Social Network Analysis
CHAPTER I: INTRODUCTION
1.1 Research Design
The proposed design for this study is a flexible mixed-method design making use of both qualitative and quantitative methods for data collection and analysis in order to understand and later analyse policy implementation networks for MSEs in Ireland, particularly concerning those in the Dublin 15 area.
This approach was required for two reasons: 1) time and resource constraints; and 2) the literature reviewed for this research indicated that such an approach would provide better answers to the questions being asked. As shown in Table 1 below, this method was also important to establish the complimentary nature of the methods being used for the enquiry. For example, qualitative methods were required initially for content analysis of documents produced by the European Commission and the Dept. of Trade, Enterprise and Employment. However, a quantitative approach is given to the treatment of the relationships under investigation namely, mandatory and voluntary cooperation for policy implementation. Therefore, the following steps were taken for this particular research:-
1. Internet research for qualitative content analysis and name generation of policy actors and policies being made at both European and Irish levels;
2. Short open-ended qualitative interview with a major policy actor in Ireland for name generation of policy implementation actors;
3. Quantitative data collection and non-random sampling and estimate number of start-up companies in the Dublin 15 area; and
4. Data collection and analysis using Social Network Analysis methodology for all names generated to measure the effects of mandatory and voluntary cooperation among policy actors. Patience
A follow-up interview with a key informant on the network composition and to confirm that the relationships drawn on the literature for policy implementation indeed existed in the manner they were in the questionnaire. In this case, a Head of Small Business Unit in the Department of Enterprise, Trade and Employment could confirm the relationship among the actors in the network from an ‘ego’ perspective and pros and cons for this approach will be discussed further in the methodology chapter.
The table below gives an illustration of the design being used and it follows the design approach suggested by Creswell et alli 2003:-
Table 1 – Sequential Research Design
Sequential – Qualitative first
At Data Collection and Interpretation
With Some Combination
As mentioned earlier, this design draws on Creswell’s (2003) work about various research design approaches – with special attention to the use of mixed-methods in the human sciences. A rather overt and systematic use of the methods have grown in popularity and it is evidenced in the work entitled Handbook of Mixed Methods in Social Behavioral Sciences, launched in 2003 by Abbas Tashakkory and Charles Teddlie and from which Creswell derives most of his useful explanations on the advantages and weaknesses of such a design.
In a flexible design, the researcher is given a greater freedom to change research questions as the research progresses since she would be rigidly attached to one philosophical paradigm or the other, but would find relevant to include those that make sense to the enquiry as it evolves. On the other hand, making use of both qualitative and quantitative methods for data collection and analysis – whether they take place sequentially or concurrently, allows for a richer understanding and, hopefully, explanation of the problematic under investigation.
Details of the methodologies and paradigms they are associated with are dealt with in the Methodology chapter. For this section is imperative now that the researcher makes the esteemed reader aware of the possible biases and the actions taken to deal with them.
It has been argued that bias is when systematic random or non-random errors are not accepted or acknowledged by the researcher at any point of her work, such as in the design, measurement, sampling, procedure, or choice of problem to be studied. Bias is also related to the accuracy of the information produced and reliability and validity may also be sacrificed, though not necessarily.
Hammersley and Gomm (1997), however, assert that nonfoundationalist realism is rather appropriate for it neither relies on foundational epistemology nor it does on relativist and standpoint theory and its aims are to favour the production of knowledge rather than one’s political or personal agenda. According to these authors, foundationalism makes bias a rather evident matter in a research once the researcher’s ideological commitments are indentified as affecting research procedures hence compromising outcomes expected credibility. But bias is not as straightforward as foundationalists, according to them, would like it to be. In a nonfoundationlist realistic perspective, social researchers are not part of the phenomena they want to describe or explain nor can they construct phenomena through their accounts, but merely account for these realities and be judgemental of the influence that what they understand for established knowledge can have on their accounts of social reality. (1997)
In this research, in a foundationalist perspective, the biases that could possibly compromise the commitment with knowledge production is that:-
1) the researcher is a micro-entrepreneur from the vicinities;
2) her commitment with epistemological views of policy networks as ineffective given the predominant hierarchical structure and political agendas;
3) Snowball sampling technique may introduce bias for it may not represent a good cross section of the population investigated; and
4) Questionnaires sent to MSEs were not returned for reasons which will be explained in the Methodology chapter. This can also lead to incorrect conclusions.
