"you are a researcher. your topic of research is "____________________". you have completed primary and secondary data collection via interviews and research articles, journals, books, government websites, etc, respectively. you have analysed you primary data which is a qualitative data, using thematic analysis. now, you have to write a research methodology. you have to choose appropriate research approach, research philosophy, and also explain how thematic analysis is used to analyse qualitative data" for anyone who is confused just as much as me....this prompt has worked for me quite well. Hope this helps XD
sir i want to cry. i've been stuck on coding for my thesis, have watched 3 or 4 tutorials about discourse analysis that were of no help then found this. my initial codes actually are more or less aligned with chatgpt's. god bless u
This was an academic breakthrough/mental relief in reference to beginning categorizing/creating themes from codes I created in NVIVO. I literally shouted with joy. God Bless you!
Thank you so much for your kind words and enthusiasm! I'm truly honored to have played a part in your academic breakthrough. It's moments like these that remind me why I do what I do. Keep up the fantastic work, and may your research journey continue to be rewarding and fulfilling. God bless you too! 🙏
Well done Dr. Adu, It's just a matter of semantics and knowing what to ask for and what to look for. It is very intuitive. What you put in is what you get back. Thank you for sharing
I am impressed with this new tool! As someone who is new to the concept of "coding," it will be invaluable in helping me understand and organize the codes that have taken me hours to input. Thanks again Dr. Adu. Side note, it seems I have cited you a few times in my dissertation :) thanks for all that you have done in this research area.
Amazing video been having trouble working my way around Nvivo and Atlas.ti. To summarize: 1. Summarize the Transcript. 2. Review Transcript, Generate Phrases (i.e. Codes) 3. Use summary to develop themes
Wow ❤ saved me so much time. Reduced my stress level. I even enjoyed the chapter I was dreading the most. You are an academic life saver sir thank you 😊
Now I found you-yes this is better than That costly program nvivo because If I can think about what I want, All I have to do is simply speak it in the chat GPT and it produces
Thanks for your excellent video. I did a single, small, quite satisfactory attempt with chatgpt 3.5, (themes were similar to those identified by a skilled human), but when I attempted with a larger set using version 4, I encountered a few problems: 1) I am trying to analyze a feedback questionnaire with relatively short answers but more of 1400 students, so I need to send them in about 20 chunks. The problems are that a) it tend to forget the directions (which are patterned over your example in the video,) after a few chunks, so I had to resort to to repeat them every time before each chunk. b) even when given the same directions, it tend to apply them in different ways every time. For example, I interact in English, wile the transcripts are in Italian. Even when instructed to write codes in English and transcripts in Italian, it keeps to do it erratically, sometimes right, sometimes translating both in either language. 2) most transcript (non just a few) are badly fabricated. Not just hallucinations, just bare cheating. Same happens for the identification of themes, some of which are clearly inconsistent with the specific context. You quite rightly warned several times about this possibility. Could you make a video suggesting how to limit and deal with this kind of problems?
Alternatively, you can use claude.ai/chats to analyze your open-ended survey data. If you have the data in excel spreadsheet, you save it as CSV file and upload to the ai tool. Another option is to first upload the data into google drive and obtain a link of the data (when you click on share). You then go to ChatGPT 4 and install 'Link Reader' Plugin. You then paste the link and ChatGPT will access the data. I'll so videos on these soon!! Philip
I must confess, this has assisted me tremendously in my research. You are doing a marvelous job with your channel. It arrived at the nick of time, just about when I was having trouble writing my report. Thank you very much. God bless you.
The big problem with ChatGPT or other AI is missing data if you feed them long text. Often, they overlook parts of the text, which results in missing important data or patterns. I faced this issue while doing historical analysis for my research, and it wasted my time because I had to go through each piece of data and review the outcomes. I hope they solve this issue soon.
Thank you for sharing your insightful observation. sometimes, it depends on the kind of AI model you are using. Trying to use one of the latest AI models such as GPT4o, Gemini 1.5 and Claude 3 would help. Another strategy is to ask the ai tool to diligently evaluate its output - making sure it has all the information you were looking for. I hope this helps!
Sir, i didn't understand addresing the research question the chatgpt generated 20 codes do we need to take all the codes in consideration? or we can choose based on our research question which code to take. please help. thanks.
Tq so much for all the tips shared..It is really an eye opener. However, I have question. Is it ethically accepted to analyse data using Chatgpt? How to report this and justify it in the thesis?
Thank you so much for your kind words! I’m glad the tips have been helpful and provided new insights. Your question is an important one, and I’m happy to clarify. Yes, it can be ethically accepted to use tools like ChatGPT for data analysis, but this depends on a few key factors: 1. Institutional Policies: Before proceeding, check your institution's policies or guidelines regarding the use of AI tools in research. Some institutions may have specific rules or recommendations that you need to follow, and aligning with them is crucial to avoid any ethical concerns. 2. Transparency: Clearly document how you used ChatGPT in your thesis. For example, specify whether it was used to assist with thematic coding, generating summaries, or performing statistical calculations. Transparency ensures your methods are both ethical and reproducible. 3. Data Privacy: Be cautious about uploading raw data with sensitive or identifiable information. Always remove or anonymize such data before using ChatGPT. 4. Supplementary Role: Emphasize that ChatGPT is a supportive tool, not a replacement for your critical thinking and expertise. Clearly show that you interpreted and validated the results generated by the tool. How to Report and Justify This in a Thesis In your methodology section, describe ChatGPT as part of your analytical tools, including: The specific tasks it performed (e.g., assisting in organizing qualitative themes or performing basic text analysis). The steps you took to ensure accuracy and reliability (e.g., cross-checking AI outputs with manual coding or statistical validation). A justification for its use, such as time efficiency, ability to handle large datasets, or aiding in identifying patterns that might be overlooked. You can also include a section in the limitations of your study, acknowledging potential biases or constraints related to using AI tools. Lastly, always consult your institution’s policies on AI use to ensure compliance. This step is critical for avoiding ethical or procedural challenges.
