project, a follow-up project of the grasslandexperiments of 1996, to evaluate the new approach in a more scien-tific manner. Additionally, an extensive scientific research project(AGRINOVIM) is also approved in this phase by the NetherlandsOrganisation for Scientific Research (NWO) and these additionalfinancial resources make it possible to involve even more scientistsin the region. A scientific council is created that brings represent-atives of the farmers and the scientific community together andstarts to coordinate the research activities in the region. In the year
2000, the landscape management programme is formally institu-tionalised with a national subsidy program that allows for farmersto manage the landscape in return for financial compensation. Over400 farmers belonging to the NFW enrolled in the programme and
616 F. Hermans et al. / Research Policy 42 (2013) 613– 623
1993 199 4 199 5 199 6 199 7 199 8 199 9 200 0 200 1 200 2 200 3 200 4 200 5 200 6 200 7 200 8
Start of VELSpring-92
Start of Va nlaNov-92
Nature agree ment with prov ince of
Friesland
VEL pres ents Pilot Plan
Feb-94
Pres enta�on of governa nce
Exe mp�on for na�onal regula�ons; sta rt of
experimentsMarch -96
Pres enta�on of new manure applica�on machine
Sep-96
Northern Fri sian Woodland s is formed
Dec-98
Na�onal nat ure an d landscape managem ent
programme sta rts "Prog ramma Beh eer" NFW get stat us of
“Na�onal landsca pe”
TransForum starts
Regional Contract is signed
1992
Publica�o n “Good Ma nure does not smell ”
Febc-01
ig. 1.
iL
wts“iUTstirg
sstaTtmgTiepoobdctf
tpaimpciWsstd
experiment plans
F
n 2003 the whole region is given the protected status of Nationalandscape (Eshuis, 2006).
The manure and nutrient management project do not fare soell. In 2001, the group of involved scientists split internally over
he interpretation of the manure application experiments. Thepark that ignited this controversy was the publication of the bookGood manure does not smell” (Eshuis et al., 2001) by a group affil-ated mainly with the rural sociology department of Wageningenniversity claiming the success of the early grassland experiments.he second group of scientists, mainly affiliated with the animalciences department of the same university, contested the claimshat were made on statistical grounds. See Stuiver (2008) for ann-depth description of this conflict. In the end, a compromise waseached that more research was necessary into the link betweenrassland quality, manure application and soil quality.
In 2004, a new national subsidy programme is set up with thepecific aim to trigger transitions to a more sustainable agriculturalector. The programme, entitled TransForum, derived its inspira-ion from transition management and SNM (Veldkamp et al., 2009)nd after some lobbying two projects related to the NFW emerged.he first project was a scientific project that places environmen-al monitoring in a more participatory regional context: instead of
onitoring on environmental pollution at the farm level it investi-ates the possibilities to shift this monitoring to the regional level.he soil scientist who had taken up a more or less neutral positionn the earlier conflict came to the forefront to lead this new sci-ntific project. The other project that was started was a practicalroject aimed at investigating the possibilities and requirementsf a regional contract as a new form of rural governance. Onef the requirements of TransForum for funding the NFW was toroaden the regional network and start making work of regionalevelopment that also included other sectors, apart from the agri-ultural dairy sector. In 2005 this regional covenant is signed byhe five municipalities, water board, province of Friesland, and thearmers.
Fig. 1 gives a timeline for the most important events in the his-ory of these two environmental cooperatives. This initial overviewrovides some preliminary support for some of our propositions in
qualitative manner. Firstly, the conflict between the scientistsnvolved that followed upon the publication of the book “Good
anure does not smell” is indicative for a loss of trust betweenarticipants. Secondly, the governance structure of the niche didhange over the years with more coordination of the network activ-ties in the form of two research councils and the regional contract.
ith the start of the Regional Contract in 2005, this new governance
tructure was also formalised. Thirdly, the development of thehared vision started with landscape management, and then fur-her evolved into nutrient management and broadened to regionalevelopment and this would indicate a shift in the need for differentpartners with different knowledge regarding these new goals andpractices, resulting in a change in the composition of the network.All in all, this case contains all the ingredients necessary to test ourpropositions regarding network development on. In the next sec-tion we will describe our methodology to construct the differentnetworks over time in some more detail.
3.2. Sources of data and data selection
We collected data from the various experimental projects fromthe foundation of VEL and Vanla in 1992 until the end of 2008using scientific descriptions of the projects, as well as archivalinformation such as project proposals, final reports, minutes ofvarious meetings, and an extended collection of over 220 newspa-per clippings detailing the founding of the VEL-Vanla cooperativesbetween 1990 and the 2000. These newspaper clippings werefurther extended with a Lexis-Nexis search between the years:2000–2010 on the topics of “NFW” and “VEL AND Vanla”. Infor-mation was structured using the timeline for the Northern FrisianWoodlands given by Van der Ploeg et al. (2007). The year 2008 func-tioned as a cut-off date for data gathering. New projects may havecommenced, but these were not included in the data set.
