How AI Can Accelerate Grounded Theory Development (Full Transcript)

[00:00:02] Speaker 1: Welcome. Today, we're talking about grounded theory development and how you can leverage AI for this. My name is Anne-Marie Brown, and I'm joined by experts in the area, Dr. Philip Adu, who has written several publications on qualitative research methods. So wherever you're joining us from, if you could just place in the chat where you're based, so we'll have an idea of the geographical representation. Just place where you are from and your name. So I guess the very first question is, Dr. Adu, for persons like me, what is grounded theory?

[00:00:47] Speaker 2: Okay. So the grounded theory is all about, you know, you are trying to develop a theory, right? And it should be based on the data that you collect, right? So that's why it's a grounded theory. The theory that you want to develop should be based on the data. And the theory should explain something, right? It should explain a phenomenon or a problem that you have identified. Let's say you are, you know, teaching a course and you realize that a lot of students didn't pass, right? They failed the course. So you can do a research and develop a theory to show why they failed the course, right? So a theory in this context is the relationship you are, it's just a statement, right? That shows relationship between concept, explaining something, explaining a problem, explaining why the student didn't pass, right? So normally it's used to explain something, right? And explain something to, you know, you call it data to explain a phenomenon. That's the simplest way that I can put.

[00:01:53] Speaker 1: All right. Thank you so much, Dr. Odu. If you're just joining us, we heard what grounded theory is. Welcome Paco, a familiar face from Spain. Paco has visited, well, joined several of the live streams already. Welcome Paco. So we have spoken of what grounded theory is. Is it the same as a hypothesis?

[00:02:16] Speaker 2: Yes. So it's the theory that you develop from the grounded theory is, it can be closer to hypothesis, but it's not really hypothesis. Because when you talk about hypothesis, you are thinking about quantitative study, right? Where you generate you, one of the hypothesis is maybe, is there a relationship between variable or is one variable contributing or affecting another variable, right? So we can generate a hypothesis from the theory that you develop, especially if you are doing a mixed method study, where you start with grounded theory, develop a specific theory based on the data, generate hypothesis, right? And then call it quantitative data to test the hypothesis. So there's a little bit of difference, but I think there's a relationship also between them because you can generate hypothesis and test it using quantitative data and from a qualitative theory that you have developed from the grounded theory.

[00:03:23] Speaker 1: All right. So thank you for clarifying that the hypothesis is more related to quantitative data and data analysis. Did I hear that correctly? Okay. And the hypothesis is when you're testing. Well, okay.

[00:03:38] Speaker 2: Using quantitative data.

[00:03:40] Speaker 1: Yes. All right. So if you're just joining us, please put in the chat your name and where you're based. And today we're speaking about grounded theory. So the next question is, are there different types of grounded theory?

[00:03:54] Speaker 2: Yeah, there are many types, but I can talk about two main types, right? We have the classical grounded theory, which is, you know, started at the beginning trying to come up with a way that you can objectively generate a statement from the data that you have. And then there is also, after having classical grounded theory, there were a group of researchers that they thought about involving participants in the process of generating knowledge. They think that knowledge generation should be a core. It should be the interaction between you and the participants, right? Then you'll be able to generate a theory for you to use it to maybe explain a phenomenon. So the second part is called constructivism or constructivist grounded theory, because you and participants are constructing knowledge. You interacted with participants. You got the information. And then now you want to make sense of the information that they gave you so that you will be able to develop a theory, which is just a statement explaining a specific phenomenon, right? So these are the two main types, but there are other ones. The classical one, which is, you know, the foundational grounded theory and also the constructivist grounded theory, which is the current one that a lot of people are using.

[00:05:28] Speaker 1: Yeah. All right. Thank you for explaining the different types of grounded theories and welcome Felix from Ghana. There's a question coming in and it's from Talahun. Hope I'm pronouncing your name correctly. And it is, please, can you summarize about explanatory sequential mixed method? Yeah.

