NVivo vs ATLAS.ti: Comparing AI Features for Coding (Full Transcript)

[00:00:00] Speaker 1: thinking about using any AI function in AtlasTI or in Vivo. And you want to decide which of the AI function that you have to use. And then this video is for you. So what I'm gonna do is that I'm going to explore all the AI function in AtlasTI and in Vivo and compare so that you can make informed decision which one you want to use, right? Especially now that we have AI incorporated in most of the qualitative data analysis software, you want to be sure about a function so that you see how best you can use in an effective way and sometimes in ethical way to help you to make sense of your data. So what we're gonna do here is to explore the AI function in AtlasTI and also in Vivo and then we make informed decisions, right? So let's start with Vivo. The data that we're going to use is about responsible innovation, right? So the purpose of the study is to how participant conceptualize and apply principles of responsible innovation and also there is to fostering responsible innovation and also opportunity for enhancing responsible innovation. And there are 10 participants, these are their roles and also their profession, their age, gender, and geographic allocation, right? So imagine that you are using in Vivo, you have uploaded all your transcripts, this is the 10th transcript and you want to use the AI function to use it to make sense of your data, right? So let's explore the AI function. Let's say you want to use AI to summarize your data, you can just right click on one of the files that you want to summarize and go to AI Assistant Summarize. Always make sure that you have logged in. So when you click here, you see my account, you click here and then you'll be able to log in because without logging in, you may not be able to use the AI function, right? So make sure that you log in before you can so that the AI function will be available for you to click. So as I said, you can use it to summarize your transcript, right? You can also use it to summarize codes. Let's say you have developed codes and then you can right click on it, it will be able to use to summarize the information that you have dropped into a specific code. But one main function that we really want to do is to explore how you can use it to develop codes, right? So before you use the AI function to help you develop codes, you have to take some actions, right? After uploading all the transcript, you have to go to codes, you have to create containers for each of the research questions. So we have the third research question, research question one about how they conceptualize and apply principles of responsible innovation, the barriers to fostering responsible innovation. And last one is opportunities for enhancing responsible innovation. So you see how I've created containers for my research question, right? After creating containers, you go back to the files, you double click to open. So we can see here that I've opened the transcript for participant one, and then you can go to codes with all the containers for the research question and just select information that are significant and drop it into their respective containers for the research question. So to repeat, if you want to use the AI function to code your data, you first have to go through all the transcripts and select information that are significant and drop them into their respective research question, right? Then after you have dropped them, let me close this place. You can see here that there are 10 files connected to this research question, and also there are 24 references, which is 24 significant information, 24 assets that have been extracted from the data and then drop them into the containers for the research question. And then 24 research question two and 26 for research question three. Now that you have finished extracting all the information that are significant, then you can use the AI assistant to help you to develop codes. So if you want to do it to develop the codes, you right click on the research question that you want to create codes and go to AI assistant suggest chart codes. You click on that, and then you always have to read the terms before it allows you to go to the next step. So you have to click on the blue link to read the terms and conditions. And then, so let me click on that. And after reviewing the terms and condition, and if you're okay, you check that and click on okay. Now you can click on generate for you to get all the codes developed by. You see here that the system has developed codes that are addressing research question one, right? So you have an option to eliminate any of the codes that you don't want. You can uncheck if you don't want that. You also have an option to make an adjustment to the codes in terms of the labels, right? But you can do that later. And these are the quotations or significant information that are connected to a specific cause. If you want to know what is connected to community involvement, this is the quotation for. If you are okay with everything that you see here, you click on okay. So you can do the same thing for research question two and three. So these are the codes that have been created by the AI. You can see here, AI, AI. So let me give you a little overview here. So before you use this AI tool, you have to first extract all the significant information that are linked to a specific research question. The limitation that I can see here is that the system doesn't know the research question that you have. But because you extracted all the significant information, there's high probability that the codes that you're going to develop are in line with the research question. Then you have an option to review and make an adjustment. If you want to change anything, you can right click on it, go to code properties. And then you can make an adjustment to the names that you have here. So you can see here that another limitation that it doesn't develop themes for you per se, develop codes. So you have to review and categorize the codes and develop themes, right? So one positive thing that I can see here is that involved in the data analysis process, you are actively involved because you decide information that is significant that will be given to the system, which is the AI, to make sense and develop codes. After that, you also have an option to review and decide which codes you want to select. And also you can make an adjustment to the labels. So you have, it's like you are actively part of the data analysis process. You are not passively involved where the AI just develop codes or themes and give it to you and then you just accept. So that's the really a positive side about this one. But the negative side is that the system is not aware of your research question. And also it may take a long time to analyze your data because you first have to identify information that is significant. So that's all about the in vivo version of the AI function. Let's go to AtlasTI and then let's explore and see. So we have already uploaded all the 10 transcripts here. So these are the 10 transcripts. And then when you go to search and code, there are AI function, there are about four AI functions here. AI coding, this is where you allow the AI system to go through the transcript and identify information that is significant or emerging themes and suggest that themes for you. For this one, the system may not be aware or is not aware of your research question. You just go through the data and select significant information and then develop codes. Intentional AI coding is quite similar to the AI