We limited the selection of the projects included in the dataset to only those where members of VEL and Vanla participated,either through actively contributing or more passively by an advi-sory role or providing data for further analysis. Interdepartmentalworking groups consisting of civil servants alone were not incor-porated in the data set. Similarly, PhD research projects were notincluded. Selected projects were checked by two long-time par-ticipants in the VEL-Vanla network for accuracy. Table 2 gives anoverview of the 21 different projects we identified. We made adistinction between four types of projects, based on their mainpurpose: nutrient management, landscape management, gover-nance and research. Nutrient management projects focussed on thereduction of nutrient losses on farms through the use of additives tothe manure, combined with a systems perspective of dairy farmingthat linked the feeds, cows, milk, manure and grasslands in an over-arching analytical framework. The landscape projects focussed onthe opportunities landscape management could provide for addi-tional income of farmers. The governance projects focussed on thedevelopment of alternatives away from the top-down environmen-tal legislation towards self-governance and a broader agenda of
regional development. Research projects were process oriented,either actively coordinating research activities in the region, orevaluating the success of the collaborative projects of farmers andresearchers in terms of innovative capacity.
F. Hermans et al. / Research Policy 42 (2013) 613– 623 617
Table 2Overview of projects and their focus.
Name (in Dutch) Purpose and description Type of project
1 Bedrijfsintern Milieuzorgsysteem Development of environmental management system at farm level Nutrients2 Onderhoudsplan landschapselementen Maintenance plan for hedges, belts and alder trees Landscape3 Beheersovereenkomst De Marren Nature conservation agreement for ‘De Marren’ Landscape4 Samenwerkende milieucooperaties Collaborating environmental cooperatives Governance5 Speerpunt Mineralen en ammoniak Combatting nutrient and ammonia emissions Nutrients6 Gebiedsvriendelijke mestmachine Development of a field and soil friendly manure application machine Nutrients7 Mineralenproject 1 Nutrient Management project 1 Nutrients8 Onderzoeksraad Mineralenprojecten Research Council Nutrient Management projects Research9 AGRINOVIM International research project on agricultural novelties Research
10 Working group experiential knowledge Communication and information exchange Research11 Slim experimenteren Experimentation to encourage the innovative capacity of farmers Research12 Mineralenproject 2 Nutrient Management project 2 Nutrients13 Ureumnet Development of software for calculation and administration of on-farm nutrient cycles Nutrients14 Wageningen Atelier Think tank on manure application advice Research15 Onderzoek Theo Spruit Monitoring the environmental performance of farmer Theo Spruit Nutrients16 TransForum IP1-NFW Feasibility of a Regional Contract as a mode of regional governance Governance17 Gebiedscontract Regional Contract for regional sustainable development Governance18 Effectiviteit Alternatieve Spoor Effectiveness of the alternative track with low input dairy farming Nutrients19 TransForum WP 3MG Regional monitoring of environmental loads Nutrients20 Onderzoeksraad NFW Research council Northern Frisian Woodlands Research21 TransForum IP2-Zelfsturing en profit Regional sustainable development by means of self-governance Governance
F rojeca f proj
3
aiencsaowpWt
tions. The first assumption has to do with the information transferbetween people. Actors can exchange information in various ways,both in person and by other means (Berends et al., 2006). In our
ig. 2. Construction of networks through projects. The figure shows the different pre running at the same time. The bottom line of the figure gives the total amount o
.3. Mapping of networks
Details of the projects, such as the persons and organisationsssociated with them, their starting and end dates were recordedn a database. Start dates and end dates were rounded to the near-st quarter as sometimes either the start point or end point wasot exactly clear. The network of the niche at any point in time isonstructed through the aggregation of all projects that run on apecific point in time, cf. Rosenkopf and Tushman (1998) and Sohnd Roberts (2003). Each network consists of a unique combinationf projects and the people and their organisations that are affiliated
ith them. As a new project starts, new organisations and peo-le enter the network and once a project stops they leave again.e can regard each of these network structures as snapshots ofhe project network of the niche at any given time. This way we
ts and their start and end dates. Each network is comprised of all the projects thatects in each network and their duration.
constructed 29 different two-mode affiliation networks 4 that rep-resent a unique configuration of different projects (see Figure 2).Playing these images quickly behind each other will eventually givea dynamic movie of the networks development over time (Moodyet al., 2005).
This method to construct the networks holds two assump-
4 A two-mode affiliation network contains two different types of nodes in thesame graph called ‘actors’ and ‘events’. In our analysis of the co-membership theactors are the individual persons in the network and the events are formed by boththe organisations and the projects they belong to.
618 F. Hermans et al. / Research Policy 42 (2013) 613– 623
Table 3Some fundamental concepts in network analysis (Wasserman and Faust, 1994).
Concept Mathematical notation Definition
Node N = {n1, n2, n3, . . .,ng} Any individual, organisation or project depicted in the graphTie £ = {l1, l2, l3, . . ., lL} Lines indicate co-membership and connect two nodes with each otherGraph G (N, £) Model for a network with a set of nodes connected by a set of tiesSize g The number of nodes in the graphNetwork density � = 2L/g(g−1) The ratio of lines present in the graph to the maximum number of lines possible in the graphNodal degree d(ni) The number of lines that are incident with a nodeMaximum degree d(n*) Largest observed nodal degree in the graphMean degree d̄ = 2L/g Average degree of all the nodes in the graph∑g
degre
asfhotsdpstett
3
tnKan
tdtmsttSitt
dpeat
3
wae(w
of networks 1–9) starts in 1993 and lasted until 1997. During thisphase different agencies related to the provincial governmenttake up an important part of the network. The second phase(networks 10–24) started in 1998 with the commencement of a
Table 4Overview of organisational classification.