[00:05:50] Speaker 2: So it's a type of mixed method approach, right? So the explanatory comes from, you are using qualitative data to try to explain the quantitative finance, right? So whenever you see explanatory, because quantitative information, you start with quantitative study, right? You call it data. You maybe find out whether there's a difference between two groups, right? But if you want to know why there's a difference, this is where you go ahead and maybe interview participants. So you are based, you wait, it's sequential because you have to wait and get information from the quantitative finance and then maybe develop interview questions. And then you can also call it data, analyze and address the research question that you have or explain the quantitative finance. So that makes it a sequential mixed methods approach.

[00:06:51] Speaker 1: All right. Thank you very much, Talahun. I hope that answers your question. And on your screen, you will see a QR code. You can scan it and join this conversation where you can ask your question or give a comment live. So use the code to join the conversation. Dr. Odu, do we have anybody lined up?

[00:07:12] Speaker 2: Yes, we have one who has joined. Let me take out a QR code here.

[00:07:27] Speaker 1: Anne-Marie, can you hear us? Please go ahead on mute. All right. Not hearing from Anne-Marie. Do we have anyone else in the lineup?

[00:07:45] Speaker 2: No one else. I'll check. All right.

[00:07:49] Speaker 1: So Anne-Marie, feel free to give your comment or question in the chat. And I open the floor to others as well. Use the chat to ask questions, see clarification and scan the QR code to join. All right. So let's move on. So I think we have an understanding of what grounded theory is. Now, this live stream is about leveraging AI. So Dr. Odu, how can AI support grounded theory development?

[00:08:23] Speaker 2: So it's the same thing as doing a qualitative analysis, right? So when you are doing a qualitative analysis, what you are doing is that you identify information that is significant and then develop codes to represent the information and try to categorize the code to develop themes to address your research question that you have. We can use AI to help us to go through that process. So the same way when you have grounded theory. So imagine you have interviewed a lot of participants. You have a huge amount of data. Sometimes it takes a time for you to go through all of them physically and manually code. So this is where AI comes in to help you. Okay. Telling AI that, okay, this is the study and this is my research question and I want to develop a theory. Can you go through the data to help me to develop? So the system will go through, identify information that is significant and then develop codes. And then the system, we call it initial coding, right? Going through the data, identify information that is significant, develop codes. And then we go to the next stage where we call it focus coding, where you can tell the system, okay, can you review all the codes that you have generated, right? And see whether you can, using the focus coding strategy, see whether you can come up with themes for me, right? And the system will be able to do that for you. And you can even ask further questions about, okay, can you make sure that the themes are in line with the significant information that you have initially generated? So, you know, asking the question, we call it chaining prompting. It's just similar to semi-structured interview, where you don't have a structured question that you're asking participants, but based on how they respond, you ask further questions. So interacting with the system, based on how the system responds, you can also ask further questions, right? So that at the end of the day, the system will help you to come up with a potential theory that you have to review and make sure that they are helping you to explain the phenomenon or the problem that you came up with.

[00:10:36] Speaker 1: All right, so well said, right? And could you repeat the type of prompting that you mentioned that you asked for follow-up? What was the name of it?

[00:10:46] Speaker 2: It's chaining prompting. So chaining. So let me see whether I can share my screen. So let me see here.

[00:11:01] Speaker 1: And in the meantime, while Dr. Odu is sharing his screen, keep the comments or questions coming.

[00:11:08] Speaker 2: Okay, so I think people are joining, but I'll wait. Let me share my screen first and see. Share screen.

[00:11:21] Speaker 3: Show presentation.

[00:11:34] Speaker 2: Okay, so can you see my screen? It's called chaining prompting.

[00:11:41] Speaker 1: Okay. Chaining prompting.

[00:11:44] Speaker 2: Yes. So chaining prompting is, you see, it's like you have a step-by-step communication with the system, right? We think that, oh, I didn't type it well here. It's chaining. Okay. And let me make it a little bigger so that everybody will see.

[00:12:04] Speaker 1: So for persons who are just listening, it's chaining as in a chain necklace. Chaining. Chaining. Chaining. Chaining prompting.