Sir plz guide if we have lots of interviews and lengthy transcripts of all the interviews, what to do then? Will we summarize each transcript sepaerately first and then combine all the summaries
Thank you for your question! It really depends on what you intend to achieve with the summaries. If your goal is to get an overall understanding of the data and identify broad themes or patterns, creating a single, overarching summary that represents all the transcripts would be ideal. However, if you’re conducting detailed analysis or need to preserve nuances specific to each participant, it might be better to start by summarizing each transcript individually. This approach allows you to retain the unique insights from each interview before combining them into a more general summary. Think of it as a two-step process: Summarize each transcript to capture individual perspectives. Combine those summaries into a broader synthesis to identify commonalities and overarching themes.
I just subscribed because this was helpful. I'm just wondering, did you copy and paste all the participants' transcripts when you said 1000 words? Participants 1-5? Did you do the transcripts, coding, and theme generation process individually?
Thank you for subscribing and for your question! At the time this video was made, the process required copying and pasting the transcripts into the tool, which meant coding the transcripts collectively. However, the tool has since been updated, and it now allows you to attach individual files. This improvement makes it much easier to develop codes and themes for each transcript individually. You can work on each participant’s data separately, ensuring a more detailed and nuanced analysis. Let me know if you have any further questions or need additional guidance! This video would help: ua-cam.com/video/45Wl6V8NRJY/v-deo.html
Thank you very much! I have recorded 50 semi-structured online interviews with open-ended questions and now I am thinking whether I should transcribe them (which is time consuming) or writing a more or less detailed summary would be enough? Can you advise?
You're welcome! Going by best practices, I suggest you transcribe all of them. You could use software to automatically transcript them for you. You could try: www.rev.com/services/ai-transcription or www.descript.com/transcription
thanks a lot sir. it was absolutely helpful. I have a question. I have gathered the data through an open-ended questionnaire written on papers. I mean they aren't in my pc, all sheets of paper. how can I analyze them using Chatgpt? Do you have any advice? Thank you again.
You're welcome! I can try this: you can scan all the sheets and upload one to ChatGPT to see what it can read the information on the sheet. If successful, you can upload all of them for it is to extract relevant excerpts and develop themes.
Thanks for your exposition. I am a researcher and as an institution, we handle different kinds of policy research undertakings. However, I have never understood the difference between Quantitative and Qualitative research. I only understand that quantitative research data is a component of Qualitative research Data.... even in medical statistics, the 68. 95, 99.7 rule derives a system's behavioral features and their ratings... else, advise us!
You're welcome! Thank you for sharing your thought about quantitative and qualitative research. For me, I see qualitative research as collecting and analyzing non-numeric data to help address research questions. This presentation may help: ua-cam.com/video/N0lg9l1_d_I/v-deo.htmlsi=gctCuKZyAmz6_YpL&t=649
Hi, when you talk about the research questions, i mean the ones you are coding and generating themes from, is the questions on the interview guide or the research questions in your research proposal?
Hello Dr. Adu, great content. Thank you for providing these UA-cam resources. I have a chatGPT entry that is longer than 3000 words. How do I go about getting the summary? I tried to navigate this, but I was having some issues
Thank you for the content. Is this GPT-simulated thematic analysis? Are there any empirical research validating GPT's efficacy of conducting aualitative analysis?
You're welcome Jenny! It can be used for assist in any kind of qualitative analysis. Since it is a new technology, not much has been done to examine it efficacy. However, it has be used as an assistant - helping researchers to make sense of their qualitative data. You could explore the resources below: www.researchgate.net/publication/377752239_ChatGPT_in_Thematic_Analysis_Can_AI_become_a_research_assistant_in_qualitative_research www.perplexity.ai/search/ChatGPT-for-qualitative-d8rxmzkDTCeoBQwGfiLvaQ
Thank for this informative video Dr Adu. How do you handle privacy issues using ChatGpt? Also, should we be transparent about your methods using AI to the client?
Thank you for your thoughtful comment! The first step in handling privacy issues when using ChatGPT is to ensure that any identifiable information is removed from the data before inputting it into the system. If you're using the paid version of ChatGPT, you can also go to the data control option in the settings and specify that your data should not be used to train the model. Regarding transparency with clients, it’s a good practice to disclose the tools you’ve used if they inquire about your methods. You can assure them that no identifiable information was shared with the system and explain that every outcome was carefully reviewed to ensure accuracy and reliability. Transparency builds trust while maintaining ethical standards in your work.
Thank you Dr. Adu for your video. For the qualitative study part of an explanatory mixed method with a semi-structured interview, which analysis type is more appropriate, thematic analysis or content analysis?
You're welcome! If you plan to generate codes and themes based on the relevant information you extract from the data then thematic analysis is appropriate: see: ua-cam.com/video/LMaz9GsTL_Q/v-deo.html
Hi Sir! I just want to ask that I have a large data set up to 38 responses and the answer in the responses are very long as well. So what should I do to make the qualitative analysis easy. 1. Should I analyze all the responses for research question 1 at one time, then for RQ 2, then for 3? 2. How should I bring the results from chatgpt to my research paper, because at the end, it will be giving plagiarism or AI, and the paper may not get accepted?
Thank you for watching my video and for your questions. My response to your first question: I suggest you analyze all responses and connect them to their respective research question. In other words, with the research questions in mind, you go through the data and extract relevant information and connect them to their research questions. My response to your second question The first step is to view ChatGPT as a data analysis tool assisting to make sense of the data. Second, you determine the role is will be playing in your data analysis task. For example, us can use it to generate initial codes and them you review them - making sure they reflect the relevant excerpts from the data and address the research questions. Alternatively, you can use it to extract relevant information from the data. This process may help you to speed up the data analysis process. If you plan not to copy the information generated from ChatGPT and paste it on your research paper, you won't have any issue with plagiarism. To avoid plagiarism, there are some of the actions you could take: 1. use ChatGPT as a data analysis tool playing an assistant role 2. Make sure you review and refine the codes/themes suggested by ChatGPT 3. Be transparent - sharing and showing how you used ChatGPT to analyze your data 4. Cite source ChatGPT when needed (see: apastyle.apa.org/blog/how-to-cite-chatgpt)
Good day, Dr. Adu. Thank you for putting out valuable content. I am struggling to analyze qualitative secondary sources data and create themes and codes using Nvivo, Atlas. ia..etc. please hope you would not mind sharing available link or guides for all work done related to my inquiry.