Aggregated organisationalcategories
Subcategories
A. Politics 1. Local political parties2. Regional political parties3. National political parties
B. Government 1. Municipalities2. Provinces3. Provincial and regional headquarters4. Water boards5. National Ministries
C. Knowledge institutes 1. University chair groups2. Research Institutes3. Pioneer and Demonstration Farms4. Schools and colleges
D. Green NGOs 1. Landscape NGOs2. Environmental NGOs3. (Renewable) Energy NGOs
E. Agrarian NGOs 1. Farmer unions
Degree centralisation CD = i=1[d(n∗)−d(ni )]
(g−1)(g−2) The variance in the
the same size
nalysis we focus on the information that is shared between per-ons within a project. The projects provide the formal opportunityor people to meet each other in person and share their ideas. Weave therefore assumed that all the people in a project know eachther and communicate. We have applied the same reasoning forhe organisations people are officially affiliated with. Large organi-ations (universities and government ministries for instance) wereivided into their smaller subdepartments or chair groups whereeople can be expected to know each other and communicate. Theecond assumption has to do with the membership over time ofhe projects. We assumed them to be constant: no people leave ornter a project once it has started. It might be possible that a par-icular person of one organisation is replaced by another person ashis does not fundamentally change the network structure.
.4. Analysis procedure: Social Network Analysis
Social Network Analysis has been used as a tool to investigatehe properties of social networks and the positions of actors in thoseetworks in a semi-quantitative manner (Degenne and Forsé, 1999;noke and Yang, 2008; Wasserman and Faust, 1994). In Table 3n overview is provided of some of the fundamental concepts ofetwork analysis that are applied in our analysis.
One of the core problems in longitudinal network studies is howo compare different sized networks with each other. Network size,ensity and centralisation are correlated, for which we have to con-rol when interpreting the results. To circumvent this problem theean degree of the nodes in the network was selected as a mea-
ure for network density: that is the average amount of ties each ofhe nodes possesses in the network. This measure has the advan-age that it is independent of network size (Anderson et al., 1999;tokman, 2001). However this is not possible for the degree central-sation and we have used the conditional uniform graph hypothesisest proposed by Anderson et al. (1999) to estimate the effects ofhis possible interference.
Network composition was measured using the organisationaliversity within the network. Organisations connected to therojects were categorised according to their institutional role: gov-rnment, non-governmental, political or commercial. Table 4 givesn overview of the categories used in the analysis of the organisa-ional level.
.5. Software
Network properties were analysed using “R” the statistical soft-are programme (version 2.8.0) (R Development Core Team, 2008)
nd more specifically its statnet-package (version 2.1) (Handcockt al., 2003). Visualisation was done using Pajek (version 1.26)Batagelj and Mrvar; De Nooy et al., 2005) and SoNIA – Social Net-ork Image Animator (Bender-DeMoll and McFarland, 2006).
es of nodes divided by the maximum degree variation possible in a network of
4. Results
We constructed 29 different networks based on the combina-tion of collaborative projects running at the same time. Space doesnot permit a full representation of all 29 networks, however thecomplete set of networks has been visualised in a short movie thatshows the growth of the network over time as well as the changein structure. This movie can be downloaded as additional informa-tion to this paper. Table 5 gives an overview of the various measuresfor size, mean degree per node and the centralisation degrees foreach of the 29 networks at different points in time. Fig. 3 depictsnetworks 1 and 16 as an example of two of these 29 networks. Thefirst network shows the first project that was organised and howit brings ten persons from nine different organisations together.The other network, number 16, shows how six projects run duringthis period and how these projects are mutually linked through thepersons that are member of the same projects.
Fig. 4 gives an overview of the organisational composition ofthe network over time. Some drastic shifts in the network com-position can be observed from one phase to the other, however inbetween these shifts the network composition remains relativelystable. Based on this figure we can thus distinguish between threedifferent phases that existed over time. The first phase (comprised
F. Environmental cooperatives 1. Environmental cooperatives
G. Business 1. Consultancy agencies2. Banks3. Companies
F. Hermans et al. / Research Policy 42 (2013) 613– 623 619
Table 5Overview of network properties over time.
Nw. nr. Starting date No. of projects Persons Organisations and projects Nodes in network Number of ties Mean degree per node Degree centralisation(d-m-yyyy) (actors) (events) (g) (L) (d̄) (CD)
1 1-1-1993 1 10 10 20 22 2.200 0.456a
2 1-7-1994 2 16 15 31 37 2.387 0.271a
3 1-10-1994 3 22 17 39 49 2.513 0.208b
4 1-4-1995 3 22 18 40 48 2.400 0.205b
5 1-7-1995 4 33 20 53 73 2.755 0.205b
6 1-10-1995 3 28 18 46 62 2.696 0.239b
7 1-1-1996 4 33 20 53 75 2.830 0.203b
8 1-4-1996 3 24 16 40 55 2.750 0.277a
9 1-10-1996 2 19 13 32 40 2.500 0.361a
10 1-1-1998 4 50 23 73 111 3.041 0.371a
11 1-4-1998 3 45 21 66 102 3.091 0.411a
12 1-1-1999 4 48 27 75 126 3.360 0.356a
13 1-1-2000 3 37 22 59 101 3.424 0.456a
14 1-10-2000 4 54 33 87 153 3.517 0.303a
15 1-1-2001 4 46 33 79 123 3.114 0.340a
16 1-10-2001 6 60 40 100 163 3.260 0.265a
17 1-1-2002 5 58 38 96 157 3.271 0.277a
18 1-7-2002 4 42 29 71 107 3.014 0.279a
19 1-7-2003 5 48 36 84 123 2.929 0.235a
20 1-10-2003 4 42 29 71 108 3.042 0.279a
21 1-1-2004 5 50 33 83 129 3.108 0.236a
22 1-7-2004 6 56 36 92 143 3.109 0.212a
23 1-10-2004 5 50 34 84 121 2.881 0.236a
24 1-1-2005 2 20 18 38 43 2.263 0.278a
25 1-4-2005 3 35 36 71 76 2.141 0.189a
26 1-1-2006 3 53 43 96 119 2.479 0.296a
27 1-4-2006 3 54 38 92 130 2.826 0.305a
28 1-1-2007 4 56 40 96 139 2.896 0.291a
29 1-1-2008 5 63 43 106 155 2.925 0.263a
randorando
noosplbai
oama
Fp
a p < 0.001, centralisation is significantly higher than the centralisation of 10,000
b p < 0.005, centralisation is significantly higher than the centralisation of 10,000
umber of research projects and lasted until early 2005. For mostf this period the research groups made up for more than 50%f the network composition. The third phase (networks 25–29)tarted in the second quarter of 2005 and was still on-going at theoint where we stopped the analysis at the end of 2008. In this
ast phase the network composition changes again into a morealanced distribution of sectors present: green-NGOs dealing withspects of environment and landscape conservation become morenvolved, as well as local municipalities.