coding, but the difference is that this time you have an option to provide a system your research question. So this is a little bit a positive side compared to the, this is quite one advantage compared to the in vivo option. So in vivo, you don't have the option to provide your research question, but this one you have an option to provide your research question, right? And then there, you can also summarize your documents. It gives you a chance to summarize your documents. And another option is that you can also interact with your documents. You can let a system go through, you can select document that you want AI to review and ask the system AI questions based on the document you have provided to the AI system. So I, in general, I can see that Atlas AI has more functions in terms of using AI than in vivo. Another thing is that you don't have to go through all the transcripts and identify information that is significant. You just have to give all the transcripts to the system and the system can help you come up with codes to address your research question. But sometimes based on my experience, it can give you a lot of codes. And I'm going to show you here. It can be overwhelming in terms of the codes that you're going to get. So you have to bear in mind that you have to, you'll be receiving a lot of codes for you to review and then see how to categorize them to address your research question that you have. So let's try to explore each of the function or maybe let's try to explore some of the functions, right? I've already told you about each of the function, but the one thing, the one that I'm very interested in is intentional coding, right? So you click on intentional AI coding and then you select the transcript that you want the system to analyze. So I want the system to analyze all my transcript. So I select all of the 10 transcripts. Then I go to next. So this is a space where you can provide information about your study. So the one thing that is important to provide is the purpose of the study and a research question that you want to address. So what I'm going to do is I'm going to copy and paste the purpose of the study and the research questions. And then let's see how the system is going to function. So I have the purpose of my study, the research question. I just, let me remove the labels for now. I have the labels for my research question. I just want to remove it and allow the system to work on what I've provided. So I go to next. This one is telling you that the information will be uploaded to OpenAI, the owners of ChatGPT. So your information will be going to ChatGPT model for it to analyze. So you click on next. And then now the system has suggested three research questions. Here you have to review and make sure that it's the research questions that you want the system to do. And then the system has provided it some labels, right? You can even adjust the level, right? So this one is for the first research question. So I can say in parentheses, RQ1, just to help me to know that it's for research question one. And then this one will be RQ2 and RQ3. You have the option to add more questions, research questions if you want to, or if you don't want the system to answer any of these questions, you can turn it off, right? So then you go ahead and then click on start coding. Well, as you can see here, the system has gone through and provided generated codes. So it has generated about 767 codes. So as I told you earlier, it provides you a lot of codes. So sometimes you have to go through and make sure that they are addressing the research question. So you can see that there's a little bit of issue here. I indicated research question one, two, and three, but I can only see research question two. Where's research question one? This is research question one, but the information I indicated in research question one doesn't show here, although it has provided me the label for research question one. So let me click on here and you can see there's a lot of codes, right? And another limitation here is that we call something system prompt. We don't know what instruction that has been given to the model, whether the model has to give X amount of codes or the number of where for each code should be maybe from one to two or one to three. We have no idea about what instruction that has been given to the AI system, right? So that's another limitation. And as a researcher, knowing all this information will help you to better understand how the codes were generated. So you can see that, okay, we have a lot of codes here to go through. They talk about access. If you don't agree with any of the codes, you can turn them off, right? You can remove them. The second research question was about barriers. So we have the ones for the barriers too, right? And you see another, sometimes I feel like I don't have much control here. I did not ask the system to provide me to information about barriers. So we have barriers here, right? And then we have another barrier here. So the barrier here has only three codes and here has a lot, right? So that's another limitation because it doesn't do what exactly what I want, right? I think that if I want to give some kind of suggestion, if there's a way that I can, they can strictly go by what I'm requesting. I requested for three research questions and I got duplicate of the second research question, right? So maybe there's gonna be improvement as time goes on, right? So that's how the system is going to provide you. And then after that, if you are okay with what the system has provided to you. So now I can decide that, oh, I have only a few codes here. Maybe I can turn this one off and only focus on the collaboration barriers. The second one that has a lot of codes. So you can always decide, turn some off or remove some of the categories or codes if it didn't meet your expectation. When you're ready, click on apply. And then your codes, it gives you a summary. You have 10 documents analyzed and 316 quotation was extracted and then 762 codes were generated. And then you can click on okay here. So when you do that, when you go to codes here, you have all the three categories based on the research question and then you can see all the information here, right? So that's what it can provide you. And then what you can do is, if you think that the codes are a lot, what you can do is to export and then send it to ChatGPT for the system to help you to categorize. So you can just go to import, click on import export, and then you can export as a code book. And then you can go to ChatGPT and access system to review all the codes based on a research question and then categorize them to develop teams and a system will be able to do that. But another limitation is that using one word as a code, sometimes there's a limitation because the word can mean many things, right? And there's no area where they have provided a description or the definition of those words. So that's another limitation that you have to think about, right? So what I'm saying is here is that, yes, the AI function in Atlassia and in Vivo, they are very good, but you also have to be aware of its limitation so that you can use in a way that will help you to make sense of your data, right? So this is what I have for you. If you have any questions, you can put in the comment section and I'll be happy to address them for you. And don't forget to subscribe. And also you can suggest a video that I have to do to help you to better understand how to use some of these software and also how to do your qualitative analysis or do your qualitative research. Thank you so much for your time.