[00:12:13] Speaker 2: Yeah, chaining prompting. So this is, you know, you would start, first of all, before you do a chaining prompting, you always have to have a background knowledge about Granite Theory because Granite Theory has a specific process. You know, one of the dominant processes are initial coding, focused coding. I can type them here. Initial coding. Focused coding. And then we have theoretical coding. So you always have to have this knowledge at a background because if we don't have this information, then how are you going to communicate with the system, right? So you start this process by first letting the system know, providing a little bit of context. What is the purpose of your study? What is the problem that you want to address? What are some of your research questions, right? And then tell the system that, okay, I'm using a Granite Theory approach to analyze this data. Can you start the process by doing an initial coding? And you can also tell the system, okay, I mean going through the data, identifying information that is significant and developing codes, right? And then the system will generate that information to you. You as a researcher have to review this information and make sure that they are truly, the codes that are generated are truly from the data that you have. Then if you are not satisfied, you can ask the system, can you review the codes again and make sure that they reflect the data that I have given you? So the system can do some kind of self reflection and present you the result. Then you can move to the next one, right? So you see how based on the information that it provides to you, you also ask further questions. At the end of the day, you'll be able to come up with a theory that explains a phenomenon. And also you can ask the system, oh, so can you review the theory and make sure that I have enough data in support of this theory? Or do you think I should collect more data? Or do you think I should go back and interview participants? Okay. If I should go back and interview participants, what kind of questions should I ask participants? You see the way continuously asking the system some questions, one question at a time, you'll be able to generate a lot of information that will help you to analyze your data and also come up with your theory.

[00:14:51] Speaker 1: Great explanation, Dr. Ado. And before we go to putting people on the live with us, if you didn't get it fully with the coding, don't worry. We have a workshop in a couple of days. The link is coming across your screen shortly, where in this workshop, we will go in depth in the different types of prompting, chain prompting, and also how to do the coding and so forth. Because then we have more time. This is a live stream with limited time, but in the workshop, you can get more guidance and more explanations. So that's the link and the QR code on your screen. Dr. Ado, do we have anyone?

[00:15:36] Speaker 2: Yes. Let me take out the... Yes. We have Dr. Shahrokh. Hello, Dr.

[00:15:49] Speaker 1: Shahrokh.

[00:16:04] Speaker 2: Yes. Your live keeps on cutting. Can you repeat what you said?

[00:16:09] Speaker 1: Hello. Hello. Thank you.

[00:16:24] Speaker 2: Do you have any questions for us or comments?

[00:16:34] Speaker 3: Oh, thank you.

[00:16:35] Speaker 1: It's so nice of you to come on. It's nice to hear your voice.

[00:16:40] Speaker 3: Thank you.

[00:16:41] Speaker 1: Thank you. Nice.

[00:16:42] Speaker 3: And there's another one.

[00:16:44] Speaker 1: Ah, Anne-Marie again. Welcome.

[00:16:47] Speaker 2: Welcome.

[00:16:48] Speaker 4: Do you hear me this time?

[00:16:50] Speaker 2: Yes, we can hear you.

[00:16:52] Speaker 4: Yeah. Yeah. Hi. Very interesting discussion around AI and grounded theory. I'm rather new to grounded-based theory. And one struggle that I have when I, for example, I try to develop a theory of change for a specific project is how to translate the complexity of interlinkages, assumptions, context in a way that it's appealing visually and clearly to the audience. And now linking back to the AI, it's very interesting that the AI can assist within the internal kitchen of developing your end product. But I wonder if it can assist also in structuring the end result in a way that it's more user friendly and not text heavy.

[00:17:57] Speaker 2: Oh, yes. It can really do that. I think that, you know, it's all about having that kind of conversation with the system. You can say that I'm going to present this theory to this group of people with this kind of background. Can you help me to come up with ways of presenting this information so that it will be easily absorbed by them? Right? And the system can suggest some information for you. They can also tell you, okay, maybe you have to break it down and also present and have an analogy, right? So that people will really understand. Or they can give you examples of ways of explaining the theory. So it will be able to do that for you. Especially chat GPT, right? The same way that, you know, just tell what exactly you want and see what you're going to get. And also there is a website called Gamma. It's G-A-M-M-A app. It can help you to develop a document, a simple document with visuals to help people, especially if you want to break things down. You can also develop a PowerPoint for you with visualization. That will help people to absorb the information or, you know, learn more about information that you want to break down. Yes. So there's always a possibility for you to do that.