Fig. 5 gives an overview of the development of the total numberf organisations and persons in the network, measured as the total
mount of nodes in the network, the connectedness of the network,easured as the mean degree: the average number of ties per nodend the centralisation of the network.
ig. 3. Project networks 1 and 16 (in January 1993 and October 2001, respectively), blarojects.
mly generated two-mode networks of dimension (actors × events) with L ties.mly generated two mode networks of dimension (actors × events) with L ties.
Our second proposition that the network will become morecentralised after a number of successful projects and trust is estab-lished between the project partners, is not supported by the dataon network structure. The network centralisation builds up beforethe year 2000 and decreases after that period but with a Pearson’sproduct moment correlation of 0.23 there is only a weak correla-tion between mean degree and centralisation. This does not meanno coordination took place. After all, the governance structure ofthe niche did change over the years with more coordination of thenetwork activities in the form of two research councils and theregional contract that was signed in 2005. Yet, these coordinating
activities did not have any effect on the any centralisation of theaffiliation network. Both research councils probably acted more asa portal to the niches network: new (research) projects first hadck nodes represent organisations, yellow nodes people and the red nodes denote
620 F. Hermans et al. / Research Policy 42 (2013) 613– 623
0
5
10
15
20
25
30
35
40
45
993
994
995
996
997
998
999
2000 00
1
002
003
004
005
006
007
008
num
ber o
f org
anis
atio
ns
kno wledgengo.greenngo - agr.governmentenv.coopbusine ss
Phase IIPhase I Phase III
ositio
tce
alnnwownatswrimt
1 1 1 1 1 1 1
Fig. 4. Organisational comp
o be approved by the research council, giving the members of theouncil the control over the influx of new people in the network,specially researchers.
The network size and connectedness (mean degree) do show significant correlation with a Pearson’s product moment corre-ation of 0.60. This means that as the niche’s network grows theetwork becomes more cohesive at the same time: the averageumber of bonds between its members increases, making the net-ork more connected. The opposite also holds: a declining number
f members in the network corresponds with a less connected net-ork. These results run contrary to our third proposition that theetwork would grow more disconnected as more and more peoplend organisations become involved in the niche. Fig. 5 shows howhe mean degree steadily declined after the year 2001 and onlytarted to grow again after 2005. This coincides with the periodhen the scientific controversy between the researchers of the
ural sociology department and animal sciences group played outn the network of the Northern Frisian Woodlands. The controversy
ade it more difficult to interest new partners in the network ashe visions within the niche did not converge anymore. The results
0
20
40
60
80
100
120
jan/93 jan/95 jan/97 jan/99 jan/0
netw
ork
size
(tot
al n
odes
)
sizemean degreecentralisation
Fig. 5. Network size, mean degree
2 2 2 2 2 2 2 2
n of the network over time.
show that the amount of trust between participants determinesthe growth and connectedness of the niche. A successful niche willgrow more connected, even when it expands.
Fig. 6 shows the boxplots for the network centralisation andthe mean degree. A one-way between-groups analysis of vari-ance (ANOVA) confirmed a statistically significant difference (atthe p < 0.05 level) in the mean degree between the phases, but thenetworks centralisation scores between the phases did not show asignificant difference. Post hoc comparison using Bonferroni’s testshowed that the average network degree of the second phase dif-fers significantly with the other two phases. Phase 1 and phase 3did not show a statistically significant difference. Our results thusshow that the different phases in the niche can not only be viewedfrom the composition of the network, but they are reflected in theconnectedness in each phase as well.
5. Discussion
The results show three phases in which the composition, theconnectedness (mean degree) and goals of the niche remained
0
1
2
3
4
5
6
1 jan/03 jan/05 jan/07
cent
ralis
atio
n an
d m
ean
degr
ee p
er n
ode
and centralisation over time.
