Five Ways to Use AI Assist in MAXQDA Coding (Full Transcript)

[00:00:00] Speaker 1: Hello, everyone. I'm going to talk to you about AI function in Maskudio. Yes, Maskudio has an AI function, and the question is, when do you have to use it? And this is what my presentation is going to be about. I'm going to talk about five main moments that you could think about using AI to help you to make sense of your data. As I always say, the best way of analyzing your data is manually going through the data, identifying information that is significant, and develop codes and themes to help you to address your research question. But in a situation where you are not sure about the kind of information that is significant, you are not sure about what kind of label do you want to use to represent the significant information, then AI function will be helpful for you. You are using it as a tool to help you to make sense of your data. So what I always say is that first, make sure that you understand the coding process so that when the AI suggests something to you, you'll be able to evaluate and make sure that the information is fairly representing the data that you have. So this is what you have to think about when you are deciding to use AI function in Maskudio to help you to make sense of your data. So let's go through the process. So the first strategy is when you are not sure about a kind of code that you want to use to help you to represent the significant information, then AI assist in Maskudio could be helpful. So let me quickly show you how you can do that. So I have my data here. As you can see, I have my five participants and the study is about burnout causes and solutions. I just want to find out what are the causes of burnout among primary health care physicians and what are the solutions of burnout. I've created two main containers for my research question here. When you go through the data, I can identify information that is significant, that is addressing my research question, but I'm not sure about the kind of code that I want to use. And this is where you use the AI assist. One example is, let's say you are going through the data and then there's this statement, right? It's about lack of balance with work and home. So you have this statement and you are not sure about a code. So a code is a label that represents the significant information. A code can be in between two to five words representing the information that you have extracted and also addressing your research question. If you want to get a practical understanding on how to code your qualitative data, then this book will be very helpful for you. So this is the second edition of my step-by-step guide to qualitative data coding. It's going to be helpful for you. You can get access to it when you go to our publisher's website. You can get access to that when you also go to Amazon. We'll provide that information in the description area in case you want to get access to this wonderful book. And if you are an instructor, you can also get it for free by going to this link. I'll provide that information there. So in this case, the information here is addressing a research question, but you are not sure about the label that you want to give to it. So you can right click on it and then go to AI Assist and then click on Suggest New Code for Selected Tests. So when you go here to give you this window, you click on OK. Right. And the system, within a few seconds, will be able to generate a code, a list of code for you. So as you can see, I have a list of codes. I have about nine suggested codes. Right. And it has been grouped into three thematic code alternatives and also interpretive code. It can be also called description focus code because it directly describes what you have selected. And then the last one, interpretation, goes beyond the description but giving you suggestions on how the AI makes meaning of that information. So you can see that the interpretive codes are more of abstract compared to the descriptive code. For me, AI is very good in giving you descriptive codes like thematic code. So I always suggest that try these thematic codes instead of interpretive code because when AI is interpreting something, it also makes assumptions that you are not aware of. This function only has access to the information that you have selected. It doesn't know the context. It doesn't know your participant. It doesn't know the research question that you are addressing. So when you move to interpretation, it may not be able to capture what exactly the participant is saying helping you to address your research questions. At this stage, I will use the code suggested under thematic code. So as you can see here, work-life imbalance might be the best. And also you can see the definition. Right. When you're okay with that, you select one. You can select more than one, but I think this one fairly captures what I selected. And I click on okay. When you do that, it will come under codes, but it's not under any of the research questions. So you have to drag and drop it to the respective research question. In this case, I'm dragging and dropping it under research question one. Right. And then if you want to know more about this code, you can right-click on it and go to memo. The memo will give you the name of the code and also the definition of the code. So you see the definition here. Right. The name of the code and the definition. Then if you decide that you want to change, maybe you don't like the name that has been given, you can always right-click and go to properties, and then you can change the name if you want to. So this is how you can get codes that is addressed in your research questions. So you can do that. Let's try another one. So you select this one, and then you can right-click on it. Select this one, right-click, AI assist, and click on suggest new code. Click on OK. And then looking at this content, it's about spending time in praise and reflection. So I think maybe the first one best reflects what I selected. And then I click on OK. And as you can see here, it's not under any of the