(p.309) Appendix 2 Coding and Analyzing Stories
(p.309) Appendix 2 Coding and Analyzing Stories
Initial qualitative coding strategies began to be developed as the research team moved deeply into the data gathering. Amy Moff Hudec, Roman Williams, Melissa Scardaville, Tracy Scott, and I gathered in Atlanta at about the midpoint to begin to identify the themes we were hearing. The analysis, in other words, began to emerge in the midst of our interaction with each other and with our participants. By the time we finished gathering data in early 2008, had transcripts of all the interviews and diaries and field notes, and had catalogued 1,311 photographs, the sheer volume and depth of information was daunting. All 330 texts were entered into the qualitative data analysis software MaxQDA, allowing us to code and eventually sort the data to look for patterns in the stories. The original research team was joined in 2008 by Emily Ronald and Kevin Taylor, graduate students in Boston University’s Religion and Society program, and we all became a team of coders, meeting in person and via Skype to compare notes as we developed the analytical structure we would use for our ordering of the stories. Coding was done in standard qualitative analysis style (Lofland and Lofland 1984; Miles and Huberman 1994), applying multiple and often overlapping thematic codes to whatever portion of text constituted a logical unit.
This initial coding laid the foundation for the next level of analysis. Different domains of everyday life were used as a primary organizing scheme for an initial sorting of the stories and organizing of this book. Within each domain, the primary unit of analysis for the second stage was the narrative itself. That, however, is not as straightforward as it might seem. People rarely told a story in a contiguous beginning-to-end fashion. They would begin and need to be prompted to elaborate. They would chase a rabbit. Or they would (p.310) pick up new details and threads much later in the conversation. The first analytical task was simply identifying and naming the stories. Even this relatively small number of participants produced literally thousands of them, and the work on each of the chapters in this book began by cataloguing and naming the narratives to be analyzed.
A central question for each narrative was the presence or absence of “spiritual” or “religious” content. There were many ways for a story to be coded as pertaining to “religion or spirituality.” It might directly mention a deity as an actor in the story. It might be explicitly describing an event as “spiritual” or “religious.” It might be a story about a named religious tradition or group. It might be about a ritual or practice that the participant considered to be religious or spiritual. In other words, at this basic yes-or-no level, we included both commonsense identification of references to religion and an expansive identification of elements that the participant named as spiritual or religious, whether or not they fit typical cultural or scholarly designations as such.
A nuanced analysis of the meanings of “spirituality” is the subject of Chapter Two. All of the things described there—from references to mysterious happenings to references to ethical commitments or awe-inspiring natural beauty—were coded as spiritual if the participant identified them as such. Specific dimensions of religious practice—service attendance, fulfilling ritual obligations, activities sponsored by churches and synagogues, displaying religious objects—were included, as well. As I show in Chapter Two, the line between “religion” and “spirituality” is distinctly blurred, and one of the aims of this study is precisely to highlight that overlap. When participants made a point of highlighting what they saw as a difference, that becomes part of the story. When they are simply describing all the many ways in which they encounter nonordinary realities, the language of “religious” and “spiritual” will be used interchangeably and in tandem. These are categories that emerge from the stories themselves.
Having selected the text sections relevant to the domain of each chapter—households, religious communities, work, and the like—the cataloguing of stories began by establishing inductively the range of content on a variety of story elements.1 For instance, what kinds of characters appeared in the story? The range in households included spouses and partners, children, extended family of all sorts, and so on. The range for work included co-workers, bosses, clients, and customers as well as friends and family. This detail allows us to see how the telling of spiritual stories may vary depending on what other characters are in the story.
(p.311) Stories were also coded for the specific content or focus of the activity, but the range obviously varied by domain. For instance, when coding spiritual practices, the range encompassed all the activities described in Chapter Three. In coding religious participation, the range included music, preaching, other ritual activities, administrative work, care work, and the like. Among stories about the body, the focus ranged from exercise and other health practices to chronic illness and death. Again, the detail about what activities were being described allowed analysis of when and how human action was sometimes sacralized.
In some domains, other aspects of the stories were added to the analysis, as well. For religious participation, I coded for whether the story was about an activity in which the person had a leadership role. For religious practices, I coded for the range of material objects and settings that were described. For public activities, I paid attention to the form of political participation and the degree of agency being expressed as well as to the issue that is the focus of the story.
Across domains, stories were coded for whether the action was a one-time memorable event in the past, an ongoing routine pattern, or something anticipated in the future. Similarly, all stories across domains were coded for the emotional tone of the account. Was the story told with positive affects such as joy or humor, pride or satisfaction, or with negative affects such as anxiety or fear, sadness or frustration?
Each story, with codes describing its various contours, was entered in an SPSS spreadsheet. All the basic demographic information on the storyteller was added, as well, so that it was possible to compare stories across cultural locations and religious traditions, across gender and age and ethnicity, and along a variety of other lines suggested as the analysis progressed. From time to time, key differences are presented in the text via tables with numbers, but most of that quantitative analysis remains behind the scenes. When the text says that more stories of a certain type were religious or spiritual, the analysis behind the scenes was a comparison of means (t-test) or cross-tabulation (chi-square test) that showed the difference to be statistically significant. With several hundred stories within each domain, this statistical sorting has made it possible to note how stories are told differently by people in different social situations. (p.312)
(1.) Some narrative analysts engage this work by doing very detailed and nuanced analysis of a few highly elaborated stories (Riessman 1993). The method adopted here seeks broader and comparative analyses, taking smaller narrative units as the object and employing quantitative analysis in conjunction with qualitative coding. This is not unlike techniques used in content analysis. See Franzosi (2004) and Abell (1987).