Prior to beginning the coding process in the first phase of my data analysis, I kept in mind Saldaña’s (2009) advice about the “bottom line criterion” when navigating through the first coding cycle. He suggests, “as you’re applying the coding method(s) to the data, are you making new discoveries, insights, and connections about your participants, their processes, or the phenomenon under investigation” (p. 51). On the broadest level, my goal in this study was to understand how TAs guide students’ reading in FYC, and to do so, I needed to remain open about which angles would be worth pursuing as I became more immersed in and attuned to the data during various phases. Selecting which of the 24 TAs to interview based on their survey responses presented an opportunity to use coding as a tool to make inductive decisions that would drive the ensuing stages of my inquiry. With that in mind, I coded my first source of data in two different ways. First, I coded the open-ended qualitative responses through an “initial coding” approach. Secondly, I used a “structural coding” approach to compare the codes that I had just generated to TAs’ responses to the closed-ended Likert scale questions. Both of these coding approaches, in tandem, enabled me to achieve both of my goals for phase one: (1) to generate an expansive set of possible codes that could inform later phases of my data analysis, namely, approaching phase three with a refined set of codes, and (2) to satisfy my subsequent methodological step of determining which TAs to interview.
Saldaña (2009) describes initial coding as “the first major stage of a grounded theory approach to the data. The method is truly open-ended for a researcher’s first review of the
corpus, and can incorporate In Vivo and Process Coding” (p. 66). The next consideration that I had to make was which of the two general approaches I should take to coding the data: splitting and lumping, wherein splitting attends to the “careful scrutiny of social action” while lumping “gets to the essence of categorizing a phenomena” (Saldaña, 2009, p. 20). Although the advantages of lumping include conserving time and establishing broad, inclusive categories, I decided that splitting would provide the most purposeful methodological approach for the initial stage of using the survey data to determine which TAs to interview data. Splitting would allow me to more thoroughly attend to the nuances of TAs’ responses and, consequently, the particularities of their expressed reading pedagogies. This approach would also permit more flexibility in my ensuing rounds of coding—after splitting the data on a more micro level, I could then consider which individual codes could most meaningfully be lumped together to form broader categories, and I could also determine which categories are best suited to the data.
During my first coding pass, I paid special attention to TA’s exact phrases and made an extensive catalog of these in vivo codes. I notated in vivo codes with quotations so that I
remembered where these words and phrases originated. I also permitted myself full flexibility to project meaning onto the data with descriptive codes and process codes. After I finished coding the 24 surveys, I separated the codes into two groups, depicted in the table below.
Table 2
Comparison of In Vivo and Descriptive Code: During First Round of Coding Open-Ended Survey Data
TA and Discipline
# of Total In Vivo Codes and Examples from Q#1
# of Total Researcher-Generated Codes and Examples from Q#1
Religious
Studies TA#1 • 127 • “the point of freewriting”
• “deemphasize the importance of grades”
• 30
• detailing/explaining specific aspects of composition Religious
Studies TA#2 • 95 • “to change how someone thinks” • “functioned effectively as steps in a
• 22
Religious Studies TA#3
• 10
• “developing reading skills in the academic context”
• 2 • None Religious
Studies TA#4 • 34 • “foundational in how I approached teaching”
• 8 • None Religious
Studies TA#5 • 13 • “the guidance of supervisors like Doug Bradley, Chris Dean and Randi”
• 4 • None Religious
Studies TA#6 • 25 • “Seeing writing as a series of choices has allowed me to consider the teaching of writing a much more holistic practice.”
• 9
• higher-order and lower-order concerns
• X-realization (about writing) allowed me to do/think Y History TA#1 • 86
• “I have benefited as a student and as an instructor from” X
• 12
• thinking through concepts History TA#2 • 14
• “it makes a compelling argument about the larger social significance of clear political writing”
• 3
• teaching practices being
influenced by a text’s argument History TA#3 • 55
• “he said something that really changed who I am as a writer”
• 10
• sponsorship
• WAC: characterizing/ describing disciplinary writing
History TA#4 • 55
• “same problems in the writing of juniors and seniors.” • 7 • consequently (implies relationship) History TA#5 • 35 • “of course” • 9 • None History TA#6 • 53
• “I couldn't figure out how to do that” • “I don't want anyone to feel like I did”
• 11
• argumentation
• guided questions for reading and reading-writing comprehension History TA#7 • 69
• “set the groundwork for my approach”
• 6 • None History TA#8 • 32
• “give my students some control and choice in the classroom”
• “I've started doing a lot more direct modeling”
• 7
• paradoxical relationship
History TA#9 • 41
• “an undergraduate philosophy course I took with Kenneth Sayre taught me more about the mechanics of writing than any other experience I've had.”
• “hates pedants”
• 10
• no mention of Writing 2 people/canon (ABSENCE OF the typical codes)
English TA#1 • 36
• “expected to have a working body of
• 10
English TA#2 • 46
• “since then
• “I underestimated, initially, the difficulty students have in viewing words as words, language as language, and writing as writing”
• 6
• biggest impact determined by students
English TA#3 • 70
• “for the research paper unit”
• 20
• if it’s useful for me, it’ll be useful for students
English TA#4 • 66
• “Genre analysis as a concept”
• “how to help students develop skills for the long term”
• 10
• Not X, but rather Y
Comparative Literature
TA#1
• 30
• “getting students to become explicitly aware of what they're doing and why they're doing it.”