Excellent, encouraging video! Is there a way or recommended 'place' to store a large, growing number of lengthy transcripts (100+) for ongoing analysis using Chat GPT or do you recommend another AI tool for that?
Thank you Jana! You could create custom GPT to have analyze your data (see: ua-cam.com/video/SiaZOPjpHS4/v-deo.html). However, I'm not sure you could upload all the 100 transcripts at a time. G
You're welcome! It depends on the kind of report you want to generate. I would try and ask it to see what you'll get. I think it can provide you the results in a Powerpoint slides form. Alternatively, you can use Gamma (gamma.app/) to generate a report.
Hello, I like this explanation very much. However, I would like to ask you for more clarification about summarizing the data by ChatGPT. For example, I have 30 transcripts, each of which is 10 pages long. This means that I have 300 pages of transcripts from different research participants. How can I summarize them by using ChatGPT? Is it possible that I can summarize the first transcript and generate initial codes by using ChatGPT and keep these initial codes generated, then do the same thing with the rest of the transcripts. Finally, I can put the initial code from each transcript in the chat GPT to generate themes?
Thank you! It allows you to upload 10 documents at a time. So, yes, you can upload the first 10 transcripts and develop initial codes and do the same for the second and third 10 transcripts. You can present all the initial codes to ChatGPT and ask it to categorize them to generate themes.
It can but it is not guaranteed that it will solely clean the data without making changes to the original content. Also, it depends on the kind of cleaning you want it to do. Is it about indicate what the interviewer said and what the participants shared?
That was helpful. But can we use this trick in data analysis(synthesis) of systematic qualitative review such as verbatim quotes of various participants converting to codes and eventually generating themes?
You're welcome! It sounds like you collected quantitative data. If this is the case, I recommend you use Statistical software like SPSS. Alternatively, you can use a free software: JASP Software: lnkd.in/eme_2V8 JASP free Textbook: lnkd.in/eyKJPBK Video: lnkd.in/e9t3_D5 JASP 'How-To' Resources: lnkd.in/eGjJn5x Jamovi Software: lnkd.in/eYnf248 Free Textbook: lnkd.in/ejxMxVZ Video: lnkd.in/eFfkfbT
You're welcome!! You can first them in chunks and submit each chunk. After that you can ask ChatGPT to review all the chunks and then ask it to analyze the data.
@@bernadettekina3408 You can start by stating, "I'm going to provide you my qualitative data in chunks for you to review. When I complete providing you all the chunks of data, I'll let you know what to do next. Should I go ahead?"
This is truly amazing! Thank you for all the insightful knowledge! Do you have a method of literature here that I can use to cite this? I would like to use this for my paper
Great video Dr. Now I have one question, do you have a video on the ethics of using chatgpt for on research?. I was thinking if not there must be otherwise these students will missuse this tool.
There is a way but an ethical way of using it for discussion and conclusion is to ask it to provide you ideas on how to write them based on your findings.
Sir can we use for qualotative data anlysis for systematic literature review? i have dowloaded R software for qualitative data analysis, but RQDA package is not available on recent R software.. idk what to do.. need your help!
Quick question, Dr Adu, if using a large in-depth interview of about 50 pages of transcript, how many words can you summarize it to? For this example, you used 1000 words. Is 5000 words ok or too long? Thank you
500-1000 words will be okay. The essence of summarizing is to gain an overall understanding of the data. Then main focus of the analysis would be extracting quotations, and developing codes and themes.
Thank you so much, Dr. Adu. This is really needed at this moment of my thesis.I see that these participants’ interview are short. I would like to ask what if you have a long interview session from about 8 participants. Do you advise I summarize each transcript separately, then resummarize all into 1000 words? Then summarize the new summarized transcripts based on the research questions. Then, generate the codes And then the themes? Thank you for your prompt response.
You're welcome!! You have two options: Option #1: You can first submit the data in chunks by informing it that you have a large qualitative data and you will be submitting the data in chunks. You then indicate 'Chunk One' and you paste the first batch of data and press enter. You then do the 2nd, 3rd, 4th ... until you complete submitting all the data. You then ask it to summarize the data, summarize it based on the research question, develop codes and generate themes. Option #2: Working on doing the summary and developing codes and themes for each of the participant's transcripts. You can compile the output and ask it to generate overall summary, codes and themes based on the outputs ChatGPT generated from each of the transcripts. Alternatively, you can use Claude.ai, see: ua-cam.com/users/shortsNKDcBe8Pw8A
@@drphilipaduThanks a lot for your prompt response. I will explore both options. I just remembered that I can use linkreader to upload the transcript. I will explore all the options. I really appreciate you, Dr. Adu. Thanks, and God bless you
After developing the themes from chatgpt, should we write the details in our work or only table form presentation is enough as you showed in video? Thank you
@@asmashahzadi3803 To maintain transparency, I suggest you show how you arrived at the final output. This video would help: ua-cam.com/video/Gm3okfqvgNs/v-deo.html
hello dr adu, thanks a lot for this video. I have a query. Im doing expert interviews of 3 types of participants. i have totally 15 participants, which means 5 in each category. i find out the perspectives of these categories of experts and To find out agreements or disagreements between them. each in-depth interview is around 2500 word transcripts. how can I use chat GPT to provide a summary after analysing the 5 sets of 2500 words.
You're welcome! ChatGPT has a word limit when it comes to processing inputs/prompts. It can process a request with a word limit of about 2500 words. In your case, I suggest you first request for a summary for each of the transcripts and then compile the summaries and ask it to provide an overall summary based on the 5 summaries.
