The Room Around the Elephant: Tackling Context-Dependency in the Social Sciences

Abstract

Context is crucial for understanding social phenomena, but is not being addressed. Contexts can become socially entrenched and acquire their own labels, allowing different social coordination systems to be developed for different kinds of situation. Three ways to avoid context are discussed. Fitting data to mathematical models which ‘explain’ the data using significance tests avoids the problems of context, but may average over different contexts inappropriately. ‘Behavioural foundationalism’, which assumes a generic model of behaviour that is valid across different contexts, avoids the context problem by producing models based on a micro-specification to see if the macro-consequences match the available data, e.g. neo-classical decision theory and some agent-based simulations. A third strategy to avoid the context problem is to retreat into specificity, providing so much detail that the context is unique with no attempt at generalisation. Three ways forward are proposed (1) using data mining techniques to look for models whose output ‘fits’ various target kinds of behaviour, (2) context-dependent simulation modelling, with the memory of the agent being context-sensitive, and context-relevant knowledge and behaviours being applied in decision-making, and (3) combining qualitative and formal approaches, with neither qualitative nor quantitative evidence being ignored. Agent-based modelling can use qualitative evidence to inform the behavioural strategies that people use in given situations. Simulations based on micro-level behaviours can produce numbers for comparison with macro-level quantitative data. This supports experimentation to understand emerging processes, and investigate the coherence of the qualitative assumptions and the quantitative evidence. Explicitly recognising and including context-dependency in formal simulation models allows for a well-founded method for integrating qualitative, quantitative and formal modelling approaches in the social sciences. Then some of the wealth of qualitative ethnographic, observational and interviewing work of the social sciences can enrich formal simulation models directly, and allow the quantitative and the qualitative to be assessed together and against each other. Before the advent of cheap computing power, analytic mathematical models were the only formal models available, but their simplicity ruled out context dependency, leading to a focus on what generic models might tell us. New information and communication technologies have resulted in a lot more data on social phenomena to distinguish different contexts and behaviours. We no longer have to fit generic models due to data paucity and limits to storage and processing, or ignore context or over-simplify what we observe to obtain and use formal models. Addressing context has huge potential for the social sciences, including: better models and understanding of human behaviour; more effective ways of collecting, integrating and analysing data; and the prospect for a well-founded means of integrating the insights from quantitative and qualitative evidence and models.