[00:19:24] Speaker 4: Thank you so much. And a second worry that I had related with AI use is sometimes the AI seems to be a sort of black box and especially for analytical processes. So it's interesting that we can learn to give more refined prompts in order to assist each step of development. But in the same time, the question that I have, what are the criteria that the black box uses in order to create a causal link or validate the causal link? Or should I also provide to the AI very clear in structure around, okay, how you should look for causal change? Well, what are the criteria for a strong causal link between the different elements?

[00:20:18] Speaker 2: Yeah. Yeah. Thank you. You are so right. You know, in terms of what happened at the background, we don't have all the information. Right. But I think that what we could do is to be clear about what the system has to do for you. Right. So you can say, you know, it's all about giving a clear command and you can also tell the system, can you do this way for me? And a system can do it. You know, they want most of the time, if you want to get a lot of information, you can give examples. Right. One example is that, okay, you want a system to look into the relationship between codes or themes. Right. Can you give an example? You can say that, okay, for this significant information, I can see the relationship between these two information. Right. The relationship between these two things, this data can support a relationship. Looking at this example, can you go to the data and also look through and extract information that can support the relationship that you are suggesting? Right. So, you know, according to research, if you give a couple of examples, the system will know what you want and they will provide it to you. I think it's the same thing as we haven't completely understand how our brain works. Right. So the same way, you know, we have not totally comprehend how the system work at the background. But what we have is when you provide more context, you provide simple questions, there's high probability for you to get answers and try to show that you have to also scrutinize the answers that you get. Right. You can say that, oh, can you really look at this information I provided to me and making sure that they are the right questions. Right. Or they are the right results. Letting the system do some kind of self-reflection, you'll be able to get rich information from it.

[00:22:21] Speaker 1: Yeah.

[00:22:21] Speaker 2: Thank you so much. You're welcome. Also using the current model, not the old ones. Current models are very good to help you to get what you want.

[00:22:32] Speaker 4: What are those? Sorry.

[00:22:35] Speaker 2: So when we look at GPT, I can quickly show you. Let me see what I can share my screen again. Okay. Share screen.

[00:22:45] Speaker 3: Let me go to. Let me see. Okay.

[00:22:54] Speaker 1: Okay. So.

[00:23:06] Speaker 3: All right.

[00:23:06] Speaker 1: So if you're just joining us while Dr. Ado shares his screen, we're speaking about grounded theory development and how you can leverage AI to assist you with that process. And at this stage, great questions by Anne-Marie. Thank you so very much. Looking at the different models. What are the latest models that is fit for this program? What are the latest models that is fit for this purpose?

[00:23:31] Speaker 2: So the latest model that is fit is GPT 4.0. It's very good. It can provide you, especially when you give clear questions and provide context, it will be able to go through your data, extract information that is significant, and help you to come up with a theory. The latest one that is also good, but it's in a preview format. So this means that it's not fully functioning. It's O1 preview. This one is normally used if you want to view the model as a consultant, right? So imagine that you are, you have met a consultant who have expertise in something and you have a higher level conversation with that person, right? This system is so good because it thinks. It reasons, right? And it takes its time to reason because when you ask a system a question, the system will go through the processes of raising step by step, right? So that gives you the best answer for your questions that you are asking. But the problem about this one is you cannot attach any documents. So it's not fully functioning. But in case you want to have top-level conversation, one conversation is like, maybe you are looking into a theory of change, right? You can say that I want to present a theory of change to this audience. Can you show me the best way to break this down so that people will understand what I mean by this concept or this theory? The system can give you step by step information. So it's so amazing how the new model can help. But now, if you want to upload the data, you have to use GPT-4.0 and then ask the system a question. Another good thing, I have come up with a chart GPT. I've generated, we call it custom GPT. It's called, let me look for that GPT and show you. You can even search if you have access to chart GPT. It's called Granite Theory Analyzer, right? I developed it and this one is so helpful. It helps you to go through all the process initial coding, focus coding, theoretical coding, and developing your theory. So the most important is to provide a system context, upload your data, and ask step by step questions. And you'll be able to get good information from that. It has been used by a lot of people and I think they have given a very good feedback. So it's going to be very helpful if you are doing a granite theory and you want to get some ideas about the theory that may reflect the data that you have. I can put the link in the chat box or when we finish, I can go to the system and then post the links there so that you'll be able to have access to this one.