Hi. I'm conducting a thematic analysis on secondary sources, peer reviewed journal articles. How can I do that using Chatgpt. I have Chatgpt plus. I need your help
Since you have Chat plus, you can use my custom GPT (chat.openai.com/g/g-v8U2B4btf-qualicoptergpt). Also, you can use the following initial Prompts: Prompt #1: Can you give me the six steps of Braun and Clarke's thematic analysis? After it has generated the steps, you then attached your data and use the following prompt: Prompt #2: Can you use the 6 steps to analyze the attached data addressing the following research question: [you can state your research question here]
Greetings! Great video. I needed an alternative because my institution does not want to sponsor NVivo for me. I am on chapter 3 of my dissertation. How would you address and/or justify the usage of chat gpt in my methods section? Should I document every step?
You can say, you plan to analyze your data manual using Microsoft Word document (see: ua-cam.com/video/DSpts_DszEY/v-deo.html). In addition, you plan to use ChatGPT to help you extract relevant information from the data, and generate codes and themes. You should emphasize that the outputs gotten from ChatGPT will be critically reviewed - making sure they reflect data collected and can assist in addressing the research questions. The critically reviewed and refined ChatGPT outputs will then be used to inform the manual qualitative data coding process. To cite ChatGPT, please go to: apastyle.apa.org/blog/how-to-cite-chatgpt Alternatively, you can use dedoose instead of conducting manual coding (see: dedoose.com/home/pricing)
"you are a researcher. your topic of research is "____________________". you have completed primary and secondary data collection via interviews and research articles, journals, books, government websites, etc, respectively. you have analysed you primary data which is a qualitative data, using thematic analysis. now, you have to write a research methodology. you have to choose appropriate research approach, research philosophy, and also explain how thematic analysis is used to analyse qualitative data" for anyone who is confused just as much as me....this prompt has worked for me quite well. Hope this helps XD
Thank you so much for sharing this helpful prompt!!
Please, when do you sue this prompt? Is it for your methodology or analysis?
@@damilolafasinu The prompts are mainly for analysis.
sir i want to cry. i've been stuck on coding for my thesis, have watched 3 or 4 tutorials about discourse analysis that were of no help then found this. my initial codes actually are more or less aligned with chatgpt's. god bless u
God bless you too! I'm so glad that my video was helpful to you. All the best with your analysis.
This was an academic breakthrough/mental relief in reference to beginning categorizing/creating themes from codes I created in NVIVO. I literally shouted with joy. God Bless you!
Thank you so much for your kind words and enthusiasm! I'm truly honored to have played a part in your academic breakthrough. It's moments like these that remind me why I do what I do. Keep up the fantastic work, and may your research journey continue to be rewarding and fulfilling. God bless you too! 🙏
Well done Dr. Adu, It's just a matter of semantics and knowing what to ask for and what to look for. It is very intuitive. What you put in is what you get back. Thank you for sharing
I agree! Providing the right prompt will lead to the generation of a useful output!
I am impressed with this new tool! As someone who is new to the concept of "coding," it will be invaluable in helping me understand and organize the codes that have taken me hours to input. Thanks again Dr. Adu. Side note, it seems I have cited you a few times in my dissertation :) thanks for all that you have done in this research area.
You're welcome @Happy-nu2lk!! I'm glad it was helpful to you. Thank you for citing my works. I really appreciate it!!
Philip
Amazing video been having trouble working my way around Nvivo and Atlas.ti.
To summarize:
1. Summarize the Transcript.
2. Review Transcript, Generate Phrases (i.e. Codes)
3. Use summary to develop themes
Thank you! I'm glad this was helpful to you!!
Amazing video & discussion to help researchers like me! Can't thank you enough for this.
Thank you!! I really appreciate your comment!!
Wow ❤ saved me so much time. Reduced my stress level. I even enjoyed the chapter I was dreading the most. You are an academic life saver sir thank you 😊
You're welcome! I'm so glad it helped :)
Now I found you-yes this is better than That costly program nvivo because If I can think about what I want, All I have to do is simply speak it in the chat GPT and it produces
Thanks for your excellent video. I did a single, small, quite satisfactory attempt with chatgpt 3.5, (themes were similar to those identified by a skilled human), but when I attempted with a larger set using version 4, I encountered a few problems: 1) I am trying to analyze a feedback questionnaire with relatively short answers but more of 1400 students, so I need to send them in about 20 chunks. The problems are that a) it tend to forget the directions (which are patterned over your example in the video,) after a few chunks, so I had to resort to to repeat them every time before each chunk. b) even when given the same directions, it tend to apply them in different ways every time. For example, I interact in English, wile the transcripts are in Italian. Even when instructed to write codes in English and transcripts in Italian, it keeps to do it erratically, sometimes right, sometimes translating both in either language. 2) most transcript (non just a few) are badly fabricated. Not just hallucinations, just bare cheating. Same happens for the identification of themes, some of which are clearly inconsistent with the specific context. You quite rightly warned several times about this possibility. Could you make a video suggesting how to limit and deal with this kind of problems?
Alternatively, you can use claude.ai/chats to analyze your open-ended survey data. If you have the data in excel spreadsheet, you save it as CSV file and upload to the ai tool.
Another option is to first upload the data into google drive and obtain a link of the data (when you click on share). You then go to ChatGPT 4 and install 'Link Reader' Plugin. You then paste the link and ChatGPT will access the data.
I'll so videos on these soon!!
Philip
I must confess, this has assisted me tremendously in my research. You are doing a marvelous job with your channel. It arrived at the nick of time, just about when I was having trouble writing my report. Thank you very much. God bless you.
You're welcome @oladelephilip8273! I'm glad my videos were helpful to you. Thank you for your kind words.
Sir, you are a national treasure, thank you!!!!
Thank you so much for your kind words.
I must admit that this video is one will assist researchers around the world. Sir, you have done a marvelous job. kudos. Sharing it immediately.
