In the last post, I discussed my efforts to impose greater structure on my code system. In doing so coding has become an easier process in two ways. First, I have a better understanding of my themes which allows me to understand when to apply a certain code. By having a greater understanding of how I am defining codes I can recognize themes faster and apply codes with greater confidence. The second aspect is concerned with my outlook on the project and the process. I find myself enjoying the process more, mainly because I have replaced the confusion that stemmed from my unstructured codes with greater clarity and confidence. I am still struggling with some aspects of my code system. It can be tough to decide how to code certain lines of text. At times this can be remedied by applying multiple codes, but at other times applying multiple codes is not appropriate and I have to deeply think about what code to assign. This leads to an important point about coding and analysis, be sure to take time to think about the “big picture” of your work. Often when I run into issues it helps to take a step back and think about how this code relates to others and fits into the project as a whole. On the surface, this idea sounds simple and apparent. However, I often find when I spend multiple hours coding, I will focus on small portions of text for long periods of time which can be a useful mindset, but it can also mean I lose track of the text segment in relation to the scope of my project. Taking breaks periodically to think about how this code fits into a theme which fits into my project as a whole can be a useful exercise. As I continue my work, I need to remind myself to periodically stop and think. No computer, just time to think. Of course, after a brainstorming session, it is important to record your thoughts so you can revisit them and build on these ideas.

Over the past few weeks, I have been using an inductive approach to coding. I began by reading through a few interviews and recording the themes I found. After continuing to read through interviews I quickly recognized that I was generating a large number of themes. I found this to be both a positive outcome, as I now have an abundance of themes to work with, and a negative outcome as I found myself drowning in themes. It quickly became apparent that I needed to address the growing list of themes. The solution lies in creating structure. It is important to establish hierarchies in coding systems. I found that once I began to identify the main themes, I could then identify and place subthemes below them. Creating structure in my coding system not only makes analyzing theme easier but also making coding interviews more efficient. While adding initial structure to my code system has helped me to gain clarity, my work is not done. Creating code structure is an ongoing process that entails constant realignment and refinement. This continuous retooling is time-consuming, however, it pales in comparison to the time and frustration I would experience by omitting structure.

Sorry for the brief absence. No excuses. I let my stress bog me down, but good news! The past week I wrote a research proposal to explore the impact of climate change on human health in Belize. This research will be conducted in Belize over a two-week period from July 15 through July 30, 2019. The field sites for this work are Belize City, Belmopan, and San Pedro. The methods used in this research will be exploratory and unstructured in an attempt to observe without taking a lead role and imposing myself, my assumptions and theoretical positioning on the study site and populations. Direct observation will be used at each field site as a means to increase my understanding of each specific context. Informal interviews will be used to gain an understanding of experiences, ideas, perceptions and cultural norms. Informal interviews go beyond casual conversation and can provide important content and context (Bernard et al., 2017). The intent of these interviews is to generate a dialogue focused on health, climate change, and other sources of vulnerability.

            Ethnographic interviews will also be used in this research. These interviews are more structured than informal interviews and will focus on the topics of climate change and health. Participants will be identified based on their willingness to consent and demonstrated knowledge of relevant topics. The idea of ethnographic interviews is to introduce a topic, then allow the individual to provide information which they deem as being important (Bernard et al., 2017). These interviews will be used to build rapport and produce initial data which will be used to design future interviews and questionnaires used during dissertation work. I anticipant these individuals may be government officials, NGO employees, community leaders, and/or other community members who demonstrate a wealth of knowledge. These interviews are critical to identifying collaborators for future work. Pinpointing key informants prior to research is difficult. Bernard and colleagues (2017) describe the relationship between researchers and key informants as one in which both parties “choose each other, over time” rather than one that is chosen by the researcher. Additionally, they warn against choosing informants too quickly, instead encouraging researchers to spend time at the field site to gain an understanding of the context as well as the roles and statuses of your informants in this context.

            The data analysis will consist of identifying and describing themes from the interviews. Identifying themes is essential to understand unifying and reoccurring concepts that convey meaning in these communities. Themes will be established using the twelve techniques for identifying themes described by Bernard and colleagues (2017). All field notes will be entered into MAXQDA daily. Transcription of interviews will occur after research has been completed using MAXQDA.

Objectives:

1. Assess if the proposed field sites are appropriate for my dissertation research.
2. Identify and build relationships with key informants and other potential research partners, such as government agencies, NGOs, and community organizations.
3. Identify relevant health concerns at each field site. 

I recognize that my ideas and the approaches that I use this summer will most likely undergo a transformation between now and July, but I am happy to have a plan. It gives me something to think about and has made me more excited to tackle the forthcoming challenges. I’m sure this will be the last time I ever stress about my research… 

This week was challenging. I have been struggling to find a data set that I want to work with. I spoke with Dr. Gravlee last week and we brainstormed some ideas. I have been trying to follow up on the thing we discussed but have been running into problems. I tried to collect articles from the four major Jamaican newspapers, but I can only access the achieves of one of them. I searched the suggested databases from the syllabus but haven’t located a data set that works for what I am trying to accomplish. The hunt continues and I am confident that I will have a data set by this time next week.
On a positive note, I’ve found that as I am going through my week, I see model and matrices everywhere. It gives me pause as I like to identify whether it is proximity or profile matrix. After which I spend a few moments examining the content, thinking about the information being conveyed but also why this individual choice to represent this data in this particular way. Using the world around me to study and engage with the materials from class has proven to be an exciting and unexpected element of this class. Additionally, everywhere I go I hear “indigenous terms.” Again, I find myself pausing to think about the phrase and trying to understand it in the context of its use. Being on campus with students who are 10-12 years younger than me provides numerous opportunities to test your skills. While I do find myself frustrated with my data set hunt, I have greatly enjoyed breaking down the wall between classroom learning and life experience. By taking what I’ve learned in class and applying it to the rest of my week, I have been able to engage with the material in a new and fun way which has led to greater learning outcomes. I intend to continue to use this approach to sharpen my skills and I look forward to adding more skills to my toolbox.