[00:26:33] Speaker 1: I think the chat box now would work for them. I was anticipating that question. Can we get the link to the Granite Theory Analyzer? It's straightforward and it's easier than this you can't get. Just use the GPT he already created. The link is in the chat.

[00:26:52] Speaker 4: Thank you so much. Very interesting. Thanks.

[00:26:56] Speaker 1: You're most welcome. Do we have anyone else on the line?

[00:27:00] Speaker 2: Yeah, we have a lot of people waiting. Let me see. We have Kinsley. Oh, Kinsley is not. Kinsley, are you there?

[00:27:12] Speaker 5: Yeah. Hello, please. Can you hear me?

[00:27:15] Speaker 2: Yes, I can hear you. Let me see. Yes, Kinsley, we can hear you.

[00:27:24] Speaker 5: Okay, so I think I just joined. Before that, I was having a meeting with my supervisor, so I couldn't get a lot of information which was shared initially. I hope that for what I've seen just a couple of minutes, it has been insightful looking at how AI can help us to perform other research work within a blink of an eye. I think it's helpful and then kudos to you as well.

[00:27:55] Speaker 2: Thank you so much and also thank you for joining us. We appreciate your time. Yes, it's so rewarding. When it comes to the qualitative analysis, it takes a long time for you to manually go through the data. Imagine you have a software that can within seconds go through and identify information that is significant, and then you can ask the system some questions. It's so rewarding, and it gives you more time for you to critically review the information that the system has given you and then ask further questions. It's really a great opportunity.

[00:28:33] Speaker 1: Thank you for joining us, Kinsley. Franklin, thanks for joining us as well. We have been discussing grounded theory, what it is, how it differs from hypotheses, and how we can leverage AI. Felix, welcome.

[00:28:51] Speaker 2: Hello, Felix. Oh, okay. I think we can go on and maybe just join in. Okay.

[00:29:01] Speaker 1: All right. Do we have anyone else on the line?

[00:29:04] Speaker 2: No, we don't. Yeah, we don't have any.

[00:29:07] Speaker 1: All right. We have been speaking about the beauty of AI in terms of speed and efficiency and coding and so forth, but what are the challenges in using AI for your grounded theory analysis? Yeah.

[00:29:25] Speaker 2: I think one main challenge is that if you don't have any background of grounded theory, you might not be able to give effective prompts. You might not be able to see whether the information that the system is giving you is the right information. So I always say that before you use this tool, just get a basic understanding of what grounded theory is all about and also the steps so that you can, you know, have very good interaction with the system, right? The same way when you interview participants, you should be knowledgeable about the kind of questions that you're going to ask participants and so that you can have a fruitful discussion with them, right? And secondly, when you are interacting with the system, one challenge is that it may not give you information that is based on the data that you provided. So you always have to prompt it. You can say that, please, you are reminded that any conversation we're going to have should be based on the data that I have attached, right? You should go based on the data. Don't go beyond that. So giving the system some kind of parameters will help you to get information and also providing some context, right? Before you start a deep conversation with the system, kind of give a little background about your study. Don't assume that they know about the data in your study, right? The same way when you are meeting somebody for a consultation, they can ask you background information, right? To help the system to provide you rich information. So if you are not clear, you might not get rich information from the system. And always have to check because it can give you wrong information, right? Especially, I don't know, but I see that the system is such a way that it wants to technically please the user, right? It wants to make you happy. So if you don't take care, it will give you what you want and you are happy that you have gotten a theory. But maybe it's not really related to the data that you have. So it's so important to be skeptical. It's so important to ask the system to review what you have. It's so important to ask the system to give you evidence in supporting the theory that it has created, right? So that's a way that you can also help. So it can also give, you know, because the system is trained on data, the data can be biased and it can also reflect on the information that you are generating. So you always have to be, I think, you know, be aware that the same way participants can give you biased information because every information that they are giving you is influenced by their background and preconceived ideas. The system can do the same thing to you. So what do you have to do? Questioning, always question the information, always ask for more information, always trying to let the system do some kind of self-reflection so that you can get rich information. And see yourself as the head of the analysis process. You are not solely dependent on the system to provide you some information. When it provides you, you always have to analyze and then make sure that information is true before you can use it. So these are the things that you can use to prevent any kind of limitations.