However, this is not an action research and the methodologies applied for enquiry on policy implementation networks aim to control for bias arising from the researchers’ own point of view regarding cooperation for the benefit of Small and Micro Enterprises in Dublin 15. Moreover, from a nonfoundationalistic realism view point, efforts were consciously made in order to avoid that knowledge so far taken as established in the field of cooperation among actors in policy networks compromises the knowledge produced by the evidence on cooperation among actors in this research.
Taking the above into consideration, one of the methodologies used is a qualitative approach considering respondents’ representations of reality through a qualitative content analysis and qualitative techniques in social network analysis.
Once data were generated, they were treated by using quantitative data analysis techniques and were converted into algebraic formulae reflecting responses given, i.e. whether a relation exists or not between two actors or more and if it is reciprocal one is a matter of replacing YES and NO answers, or in this case BLUE and RED arrows, for 1 and 0, respectively. This is not to say that social network analysis will serve as a complete assurance for validity and credibility, as questionnaire return rates and responses can indeed compromise it. It is hoped, however, that the weaknesses evident on this work are rather derivative from the very limited time afforded to Taught Masters students, effectively from May to August 2009, than from research bias.
This work has a scientific and practical relevance and they are equally important as it is hoped that its findings and conclusion will not just add to the scientific knowledge but also reach the actors involved in policy implementation for micro and small enterprises in Ireland as well as those concerned with aspects of economic development.
In the scientific sphere, this study will be initially guided by O’Toole’s (1997) work stressing the increasing complexity of networked settings in the public administration where he also urges for more empirical research to be conducted in order to comprehend the impact of such settings on policy implementation. This is also Blair’s (2002) views in his work on implementation networks in which he infers that complex systems formed by actors both coming from both the public and private sectors are assumed in now-a-days public administration.
According to Blair, this will help measure the level of direct action by the government so to assess the complexity of such networks and he goes on to borrow from Alexander (1995) the concept of compulsory and voluntary links as the essence of policy implementation systems. Further to such concepts, the focus of this study is on mandatory and voluntary cooperation amongst MSEs stakeholders both in the public and private sectors.
Having it all said, measuring service integration among agencies is not the objective of this study. Instead, it will try to leverage on social network analysis practice and theory to measure centrality, closeness, structural holes and brokerage (Freeman 1979; Burt 1995 and 2005). Other theorists, especially those whose studies are within social network analysis (Simmel 1995; Wellman 1983; Hanneman 2005), and regarding the bureaucratic aspect of the public administration (Weber 1978 and Blau 2006) and the (potential) social capital resulting from networks (Bourdieu 1986; Putnam 1995; and Fukuyama 1996) will also be used in the hope to complement existing literature on the topic. The above will, indeed, guide this study all the way through.
1.3 Chapter Outline
CHAPTER III: LITERATURE REVIEW
Defining the size of a business
This research does not aim to be seen under the umbrella of business studies, but a sociological study of cooperation among policy implementation actors for policies in benefit of micro and small businesses in Ireland. Nevertheless, it is found pertinent to use business literature in order to outline the most commonly used definitions of business’ size according to factor inherent to its life cycle.
The literature on organisation life cycle (OLC) has been found to be the most appropriate to define the size of a business. However, it is important to note that professionals used to working with businesses may find their own ways of defining business’ size and may as well widely disagree from the scholarly models created and widely used to date.
The traditional models for sizing a company were given mostly in the 1980s first through conceptual works and eventually through empirical research which attempted to establish the phases of an organisation according to the environmental issues it faced throughout its existence (Rutherford et al. 2004).
The literature consulted for this section has shown that one of the models that has been most cited and used as a reference is that created by Miller and Friesen (1984) whereby the organisation life cycle was divided into five phases: birth, growth, maturity, revival and decline (1984: 1162). These phases can be summarised according to the age, size and growth rate of the organisation. In summary, these phases were the birth phase for small companies, the growth phase for medium-sized companies, the maturity phase for larger organisations, the revival for very large and finally the decline phase for market size companies. The latter phase is linked with the decline in innovation and with environmental factors such as market dry-up. Meanwhile the earlier phases of the business are linked with the most innovative, although unstructured, stages of the organisation (1984: idem). Figure 2 below exemplifies the above:-
Figure 2. Organisation life cycle (OLC) and its expected size per phase.
The head count for the traditional models is shows as 0-4 persons employed in the birth phase, then 5-19 employees in the medium-sized firms and finally above 20 in the larger ones (Huang and Brown 1999 cf. Rutherford et al. 2004).
