CAPÍTULO 5. CONCLUSIONES Y RECOMENDACIONES
5.3 Aspectos a continuar investigando
All interviews and focus groups were translated by the research assistant during the interview, and recorded. After each interview I transcribed the data and read through the interview text to garner a thorough understanding of the emerging themes and concepts and uncover commonalities and distinctions in women’s experiences. This allowed me to become fully immersed in the data before I started coding for themes. In accordance with grounded theory, research and initial analysis was conducted simultaneously. Though a rough sample of 20- 30 was helpful for planning purposes, interviewing and analysis was conducted until the
saturation of concepts was complete, where no new information about the core processes is forthcoming from ongoing data collection (Strauss & Corbin, 1998). Based on the literature it was probable that between 20 and 30 interviews would be needed to reach this point of saturation (Creswell, 2007).
Following, the interviews and focus groups transcripts were then printed into hard copies which were thoroughly read once again in order to get a general “feel” for the data. Once I felt comfortable with the material and questions, I began going through the interviews in a more formal manner, using the procedure of open coding. Open coding allowed the highlighting of particular words or chunks of responses that indicated concepts and categories that fit the data.
Using Strauss’ (1990) conception of this procedure, this stage involved unrestricted coding of the data and the understanding that all concepts uncovered are tentative. At this stage, the interviews were reviewed independently and taken for their individual contributions. By the time I had reached the last of the interviews, it was clear that many of the questions generated similar responses, and moreover, overall themes were emerging across questions. At this stage of analysis, data saturation was evident. However, it was also important to note any of the unique answers, as these were never excluded from the analysis.
Upon the completion of open coding, I consulted with Dr. Luginaah about the generated codes. We compared and contrasted ideas, which was helpful as this was my first time conducting this kind of work. Viewing Dr. Luginaah’s understanding of the codes affirmed that I was on the right track with the analysis. The occurrence of similar coding by two separate people, as was the case here, lends credibility and reliability to the study.
After our meeting, I created an axial coding model, which is a multileveled coding model. This step in the content analysis is used to organize the data and identify findings after open coding has been completed. I read through each transcript again and tried to group the open codes previously collected. This allowed for analysis of recurring themes and for quick reference when identifying commonalities and comparing answers between participants. This proved to be a very effective way to organize and analyze the data.
Often, qualitative researchers use computer programs for this type of work. However, I chose not to do this for three reasons. First, I was eager to complete the analysis in order to keep with the projected timeline outlined for my thesis. Using a computer program would have required a greater investment in learning a new program and thus would have delayed my work. Secondly, I did not have any access to the computer program while I was in the field, and had restricted temporal and locational access to the program once arriving in Canada. Lastly, the total number of interviews was small enough to make manual coding manageable. As well, when I consulted others, they agreed that this was reasonable.
While all questions posed during the interview process could provide context and insight into my thesis, I was primarily interested in developing an understanding of the thesis’ five objectives. The exercise in coding the entirety of the interview was not a fruitless effort. It was first and foremost, an excellent way for me to immerse myself into the interviews and get a feel of the information. If I had only analyzed my five questions I would have missed a great breadth of information and misunderstood responses given in the narrow context. The interviews were conducted as a whole entity and thus I felt that any analysis of the information needed to be performed in its entirety. It also gave me a greater understanding as to women’s knowledge, access, and utilization of maternal health care within their socioeconomic, geographic and cultural context. Finally, it was a necessary step in conceiving overarching themes of the data and demonstrates connecting aspects of the research questions.
While it was important to analyze all of the interviews, it was also important to focus on the five key objectives for this study. As such, my final step in the analysis was to pick apart the responses for those particular responses related to my objectives. I did this by creating a chart, which indicated the location of the respondent, the age of the participant, their enrollment status, location of delivery, number of children, and role of the participant (mother, community nurse, general nurse, midwife, TBA, or health administrator). The chart contained all of the relevant information generated by respondents’ responses to the objectives. This included both responses that fit well with the axial coding created earlier and unique answers that highlighted alternative viewpoints. Since the total number of participants in each district or category was limited, a unique response could potentially be quite telling. The data were entered into the chart in concise point form method, which gave myself detailed referencing so I could easily go back to the transcripts and understand the point in the proper context of the interview. Organizing the findings into the chart provided significant assistance when I shifted into the analysis stage. By using this “selective coding”, the findings were organized in relation to other variables which were of interest. I could therefore interpret the results accordingly. For instance, the women’s life
stage, policy status, and location of delivery were all laid out. While it can be a mistake to assume a variable has relevance (Strauss, 1990), this method of organization allowed the variable to prove its relevance in the larger picture of the research. For example, understanding the role of exemption status could explain the location of delivery or prove beneficial for locations otherwise different.
After the data was organized and analyzed, my consultation with Dr. Luginaah helped to focus and articulate the main ideas. Taken together, the use of open coding, axial coding, and point form notes for the five objectives provided a solid base for analysis and interpretation. See Figure 13 for data analysis diagram.
Figure 13: Data Analysis Techniques
Source: Created by Author
4.12
Chapter Summary
This chapter described the study design and methods utilized to obtain the results of the study. Specifically, this study adopted the qualitative case study methodology. This chapter outlined the logic of qualitative methods and qualitative case study design, as they work particularly well in addressing the research objectives. Following this description, the chapter
outlines the multiple methods utilized noting their strengths as well as the choice of qualitative methods for this particular study. The section discussed the details of participant selection and how the interviews, focus group discussions, key informant interviews, and field
observation/notes assisted in the gather of data for the study. The chapter concludes by discussing data analysis techniques which were utilized to produce the themes presented in the results section.
Chapter 5
5
Results of the In-Depth Interviews, Focus Group
Discussions, and Key Informant Interviews
5.1
Introduction
The interview questions explored participant’s prior enrollment in NHIS, the number of children they have birthed and the location of child delivery (ies). Further discussion topics were derived from the theoretical framework of this thesis, incorporating concepts relating to political ecology of health, women’s autonomy, and factors which influence access to health facilities. The results are presented around key themes examining women’s knowledge of the MEP, the
influence of the NHIS enrollment on use of the MEP, certain geographic, socioeconomic, and cultural influences on maternal health care utilization, confusion regarding appropriate care, and treatment women receive from health professionals. Additional concepts which emerged from the research is also presented covering concepts such as anemia, blood transfusions, complicated delivery, inadequate food supplies, coverage of drugs, MEP expiration, desire for ambulance services, and lack of male involvement. Under each theme, sub-themes are provided with
quotations used to help contextualize the findings. Respondents’ quotations were selected in order to ensure not only “representativeness” but also vividness in reporting of themes (Seale &
Silverman, 1997). Each quotation is identified with a pseudonym, focus group discussion (FGD) or in-depth interview (IDI), their occupation when a key informant, and the age of the participant.