research questions. So you have to drag and drop it into the respective research question. Oh, this one belongs to research question two, right, about the solutions of burnout. So this is how you're going to do it. Go through all the transcript, identify information that are significant, and ask AI assist to help you to develop a code that best represents this information. I like this process because the system provides you suggestions. And it's very important for you to know the strength and limitation of this function. As I indicated, the strength is that it gives you, based on the information that you have selected, it will give you options for you to choose from. And you can decide to choose the one that best reflects what you have selected. The limitation is that it doesn't know the context, right. It doesn't know the question that you ask participant. It just knows the information that you have selected. It doesn't know your research question. Because of that, you have to be very careful and make sure that you review and decide the best one that best represents the information that you are selecting. The second way is when you are not sure about the information that is significant. If you cannot identify any information that is significant addressing a research question, then you can use the system to help you. And how are you going to do that? We're going to go through the process and then see what we can do here. As you can see, I have my five participants and these are my research questions. I created containers for the research questions, burnout courses, and solutions. So I want to go through the data, identify information that are significant, and develop codes to address my research question. Imagine that I don't know the information that is significant. What should I do? When you are creating your container, it's very important for you to... let me create one container for you to see what I mean by that. So let me show you how I created a container for research question one. So I went here, click on the plus sign, and then I indicated the research question one, right? I have already the research question one here, so I'm just going to type research question one here, right? And then remember you can define what a research question is. The research question is that what are the causes of burnout. So normally what you're going to do is you can just indicate here, the system will view this as code. So that's why I'm typing this way. This code represents the research question, and then I bring the research question here, right? And then under that you can provide information about... in the normal situation you can just indicate this one and move on, but we're going to use this one in the next step. So just follow me and I will show you why we are providing that. What you have to provide here is that you have to provide an example, right? The reason why you are providing examples are you are showing... going to... later you're going to use this to show the AI system that these are my questions and these are the examples of quotations that you have to extract addressing the research question, because we're going to let the system develop quotations for you. So you can go to one... it's not required if you don't have any quotation that is addressing a research question or you don't know where to find an example of the quotation, that's fine, but providing an example helps the system to know what exactly you want it to do. So what I will do is that I open one of my transcripts. So as you can see, this is one of the transcripts, right? So what I'm going to do is that in terms of burnout courses, I'm going to also select some quotations for the system to know example of quotation that is addressing a research question. So I'm selecting this one. I will say quotation one, and then here I will say quotation two, quotation two. So I go and then I have another quotation that I can do about course of burnout numerals, clinical and administrative tasks. So I bring the quotation here, right? You can bring dots, yeah, three dots to show that there's some information before that. So now I'll bring the quotation. The last one that you have to state is that it's called delimitation. You know, the reason why you have to have this is where you want to tame the system. You want a system to know what exactly the boundaries, right? What are the boundaries, right? So what do you want a system to do? What do you don't want a system to do? So boundaries, establishing boundaries is very great so that, you know, the system will do what you expect it to do. If, you know, you want to show the boundary. So this is a strategy that you can use. You can use a word like always. You can start with a word like always extract information that is addressing the research question. Or you can also say never extract information that doesn't address the research question. Or so you can use never, always. Yeah, so to draw a little bit of value. So I can say that always extract relevant quotation from the data addressing the research question, right? So that's the strategy. Just present a boundary so that the system will work within those boundaries. So when you are done, you click on OK, right? So that's what you're going to do. And then if you want to change the instructions, you can always go back and do that by right-clicking on this. If you go back, right-click on it and go to purposes, you will not see what you type in terms of the instruction that you're going to give to the system or information that related to the research question that we typed, right? So this is where you're going to go. Right-click on it and go to memo. Memo has all the information that I type. You see here it has my research question that I type, examples, and also delimitation, right? So you can also add or adjust any information that you want to adjust before you start coding the process. What if we just created a code without giving information about a code under memo? This is what is going to happen. The system may not know what the code represents, so the system may guess and