F. Hermans et al. / Research Po
rnetofttvmtrIlfinUdecspts
TC
Fig. 6. Boxplots for network centralisation and average degree per phase.
elatively constant. Table 6 summarizes the most importantiche internal processes within each of the different phases. Thenvironmental cooperatives were the result of farmers unitinghemselves against the threat of a top-down implementationf national environmental legislation unsuitable to their way ofarming in a small-scale landscape. This initial pre-niche phasehus was characterised by the self-organisation of farmers intohe two environmental cooperatives. In phase 1 the vision con-erges as the farmers work out an alternative vision on landscapeanagement and nutrient management on dairy farms. To spread
heir local vision, farmers actively lobbied the national authoritiesesulting in an exemption from the environmental legislation.t gave them the possibility to put their alternative manure andandscape management practices to the test during a series ofeld experiments that were conducted under the supervision of aumber of researchers from different departments of Wageningenniversity. Half way during this phase (phase 2) the researcherseveloped a major conflict on the statistical interpretation of thexperiments and the vision in the niche became fragmented. Thisonflict lingered on in the network leading to a decrease in the
ize of the niches network. Unable to prove beyond dispute theositive results of their experiments, the NFW farmers were forcedo broaden their initial goals to include new goals of regionalustainable development. The network was broadened and theable 6haracterisation of niche internal processes and external support.
Phase Dates Internal niche processes
Network composition(Leading actors)
Vision alignment and tru
0 1990–1992 Local farmers Establishment of twoenvironmental cooperativ
1 1993–1997 Environmentalcooperatives & provincialgovernment
Alignment: developmentpilot plans
2 1998–2005 WUR scientists &environmentalcooperatives
Vision fragmentation andbreak down of trust
3 2005–2008a Municipalities, regionalgreen NGOs and scientists
New vision on regionaldevelopment and govern
a End of data.
licy 42 (2013) 613– 623 621
farmers were one of the partners, but not the most importantones.
The composition of the niches network depends on the projectsthat it undertakes as part of the learning and experimentationprocess. The composition thus remains fairly constant withineach phase. At first the niche had two main goals: landscapemanagement and nutrient management. Although the landscapemanagement initially faced strong opposition, the experimentswith landscape management by farmers were successful and in theyear 2000 it became official government policy. After this year nonew landscape management projects were initiated anymore andthis is a sign of success. The nutrient management projects werefar more controversial and the experiments only raised new ques-tions instead of answering them. The experiments done in the nichedepended on exemptions from existing environmental legislationand each new project therefore had to obtain a new exemption forfurther testing and experimentation. So far the NFW has been ableto secure the political backing but it has to be expected that this sit-uation will not last if the nutrient management approach remainsscientifically controversial.
This brings us to the question the relationships of the localniche and supporting organisations such as government institu-tions and research funds that are not necessarily part of the projectnetworks. Table 6 also shows the influence of some of these actorswithin the different phases. Funding from the Ministry of Hous-ing, the Environment and Spatial Planning provided the necessaryfinancial means that sustained the network in the first phase. Thesecond phase of the niche started when the Minister approvedthe exemption of national environmental legislation under thecondition that the field experiments would be conducted underthe supervision of a number of scientists. Additional funds fromthe Dutch organisation for scientific research made it possibleto include even more researchers in the experiments. In the lastphase, the network composition was strongly influenced by therequirements of TransForum. The shifts from one phase to anothercan be attributed to these powerful organisations that influencedthe niche through the conditions they set on the collaboratingpartners in return for their financial or legislative support. Thenetwork expands and decreases in time along with the financesprovided by various governmental subsidies that sustain it. Furtherresearch in the evolution of collaborative networks should focuson quantifying this effect, not only in SNM cases where govern-
ment is very influential, but also in more commercial cases wherethe collaborating partners themselves or banks or venture capi-tal provide the funds and resources necessary for the network toexpand.External support
st Learning and experimentation(projects)
esNone
of Nutrient management andlandscape management
Award and grant from Ministryof Housing SpatialDevelopment and theEnvironment
Nutrient management andscientific projects, proving theeffectiveness of nutrientmanagement
Exemption Ministry ofAgriculture NatureConservation and Fisheries;subsidy from the DutchOrganisation of ScientificResearch (NWO)
anceNew governance forms Subsidy from TRANSFORUM
6 rch Po
tanwunmoMrypsetaitiatnac
cwtpitBbtpatdraf
rmbgobtawsinnip
6
tiu
22 F. Hermans et al. / Resea
A limitation of our study is the fact that we have only inves-igated the ties as expressed through the official membership of
multidisciplinary project. This has the disadvantage that it doesot look at the role of weak ties in the development of the net-ork. The decision who to invite for collaboration in the nichesually starts with some informal contacts between possible part-ers (Ahuja, 2000). However, the choice of our data gatheringethod based on archival information limited the possibilities
f exploring these important mechanisms. For instance, after theinister of agriculture gives the exemption for the national envi-
onmental legislation at the end of 1996, it still takes anotherear before the cooperatives bring in new partners and the newrojects commence (in 1998). Similarly it also takes a while toet up the new projects funded by TransForum and it is at thend of the different phases that these informal contacts are likelyo play an important role that currently remains hidden in thenalysis. Further research should therefore focus on the partner-ng process in niches in more detail by incorporating also otherypes of relationship ties. This might also be helpful to furthernvestigate the centralisation process in the niches network. Usingffiliations did not reveal any significant centralisation process inhe niche (proposition 2). However, applying another measure foretwork ties (like money flows or formal authority ties betweenctors) might still shed more information on the centralisation pro-ess.
