Data Analysis Strategies for Grounded Theory

Data Analysis Strategies for Grounded Theory

I think this is a grounded theory research design. My study seems to qualify as GT according to the following criteria (Creswell, Ch. 13):

GT is a…

• qualitative procedure used to generate a theory that explains, at a broad conceptual level, a process, an action, or an interaction about a substantive topic.
• “process” theory—it explains an educational process of events, activities, actions, and interactions that occur over time.
• theory in which the researcher proceeds through systematic procedures of collecting data, identifying categories (also referred to as themes), connecting these categories, and forming a theory that explains the process

According to the research from Cresswell and Merriam, grounded theory is used when you want to study some process or to explain actions of people, or an interaction among people.

In trying to determine if grounded theory is an appropriate research construct for my study, I have been asking myself if my question requires a broad explanation of process.  Is there an existing theory around science or STEM teaching efficacy and the collaboration between generalist teachers and STEM specialist? I have not found anything in the literature that offers more in terms of teacher development and support outside of purchased curricula and non-specific “professional development”.  Perhaps this is too defined a question to merit the development of a theory. Some of that may be my thinking of a theory. In terms of my science training, a theory is is a well-substantiated explanation of some aspect of how the world works. This explanation is based on a body of evidence that have been repeatedly confirmed through observation and experiment. In those terms, my Plan B does not represent an opportunity to do anything more than propose further lines of inquiry on my topic and the subjects of my research.  In terms of educational research, however, this study does offer a glimpse into the impact that STEAM specialists have in the general education teacher population and how that impact may translate to instruction and student achievement.

This type of social science data analysis seems alien to me.  As much as I fear statistics, a strict mathematical analysis of quantitative data feels more like the “devil I know”... The systematic, step-by-step nature of GT analysis might make things easier.

Do I have a large enough sample size to make any categories I can parse from data meaningful?

What about this idea that there are three different pathways for Grounded theory- Strauss, systematic (1987) and Glaser, emerging design (1992).  The latter calls for a theory development process that is less structured at the outset and allows the research study to flow in ways that the data reveal.  The former advocates for a theory to be developed prior to the analysis of data- sort of like developing a hypothesis before having any experience with the phenomenon under scrutiny. This systematic approach is more theory verification through the data than theory generating. It seems like the Charmaz (2006) provides a pedagogy which fits with the overall concept of “grounding” a theory in the data. She outlines a constructivist methodology where researchers use a more flexible approach and allow for the meaning participants ascribe to situations. While reading more about the three schools of thought in grounded theory design, I was drawn to one of the criteria used for Glaser’s emerging design approach:

“A good grounded theory must meet four central criteria: fit, work, relevance, and modifiability. By carefully inducing the theory from a substantive area, it will fit the realities in the eyes of participants, practitioners, and researchers. If a grounded theory works, it will explain the variations in behavior of participants. If it works, it has relevance. The theory should not be “written in stone”and should be modified when new data are present” ( Glaser, 1992 , p. 15) .  YES!, I say- a scientific perspective- if a theory works (i.e. if tested and the theory supports the current explanation), then the explanation becomes stronger and more able to be used to further additional understandings through inquiry.

Charmaz allows for the place of the researcher in the theory which makes sense to me. “a grounded theory procedure does not minimize the role of the researcher in the process. The researcher makes decisions about the categories throughout the study ( Charmaz, 1990 ). The researcher brings certain questions to the data, along with a “store of sociological concepts” (p. 1165). The researcher also brings values, experiences, and priorities.

Given that my study is looking at self-reported teaching efficacy, I feel the approach outlined by Charmaz will be most helpful in developing a theory from my data. In applying Charmaz’s approach, the theorist explains the feelings of individuals as they experience a phenomenon or process. The constructivist study mentions the beliefs and values of the researcher and eschews predetermined categories, such as those found in the systematic coding process of Strauss.. The narrative is written to be more explanatory, more discursive, and more probing of the assumptions and meanings for individuals in the study. Analysis of the data involve a coding of the experiences of the research participants in active verbs (i.e. awakening, preserving) that signal basic processes of the respondents.  The researcher’s task is to interrelated these experiences, their conditions, and their consequences in a narrative discussion without the use of diagrams or figures to summarize these processes. In one particular example given, Charmaz ended with thoughts such as “What are the conditions that shape whether a man will reconstruct a positive identity or sink into depression?” I can imagine that my data may lead me to a similar question, such as, “In what ways are gen. ed. teachers STEM teaching efficacy influenced through collaboration with me?” In many ways, the answers are more suggestive and questioning of the data than conclusive.