The first step was to create a basic map of the entire decision-making system
contained within GMRSD school committee business over the course of the 2016-17 school year by identifying to the extent possible all decisions, problems, solutions, participants, and choice opportunities, and their entry/exit timing. This answered the first sub-question: “What problems, solutions, participants, choice opportunities, and decisions appeared in GMRSD school committee meetings and materials between July 1, 2016 and June 30, 2017?”
Before identifying the components, I defined the properties and dimensions of the system components as described next.
Identifying system elements. The Phase 1 data set includes all GMRSD school committee meeting minutes, agenda packets, and videos where available (most meetings were recorded, and these are stored in the Montague Community Television’s online Vimeo account) from this time period (see Appendix B). I converted text from meeting minutes into fieldnotes for easier reading, coding, and text searching. Fieldnotes are the bridge
between data collection and data analysis. They are the place where “thick description” about physical settings, timing, people, interactions, and so forth is captured (Geertz, 1973). I used a fieldnote template outlined by Emerson, Fretz, and Shaw (2011) that includes columns for description, reflexivity, and ongoing analysis. Each digital fieldnote I created included the event date (and/or retrieval date for artifacts), writing date(s), and links to relevant resources (e.g., original artifacts, associated photographs, websites, etc.). The description column for artifacts primarily consisted of text converted from the artifacts themselves (i.e., I converted .pdf files into text, which I then pasted into the description column and checked for
accuracy against the original document). I pasted screenshots of relevant tables or images that did not convert in a usable format into the description columns.
Converting text-based artifacts to fieldnotes created consistency, and facilitated coding and word/phrase searching during analysis. I also frequently referred to the original documents. In the reflexivity column, I wrote notes regarding my personal responses to the data, and reflections on my positionality. In the ongoing analysis column, I documented emerging ideas, questions, and connections as I created, read, and re-read the description column. I added to the reflexivity and analysis columns throughout the entire research process, and also kept personal journals for hand-written notes and ideas. All digital data
were stored in an organized system that was password-protected. Hand-written notes, documents, and journals were stored in a locked filing cabinet to which I had the only key.
To ensure I accurately and comprehensively identified and described the system’s problems, solutions, participants, choice opportunities, and decisions in the school
committee meeting data, and the timing of these elements’ appearances in the system, I used a standard qualitative data analysis method that included reading, coding, and interpreting phase one data based on these five deductive categories (Creswell, 2014). While the basic categories were established in advance based on the garbage can model (Cohen, et al., 1972), the process to define each element’s properties and dimensions shifted between inductive and deductive reasoning, and required multiple readings to accurately identify, categorize, and interpret them (Rossman & Rallis, 2012; Strauss & Corbin, 1998). I outlined these properties and dimensions above.
This process included open coding, or what Charmaz (2014) calls initial coding, which is a dynamic and relatively fluid process to discover the properties and dimensions of each category in order to operationalize them and be able to recognize them in the data. The process relies on the “constant comparative method,” which was first developed by Glaser and Strauss (cited in Charmaz, 2014, p. 132), and involves an iterative process of comparing related pieces of the data set to each other to find similarities and differences in order to achieve greater clarity. Once this was complete (although I made small adjustments throughout most of the analysis process as my understanding grew), I engaged in focused coding in which I applied these operationalized categories to the entire data set (see Appendix C for a sample).
This generated a relatively complete and accurate accounting of all elements in the system, which I transferred into a spreadsheet to enable sorting. I ensured any adjustments I
made in later phases were also adjusted in this spreadsheet. After coding for elements, I created deductive categories that included: adult learning and culture; costs, budgeting, and resources; curriculum, instruction, and assessment; governance, leadership, and management; operations and services; parents and community engagement; performance and state
accountability; student conduct, social and emotional learning, and school climate; and vision, mission, and values. These categories were loosely based on categories that the GMRSD superintendent used in his Entry Report of February 2014, although I adjusted them based on the topics that appeared frequently in school committee meetings. This enabled me to sort the spreadsheet by categories to look for emerging patterns.
I hand-coded all data throughout this study. Emerson, et al. (2011) cite limitations to computer-assisted qualitative data analysis programs due to the ways in which they tend to lock researchers into early categories, and encourage fitting all data neatly into existing codes. Hand coding is more laborious, but allows one to read and re-read the data set, and become intimately familiar with it (Michael Burawoy, Keynote Address, Unbounding Ethnography Conference, UMass Amherst, November 4, 2016, personal notes). Based on the garbage can model (Cohen, et al., 1972), some of the connections between elements were essential to identify to the extent possible as I mapped the system, including who brought elements into the system, who made decisions, and entry/exit timing. The initial and focused coding processes necessarily entailed analyzing and interpreting the same data from multiple
stakeholder perspectives, as this is essential in systems analysis (Williams & Hummelbrunner, 2011). Throughout this entire process, I added to the reflexivity and ongoing analysis
columns in the fieldnotes, and wrote analytical memos in digital documents or my hand- written journals in order to create a record of my ideas, interpretations, questions, and
potential analytical themes and directions, and to engage in writing as a method of inquiry (Richardson, 1994).