With regard to general transition theory, the application of SNAomplements the multi-level perspective that is commonly in usehen studying socio-technological transitions. More research in
his area is necessary but the network evolution perspective weresent in this paper has the potential to allow for more detail
n the study of the relation between niche and regime actorshan is currently possible with the multi-level perspective alone.y reframing the interactions between competing niches andetween niches and regimes in network terms it becomes possibleo study the networks that are formed around different ideas andractices and analyse the different positions organisations andctors have within these network and their relationship towardshe niche in more detail. Applying this perspective reframes theevelopment and spread of socio-technical innovations as theesult of a process in which many different actors and organisationsre linked together by the different projects that they cooperate,und or provide legislative space for.
On a practical level the results show the importance of theecognition of the different possible phases in a niches develop-ent for policy makers. Institutional actors have been shown to
e very influential in determining both the composition and theoals of the niche. For instance, the requirement of the Ministryf Agriculture to involve scientists led to a network dominatedy researchers, while the requirement of TransForum to move ono regional development saw the inclusion of new actors and thedoption of a new discourse focussing on regional developmentithin the network. Whether intentional or not, funding criteria
hape the room for a niche to develop in. Government policies aim-ng to support a niche’s growth and development should thereforeot only cover a range of policy instruments (from shielding, tourturing to empowerment, see Smith and Raven, 2012) but more
mportantly implement these instruments according to the kind ofhase the niche is in.
. Conclusions
In this paper we have used a simple and elegant method to maphe various network configurations in a niche over time by focus-ng on the flow of (multidisciplinary) innovation projects that arendertaken by a changing group of people and their organisations.
licy 42 (2013) 613– 623
As projects start or end the network configuration changes accord-ingly. We have applied this method on the case of the NorthernFrisian Woodlands and we have shown that we can distinguishbetween three different phases in which the vision in the nicheand subsequently also the projects and experiments done in theniche differed. We have shown how the structural characteristicsof the network of a niche evolve over time and how its size anddensity are related to the building of trust as a result of success-ful experimentation. Successful experiments will not only increasethe size of the network but will also increase its connectednessat the same time. However, not all experiments lead to consensusand a fragmentation of visions between niche partners triggers theopposite: it leads to the erosion of trust within the network anda shrinking and a more disconnected network. The compositionof the network and its mean degree remain (relatively) constantwithin each phase, but differ significantly between the phases. Theshifts from one phase to another can therefore be attributed to thesepowerful organisations that are not necessarily part of the niche,but influence it through the conditions they set on the collaborat-ing partners in return for their financial or legislative support. Thenetwork expands and decreases in time along with the financesprovided by various governmental subsidies that sustain it.
Acknowledgements
This research has received funding from TransForum and wascarried out as part of TransForum’s Scientific Programme on Imagesof Sustainable Agriculture. All decisions regarding the study’sdesign and the collection, analysis and interpretation of data andthe writing of this paper were taken by the authors.
Appendix A. Supplementary data
Supplementary data associated with this arti-cle can be found, in the online version, athttp://dx.doi.org/10.1016/j.respol.2012.10.004.
References
Ahuja, G., 2000. Collaboration networks, structural holes, and innovation: a longi-tudinal study. Administrative Science Quarterly 45, 425–455.
Anderson, B.S., Butts, C.T., Carley, K., 1999. The interaction of size and density withgraph-level indices. Social Networks 21, 239–267.
Argyris, C., Schön, D.A., 1978. Organisational Learning: A Theory of Action Perspec-tive. Addison-Wesley, Reading, MA.
Batagelj, V., Mrvar, A., Pajek – Program for Large Network Analysis, 1.26 ed.http://vlado.fmf.uni-lj.si/pub/networks/pajek
Beers, P.J., Veldkamp, A., Hermans, F., van Apeldoorn, D., Vervoort, J.M., Kok, K., 2010.Future sustainability and images. Futures 42, 723–732.
Bender-DeMoll, S., McFarland, D.A., 2006. The art and science of dynamic networkvisualization. Journal of Social Structure 7.
Berends, H., Van der Bij, H., Debackere, K., Weggeman, M., 2006. Knowledge sharingmechanisms in industrial research. R&D Management 36, 85–95.
Burgess, J., Chilvers, J., 2006. Upping the ante: a conceptual framework for designingand evaluating participatory technology assessments. Science and Public Policy33, 713–728.
Burt, R.S., 2005. Brokerage and Closure; An Introduction to Social Capital. OxfordUniversity Press, Oxford, UK.
Buskens, V., 1998. The social structure of trust. Social Networks 20,265–289.
Caniëls, M.C.J., Romijn, H.A., 2008. Actor networks in Strategic Niche Management:insights from social network theory. Futures 40, 613–629.
Coleman, J.S., 1988. Social capital in the creation of human capital. Amer-ican Journal of Sociology 94, Supplement: Organizations and InstitutionsSociological and Economic Approaches to the Analysis of Social Structure,S95–S120.
De Nooy, W., Mrvar, A., Batagelj, V., 2005. Exploratory Social Network Analysis withPajek. Cambridge University Press, New York.
Degenne, A., Forsé, M., 1999. Introducing Social Networks. Sage Publications, Lon-
don/Thousand Oaks/New Delhi.Elzen, B., Leeuwis, C., Van Mierlo, B., 2008. Anchorage of Innovations: AssessingDutch efforts to use the greenhouse effect as an energy source, Acting withscience, technology and medicine. Society for Social Studies of Science and theEuropean Association for the Study of Science and Technology, Rotterdam.
rch Po
E
E
E
G
G
G
G
G
G
G
G
H
H
H
H
K
K
K
K
L
L
L
M
M
P
R
R
87–96.