Weaknesses of this model have been pointed by more recent works (Stubbart and Smalley 1990; Tichey 1980 cf. Rutherford, Buller and McMullen 2004) which took different views and approaches to establishing the different phases an organisation goes through. Other works making use of open-ended choices rather than forced-choice methods to identify an organisation’s problems according to its size, have found that obtaining external financing, internal financial management, economic environment and regulatory environment are problems often experienced by organisations in their start-up and growth phase (Terpstra and Olson 1993 cf. Rutherford, Buller and McMullen 2004)
Taking definition from a different angle, the authors Rutherford et al. (2004) infer that other models have been created to overcome the weaknesses presented in previous models and the model by Kihenen (1990), which makes use of Artificial Neural Network (ANN) in order to depict rather realistic stages of an organisation given a series of other factors not limited to age, size and growth rate and it is similar to a cluster analysis and named as self-organising map (SOM). This model, different from traditional models, including the model created by Miller and Friesen (1984), does not force firms to fit into predetermined groups but allows for stages to fall into categories given by the data under study (2004: 328). Nevertheless, the traditional model remains imperative throughout their work and so it will throughout this work. The reason is twofold: a) the traditional definitions have been confirmed through empirical work and across countries; and b) it is assumed here that the problems faced by an organisation in each phase of its existence can be logically linked with the needs of these organisations for growth.
Hence, policy-making and implementation, as it will be demonstrated later on, is more likely to follow traditional models than novel models. Though this is not to say that new findings on an organisation’s functioning will not be appreciated by policy actors. Moreover, taking into consideration the criticism on the traditional models it is therefore acknowledged that the adoption of traditional model is merely as starting point for reference given the insufficient time for the maturing of this study. Table 3 below shows the most common problems in organisations given their size and phase found by traditional models which have also been confirmed by more current studies utilising different organisational typology techniques:-
Table 3. Problems helping establish the size and stage of an organisation.
|Newness, obtaining external finance, internal finance management, unstructured, informal, product development, economic and regulation environment, weak client base, owner/manager dominated.||Birth/Start-up (Small firms)||0-4 employees|
|Production efficiency and effectiveness, human resources management and re-organisation, economic and regulation environment.||Growth
|Maintaining growth momentum and market position, economic environment, production stagnation and lack of innovation, and employee retention.||Stability/Maturity and Decline
Source:Miller and Friesen 1984; Kazanjian 1988; Terpstra and Olson 1993 and Huang and Brown 1999 (cf. Rutherford et al. 2004).
Conceptualising the phases and problems of organisations will help us understand the policies made in benefit of micro and small businesses in the European Union. However, a point in time must be made regarding the micro-enterprises that may not fit into the typology above given the ethos of their existence.
In Europe, the crafts industry is within the definition for micro-enterprises as well as entrepreneurs in the rural areas. However, one should also consider the large number of self-employed professionals such as plumbers, carpenters, painters and also accountants whom may never expand their business, formalising it into a company hence adhering to the organisational life cycle above. Nevertheless, they too generate employment and tax revenue on top of their networked oriented business behaviour hence contributing to the country’s economy but as well as for social and local integration in Europe (European Commission’s SME Portal – Observatory Report of European Enterprises 2003/7).
Given the above, the needs for micro and small enterprises will vary according to their problems and other human aspects inherent of the cultural and social settings and mindset businesses are found. These settings have been acknowledged by the European Commission in providing assistance to groups widely seen as minorities in large scale businesses, but forming a large number among micro and small entrepreneurs, namely women and immigrants.
The Concept of Policy Implementation Networks
In policy network theory, implementation networks have been categorised and conceptualised mostly according to the context in which they have emerged in order to either reduce transaction costs or resolve problems effectively. Such a context will determine the resources and power that members will share on their way to achieve their shared goals or resolve their common problems (Provan and Milward 1995 and Bozer 1996). The subject of context is here defined as the ethos and habitus of policy implementation networks in general and they will be explained later on in this chapter.
Modern states are now dealing with problems of greater complexity and this requires a superior level of negotiation and public participation in order for services to be delivered to the public as effectively as it is possible. For this reason, O’Toole (1997) asks researchers and public administrators to take networks more seriously and calls for a whole research agenda on the effects they have on hierarchical structures common to public agencies and ultimately affecting policy making, implementation and evaluation (O’Toole 1997; Provan and Milward 1995, 1998 and 2001; Bolzer 1996; and Sandström and Carlssson 2008, among others).
The last decade saw a particular growth in research interest on policy implementation networks. Most of the work produced, once satisfied with the working concepts and theoretical frameworks on the subject, now concentrates on measuring implementation effectiveness from different angles and using different methods (idem).