Thank you so much for your kind words! I also appreciate you sharing the video!!
Absolutely beautiful. Many thanks and appreciation for sharing. God richly bless you and your precious family members.
Thank you for your kind words! God bless you and your family too!
The big problem with ChatGPT or other AI is missing data if you feed them long text. Often, they overlook parts of the text, which results in missing important data or patterns. I faced this issue while doing historical analysis for my research, and it wasted my time because I had to go through each piece of data and review the outcomes. I hope they solve this issue soon.
Thank you for sharing your insightful observation. sometimes, it depends on the kind of AI model you are using. Trying to use one of the latest AI models such as GPT4o, Gemini 1.5 and Claude 3 would help. Another strategy is to ask the ai tool to diligently evaluate its output - making sure it has all the information you were looking for. I hope this helps!
This was so helpful and you are a great teacher! This video will be useful for my research
Thank you!! I’m glad it is useful for you!!
Thank you for an excellent eye-opening demonstration of how to use chat-gpt.
You're welcome! I'm glad it was helpful to you.
This was concise, to the point and ever so helpful. Thank you.
You're welcome! Thank you for your kind words.
I found this tutorial very helpful and a time saver for analyzing qualitative data.
Thank you!!
Thank's!!! Its a very useful information for we researchers!!
You're welcome :)
If you know how I like watching your video.
Keep going dear doctor
Thank you for watching my videos :)
Very helpful, Dr. Philip. Simple to follow
Thank you! I'm glad it was helpful!
Sir, i didn't understand addresing the research question the chatgpt generated 20 codes do we need to take all the codes in consideration? or we can choose based on our research question which code to take. please help. thanks.
See the codes generated by ChatGPT as suggestions. Our role is to evaluate the codes and select the ones that are addressing the research question.
Tq so much for all the tips shared..It is really an eye opener. However, I have question. Is it ethically accepted to analyse data using Chatgpt? How to report this and justify it in the thesis?
Thank you so much for your kind words! I’m glad the tips have been helpful and provided new insights. Your question is an important one, and I’m happy to clarify.
Yes, it can be ethically accepted to use tools like ChatGPT for data analysis, but this depends on a few key factors:
1. Institutional Policies: Before proceeding, check your institution's policies or guidelines regarding the use of AI tools in research. Some institutions may have specific rules or recommendations that you need to follow, and aligning with them is crucial to avoid any ethical concerns.
2. Transparency: Clearly document how you used ChatGPT in your thesis. For example, specify whether it was used to assist with thematic coding, generating summaries, or performing statistical calculations. Transparency ensures your methods are both ethical and reproducible.
3. Data Privacy: Be cautious about uploading raw data with sensitive or identifiable information. Always remove or anonymize such data before using ChatGPT.
4. Supplementary Role: Emphasize that ChatGPT is a supportive tool, not a replacement for your critical thinking and expertise. Clearly show that you interpreted and validated the results generated by the tool.
How to Report and Justify This in a Thesis
In your methodology section, describe ChatGPT as part of your analytical tools, including:
The specific tasks it performed (e.g., assisting in organizing qualitative themes or performing basic text analysis).
The steps you took to ensure accuracy and reliability (e.g., cross-checking AI outputs with manual coding or statistical validation).
A justification for its use, such as time efficiency, ability to handle large datasets, or aiding in identifying patterns that might be overlooked.
You can also include a section in the limitations of your study, acknowledging potential biases or constraints related to using AI tools.
Lastly, always consult your institution’s policies on AI use to ensure compliance. This step is critical for avoiding ethical or procedural challenges.
Sir plz guide if we have lots of interviews and lengthy transcripts of all the interviews, what to do then? Will we summarize each transcript sepaerately first and then combine all the summaries
Thank you for your question! It really depends on what you intend to achieve with the summaries. If your goal is to get an overall understanding of the data and identify broad themes or patterns, creating a single, overarching summary that represents all the transcripts would be ideal.
However, if you’re conducting detailed analysis or need to preserve nuances specific to each participant, it might be better to start by summarizing each transcript individually. This approach allows you to retain the unique insights from each interview before combining them into a more general summary.
Think of it as a two-step process:
Summarize each transcript to capture individual perspectives.
Combine those summaries into a broader synthesis to identify commonalities and overarching themes.
Please tell me how to do mixed method study especially the qualitative part
This video will be ua-cam.com/play/PLSOjmgd2Fg_lFm4P1zqSiLUeyR58mhhcl.html&si=HwlCltcfwktO6d6k
wow! that's an amazing and insightful presentation. Thank you, Dr. Adu
You're welcome! Thank you for your kind words!
Wow wow 😊
God bless you Dr. Adu 🙌
Thank you! God bless you too!
You are an exceptional teacher. Great work! I learnt a lot
Thank you! I’m glad it was helpful!
Thank you so much for this information. Really helps with my research analysis
You're welcome! I'm glad it was helpful!
This is very powerful. I will be trying this today. Thanks a lot.
You're welcome!!
I just subscribed because this was helpful. I'm just wondering, did you copy and paste all the participants' transcripts when you said 1000 words? Participants 1-5? Did you do the transcripts, coding, and theme generation process individually?
Thank you for subscribing and for your question! At the time this video was made, the process required copying and pasting the transcripts into the tool, which meant coding the transcripts collectively. However, the tool has since been updated, and it now allows you to attach individual files.
This improvement makes it much easier to develop codes and themes for each transcript individually. You can work on each participant’s data separately, ensuring a more detailed and nuanced analysis. Let me know if you have any further questions or need additional guidance! This video would help: ua-cam.com/video/45Wl6V8NRJY/v-deo.html
please iam expecting your feedback, the transcript you put in there is it individually? or participant 1-5
Thank you Dr. Adu. This was really helpful.
You're welcome! I'm glad it was helpful!!
Thank you very much! I have recorded 50 semi-structured online interviews with open-ended questions and now I am thinking whether I should transcribe them (which is time consuming) or writing a more or less detailed summary would be enough? Can you advise?
