Analysing qualitative data with AI
Inductive and deductive approaches to try
This week, I’m facilitating a session on data analysis and realised that our content could be updated to include some approaches that use AI. The following is a technique that I was initially apprehensive about due to the sensitivity of research data and the way that in our ethics documents, we usually write things like ‘the research data will be stored on a password-protected file, and only myself and the research team will be able to access the data’. Of course, uploading to openAI is not keeping that data securely on your laptop. However, with the option to restrict data use on ChatGPT and modify future ethics documents to include AI considerations, it is now more appealing. Some of our students have already been using AI to analyse their data, so I thought I’d share this short video and some prompts to help students do this perhaps more consciously and ethically.
A key point is that if you are going to use this approach, please attempt to understand and analyse the data yourself prior to analysing it with any AI tools, as we know errors are common with these tools, and they cannot be relied on for accuracy. Also, understand the ethical concerns of using AI tools to analyse your data. Always check the output for accuracy. I’d probably use this technique more as an idea generator or as a check for further themes once you have completed your initial analysis.
Here is a link to the document that includes the prompt ideas and fictional data you can experiment with to fine-tune the prompts to suit your needs.
**Update August 2024: I have now created a GPT that will analyse the data for you, it was built on the prompt linked in the document above. If you’d like more control and would like to understand a bit more about the process use the prompt, if you’d like to just throw in the data and watch GenAI work it’s magic then use the GPT!**
Analyzing Qualitative Data: Two Approaches - Watch Video
As you can see in the video, I only have the 5-minute budget version of Loom, so the video is nice and short, but it cuts off before I finish talking about how I would have worked through different iterations of the thematic analysis. Here, you want to bring in your identified themes or research questions to ask if any of these themes can be identified - an iterative approach is important here to ensure you can find what you are looking for.
I’d like to see if anyone can modify and improve these prompts, so let me know in the comments if you have a go with this approach to using AI to help with qualitative data analysis.
Have a look at my previous guide on how to collaborate with ChatGPT.
… and, of course…