While O’Toole (1997 and 1999) remained mostly in the theoretical field hypothesising on management action in hierarchical structures (O’Toole and Meier 1999) and Bozer (1997) strived to conceptualise networks as forms of governance, others like Blair (2002) took the empirical road either by putting to test models created by O’Toole and Meier (1999) or utilised Social Network Analysis in order to analyse networks for what they really are, social structures with interdependent interests (ibidem).
As to the initial theoretical framework for policy implementation networks, O’Toole and Meier (1999) created models for managerial action in public administration taking into account the prevailing hierarchical structure of public agencies. For them, networks are of great help to public servants and they see the two structures, linear and non-linear, as poles of a continuum related to buffering in the interests of stability (1999: 510).
However, it is important to note that the emphasis of their work is on improving hierarchical systems rather than replacing them for networks since the authors see the informal and inconsistent nature of networks as a weak structure for buffering environmental issues common to policy implementation or service delivery while they also believe that a hierarchical system can maximise management resources more than networks as the latter demands more of such resources in order to maintain reciprocal ties. Though, they do acknowledge that not all networks have inconsistent relationships and have achieved stability by sharing a consistent policy agenda away from a hierarchical habitus, e.g. US farmers’ network, borrowing the term in italics from Tocqueville (2004: 329).
Blair (2002), on the other hand, deals with empirical quantitative research on policy implementation in the field of economic development by studying the so-called Enterprise Zones in Southern United States, very popular in the 1990s, especially during Reagan’s second term in presidency. He makes use of policy tools theory and methodology while also drawing from O’Toole and Meier’s (1999) conceptual framework and models for public management in policy systems to formulate his three major hypotheses that aim at measuring the level of direct government intervention in service delivery systems for Enterprise Zones in distressed areas.
The first hypothesis states that less government direct intervention means more complex implementation networks, while the second hypothesis continues from the first by asserting that less government participation also means that more actors are involved in implementation, especially private and non-profit organisations. Finally, the third hypothesis, which will later provide the most confounding findings, posits that the private sector takes a greater role in implementing programs in economic development policy networks (2002:170).
The first two hypotheses were validated by the findings and agreed with the theoretical concept framework by O’Toole and Meier (1999) on the structure of new policy implementation systems. However, the third hypothesis for which he expected a negative correlation between private and public participation in implementation, he, in fact, obtained a positive one. His findings indicated that the more active is government intervention in the EZs investigated, the more participative businesses were in implementing the programs. However, this finding, according to Blair, does not contradict what had been theorised by O’Toole and Meier’s (1999) (cf. Blair 2002), but he classifies O’Toole’s (1997) system as the old implementation system and the later as the new system.
Although policy tools theory is one good way of measuring government intervention in economic development, other approaches may also provide fruitful findings and ways of analysing policy implementation networks.
Bozer (1997), for example, explains that policy networks are conceived as a particular form of governance in modern policy systems (Kenis and Schneider 1991; Kooiman 1993; Mayntz 1994, cf. Bozer 1997). In fact, she notes that their main focus is on problem-solving and not on intermediating interests. Any cooperative game members play in order to negotiate strategies to arrive at a solution and based not only on mutual goals, but mostly on communication and trust while acknowledging that their values, ideas and identities are essential to the network. Therefore, Bozer understands:-
Policy networks as ‘webs of relatively stable and ongoing relationships which mobilise and pool dispersed resources so that collective (or parallel) action can be orchestrated toward the solution of a common policy’ (Kenis & Schneide 1993: 36 cf. Bozer 1993:5). A policy network includes all actors involved in the formulation and implementation of a policy in a policy sector. They are characterised by predominantly informal interactions between public and private actors with distinct but interdependent interests, who strive to solve problems of collective action on a central, non-hierarchical level. (Bozer 1997: idem)
She asserts that scholars whom have dedicated their work to see these network members as rational actors, following a horizontal self-coordination, may forget to include some of the main aspects making for such networks in the first place. For Bozer, looking at the intermediation position that networks can have, one may realise that policy-making can be ‘blocked by dissent’ and create what she sees as the ‘bargaining dilemma’ (or prisoner’s dilemma) where ‘defection from cooperation is more rewarding than compliance for a rational actor, owing to the risk of being cheated’. (1997:4) The exception exists when ‘bargaining is voluntary’.
