3. Methodology
3.1 Research design
22 To be further discussed in Chapters 6 and 7.
3.1 Research design
development approach, and it aims to contribute to the design of models for application in the international development field. Figure 3.1 highlights the context of this study.
Figure 3.1 Context of study Source: (Davidsson, 2002)
As the figure illustrates, three inter-relationships constitute agent-based social simulations (this study lies in the intersection of all three fields). Grounded in the dynamics of human interactions, the social sciences provide a context for studying the interactions of social entities in this research.
The field is typically “messy”, with ill-defined boundaries and limited individual capacities (Moss and Davidsson, 2000). Agent-based computing emerges from the field of artificial intelligence as a systems engineering platform (Jennings, 2006) linking the technical and social paradigms. From computer simulations, the techniques of simulating phenomena in a controlled space is provided.
This research operates in the intersection of these fields, drawing from, and aiming to contribute to each. To develop the agent-based model, empirical data is collected adopting tools and approaches from the social sciences i.e. data is formalised for modelling using a conceptual framework of adaptive strategies developed in the anthropological field. To define the governing dynamics of the model, technical infrastructure from systems engineering and computational operations are employed. As such, the existing libraries of agent-based computing models are surveyed, and codes from computing simulations are referenced in an attempt to build (/re-create) the observed social system.
With these fields informing the overall research design, the methodology is structured into two parts. The first part identifies a suitable test case which can speak to the objectives of this thesis i.e. a programme that operationalises the theories of complexity in the education sector (Chapter 2 has already unpacked how the dynamics of the sector is characterised by complexity). With the selection of the system to be modelled, a preliminary profile of the various stakeholders in the system is created and the social, economic and political context can be mapped out. This part also centres around the nature of the data required to satisfactorily model the social system (presented in Figure 3.2) – this aspect encapsulates the granulity of data which will enable the modeller to build a rich and dynamic simulation. Thus, the quality of data is an essential consideration because it determines the degree to which this study can successfully apply the concepts of complex adaptive systems in a model, which in turn can reasonably produce insights into changes in the real system. Accordingly, the methods utilised to gather information are discussed with regards to the level of detail required.
Figure 3.2 Conceptualisation of the target social system Source: Author’s conception
The second part of the methodology involves developing a modelling framework of the social system presented in Figure 3.2. Here, the data collected is formalised through a knowledge engineering process that defines (and continually refines) the governing dynamics of the model from the ground up. Because this research is in part an exploration of a methodological approach,
methodology is weaved into model design and analysis of the model results.23 Consequently, this chapter shall focus more on the first part of the methodology, and will delve more deeply into the second part in subsequent chapters. The end of this section discusses specific challenges relating to data collection, the potential implications of these challenges.
As already discussed in Chapters 1 and 2, this study has three strands of interest: development, education, and modelling complex social systems. To this end, these elements are necessary criteria in selecting the case-study on which to base this research. Early on, the Education Sector Support Programme in Nigeria (ESSPIN) was identified as meeting these primary requisites. In other words, it is:
A development programme in the education sector.
Implemented in six states to date. This provides a rich, diverse and large scope in which to study variation and similarities across contexts. It also allows space for the researcher to choose at which scale to model. This is a critical consideration – insufficient information may result in the researcher making too many unsubstantiated assumptions to fill in gaps, and too much data can sometimes wash out the uniqueness or peculiarities of states by over-abstracting individual experiences. Consequently, it is for the researcher to determine the ideal balance (in this research, this was achieved through a collaborative process with stakeholders).
A programme which involves a variety of stakeholders who work together to successfully implement the programme. It is a dynamic programme which has been greatly influenced by the context of each state. As such, it has performed differently geographically, as well as longitudinally. It exists within a demonstrably a complex social system and transfers well into the agent-based modelling approach.
The scope of ESSPIN is very large; it is implemented in over 16,000 schools and its associated teachers, pupils, communities, education authorities (at federal, state and local government levels), and civil society organisations, among others. Designed as a development programme, it is jointly funded by UK Government and the Nigerian Federal and State Governments and is implemented in six states: Enugu, Jigawa, Kaduna, Kano, Kwara and Lagos. For now, the process of selecting
23 Model design is discussed in Chapter 7, and the model results are presented and analysed in Chapter 8.