then extract anything that it thinks may help. So this is where we call it providing contest. You provide examples of the quotation and then you give a boundary that will help the system to give you what you want, right? So as you can see here, the old one that I did, when you go to right-click here and you go to memo, you'll see what I wrote. So it's similar to what I example I gave you about this code for the research question one new. This is the old one. You can see that I've provided examples. I indicated what a code, what the research question is, and also provided the limitation. Only HVAC quotation addressing the research question, right? So now that we're finished, we're going to start the process. So I have my two research questions and I want the system to go through all my transcripts, extracting significant information that is addressing my first research question. You go here under document and select all of the transcript that you are interested in, right-click on it and go to AI assist and go to AI coding. Right. And then you drag and drop the research question that you are interested in into this place. When I drag and drop memos, the information memo, code memo is also the information that I type about. If you just created a container for the research question, this is a very good time for you to make it type what this code represents. See, this code represent all quotation addressing the following research question. And I have my research question here. And also I provided the system an example and I provided the limitation, right? So we click on, okay, if everything is fine and make sure that all the documents were selected right before you go to the process. So you see here that 10 significant information was extracted from the five documents, right? And addressing the research question, right? So I click on okay, and then you see here, it has created a code under the main research question, another container for that. And it is showing that it has gone to five documents and then 10 means that 10 significant information was extracted. And when you scroll down here, you can see that the 10 significant information that was extracted, this one was extracted. And then the same thing, this one was extracted. And if you want to know the list of all the information, you can double click on that entry to show you the first one to show you all of them. This is a chance for you to review. And if you don't like the quotation, if you realize that, oh, this one is not addressing the research question, what are you going to do? You can just check here, right? And then you can delete. So if you always have a chance to review, make sure that everything is addressing the research question before you go ahead. And this is also a good time if you want to code everything by yourself, after you have selected significant information, this is a good time to code. You can always generate a code here, right? So you can select, and then you can click on retrieve segment with a new code. And you click here and then you type a new code and you type okay. So this one code can be called having long hours. And you can give a description if you want. You can say this code represent participant expression of doing a lot of work in a shorter period of time, right? So you can give a description here and a code memo. It's not required, but if you want to define the code, this is a place that you have to do that. And then click on okay. When you do that, that information comes here. So you have to drag and drop it into the respective research question. So you see I have a drag and drop it into the AI generated segment for research question one. So that's how you can do the coding. If you see that you go to the next one, you click and you can also code if you want to code, right? So this is how you can allow the system to go through the data for you to help you to identify information that is significant and help you to address your research question that you have. And as you are doing that, you have to be systematic and you have to be ready to be transparent and report how you use AI in Mask UDA to help you to make sense of your data. Next step is when you have your research question and you want to develop subcodes under the research question or you want to develop codes under the research question, this is the strategy that you have to use, right? So in this situation, you have already gone through the data, identify information that are significant, and then you want to develop codes, right? So let's go to Mask UDA. So as you can see, I have my five participants here. You can see that for research question one, I have 14 quotations that I have extracted from the data. And for the second research question, I have 12. Imagine that you have already created containers for the research question and then gone through the data, extracted information that is significant, but you haven't developed any codes yet. What are you going to do? You can use AI Assist to help you to develop codes under each of the research questions. So how do you do that? You can easily right-click on the research question, but the step that you have to take before you do this one is to make sure that you have extracted all the significant information and connected to their respective research questions. As you can see here, I have 14 significant information. I right-click here and I go to AI Assist, and then I go to Suggest Subcodes. Click on that, and then what the system is going to do is going to go through all the 14 significant information and then generate codes that are representing the significant information. So you click on OK here, and then within a few seconds, you'll be able to get your results. So as you can see here, the system has suggested a lot of codes for us. We have 1, 2, 3, 4, 5, and also Other. And then each of them, when you click on it, it also has the definition of the code and examples of the significant information that are really connected to the code. And then it also gives information about delimitation. So it has given you all the description