• 2
• personal writing experiences/ development Comparative Literature TA#2 • 65 • None • 5 • None Classics
TA#1 • 22 • “bridging the gap between theories and daily in-class practice”
• 7 • None Music TA#1 • 28
• “my best students of the quarter” do X”
• 9
• connecting a TC to what they’ve experienced
• evolving expectations over the course (students’)
Music TA#2 • 5
• “talking with colleagues casually around the cube farm”
• 0 • None
Although the total number of codes might be considered relatively large, especially for a survey, I wanted to use this first phase of data analysis as an opportunity to “build a foundation for [my] future coding cycles” (Saldaña, p. 66). On one hand, these codes helped me to
understand the ideas and practices that TAs brought to their teaching practices, along with their attitudes and feelings. On another, more practical level, these codes could give me a sense of the major categories within the data, which would, later, help me bring a more refined approach to phrase three.
Once I examined the list of in vivo and descriptive codes, I noticed that some unexpected distinctions emerged across TAs. TAs tended to fall into groups based on their
practices or attitudes. I affectionately labeled some TAs “Kool-Aid drinkers,” which
encompassed TAs who acknowledged being meaningfully influenced by the Writing Program’s TA training in some way. Other TAs, though, seemed to be relatively apathetic about their Writing Program training, and I labeled these participants “Kool-Aid abstainers.” However, after continuing to analyze the data, I considered that my “Kool-Aid drinkers” designation didn’t quite capture the extent of one TA’s admiration for the Writing Program who wrote “I agree, and agree passionately, with the other ethos of UCSB's Writing,” so I labeled this participant a “Kool-Aid pourer.” That decision, in turn, required reconfiguring the members in my previous categories. In this iterative way, different groups emerged from TAs’ responses, creating natural tensions amongst my TA-participants, which I could consider when making decisions about who to interview. At times, these categories would also come in handy during later phases of my data collection.
I formalized these emergent groups, a decision which reflects what Namey et al. (2008) refer to as structural coding: “a labeling and indexing device, allowing researchers to quickly access data likely to be relevant to a particular analysis from a larger data set" (p. 141, as cited by Saldaña, 2009, p. 67). MacQueen et al. (2008) acknowledge the value that this approach could afford studies like mine, specifically, for the decisions that I would need to make in my ensuing selection of interviewees. They state that, "Structural Coding generally results in the
identification of large segments of text on broad topics; these segments can then form the basis for an in-depth analysis within or across topics" (p. 125, as cited by Saldaña, 2009, p. 68).
Structural codes, or groups, emerged from the survey data in two ways: via my initial coding of the open-ended data, as well as through TAs’ Likert scale responses in the quantitative portion of the survey. In a sense, in the former scenario, I placed TAs into groups, and in the second scenario, TAs self-placed into groups based on their alignments with various perceptions
and practices. Examples of groups that I placed TAs into included those who felt strongly about TA training and the Writing Program’s ethos, those who made explicit acknowledgement of threshold concepts, and those whose comments assumed significant student-centeredness. Structural codes also took shape when I collapsed TA’s Likert item statements. I consolidated
agree and strongly agree responses into a “definitively agree” group; conversely, I consolidated disagree and strongly disagree responses into a “definitively disagree” group. Thus, after collapsing
TAs’ self-reported measures on “I experience difficulty explicitly addressing comprehension in my WRIT 2 teaching practices” (Q#11), TAs could be categorized intro three groups: those who
definitively agreed that explicitly addressing comprehension was difficulty, those who definitively disagreed, and those who felt relatively apathetic. The two groups on polar sides of the spectrum,
again, presented an inherent tension across TAs’ perceptions and practices that might merit further exploration. Once I had these structural codes in place, I looked for areas of consensus across the TAs—perceptions and practices that seemed to be shared by a large majority of TAs. Then, with this majority group accounted for, I sought out points of dissensus.
In this way, my two approaches to data analysis in phase one allowed me considerable flexibility for selecting TAs for interviews. The initial coding allowed me to understand the different themes that arose through TAs’ articulations of reading, while structural coding presented an opportunity to draw comparisons across each individual TA’s relationships to those themes and make further methodological decisions accordingly. On one hand, I could examine the initial codes that emerged from TAs’ open-ended qualitative responses to find TAs whose responses either represented sophisticated notions of guiding students’ reading, elicited an unusual theme, or contained some other noteworthy aspect. On the other hand, I could rely on the more systematic structural codes to isolate particular TAs whose responses presented anomalies to the greater trends that I observed across TAs. My hope was to find a balance of the
above distinctions to gain a more holistic perspective of the ways in which TAs might conceptualize “good reading” in FYC and attempt to guide students’ reading.
Phase II – Examination of Classroom Resources to Generate Interview Questions