You're welcome! Going by best practices, I suggest you transcribe all of them. You could use software to automatically transcript them for you. You could try: www.rev.com/services/ai-transcription or www.descript.com/transcription
@@drphilipadu thank you very much for a quick reply and useful links. What you do is priceless!
thanks a lot sir. it was absolutely helpful. I have a question. I have gathered the data through an open-ended questionnaire written on papers. I mean they aren't in my pc, all sheets of paper. how can I analyze them using Chatgpt? Do you have any advice? Thank you again.
You're welcome!
I can try this: you can scan all the sheets and upload one to ChatGPT to see what it can read the information on the sheet. If successful, you can upload all of them for it is to extract relevant excerpts and develop themes.
Amazing. It's getting interesting. Thank you, Dr. Adu.
You're welcome Patrick!!
This was so helpful. Well done Dr. Adu
Thank you John!!
Thank you. A very clear and concise presentation 🙂
You’re welcome! Thank you for your kind words!!
Thanks for your exposition. I am a researcher and as an institution, we handle different kinds of policy research undertakings. However, I have never understood the difference between Quantitative and Qualitative research. I only understand that quantitative research data is a component of Qualitative research Data.... even in medical statistics, the 68. 95, 99.7 rule derives a system's behavioral features and their ratings... else, advise us!
You're welcome! Thank you for sharing your thought about quantitative and qualitative research. For me, I see qualitative research as collecting and analyzing non-numeric data to help address research questions. This presentation may help: ua-cam.com/video/N0lg9l1_d_I/v-deo.htmlsi=gctCuKZyAmz6_YpL&t=649
Hi, when you talk about the research questions, i mean the ones you are coding and generating themes from, is the questions on the interview guide or the research questions in your research proposal?
They are research questions in your proposal. We normally generate codes and themes to address the research questions but not interview questions.
Its amazing Dr. Philip, I’ll try to practice it, hope it works well like u especially that I have lengthy data😢,
Thank you! This video would also be helpful: ua-cam.com/video/45Wl6V8NRJY/v-deo.html
Thanks for sharing this tutorial, doctor Philip.
You're welcome!!
It's a wonderful video ...
Thanks...
Can u please tell if we generate graphs, bar charts for the qualitative data similar to qda miner lite ?
You can but you first need to prompt ChatGPT to generate a list of all relevant excerpts and their respective themes.
Hello Dr. Adu, great content. Thank you for providing these UA-cam resources. I have a chatGPT entry that is longer than 3000 words. How do I go about getting the summary? I tried to navigate this, but I was having some issues
Thank you!! I suggestion you submit the data in chunks, then you ask it to review all the chunks and summarize them.
Thank you for the content. Is this GPT-simulated thematic analysis? Are there any empirical research validating GPT's efficacy of conducting aualitative analysis?
You're welcome Jenny! It can be used for assist in any kind of qualitative analysis. Since it is a new technology, not much has been done to examine it efficacy. However, it has be used as an assistant - helping researchers to make sense of their qualitative data. You could explore the resources below:
www.researchgate.net/publication/377752239_ChatGPT_in_Thematic_Analysis_Can_AI_become_a_research_assistant_in_qualitative_research
www.perplexity.ai/search/ChatGPT-for-qualitative-d8rxmzkDTCeoBQwGfiLvaQ
@@drphilipadu thank you so much!
Thank for this informative video Dr Adu. How do you handle privacy issues using ChatGpt? Also, should we be transparent about your methods using AI to the client?
Thank you for your thoughtful comment! The first step in handling privacy issues when using ChatGPT is to ensure that any identifiable information is removed from the data before inputting it into the system. If you're using the paid version of ChatGPT, you can also go to the data control option in the settings and specify that your data should not be used to train the model.
Regarding transparency with clients, it’s a good practice to disclose the tools you’ve used if they inquire about your methods. You can assure them that no identifiable information was shared with the system and explain that every outcome was carefully reviewed to ensure accuracy and reliability. Transparency builds trust while maintaining ethical standards in your work.
@@drphilipaduThank you for your reply!
Very clear and easy to understand. Thanks.
The link for PowerPoint slides does'nt work fine
Thank you!! Here is a link to the document: drive.google.com/file/d/1VpNzdvQNm9JviKYqU_VtJVzGx4MaQUTy/view?usp=sharing
@@drphilipadu Thank you
Thank you Dr. Adu for your video. For the qualitative study part of an explanatory mixed method with a semi-structured interview, which analysis type is more appropriate, thematic analysis or content analysis?
You're welcome! If you plan to generate codes and themes based on the relevant information you extract from the data then thematic analysis is appropriate: see: ua-cam.com/video/LMaz9GsTL_Q/v-deo.html
Hi Sir! I just want to ask that I have a large data set up to 38 responses and the answer in the responses are very long as well. So what should I do to make the qualitative analysis easy.
1. Should I analyze all the responses for research question 1 at one time, then for RQ 2, then for 3?
2. How should I bring the results from chatgpt to my research paper, because at the end, it will be giving plagiarism or AI, and the paper may not get accepted?
Thank you for watching my video and for your questions.
My response to your first question:
I suggest you analyze all responses and connect them to their respective research question. In other words, with the research questions in mind, you go through the data and extract relevant information and connect them to their research questions.
My response to your second question
The first step is to view ChatGPT as a data analysis tool assisting to make sense of the data. Second, you determine the role is will be playing in your data analysis task. For example, us can use it to generate initial codes and them you review them - making sure they reflect the relevant excerpts from the data and address the research questions. Alternatively, you can use it to extract relevant information from the data. This process may help you to speed up the data analysis process.
If you plan not to copy the information generated from ChatGPT and paste it on your research paper, you won't have any issue with plagiarism.