about a code, including delimitation. You see here, it has given you work-life imbalance, the definition of that, and also examples, and also delimitation. So if you are okay with them, you can click on All, and then it will move to the side, right? You can always delete what you don't want. Let's say you don't want Other, you can click on it to delete it. If you don't want it, you just click on it to delete, right? So you have the chance to delete what you don't want. And when you are done, you click on Create Codes. And then when you do that, what will happen is that the information will come right under the research question here, right? So I have all the codes here. There's a little limitation here that there's no information or significant information that has been dropped into all these codes, right? So we don't know out of the 14 which of the quotations are connected to each of the codes that we have here, because the system wasn't able to do that. What you could do is that you can double click on the research question container and then go through them one by one and then drop it into their respective codes, right? So you can see that long hours, you can select and drag and drop it into this code, which is called workload and time pressure, right? You go to the next one, this one, numerous clinical and administrative tasks is related to administrative burden. So you see how you can easily go through and manually select information that are significant and drop it into their respective code. I like this process because it really forces you to be active in data analysis process, right? The system has suggested codes for you, but you have to go through all the significant information and drag and drop them into their respective code, right? So it helps you to play an active role in the data analysis process. That's what I like about it. But if you really want the system to go through the transcript and extract information and connect them to the code, you can also do it by, you can select all the codes here, right? And then let's say I want the system to go through all the transcripts here and extract any significant information addressing the code workload and time pressure, right? What are you going to do? You first select all the information that is significant, all the transcript data you want the system to go through. You right-click on it, you click on AI Assist. So you right-click and go to AI Assist and go to AI Coding. And here, make sure that all the document of interest has been selected. You click on OK, and then you drag and drop workload and time pressure here. You see here that the system has also brought information about what this code represented, examples of that, and delimitation. This one will help the system to look through and extract information that is truly addressing or extract information that are truly linked to this code. So what will happen is the system will go through all the quotations, all the transcripts, and extract information that are significant and then connect it to their code workload and time pressure. So you can see that 10 significant information have been extracted from the five documents. You click on OK, and then you can see here 10 camps here. You can double-click on the review, double-click on the code, and review all the quotations and see whether everything is addressing the research question that you have. If you want to learn more about MassQDA, this webinar, House on MassQDA Training, is a two-day training from April 27 to April 28 from 10 a.m. to 2 p.m. And then if you want to be part, you can click on a link in my description and you'll be able to register. And then I'll put a promo code there. You get 25% off if you use that promo code. So this is where you have gone through the data, identify information that is significant, you have already developed codes, but you have a lot of documents to go through. So let's say you have maybe 20 transcripts and then you're able to go through five of the transcripts, right? You already developed initial codes and then you want to use the initial codes to go through the document that you have not coded so that the AI can code it for you. How do you do that? So let me show you how you can do that. As you can see here, I have five participants and the transcripts from them, right? Five transcripts. And I have my two research questions and then let's look at the research question two. I've gone through the data. I think I've gone through one or two of the transcripts and I'll be able to develop these codes, right? And then you want the system to use a code to go through the rest of the transcript that you haven't gone through yet, right? So what you're going to do is you can select the ones that you are interested in. So I'm going to select, let's say I'm interested in three, four and five, right? And then I right click on them and then go to AI assist and then I go to AI coding, right? Now make sure here under document, make sure that you selected all of them, right? Sometimes initially it doesn't select everything. So you have to make sure that you selected the ones that you are interested in and click on okay. Now, what code do you want to use for you to go through the significant form? Let's say I want to use this code. So I bring it here, right? This is what is very important. It's very important. This place, you have to provide definition of the code. What does the code represent, right? And then, so you have to define the code. So definition, because without doing that, the system might not know what do you mean by engaging and exercising, right? So defining the code, right? And then after defining the code, you also have to provide example. And then you can provide boundaries with respect to what you want the AI to do to delimitation. If you don't have any boundaries description, you can leave that place alone. But these are the three things that you need, right? In this case, you want the system to go through the data, right? So you are defining this concept, engaging in exercise. So we can say this code represent participant