3 Methodology
3.1 Research design 53
Approaches to assessing the impact of the environment on migration can be divided into two broad groups within the literature: first, an identification of environmental hotspots and some form of integrated assessment to establish those at risk, of which all or a proportion will be expected to migrate. The second, more analytical approach, attempts to discern the environmental signal from other drivers of migration through a variety of methods including ecological interference, individual analysis, time series analysis, multi-level analysis, agent-based modelling and ethnographies (Findlay, 2011; Kniveton et al., 2008; Piguet et al., 2011). For the purposes of this research, I will
undertake a fine-grained, multi-level analysis of migration under conditions of environmental change that draws on a number of the analytical methods outlined above.
I use an analogue comparative case study approach (Morrissey, 2009; Yin, 2011). Adger et al (2003) differentiate analogues by space and time. Temporal analogues involve taking information from past or present responses to climate change and making inferences about a current or future point in time. Spatial analogues examine present day behaviour and apply the learning to other locations. The use of analogues is widespread in research on climate change adaptation providing insight into real-life events and responses. However, analogues are not without their drawbacks: chief among these is their inability to provide information on changes beyond present day norms13 and the unpredictability of future human behaviours and responses to stimuli.
13 This encapsulates the issue of stationarity whereby it is assumed the natural systems fluctuate within
This is particularly relevant in relation to climate change where no past analogues will exist for certain events (Kniveton et al., 2009; Patt et al., 2005; Williams and Jackson, 2007). Although imperfect, analogues remain one of the only ways to generate
empirical insights into human behaviour under conditions of climatic variability that are likely to mirror to some extent future conditions and thus remain a valid research approach. Accordingly, this research uses a temporal analogue to understand the response of human systems to current climatic perturbations such as droughts and floods in order to make inferences about what might happen in the future (Ford et al., 2010; Glantz, 1990; Glantz, 1991; Glantz, 1996; Goulden, 2006).
When designing the research I was aware of the trade-offs between detail and extent. Migration research tends to fracture along lines of scale. On the one hand extensive approaches focus on regional surveys to identify risk and hence vulnerability
(Gemenne, 2011; Piguet, 2010) and on the other intensive case studies are employed to focus on a set of individuals, households or communities with the intention of establishing how they are responding to a change in circumstances (be it political, social, environmental). For this research, I adopted an intensive approach for a number of reasons.
First, fine-scaled analysis is more appropriate to understand the impact of shorter episodic events. Whereas working at a coarser resolution can reduce the sensitivity of an investigation to such an extent that the impact of finer grained variations are not captured or well represented (Poston and Zhang, 2008; Slingo et al., 2005). Second, extensive analysis limits the extent to which one can ask why, dealing as it does in aggregated units (Ford et al., 2010). I was more interested in understanding the impacts of climatic variability on individuals, households and communities and how they responded to such events. Third, the issue of data access and reliability was a significant cause for concern (Simelton, 2011: 37-38; Zhang, 2008; Zhang, 2011). Therefore, I determined that a research approach that generated the majority of the data (with its limitations known to me) was a more reliable and robust approach14.
As discussed earlier (see section 2.1.2), Sayre (2005) (amongst others) highlights the importance of links between levels arguing that this necessitates cross-scale and
between the individual, the household and the village and use methods (specifically the questionnaire survey, semi-structured interviews and RRA activities) that generate a degree of time-varying data; thus responding to the call for cross-scale and nested research that are often at the core of explanations for people environment interactions (Adger et al., 2009a; Birkenholtz, 2011; Mauro, 2009). Links with larger scales were taken into account during the research but were not explicitly addressed. For example, the conceptual framework was designed with the specific Chinese institutional context in mind, some of the research instruments captured influences of larger scales and the analysis drew on issues and processes that were occurring at a regional or national level.
The mixed methods approach I employed has been utilised in recent research (specifically the EACH-FOR and Rainfalls projects) on the migration-climate change nexus (Piguet, 2010; Warner, 2011; Warner et al., 2012; Warner et al., 2009) and interdisciplinary research more generally (Nuijten, 2011). One of the key strengths of this sort of approach is the different perspectives and layers of understanding that are generated through the data collection and analysis. The multiple methods (such as biophysical climate data, surveying, rural appraisal activities, and interviews) used, enabled me to gain a broad understanding of the ways in which individuals,
households, and communities responded to floods and droughts and role of mobility within this response. A further important benefit of the multi-method approach is the enhanced level of analytical rigour engendered by working iteratively across the
different data. In sum, the methodology permitted me to interrogate the ‘how’ and ‘why’ – a crucial strength of case study research (Yin, 2011).