To avoid plagiarism, there are some of the actions you could take:
1. use ChatGPT as a data analysis tool playing an assistant role
2. Make sure you review and refine the codes/themes suggested by ChatGPT
3. Be transparent - sharing and showing how you used ChatGPT to analyze your data
4. Cite source ChatGPT when needed (see: apastyle.apa.org/blog/how-to-cite-chatgpt)
Good day, Dr. Adu. Thank you for putting out valuable content. I am struggling to analyze qualitative secondary sources data and create themes and codes using Nvivo, Atlas. ia..etc. please hope you would not mind sharing available link or guides for all work done related to my inquiry.
You're welcome! The following links would help: ua-cam.com/video/NwY_XXrpdl4/v-deo.html
ua-cam.com/video/G7PjAzIf-RQ/v-deo.html
Thanks so much this is quite helpful. God bless
You're welcome! God bless you too!!
HI Kindly guide me how to use NVIVO to generate codes themes and analyze data and then get report
This playlist would help ua-cam.com/play/PLSOjmgd2Fg_looShtg4Qfrq-xFHg4nXhN.html&si=wfjnWCP28dCirLd2
Excellent, encouraging video! Is there a way or recommended 'place' to store a large, growing number of lengthy transcripts (100+) for ongoing analysis using Chat GPT or do you recommend another AI tool for that?
Thank you Jana! You could create custom GPT to have analyze your data (see: ua-cam.com/video/SiaZOPjpHS4/v-deo.html). However, I'm not sure you could upload all the 100 transcripts at a time.
G
This is helpful to me. Thank you. How do i generate report?
You're welcome! It depends on the kind of report you want to generate. I would try and ask it to see what you'll get. I think it can provide you the results in a Powerpoint slides form. Alternatively, you can use Gamma (gamma.app/) to generate a report.
Hi there, can you please help me, do you have a literature method for this? so I can reference this in my paper?
This resource would help: www.perplexity.ai/search/ChatGPT-for-qualitative-d8rxmzkDTCeoBQwGfiLvaQ
Hello, I like this explanation very much. However, I would like to ask you for more clarification about summarizing the data by ChatGPT. For example, I have 30 transcripts, each of which is 10 pages long. This means that I have 300 pages of transcripts from different research participants. How can I summarize them by using ChatGPT? Is it possible that I can summarize the first transcript and generate initial codes by using ChatGPT and keep these initial codes generated, then do the same thing with the rest of the transcripts. Finally, I can put the initial code from each transcript in the chat GPT to generate themes?
Thank you! It allows you to upload 10 documents at a time. So, yes, you can upload the first 10 transcripts and develop initial codes and do the same for the second and third 10 transcripts. You can present all the initial codes to ChatGPT and ask it to categorize them to generate themes.
hi i have transcripts AI generated from the interview session. can chatgpt helps to make the transcript clean?
It can but it is not guaranteed that it will solely clean the data without making changes to the original content. Also, it depends on the kind of cleaning you want it to do. Is it about indicate what the interviewer said and what the participants shared?
Awesome, It help me to understand about coding.
Thank you!!
Very clear and easy to understand. Thanks.
You're welcome! I'm glad it is helpful!
That was helpful. But can we use this trick in data analysis(synthesis) of systematic qualitative review such as verbatim quotes of various participants converting to codes and eventually generating themes?
Yes, it could be done!!
Wooow this video help me a lot! thank you so much.
You’re welcome!! I’m glad that it was helpful
Love it ♥
Thank you :)
You have just save my life thank you!!
You're welcome! I'm so glad the video was helpful!! Thank you!!
thank you this is really helpful. I am working on data with sample size 85 and close ended questions. Any idea how can i do that please
You're welcome! It sounds like you collected quantitative data. If this is the case, I recommend you use Statistical software like SPSS. Alternatively, you can use a free software:
JASP Software: lnkd.in/eme_2V8
JASP free Textbook: lnkd.in/eyKJPBK
Video: lnkd.in/e9t3_D5
JASP 'How-To' Resources: lnkd.in/eGjJn5x
Jamovi Software: lnkd.in/eYnf248
Free Textbook: lnkd.in/ejxMxVZ
Video: lnkd.in/eFfkfbT
Any chance we can have the slides? The forms in the link are not sending slides to email
Here is the link: drive.google.com/file/d/1VpNzdvQNm9JviKYqU_VtJVzGx4MaQUTy/view?usp=drive_link
THANKS SO MUCH DR!
Thank you for these helpful tips. I would like to see how you analyze multiple transcripts that are all longer than 3000 words.
You're welcome!! You can first them in chunks and submit each chunk. After that you can ask ChatGPT to review all the chunks and then ask it to analyze the data.
@@drphilipadu What command will I use for the separate chunks?
@@bernadettekina3408 You can start by stating, "I'm going to provide you my qualitative data in chunks for you to review. When I complete providing you all the chunks of data, I'll let you know what to do next. Should I go ahead?"
This is truly amazing! Thank you for all the insightful knowledge! Do you have a method of literature here that I can use to cite this? I would like to use this for my paper
Thank you! This resource would help: www.perplexity.ai/search/ChatGPT-for-qualitative-d8rxmzkDTCeoBQwGfiLvaQ
Very useful. Thank you very much Philip Adu, Ph.D😀
You're welcome!!
What a guy man so so useful. Thank you
You're welcome!!
Thank you for this tutorial. It was super helpful.
You're welcome!! I'm glad it was helpful to you!
Very useful. Thank you so much!
You're welcome!!
Great video Dr. Now I have one question, do you have a video on the ethics of using chatgpt for on research?. I was thinking if not there must be otherwise these students will missuse this tool.
This video would be helpful: ua-cam.com/video/stB9oeKvw2Y/v-deo.html
This is amazing!!! thank you so much!
You’re welcome!!
Hi Dr Adu, is there any way to get Chatgpt to write a clear discussion and conclusion from my thematic analysis?
There is a way but an ethical way of using it for discussion and conclusion is to ask it to provide you ideas on how to write them based on your findings.
Now how do I present my data in my chapter after coming up with the themes?
This video may help: ua-cam.com/video/lsxcpY4ghxE/v-deo.html
Thank you very much. This is very helpful.