expression of engaging in physical activities to reduce burnout. And here you provide example. You think about the tasks that you want the AI to do. You want the AI to go through the transcript, identify information that is significant. So you could go to your transcript and identify information that you think is significant in relation to engaging, exercising. This is participant one document. And you can choose any of the quotation from the transcript that you have coded. So I can just copy this one and then bring it here. I can type quotation one, and then I'll say exercising. So you can bring about two quotations just to provide the system an example of the kind of quotation that are linked to this code. So let me check another transcript and see whether I can find something. Let me see. This is second transcript. I haven't coded for this example. I haven't coded three, but if you can find anything that could help the system to know what you're looking for, that would be great. It looks like I cannot find anything here too. So maybe a few people talk about exercising. So normally you give two or three, but if you don't have any example, you can give what you have. In this case, I don't have much. And then here you are just trying to tell the system that we can say make sure all quotations are related to the code provided. So providing these three kinds of information will help you to get very good results. So when you are done, you click on OK. And let's see what the system is going to do. So the system is going to find any quotation that's related to exercising. Maybe there might not be any quotation. You see here, there's no quotation that is related to exercising based on the three documents. So not a lot of people talk about exercising. You can try another one and see. Let me go here. Go to AI. Make sure you select the ones that you are interested in. Maybe let's select everything, right? Select the ones that you are interested in. Click on OK. And then I bring the engaging and religious activity here. Click on OK. And then let's see what the system will do for us. OK. So there are two segments connected to the five documents, right? So click on OK here. And then you see here that the system has created a code called AI Engaging in Religious Activities. And when you double click on it, you can see the two quotations, right? So one said they engage in prayer. And one also said they practice positive thinking, right? And they also attend church. So you see how it is done. The strategy here is that you first code part of the data. And then based on the quotation that you have, you can use it to go through the rest of the data, right? And when you are providing a systemic code, you have to also define the code, provide examples of quotations that are related to the code, and also delimitation. They are very important for you to get a very good result. So that's how this one is done. The last strategy that I want to talk about is developing themes from the data, right? Using AI Assist. And this one is very straightforward, right? So imagine that you are going through the data. You have identified information that are significant and developed codes, right? But you are having difficulty categorizing the codes to develop themes. AI Assist could help. So what are you going to do? So what you can do is that you first select, we call it activate. So you click here and click on document to activate all the files. And then you also activate the codes that you are interested in. In this case, I'm interested in all the codes under the first research question, right? Now what you have to do is to, you see this icon? You're going to chat with the retrieved segment. So you click on that here. A window will pop up and then you start chatting. So we see that 40 segments have been selected, right? And this means that all the place that we have activated have been selected. So we activated the transcript and also we activated the codes, right, under the research question. So you can say that, can you review all the codes under the burnout causes RK1 and categorize them to help develop themes addressing the following research question. It's always important for you to set a research question for the system to know the context because just a label for the research question like RK1 doesn't give the system a lot of information that it has to work with. Sometimes you can also provide a little background of your study so that the system will know what the study is about. So you can say this is the purpose of the study. The more you provide a good context, the richer the output. So I'm going to bring the research, the purpose of the study, and then I'm going to click on enter. So within some few seconds you'll be able to get all the themes addressing the research question and also the codes under them. So as you can see here, the system was able to go through and then I have about four themes and then the system has grouped the codes under them. So with this information, you can review and make sure that everything is right. Are the themes addressing the research question that you have? Are the codes related to the themes? If you are satisfied, you can come here. You see here, you can first copy the theme. So I'm going to copy that and then I go right click here on my research question, click on new code, and then paste that information here. You can give a description if you want to. Click on okay and now what I'm going to do is I'm going to drag and drop all the codes that are related to the first theme. So you see how you can copy the theme and when you copy the theme, you right click on the research question, go to new code, put the theme there, and then you can drag and drop the codes under their respective themes. The AI will help you to get themes. You review the theme, making sure everything is right. You create containers for the theme and drag and drop codes under them. So I hope this one is helpful and if you have any questions, you can put in a comment section and I'll be happy to address them for you. Thank you so much for your time.