INTEGRATION PARTICIPATION
5.3.6 Data analysis: the ‘Framework’ approach
The conceptualisation of rural governance as a symbolic process of integration, participation and empowerment shaped by constant action and interaction among people has implications for the research design in general, as well as for the method of data analysis in particular. At the centre of the research are narratives of local policymakers revealing their perspective on rural governance through their everyday work experiences, which simultaneously affect and are affected by, policy decisions. Hence, a bottom-up approach of data analysis was employed which allows generating theory from a systematic analysis of row, unstructured interview data.
6
Much of qualitative research uses sequential (Becker, 1971) or as Miles (1994) calls it, interim data analysis, in which the analytical process starts during data collection as the data already gathered are analysed and shape the ongoing data collection (Pope, 2000). As evidenced throughout the data collection of this research, the researcher goes back and forth between empirical data and theory, refine questions and develop hypotheses in pursuit of patterns and depth.
Considering the research questions and objectives, the research population and the method of data collection, a relatively recent inductive approach, Framework Analysis (Ritchie & Spencer, 1994) was found to be the most suitable method for data analysis. This approach was developed in the context of applied policy research by the Qualitative Research Unit of the National Centre for Social Research, which is Britain’s largest independent social research institute (Ritchie & Lewis, 2003). This is particularly important because this research resembles, in many of its characteristics, applied policy research.
The basic distinction between theoretical and applied policy research is that the former refers to the traditional academic research which is guided by disciplinary departments of universities, whereas the latter is driven by the specific information requirements and needs of the funding body, typically a public agency, to aid decision-making and/or evaluate policies or programmes (Haas & Springer, 1998; Majchrzak, 1984). Applied policy research is therefore responsive to the study users and provides them with action- oriented recommendations. It is multi-dimensional and empirico-inductive research, which is concerned with policy-manipulable factors and incorporates numerous, sometimes conflicting values (Majchrzak, 1984).
Although Framework Analysis shares many of the common features of much qualitative analysis, particularly Grounded Theory (Glaser, 1967/1999), there are significant differences between them (Lacey & Luff, 2007). Grounded theory sets a broad, general concept as a starting point with no a priori issues or a specific sample assigned (Lacey & Luff, 2007). Theory is generated systematically from the new emerging themes and the sample is identified and expanded gradually by ‘theoretical sampling’, which is not concerned with drawing samples of specific units of analysis such as groups of individuals, but it is driven by concepts, incidents and events, thus usually interrogates a diverse group of people.
In contrast, framework analysis allows the inclusion of a priori concepts in addition to the emergent themes, at various stages of data analysis, for example during the development of the thematic framework and the coding process. This can be particularly important in studies where there are more specific information requirements and pre-defined samples of professional actors to be addressed. The ‘Framework’ approach is a systematic data analysis method based on data reduction by the development and continuous refinement of a thematic framework, which allows the identification of patterns and clusters in the data.
The method draws on the theory of ‘social representations’ (Yuksel, et al., 1999), which is a social-psychological framework to explain collective psychosocial phenomena in modern societies. A social representation is understood as a fundamental organisational principle of the human society, which constructs a stable, predictable world and social order from the diversity of individuals, attitudes and phenomena (Moscovici, 1984). It is based upon consensual understandings, emerging through informal everyday communication and action between group members (Hogg & Vaughan, 2008). These cognitive patterns shape the social interactions within and between groups and are in turn shaped by those interactions (Yuksel, et al., 1999). Thus, the identification of patterns and clusters in social activities within a group representative to a social phenomenon helps the global understanding of the phenomenon.
There are five key stages of the method, which are presented in Table 5.4, based on Lacey & Luff (2007). These five stages are: (1) Familiarisation; (2) Identification of a thematic framework; (3) Indexing; (4) Charting; (5) Mapping and interpretation. Before presenting how these stages of data analysis have been applied in the context of this research, first a brief description follows of the considerations on selecting the computer- based qualitative data analysis tool to assist the process.
Table 5.4: Key stages of Framework Analysis
Familiarisation Transcription and several readings of data Identification of a thematic
framework
Designing an initial coding framework both from a priori issues and from emerging issues from the familiarisation stage. This thematic framework should be developed and refined during subsequent stages.
Indexing The process of systematically applying the
thematic framework to the data, using numerical or textual codes to identify specific pieces of data which correspond to differing themes (this process is more commonly known as coding from Grounded Theory analysis)
Charting Headings from the thematic framework are
used to create charts of the data so that one can easily read across the whole dataset. Charts can be either thematic or case charts. Mapping and interpretation Searching for second-level orders in the data such as patterns, associations, concepts, and explanations, aided by visual displays and plots. The aim is to define concepts, map the range and nature of phenomena, create typologies, find relationships and provide explanations.
Source: Lacey & Luff (2007)