A Practical Guide to Qualitative Coding and Themes (Full Transcript)

[00:00:00] Speaker 1: Today we're going to talk about understanding qualitative coding and I'm going to try and answer 10 common questions that you need to address before you start coding your data and if you have any questions you can put it in the chat and I'll be happy to address them for you and also you can answer some of the questions right. So what is qualitative coding? Qualitative coding is all about going through the data, identifying information that you think is significant and then you develop a code. It's like a label that represents that significant information so that it helps you to address your research question that you have. That's the simplest way to define qualitative coding. The question that you may ask yourself is what information is significant? Any information that will help you to address your research question. Sometimes it may not directly address your research question. It can give you some contextual information. Yes, you can code that significant information. So let's go through the 10 questions and let's see. So the first question is why do researchers code qualitative data? Why should we code it? Why don't we just collect data from participants and then we just present that information to our audience? Why should we code the data? We code the data first because the information that we have collected, not all of them are significant. Not all of them are important, right? You talk to them about a lot of things. Let's say you are doing a study about burnout among primary health care physicians and you are interviewing participants. They may talk about things that may not be related to burnout. So coding helps you to separate the information that is useful that can help you to address your research question so that you'll be able to understand the phenomenon that you are studying. So first, that's the reason why we code. Secondly, you want to make sure that you address the research question. So you go to the data with the research question in mind and then you can identify information that you think can address the research question and then develop code. You can think about this like data reduction. You are trying to reduce the data. How do you reduce the data? By first extracting the information that you want, labeling them, right? Sometimes categorizing them and then develop themes that best represent the significant information and also address the research question, right? So you are reducing the data so that it will be meaningful for it to address your research question. And it also helps you to communicate that information. So when you try to reduce the data in a very meaningful way, you can also share that information to your audience for them to consume that information. So technically what you are doing is that you take raw information, you break them down, right? And then extract what you want, reduce it, and then provide that information to your audience so that they can easily consume that information. So that's all about coding and why we have to code, right? What makes a good qualitative code? So a good qualitative code is any code that represents significant information from the data, right? So let me quickly show you an example. So as you can see here, let me see whether I can make it a little bit larger here. So as you can see here, one of the research questions is about the solutions of burnouts, right? So you can see here that it says engaging in exercise. This is a code representing what participants said that they have to engage in exercise to reduce stress, right? So a good code is any code that represents information that is significant from the data. And also it's addressing a research question that you have. So you can see here that this one is engaging in exercise and it's addressing the research question that you have. A good code should be also meaningful. So a good code shouldn't be too broad. Let's say you are looking into the causes of burnouts and then you collected your data. You want to identify information that are significant. And you got a code, but you said external causes. And then you got another quotation and you said internal causes. That is not a very good code. It's because it's so broad. What kind of internal causes is that? What kind of external causes is that? So it's very important for your code to directly represent information that is significant and also address the research question that you have, right? So that's all about the second question. Let's move on and talk about the third question. How do researchers decide what information is significant enough to code? How do you decide that, okay, this information is significant? When that information can address the research question. So as you can see here, let me show you again. I know that, let's go here. I know that this one is going to address the causes of burnouts, right? So this information is significant. Sometimes that information may not directly address the research question, but it can give you some background information, right? So it doesn't sometimes have to directly address the research question. It can give you some context. What situation do they find themselves in that caused them create the situation like having numerous work-related hours, right? So you see that maybe that person is the only person there taking care of patient, right? Then the person might have a lot of work to do. So you can see that significant information can be information that is addressing the research question or provide context to support the information, the main ideas that you're going to use to address the research question, right? So that's that answer to this question. If you have any questions you can put in the chat, I will be happy to address them for you. So let's try the next one. Should researchers develop codes before reading the data? Should we develop a code before we read the data? No, first you have to review after you have transcribed the data, you have to first familiarize yourself with the data by reading and rereading so that you can get a big picture about the data before you do the analysis, right? But when you study analysis, there are different kinds of strategies that you can use to code your data. If you are doing, you are coding it inductively, right? Then you can, like you are using thematic analysis, then you have to read the data, identify information that is significant and then develop codes and themes. But if you are coding it deductively, like content analysis, then you already have a list of codes or themes, so you are going to the data and identify information that are evident in supporting the theme and then you connect those evidence to the theme or the codes, right? So it all depends on the strategy that you are using. Are you using content