You’re welcome Benjamin!
Thank you so much 💕
You're welcome!!
Sir can we use for qualotative data anlysis for systematic literature review? i have dowloaded R software for qualitative data analysis, but RQDA package is not available on recent R software.. idk what to do.. need your help!
Yes, you can use qualitative analysis techniques for systematic review. I’m sorry I’m not familiar with R software.
That is great piece of teaching
Thank you for the kind words!!
wow, that was just what i needed, thanks :)
You’re welcome!!
Quick question, Dr Adu, if using a large in-depth interview of about 50 pages of transcript, how many words can you summarize it to? For this example, you used 1000 words. Is 5000 words ok or too long? Thank you
500-1000 words will be okay. The essence of summarizing is to gain an overall understanding of the data. Then main focus of the analysis would be extracting quotations, and developing codes and themes.
simply and easy to follow👌
Thank you!!
Very helpful video.
Thank you!!
Thank you so much Doctor.
You're welcome!!
Sir, what is the recommended sample size for qualitative data analysis?
It depends on a couple of factors but having 8-15 would be great. Resource: eprints.ncrm.ac.uk/id/eprint/2273/4/how_many_interviews.pdf
Thanks so much sir@@drphilipadu
thank you Dr Adu your explanation surely worth it
My pleasure!!
thank you very much Doctor, it was so helpful for me
You're welcome! I'm glad it was helpful to you!
You are awesome, thanks for your help!
You're welcome @daniazavala397!!
helo! can i also do this through claude instead of chatgpt?
Yes, and you may get a better results.
Thank you so much, Dr. Adu. This is really needed at this moment of my thesis.I see that these participants’ interview are short. I would like to ask what if you have a long interview session from about 8 participants.
Do you advise I summarize each transcript separately, then resummarize all into 1000 words?
Then summarize the new summarized transcripts based on the research questions.
Then, generate the codes
And then the themes?
Thank you for your prompt response.
You're welcome!! You have two options:
Option #1: You can first submit the data in chunks by informing it that you have a large qualitative data and you will be submitting the data in chunks. You then indicate 'Chunk One' and you paste the first batch of data and press enter. You then do the 2nd, 3rd, 4th ... until you complete submitting all the data. You then ask it to summarize the data, summarize it based on the research question, develop codes and generate themes.
Option #2: Working on doing the summary and developing codes and themes for each of the participant's transcripts. You can compile the output and ask it to generate overall summary, codes and themes based on the outputs ChatGPT generated from each of the transcripts.
Alternatively, you can use Claude.ai, see: ua-cam.com/users/shortsNKDcBe8Pw8A
@@drphilipaduThanks a lot for your prompt response. I will explore both options. I just remembered that I can use linkreader to upload the transcript. I will explore all the options.
I really appreciate you, Dr. Adu. Thanks, and God bless you
@@damilolafasinu You’re welcome 😊
After developing the themes from chatgpt, should we write the details in our work or only table form presentation is enough as you showed in video?
Thank you
@@asmashahzadi3803 To maintain transparency, I suggest you show how you arrived at the final output. This video would help: ua-cam.com/video/Gm3okfqvgNs/v-deo.html
Very much helpful.
Thank you!!
absolutely helpful! Thank you so much
You're welcome! I'm glad it is helpful!!
hello dr adu, thanks a lot for this video. I have a query. Im doing expert interviews of 3 types of participants. i have totally 15 participants, which means 5 in each category. i find out the perspectives of these categories of experts and To find out agreements or disagreements between them. each in-depth interview is around 2500 word transcripts. how can I use chat GPT to provide a summary after analysing the 5 sets of 2500 words.
You're welcome! ChatGPT has a word limit when it comes to processing inputs/prompts. It can process a request with a word limit of about 2500 words. In your case, I suggest you first request for a summary for each of the transcripts and then compile the summaries and ask it to provide an overall summary based on the 5 summaries.
Hi. I'm conducting a thematic analysis on secondary sources, peer reviewed journal articles. How can I do that using Chatgpt. I have Chatgpt plus. I need your help
Since you have Chat plus, you can use my custom GPT (chat.openai.com/g/g-v8U2B4btf-qualicoptergpt).
Also, you can use the following initial Prompts:
Prompt #1:
Can you give me the six steps of Braun and Clarke's thematic analysis?
After it has generated the steps, you then attached your data and use the following prompt:
Prompt #2:
Can you use the 6 steps to analyze the attached data addressing the following research question: [you can state your research question here]
thank you very much G-D bless@@drphilipadu
Hi, can you please share an alternative link to the ppt slides?
Here is the link: drive.google.com/file/d/1VpNzdvQNm9JviKYqU_VtJVzGx4MaQUTy/view?usp=sharing
Greetings! Great video. I needed an alternative because my institution does not want to sponsor NVivo for me. I am on chapter 3 of my dissertation. How would you address and/or justify the usage of chat gpt in my methods section? Should I document every step?
You can say, you plan to analyze your data manual using Microsoft Word document (see: ua-cam.com/video/DSpts_DszEY/v-deo.html). In addition, you plan to use ChatGPT to help you extract relevant information from the data, and generate codes and themes.
You should emphasize that the outputs gotten from ChatGPT will be critically reviewed - making sure they reflect data collected and can assist in addressing the research questions. The critically reviewed and refined ChatGPT outputs will then be used to inform the manual qualitative data coding process. To cite ChatGPT, please go to: apastyle.apa.org/blog/how-to-cite-chatgpt
Alternatively, you can use dedoose instead of conducting manual coding (see: dedoose.com/home/pricing)
Such a lifesaver! I will summarize with chatgpt and use dedoose for coding. Wish me luck!
Thank you for this great elaboration
You're welcome!!
Braaaaaaaavooooo!!!! THANK YOU SOOOOOOOOOOOO MCUH FOR THIS !!!!
You're so welcome :)
what if you use quirkos instead of Nvivo?
Yes, you can use quirkos!
Thanks a lot for this!
You’re welcome!!