As approaches to research change over time and so do concepts and theoretical frameworks on policy implementation networks. Though, most references analysed for this work, showed that there is more of a convergence of knowledge on the topic rather than any radical change. It was noted, however, that works on the subject matter have become more and more empirical and it was found that using structural analysis has proved rather fruitful for a much needed progression on studying it.
Aside policy tools, strategic decisions and interest intermediation to share resources and increase productivity and effectiveness, networks are social structures where norms are created and crystallised or otherwise and as such they must be analysed for their properties, the opportunities (advantages) and constraints (disadvantages) that these properties can originate. In other words, since networks are dynamic social structures, the questions asked, either to understand their modus operandi or more so to measure their effectiveness, refer to how actors organise themselves to problem-solving and what are the properties that this can originate. This seems to be the approach taken by a recent study on four policy networks in the educational system in Sweden by Sandström and Carlssson (2008).
They acknowledge the organisational roots of policy network theory and from which important concepts have emerged, such as advocacy coalitions, implementation structures, iron triangles, issue networks, policy communities, and subgovernments (Sabatier & Jenkins-Smith 1993; Hjern & Porter 1993; Jordan & Schubert 1992; Heclo 1978; Jordan 1990; Rhodes 1990; cf. Sandström and Carlssson 2008). Nevertheless, their research is justified by the reasons mentioned earlier, being a lack of substantive empirical work using the methods of social network analysis to study policy networks, since only network theorists in different disciplines (Sociology, Political Science, Business Administration etc.) have been interested in the social capital derived from network properties and which is capable of evaluating outcomes and performance.
The authors concluded that their theoretical and methodological approach, as aforementioned, proved fruitful for the understanding of policy networks as a structural process where heterogeneity and centralisation are beneficial to network performance whereas structural holes (poorly or non-connected actors) should be avoided when the objective is to achieve efficiency (2008:517). In this view, concepts and knowledge generated by their study converge with those generated by previous works mentioned earlier in a complimentary manner.
Evidently, there are several theoretical and empirical works making valuable contribution to the conceptualisation and setting an ever richer theoretical framework to rely on. However, given the short time given for this research, we will settle for what was seen as some of the most useful works with which to begin with.
The ethos and habitus of policy implementation networks
After contemplating some of the most prominent and recent concepts and theories on policy implementation networks and having touched on the subject of context, we now proceed to further the discussion on the latter beginning with seminal works on bureaucracy, the original ethos of public policy making and an aspect of the context which policy networks can be found. After that it will be discussed the idea of habitus as conceived by Tocqueville on his work about the associational ability of North Americans in the United States of his times.
Beginning with the former, Max Webber interpreted the bureaucratic organisation by using terms such as money economy (for its existence), but also stability and rigidity as forming the ‘mechanised bureaucratic apparatus’ while emphasising its ‘technical superiority over every other form’. (1978:345-50) In parallel to contemporary public administration systems, the bureaucratic systems conceived by Weber had similar reasons to evolve and proliferate, being the always increasing complexity in administration, pressure from interested parties for social policy and willingness by the state to increase efficiency while reducing transaction costs. Still according to Weber, authority and hierarchy are also essential to the universe occupied by the public office. However, networked systems, although still embedded in bureaucracy, being hierarchy its essence, would not posses the same strictness and uniformity. As we have seen earlier, this is not to say that today’s public administration cannot achieve stability through ways other than hierarchy. (O’Toole and Meier 1999)
However, it is well known that Weber came from a rather theoretical perspective, which is the opposite of Peter Blau’s (1963) case, whom has written an entire book on the American bureaucratic system as a result of his ethnographical work in the 1950s. For Blau, this type of social organisation could not be defined merely through hypotheses, but was rather the subject of empirical investigation. Moreover, he opted to see bureaucracy as an orchestrated effort of public officials to ‘transform exceptional problems into routine duties of experts’. (Hughes 1951, cf. Blau, 1963:251)
In fact, Blau acknowledges that the bureaucratic system in Germany during Weber’s times were very different from the same system on Blau’s times, especially if we consider they also refer to different countries, hence of different cultures. Even if we accept that expertise, or specialism, has been the raison d’être of public administration (again, for transaction cost reduction and efficiency’s sake), social interaction among servants are always changing and recurrent cooperation among workmates and just not friends would help contribute to the emergence of the new ethos of policy implementation networks. (Blau 1963:259; Bozer 1997)
Therefore, one can hypothesise that repeated social interaction among individuals within the same group or organisation will lead to cooperative behaviour and as Bozer (1997) has inferred, such interactions combined with the notion of good communication