After careful consideration, I decided to change the focus of my text analysis semester project. The decision is two-part, first I do not feel the data from my Jamaica project contains enough text to engage in the type of analysis asked from this class. Secondly, I recently accepted a research assistant position which includes a text analysis component. For now, the project specifics are somewhat vague. What is clear is that part of my responsibilities will be examining E. Coli references in local, regional, and national papers. The project is a proof of concept funded by the University of Florida as part of its “Moonshot” grants. As part of my work this week I have begun the process of collecting articles. My intent is to finish the collection process today and begin analyzing. I recognize this shift has put me slightly behind where I would like to be at this point in the semester and I will use that as motivation over the next week.

This week we explored the process of coding and creating codebooks. I have found text analysis to be a more creative process that I initially thought. Creating codes and recognizing themes is an enjoyable task which allows the researcher to explore text in a manner which reveals potentially extraordinary amounts of information. This process can become problematic if the researcher creates too many themes to the point where analysis is bogged down, and the researcher suffers from fatigue or becomes overwhelmed. Miles and Huberman (1994) recommend 50-80 codes. This number does not represent the amount you will most likely start with. Codebooks are subject to change and undergo a sort of evolution throughout the research and analysis. The original theme lists tend to be comprised of a large number of themes that over time are redefined, adapted, or in some cases dropped or merged with another theme.

There are a number of approaches which can be utilized depending on resource constraints and research design. I started by reading some of the open-ended text from my interviews to see what strategies I naturally began to use. I found myself frequently recognizing linguistic connectors, word repetition, and transitions. Themes that emerged included vulnerability, access, trust, change, cash crop, thief, seasonal, and farm inputs to name a few. One technique that I struggled with was Missing Data. I think that this technique will become easier to utilize as my familiarity with the text increases. To help myself recognize missing data I used an approach that we learned in class last week, which is to specifically examine the text using a single technique. In this case, I re-read the text and looked only for missing information. Using this approach helped me explore the text through a specific lens and exposed themes I had not noticed on my first review. One interesting theme that emerged from using the missing data technique was the discovery that individuals who participated in the farmer field school did not have critiques of that program. I suspect that they may be unwilling to criticize the project because they fear that doing so may jeopardize their inclusion in the project or result in criticism from project managers or fellow participants. Up until this point I hadn’t given much consideration to this possibility. Recognizing information that may be missing is a powerful tool. Having said that I can also see how it could be used incorrectly. This is where having multiple coders would be useful, a point that was made several times in this week’s readings.

With respects to creating codes and codebooks, it is clear that regardless of expertise the process is time-consuming and continuous. While the use of specific software has made certain elements easier, creating a codebook with strong validity is a marathon, not a sprint. With that in mind, I am looking forward to taking the next step.

This blog is designed to serve as a platform for tracking my research ideas and growth as a scientist and an anthropologist. The first several posts will be directly tied to the work I am undertaking for Dr. Gravlee’s Text Analysis class at the University of Florida. My intent is to continue this blog beyond the conclusion of that class. My hope is that by engaging in a weekly write up of my research I can document my struggles and growth while forcing myself to thoughtfully reflect on my research.

Each week for class I will answer four primary questions:
a. What new steps have you taken this week in the analysis of data for your project?
b. What findings are you coming up with?
c. What challenges or difficulties have you encountered?
d. What are the strengths and weaknesses of the specific methods you have used this week?

My intent is to use the 72 semi-structured interviews my research partner and I collected in 2014 in Jamaica as the text to study for my class. These interviews begin in a structured format which focused on collecting household data. By design, as the interviews unfolded the nature of the interviews became more unstructured. The focus of the research was to understand how small-scale farmers in Jamaica were accessing and utilizing climate services (drought predictions, mitigation strategies, etc.). Additionally, we sought to understand sources of vulnerability in the communities we studied, Ballard’s River and Redland to determine how climate change fit into the overall causes of vulnerability. The assumption made by climate researchers and government entities was that climate change was a major threat to the island and as such small-scale farmers farmer recognized and prioritized climate change in a similar manner. We studied community vulnerability to understand if this assumption was true.

The specific techniques which will be used to analyze this data have not been determined yet. However, based on the approaches used to collect the data some techniques are better suited than others, while some are not appropriate to use at all. Using Figure 5.1 in Analyzing Qualitative Data by Bernard et al. (2017) provides a flow chart which has helped me narrow down my list of possible techniques. The data I have is textual and is not verbatim. The flow chart suggests that repetitions, similarities, and differences, and cutting and sorting are techniques which are appropriate for data which was collected in a manner like mine. I look forward to discussing the strengths and limitations of each of these techniques tomorrow morning in class.

Thanks for joining me!

Good company in a journey makes the way seem shorter. — Izaak Walton