[00:33:01] Speaker 1: Okay, fantastic. Quick question though, Dr. Adu. Hi. Is it called socially desirable? What is it when respondents tell you what they think you want to hear? I forgot the term for it.

[00:33:19] Speaker 2: Yes, yeah, yeah, socially, yes, I think that's the term, socially desirable response, right? Oh, yes, because they want you to be happy, right? They are assisting you. They want your work to be easy, right? So one example is that when you have a research question and the data that you have, and based on the research question, the data is not adequate enough to address a research question, but a system can come up with things, right? Because it's like, I want to assist you, so I'm being technically forced to please you. So if you don't tell the system that, okay, every information that you are going to give me should be solely from the data that I've given you, it will go beyond the data and focus on other things that have been trained just to please you. So you have to be very careful when you are interacting with the system.

[00:34:19] Speaker 1: Yes. Okay, with that said, we are wrapping up, but before I go into the last segment, please remember to put your comments, your questions in the chat. This is one of your last opportunities to do so. We were quick and furious today, but we do have a workshop on AI that we will go into more details if you want to know more about proper prompting, coding, and using AI, and some of these things, the hallucinations and the desire to please, and how you can prevent some of this in your work so that you don't get false data. Join the workshop where we have three days to go in depth with you. So we're in the final stages to wrap up. What are some of the ethical considerations, Dr. Adu, when using AI for your qualitative data analysis?

[00:35:17] Speaker 2: Yes. So I think the first one is you have to be careful about the data that you upload into the system, right? So when you are using ChatGPT, Cloud AI, and any AI tool, you have to make sure that you are protecting participants. So any identifiable information in the data, you have to make sure that you remove them before you upload. And for ChatGPT, you have an option to indicate that they shouldn't use your data to train the system, right? Because sometimes they use it to train the system so that the system will do well, do improve the model so that you can use it and get what you have. But you just have to be very careful. If you are using any other AI tool, you always have to go and read about their privacy information and see what they use the data for. Are they going to use the data to train your system? Are you okay for them to use the data? There's a little positive side. If they use the data to train the system, you get good models to use, right? But at the same time, do you want participant information to be out there, right? Or somewhere where you don't have any control. So you have to balance and see a way that you can do it in a way that is ethical. Another thing that you have to think about is that you have to be transparent, right? In doing research, you know, if you are not transparent, people will not believe what you found, right? In qualitative analysis, you know, you can just say that you use specific AI tool to help you, not to do the analysis for you, to help you to develop the theory. And you can also give the processes how you did it, right? The initial coding, focused coding, theoretical coding, and all these things to help you to be transparent in the process. So just also be aware of the system bias. Don't believe everything that you got from the system. Because as I said, it's made to please you. When you're having a conversation, you feel like the person wants to make you happy and support you, but you have to be very careful. Is that information truly from the data that you have? So these are the things that you have to think about.

[00:37:34] Speaker 1: All right. Great. I hope that you have taken notes because AI is great, but there are limitations and ethical considerations. Last question, Dr. Ado. And the last question is, this one came in, is where do you see the future of AI in Grounded Theory Analysis heading over the next few years?