analysis or are you using thematic analysis? That will help you to determine whether you have to review or go through the data before you decide, before you start coding or go to the data and then after that you code, you develop codes. So I hope this one is clear. So let's go to the next one. Can the same excerpt receive more than one code? Yes, so the same significant information can receive more than one code. Significant information can be a sentence or two sentences or a paragraph and a sentence can have multiple meanings, right? So yes, you can code an excerpt into more than one code. The most important thing is what kind of meaning is the excerpt providing to you and that will cause you to determine whether you have to code it once or code it and connect it to one code or multiple codes. So the next question, how do researchers know when two codes should be combined? When they are similar in meaning. So this one is useful when you are finished coding and you are moving into developing themes, right? So you have to evaluate each of the codes and see whether there are similarities and then if they have similar meanings, you can combine them. If two codes have the same underlying meaning, yes, you can combine them. Let me show you an example. Let me go down here. So as you can see here, this code, being a young physician and feeling inadequate, they have something in common. When you are young, you feel like you haven't had much skills, so you feel inadequate. So you see here that I combined these two codes and then the theme will be, which is that underlying meaning, will be being new to the profession. Because of this one, you can see here that because they are similar in meaning, you can combine them. But everything that you do in terms of combining them, most of the time you have to think about what quotations are linked to that code. So if you have similar in quotation linked to the code, then you can decide, okay, let me combine the codes. So each code will have quotation connected to that. So this code A has quotation code B has quotation. The quotation is similar. When they are similar, you can combine them. So that's another strategy that you can do. So back to the question, what is the difference between a code and a theme? There are a lot of differences. So one simple one is a code, most of the time you develop a code first before you think about a theme, right? So a code is a label that represents significant information from the data. And then a theme most of the time is a concept or abstract. It's more abstract than a code because you have to review the codes and combine them and to develop a theme, right? So the code is most of the time connected to the significant information. A theme is, it represents a group of quotes, right? And sometimes can represent a pattern in the data too. So think about a theme as a broad kind of meaning that represent a group of quotes, and then think about codes are labels or meanings that directly connected to significant information that you have. So that's the two main, the difference between a code and a theme. So let's go to the next one. How do researchers move from code to themes? Okay, one way of doing that is sorting, right? So this means that you are evaluating each of the quotes, finding out the similarities, you know, between them based on the uniqueness of the codes. And then you combine the codes that are similar. And based on that, you'll be able to come up with a label that represent the groups that you have the clusters and that will be the theme. I can show you an example here. So as you can see here, I have a list of quotes that I put in cluster one, I put some in cluster two, cluster three. So you can see here that they have something in common. That's why I put in a cluster one, they have something in common that I put in cluster two. Then after that, I review the characteristics of each of the codes. And then based on that, I label it. So this one becomes a theme, right? Representing a group of codes here, the same thing that I did here. So that's how you move from a code to a theme. So let's move on to the next one. How do researchers know when a theme is strong enough to keep? When a theme is unique and is representing significant pattern in the data, you can keep that. When a theme is also representing a group of codes, you can keep that information. So that's the thing that will help you to keep. Another way is when like a lot of participants are linked to the theme, and also a lot of quotations from the data are linked to the theme is strong enough to keep. But you don't focus on only these two, like the number of participants or number of quotations, it's all about the meaning. The meaning of the theme can be supported by evidence that you have in the data or the patterns that you see in the data that can make a theme strong enough to keep. So what is the next one? Last question. What role should AI play in qualitative coding? AI can play a supportive role. So this means that AI can go through the data, identify information that are significant, and then you as a researcher, you have to review all the significant information that AI has identified, making sure that they are addressing the research question. And AI can also suggest codes for you, labels representing the significant information. But the most important is that your role is to evaluate all the AI outputs, making sure they are correct, because AI can make mistakes. So it's very important for you to always evaluate. You are always leading the analysis. AI is just suggesting information to you. So that's what I have for you today. These are the 10 questions. If you want to learn more about qualitative analysis, my book, A Step-by-Step Guide to Qualitative Coding, will be very helpful for you. And then if you are doing your dissertation and you want to know step by step, how do you have to, what is expected of you for each of the chapters, whether you're doing a qualitative study or mixed method or quantitative, this book will be so helpful for you. If you are doing a phenomenological study and you want to know more about what steps, specific steps that you have to take in collecting and analyzing your data and also incorporating a framework into the analysis, this book will be so helpful for you. And then if you want to learn from people or students who have already completed their dissertation, their experience and their struggles and how they were able to overcome their struggles, this book of a PhD will be very helpful for you. I also provide consultation, one-on-one consultation. If you are doing a qualitative research and you need help, you can contact me. This is my email address, info at drphilipadu.com. And then I'll be happy to provide you all the support that you need. Thank you so much for your time. And I will see you another time.