F. Hermans et al. / Resea
shuis, J., 2006. Kostbaar Vertrouwen: Een Studie Naar Proceskosten en Pro-cesvertrouwen in Beleid Voor Agrarisch Natuurbeheer Mansholt GraduateSchool of Social Sciences. Wageningen University and Research Centre, Eburon,Delft.
shuis, J., Stuiver, M., 2005. Learning in context through conflict and alignment:farmers and scientists in search of sustainable agriculture. Agriculture andHuman Values 22, 137–148.
shuis, J., Stuiver, M., Verhoeven, F.P.M., Van der Ploeg, J.D., 2001. Goede Mest StinktNiet; Een Studie Over Drijfmest, Ervaringskennis en Het Terugdringen Van Min-eralenverliezen in De Melkveehouderij/Good Manure Does Not Smell; A Studyon Slurry, Experiential Knowledge and the Reduction of Nutrient Losses in DairyFarming. Rurale Sociologie, Wageningen universiteit, Wageningen.
eels, F.W., 2002. Technological transitions as evolutionary reconfiguration pro-cesses: a multi-level perspective and a case-study. Research Policy 31,1257–1274.
eels, F.W., 2005. The dynamics of transitions in socio-technical systems: amulti-level analysis of the transition pathway from horse-drawn carriages toautomobiles (1860–1930). Technology Analysis & Strategic Management 17,445–476.
eels, F.W., 2006. The hygienic transition from cesspools to sewer systems(1840–1930): the dynamics of regime transformation. Research Policy 35,1069–1082.
eels, F.W., Raven, R., 2006. Non-linearity and expectations in Niche-DevelopmentTrajectories: ups and downs in Dutch Biogas Development (1973–2003). Tech-nology Analysis & Strategic Management 18, 375–392.
eels, F.W., Schot, J., 2007. Typology of sociotechnical transition pathways. ResearchPolicy 36, 399–417.
enus, A., Coles, A.-M., 2008. Rethinking the multi-level perspective of technologicaltransitions. Research Policy 37, 1436–1445.
ranovetter, M., 1985. Economic action and social structure: the problem of embed-dedness. American Journal of Sociology 91, 481.
ranovetter, M., 1992. Problems of explanation in economic sociology. In: Nohria, N.,Eccles, R.G. (Eds.), Networks and Organization. Harvard Business School Press,Boston, MA, pp. 269–299.
andcock, M.S., Hunter, D.R., Butts, C.T., Goodreau, S.M., Morris, M., 2003. SoftwareTools for the Statistical Modeling of Network Data, 2.1 ed.
ead, B., 2008. Assessing network-based collaborations. Public Management Review10, 733–749.
ermans, F., Stuiver, M., Beers, P.J., Kok, K. The distribution of roles and functionsfor upscaling and outscaling innovations in agricultural innovation systems.Agricultural Systems, http://dx.doi.org/10.1016/j.agsy.2012.09.006, in press.
ermans, F.L.P., Haarmann, W.M.F., Dagevos, J., 2011. Evaluation of stakeholderparticipation in monitoring regional sustainable development. Regional Envi-ronmental Change 11, 805–815.
emp, R., Rip, A., Schot, J., 2001. Constructing transition paths through the manage-ment of niches. In: Garud, R., Karnøe, P. (Eds.), Path Dependence and Creation.Lawrence Erlbaum Associates, Publishers, London, pp. 269–299.
emp, R., Schot, J., Hoogma, R., 1998. Regime shifts to sustainability throughprocesses of niche formation: the approach of strategic niche management.Technology Analysis & Strategic Management 10, 175–198.
noben, J., Oerlemans, L.A.G., Rutten, R.P.J.H., 2006. Radical changes in inter-organizational network structures: the longitudinal gap. TechnologicalForecasting & Social Change 73, 390–404.
noke, D., Yang, S., 2008. Social Network Analysis; Second Edition, 2nd ed. SagePublications, Los Angeles.
eeuwis, C., Pyburn, R., 2002. Wheelbarrows Full of Frogs; Social Learning in RuralResource Management. Koninklijke Van Gorcum BV, Assen.
oorbach, D., Rotmans, J., 2006. Managing transitions for sustainable development.In: Olshoorn, X., Wieczorek, A. (Eds.), Understanding Industrial Transformation:Views from Different Disciplines. Springer, pp. 187–206.
opolito, A., Morone, P., Sisto, R., 2011. Innovation niches and socio-technical tran-sitions: a case study of bio-refinery production. Futures 43, 27–38.
arkard, J., Truffer, B., 2008. Technological innovation systems and the multi-levelperspective: towards an integrated framework. Research Policy 37, 596–615.
oody, J., McFarland, D.A., Bender-DeMoll, S., 2005. Dynamic network visualization.American Journal of Sociology 110, 1206–1241.
ahl-Wostl, C., Craps, M., Dewulf, A., Mostert, E., Tàbara, D., Taillieu, T., 2007. Sociallearning and water resources management. Ecology and Society 12.
