Multiple Case Study Analysis

In deciding upon a suitable method of case study analysis, one which encapsulated a constructivist approach both in relation to theory (guide vs.

determinant) and meaning creation beyond the singular case, I opted for multiple case study analysis. The purpose of the Multiple Case Narrative is not to present each case narrative separately, but to “bring to light similar or distinct characteristics that have become apparent from comparisons between several case narratives” (Shkedi 2005, p.165). While comparative case study analyses also seek to elucidate commonalities, they do so on the basis of choosing few and distinct variables which are then “tested” in an analysis of cross-case data. This can conceal case atypicalities and complexities in delivering more simplistic rather than thick descriptions.

Multiple case analyses, on the other hand, are more exploratory in nature.

A generalization focus is replaced by a perspectival one which seeks to advance understanding by identifying common patterns or themes that transcend the boundaries of single cases and illuminate certain phenomena.

This is also reflected in the structure of my findings where there are no separate chapters or sections devoted to the individual case narratives.

Rather, themes derived from a multiple cross-case analysis are presented with empirical evidence drawn from the entire data archive. Thus, I chose a structure highlighting communicability and readability while avoiding the theme/findings overlaps common in approaches which present several single case analyses in juxtaposition.

6.6.1 THE MAPPING STAGE

In discussing the process of multiple case analysis with the ATLAS.ti software, the main goal of the initial recoding of all data documents was to establish the orientation and direction of the following stages of the analysis. As there are no specific rules of organizing the data during this mapping stage, one’s individual research choices drive this process. My focus in analysis rested on direct interpretation or inductive sense-making of source narratives instead of on categorical aggregation where one seeks to map frequencies of occurrences of specific interactions, behaviours, spoken words or phrases, etc. in creating and interconnecting categories (Stake 1995, p.74). Thus, in my study the more than 1,500 single codes which emerged from a broad inductive content analysis of all data were grouped into code groups. The emphasis of the mapping stage rests on creating emerging code groups or categories (Pidgeon & Henwood 1996).

In this grouping, although codes were taken from the entire data base in ATLAS.ti, a tag was made next to each element of data to indicate which case narrative it was taken from. It became obvious by looking at the initial categorizations emerging from an analysis of the NorQuest data, that these reflected almost intuitively on later cases, in an outcome described as category congruence (Shkedi 2005, p.101). For example, during my NorQuest fieldwork, I became increasingly aware of the specific structural factors both within the College and those from without which circumscribed curricular choices as well as interactions between students and staff and ultimately inclusion efforts. This recognition was reflected in the many codes referring to structural factors in various guises. Resultingly, these

“structural spectacles” refracted impressions from fieldwork at Arbis and Medis and informed coding and code labels. Shkedi (Ibid) writes that such intuition can be fruitful providing researchers are critical and do not impose code labels in an effort to make the material fit them.

Having established initial tentative code clusters, my next step in mapping involved making links between them by specifying what relationship they have to each other. Code group/category labels then denote meaningful connections rather than simply labelling disparate topics. For example, the code cluster, student attendance in my study included formal procedural guidelines regulating the number of unexplained student absences as well as informal rules of enforcement depending on teachers’

differing interpretation of attendance regulations. However, student attendance also connected to the cluster of term structures where, for example, NorQuest’s whole year term structure meant that in the summer term, female students especially had greater difficulties in attending because their children were on school holidays. In addition, the low level of social security benefits available to migrant students dictated that many were forced to work during studies and this sometimes conflicted with school attendance. Such associations linked student attendance to larger code groups including both internal (school) structural factors as well as external (societal) structural factors.

Some approaches to multiple case study analysis suggest creating data organizational diagrams in which code categories are managed by creating horizontal and vertical diagrams where the more overarching categories sit higher (Ryan & Bernard 2000). For my study, ATLAS.ti. included a component which allowed for the creation of code family diagrams facilitating the depiction of relationships between groups/categories and themes. The flexible nature of the code family program also allowed for a creative reordering and re-configuring of code groups and interlinkages in arriving at themes.

6.6.2 THE THEMING STAGE

The theming stage entails amalgamating larger data categories around core themes. It is here the researcher selects what s/he considers to be the most meaningful and central themes and elucidates their relationships to other categories while mentally juxtaposing them with the theoretical background. Ostensibly, theming in my study began rather early. Even during data collection and in the first stages of analysis, certain subjective truths embedded in the material started to present themselves. This fledgling cognitive readiness was honed and refined during the initial and mapping stages of subsequent fieldwork periods. It is, however, recommended to wait until the theming stage before making final choices.