[00:37:59] Speaker 2: I think the future is we will not spend a lot of time going through the data because the system can do that for us and do it perfectly, right? And we will spend more time on reviewing the results and making sure that they reflect the data that we have. Is that the same thing as you give a work to an assistant to analyze your data, and then now that your assistant finished and presented information, you have to review because everything that you put out there is you are responsible, right? So AI will be doing a lot of work for us in terms of data analysis. Our role is to make sure that we are transparent and we are providing step by step how we have a conversation with the system to get what we want. So that's what I see the future. I also see that I was even thinking, I don't know whether it's even happening now. There's a software that can help you to interview participants, right? So you can just give a link to the participant, the participant will have communication, and then as the participant is having communication with the AI, the system will be recording and transcribing at the same time. So there's a software called AI list that is doing that thing where you participant can, you know, you just give the link to the participant and the participant have a communication interview on your behalf. It can do structured, both structured and also unstructured interviews. So the future is bright, and I think it's also giving us opening a door for everyone to generate knowledge in a faster time. So we'll be generating knowledge for in a shorter period of time, and the sky is the limit.

[00:39:58] Speaker 1: Indeed, and I think that's a great place to end on. If you want more information, where we have more time, please scan the QR code, join the workshop, three days. We have different levels on Monday, October 21st. We have the beginner, the intermediary level, for those who want a little bit more advanced.

[00:40:22] Speaker 2: Oh, go ahead.

[00:40:23] Speaker 1: Yes, thank you, I'm wrapping up, so go ahead.

[00:40:27] Speaker 2: Yeah, I just want to quickly share what I did with the granite analyzer so that you see how you can use it, right? Just a brief information. So I wanted a system to analyze this data for me. I have 10 interviews about people's experience on online courses, right? And then I started by giving the system some background information about my study, right? So I gave a little background, and also the purpose of the study, and also the research question I want to address. That's very important. Whenever you want to start a conversation, provide background information. And the system, you know, gave me an idea about, okay, I know what you want. I'm going to do initial coding, identify key concepts. I'm going to do focused coding. I'm also going to do theoretical coding. And then I attach my transcript. Okay. And then the system, you know, review all the transcripts and came up with the initial coding results. So you see here that we have a theme here, and also code under them, right? So the system also provided explanation of all the codes that has been generated. And I said, okay, before you proceed to do the focused coding, can you provide evidence from the data in support of the themes and code generated? Now, at the beginning, when the system gave me code, it didn't provide me quotation from participants. It mentioned some of the participants, what they said, but not a quotation. So in order to be sure, I asked the system to provide me code from the data, right? And the system did that. Sometimes you have to also check the document and make sure that this quotation is truly from the data, right? Now, after that, I said, okay, can you conduct focused coding, right? And then the system went and did the focused coding. The focused coding is trying to establish relationship between the themes that the system brought, and also gave me evidence, right? And then I asked further, the system gave me the next step. Okay, the system said, okay, so I finished with the focused coding. I want to move ahead and do the theoretical coding. And I said, oh, okay, proceed with theoretical coding. So you see how I'm using, we call it chaining prompting. One question at a time, making sure that the information is right before I move to the next step, right? And the system did the theoretical coding for me, and then provide some evidence. And then lastly, the system came up with a theory, right? And then I asked the system, so do you think I should consider collecting more data to see whether they support the theory you have generated, right? And the system said, okay, yes, you know, the system wants me to collect more data so that I'll be able to make sure that the theory that the system has generated is truly from the data that is out there. So and then I asked the system, what kind of questions should I ask participants? What kind of information? And the system gave me some of the questions I have to ask participants to get more information. It's so amazing how, you know, technology can make the process so easy. So this is what I have for you. I hope you will try it out and also let me hear from you what you will learn from this process.

[00:44:07] Speaker 1: Great. And as I say, you can get further demonstration, get to use the tools, see more in the workshop that we have coming up. It's been great. Thank you for your time today. Dr. Adu, thank you very much. And until next time.