Development Core Team, 2008. R: A language and environment for statisticalcomputing, 2.8.0 ed. Foundation for Statistical Computing, Vienna, Austria.
aven, R., Van den Bosch, S., Weterings, R., 2010. Transitions and strategic nichemanagement: towards a competence kit for practitioners. International Journalof Technology Management 51, 57–74.
licy 42 (2013) 613– 623 623
Raven, R.P.J.M., 2005. Strategic Niche Management for Biomass: A ComparativeStudy on the Experimental Introduction of Bioenergy Technologies in theNetherlands and Denmark. Eindhoven University of Technology, Eindhoven.
Renting, H., Van Der Ploeg, J.D., 2001. Reconnecting nature, farming and society:environmental cooperatives in the Netherlands as institutional arrangementsfor creating coherence. Journal of Environmental Policy and Planning 3,85–101.
Rip, A., Kemp, R., 1998. Technological change. In: Rayner, S., Malone, E.L. (Eds.),Human Choice and Climate Change. , pp. 327–399.
Roep, D., Ploeg J.D. v. d., Wiskerke, J.S.C., 2003. Managing technical-institutionaldesign processes: some strategic lessons from environmental co-operativesin the Netherlands. Netherlands Journal of Agricultural Science 51,195–217.
Rosenkopf, L., Tushman, M., 1998. The co-evolution of community networks andtechnology: lessons from the flight simulation industry. Industrial and CorporateChange 7, 311–346.
Schot, J., Geels, F.W., 2007. Niches in evolutionary theories of technical change.Journal of Evolutionary Economics 17, 605–622.
Schot, J., Geels, F.W., 2008. Strategic niche management and sustainable innovationjourneys: theory findings, research agenda en policy. Technology Analysis &Strategic Management 20, 537–554.
Smith, A., 2006. Green niches in sustainable development: the case of organic foodin the United Kingdom. Environment and Planning C: Government and Policy24, 439–458.
Smith, A., 2007. Translating sustainabilities between green niches and socio-technical regimes. Technology Analysis & Strategic Management 19, 427–450.
Smith, A., Raven, R., 2012. What is protective space? Reconsidering niches in tran-sitions to sustainability. Research Policy 41, 1025–1036.
Smith, A., Voss, J.-P., Grin, J., 2010. Innovation studies and sustainability transitions:the allure of the multi-level perspective and its challenges. Research Policy 39,435–448.
Soh, P.-H., Roberts, E.B., 2003. Networks of innovators: a longitudinal perspective.Research Policy 32, 1569–1588.
Spielman, D., Davis, K., Negash, M., Ayele, G., 2010. Rural innovation systems andnetworks: findings from a study of Ethiopian smallholders. Agriculture andHuman Values, 1–18.
Stokman, F.N., 2001. Networks: social. In: International Encyclopedia for the Socialand Behavioural Sciences, pp. 10509–10514.
Stokman, F.N., Doreian, P., 1997. Evolution of social networks: processes and princi-ples. In: Doreian, P., Stokman, F.N. (Eds.), Evolution of Social Networks. Gordonand Breach Publishers, Amsterdam.
Stuiver, M., 2008. Regime Change and Storylines, A Sociological Analysis of ManurePractices in Contemporary Dutch Dairy Farming, Mansholt Graduate School ofSocial Sciences. Wageningen University and Research Center, Wageningen.
Stuiver, M., Van der Ploeg, J.D., Leeuwis, C., 2003. The VEL and VANLA environmentalco-operatives as field laboratories. Netherlands Journal of Agricultural Science51, 27–39.
Stuiver, M., Wiskerke, J.S.C., 2004. The VEL and VANLA environmental co-operativesas a niche for sustainable development. In: Wiskerke, J.S.C., Van der Ploeg, J.D.(Eds.), Seeds of Transition: Essays on Novelty Production, NIches and Regimesin Agriculture. Royal Van Gorcum, Assen.
Van Apeldoorn, D.F., Sonneveld, M.P.W., Kok, K., 2011. Landscape asymmetry of soilorganic matter as a source of agro-ecosystem resilience. Agriculture. Ecosystems& Environment 140, 401–410.
Van de Poel, I., 2000. On the role of outsiders in technical development. TechnologyAnalysis & Strategic Management 12, 383–397.
Van der Ploeg, J.D., Bouma, J., Rip, A., Rijkenberg, F.H.J., Flaminia, V., Wiskerke, J.S.C.,2004. On regimes, novelties, niches and co-production. In: Wiskerke, J.S.C., Vander Ploeg, J.D. (Eds.), Seeds of Transition. Royal Van Gorcum, Assen.
Van der Ploeg, J.D., Strijker, D., Hoofwijk, H., 2007. Noordelijke Friese Wouden:Leerervaringen. TransForum, Zoetermeer.
Van der Valk, T., Chappin, M.M.H., Gijsbers, G., 2011. Evaluating innovation networksin emerging technologies. Technological Forecasting & Social Change 78, 25–39.
Veldkamp, A., Altvorst, A.C.V., Eweg, R., Jacobsen, E., Kleef, A.V., Latesteijn, H.V.,Mager, S., Mommaas, H., Smeets, P.J.A.M., Spaans, L., Trijp, J.C.M.V., 2009. Trigg-ering transitions towards sustainable development of the Dutch agriculturalsector: TransForum’s approach. Agronomy for Sustainable Development 29,
Wasserman, S., Faust, K., 1994. Social Network Analysis. Cambridge University Press,Cambridge.
Wiskerke, J.S.C., Van Der Ploeg, J.D., 2004. Seeds of Transition; essays on noveltyproduction, niches and regimes in agriculture. Royal Van Gorcum, Assen.