(Shkedi 2005). The code group family diagrams created as organizational tools in mapping served as the means for crystallizing final themes, entitled

“network groups” in ATLAS.ti. Themes were often portrayed as interrelationship nexuses from which arms extended outwards to other

category clusters of varying proximities. Theme labels were also more abstract, oscillating as they were between the empirical and the theoretical.

In making theme choices, the sheer frequency of particular codes or volume of particular associations did not necessarily dictate final theme selection.

In fact, it was sometimes the atypical or marginal categories which encapsulated a particular poignancy in defining, describing and truly

“naming” the phenomenon of social inclusion. The Inclusion Within the Walls theme diagram (see appendix 1) illustrating the code group family/category arrangements depicting social inclusion within the schools, explicates how this theme was structured. It is constituted of code group families describing interpersonal factors (eg. Critical Consciousness/Reflexivity and Cultural Negotiation), curricular factors (e.g. Critical citizenship, and Curricular Teaching/inclusion) and institutional factors (e.g. Structural Factors/teachers/students, Program Aims and Development Needs).

However, the code group families are not exclusive as each also subsume code clusters that describe interpersonal, curricular as well as institutional factors. Separate code group families are identified by different colours, but they also include code clusters which are shared among several code group families or several themes. Thus, some code group families are composed of a multicolour selection of code clusters illustrating the categories’

interconnectedness and malleability.

Some considerations which helped me in theming were thinking about how themes linked up with research questions and how they reflected the bordered reality of single cases. For example, is the chosen theme one which holds true for all three case studies? Another consideration was that any identified theme(s) should bind a thesis together and provide focal points for dialectic and analysis. Therefore, themes which inherently unify a thesis help by avoiding the risk of separation between cases within analysis. Subsequently, it also becomes easier the explore the cross-case similarities and differences, thereby facilitating the process of arriving at fuzzy generalizations or cross-case assertions (Carey 2012). Before making a final selection then, I revisited the NorQuest, Arbis and Medis material separately and compared the associations the themes had with the data by asking, do they “hold true” when looking at the individual case? Ultimately, multiple case study analysis in my study yielded the following final themes:

Inclusion Within the Walls, Inclusion Beyond the Walls, and (Colour) Blind Spots.

6.6.3 WRITE-UP CHOICES

Lastly, decisions had to be made regarding which written format for multiple case study analyses best enhances the theoretical and empirical communicability of the text in the final report. For me this represented a choice between the category-focused-narrative report or the narrative-based theory report (Shkedi 2005). In the former, data is displayed thematically and explained within the context of emerging empirical findings without embedding it within a theoretical dialogue. Such a structure has been deemed most useful for studies dealing with many case narratives as there may be no separate chapters or sections devoted to individual case narratives (Merriam & Simpson 1984). Instead, each chapter is devoted to a separate cross-case issue, or theme; these being organized according to the logic of findings within a researcher’s narrative.

The narrative-based theory report too presents a holistic discussion of all cases without analyses being compartmentalized within single case studies. The crucial difference, however, is that theoretical reflections are interwoven with the narratives including many descriptive, first-hand quotes emanating from the empirical material. Thus, it represents a hybrid text capitalizing on the synergy between informants’ voices and conceptual-theoretical language. Narrative-based theory reports generally include key concepts, bibliographic references, and quotations from relevant academic sources. In deciding between these two written formats, I reviewed my methodological and theoretical choices to ascertain which approach best reflected them. Reflective questions that facilitated this process were,

“Which format best communicates contestations of social inclusion in integration educations? What is the best way of conveying connections between case narratives? and How can I give participants’ voices the most impact?” Ultimately, I opted for the narrative-based theory approach. By allowing for a creative juxtapositioning of participants’ quotations with explanatory theory, it holds the promise of being able to maximize the communicative impact of themes. It further amplifies the visceral immediacy of research results which may help in disseminating findings and coupling them to social change objectives. In addition, one Leitmotif in anti-oppressive research is the focus on ongoing dialectic – between researcher and participant, theory and data, and the individual and society. Narrative

based theory reports continue this conversation as a recurring theme throughout the monograph.