"One cannot ordinarily follow how a researcher got from 3600 pages of field notes to the final conclusions, sprinkled with vivid quotes though they may be."
(Miles and Huberman 1984, p.16)
“If selecting your case(s) to be studied is the most critical step in doing case study research, analyzing your case study data is probably the most troublesome.”
(Yin 2012, p.15)
There is no generally accepted method to follow for analyzing case study data; however, it is crucial to have a clear strategy for analyzing large amounts of qualitative data (Yin 2014).
Qualitative analysis is a non-linear and recursive process (Yin 2014), and often requires going back to the data and analyzing it again. In my multiple-case study research, there are two fundamental stages of data analysis. First I conduct a within-case analysis to gain familiarity with my data and also generate preliminary results for each case. Thereafter, I conduct a cross-case pattern search. The cross-cross-case pattern search is to look beyond preliminary results of each case and finally come up with broader conclusions and answers for the central research questions (Eisenhardt 1989).
3.6.1 Analyzing Within-Case Data
My strategy for analyzing within-case data is twofold. First I develop a description for each case. The descriptive framework for each case is based on the first and the second set of sub-questions about the actors involved in urban renewal, their objectives of renewal and the implementation of the renewal program. After the first stage of within-case analysis by describing each case, I move to the third set of sub-question that is based on a theoretical proposition. Based on the theories of forced migration and urban redevelopment, I propose that urban renewal program is burdening existing residents (Yin 2014, pp. 136-139). Pattern-matching is basically comparing an empirically based pattern (the one that arises from the data) with a predicted one (the one that is based on existing theories). The purpose of the pattern-matching is “building explanations on whether and why the patterns are matched or not”
(Almutairi 2014).
I codify my data to arrange the information in a systematic order and organize answers around each sub-question. For each within-case analysis, I code my data through two cycles (see
Data Analysis
Figure 3.2. Summary of Data Analysis Strategy
Saldana 2009). There is an agreement among researchers conducting the qualitative study on the idea that there is no clear-cut answer on which coding method(s) to be used in qualitative research (Saldana 2009): “Because each qualitative study is unique, the analytical approach used will be unique” (Patton 2002, p.433). I follow Saldana’s formulation of coding methods for the first and the second cycle coding.
In the first coding cycle, I use “descriptive coding” method. Descriptive coding is to summarize data in a word or short phrase to come up with the basic topics; the procedure also helps to develop a basic vocabulary of data for further analytic work (Turner 1994, p.1999) and categorized inventory and summary of the data (Saldana 2009, p.72). Descriptive coding works specifically well for addressing the first and the second set of research questions on the actors, objectives and the implementation. I use subcodes to generate a more detailed description of actors, objectives and the implementation. Descriptive coding also fits well with my primary analytic strategy to describe each case before moving on to the pattern-matching between the theoretical proposition and the empirical data.
In the second coding cycle of each within-case analysis, I use “pattern coding” to identify emergent themes and explanations. Miles and Huberman suggest that “pattern-coding is a way of grouping those summaries into a smaller number of sets, themes or constructs” (1994). I review the first cycle codes generated through descriptive coding, asses commonalities between these codes and assign a “pattern code.” Hence, the main purpose of pattern codes is to describe a major theme and a pattern of action from the data. Within each case, I use the descriptive codes from the first cycle of coding to identify “patterns” of how urban renewal is burdening existing residents. Identification of pattern codes completes the second cycle of coding for each within-case analysis.
3.6.2 Cross-Case Pattern Search
I use pattern-matching logic as the analytic technique for the cross-case analysis. After I complete each within-case analysis, I select the main themes arising from each case analysis.
Based on these dimensions, I look for within-group similarities as well as intergroup differences (Eisenhardt 1989). The cross-case pattern search helps to identify similarities, differences and regularities among cases (Saldana 2008, p.8). Searching for patterns among cases helps to revisit
conclusions from individual cases and bring findings from each case together to explain the point of interest. Based on the categories of actors, objectives and implementation choices followed by the actors, I search for clear patterns and relationships among concepts. In order to visualize the cross-pattern search, I generate word matrices to tabulate codes generated at each cycles of coding.
The result of within-case and cross-case analysis identifies the patterns of relationships between the actors implementing urban renewal projects across cities. Connecting these with the ways in which urban renewal planning is burdening existing residents provides conclusions for the central research question. Answering how urban renewal program is burdening the existing residents sheds light on potential policy interventions for planning redevelopment projects that modernize cities without burdening existing residents in designated renewal areas.