24 Qualitative Analysis
24.1 Analyse and interpret qualitative data from open-ended survey responses or interviews.
In research, qualitative data plays a vital role in understanding the richness, depth, and complexity of human experiences and perspectives. It provides us with narrative, contextual, and explanatory insights that quantitative data often fails to capture. But analysing such data, especially in large quantities, can be challenging. ChatGPT can be an invaluable tool in helping researchers navigate this rich landscape of narrative data, from open-ended survey responses to detailed interview transcripts.
In an AI in Higher Education Masterclass, I asked participants the following three questions:
If AI could flawlessly grade student work and provide detailed feedback, eliminating the need for professors’ involvement in assessment, would you be comfortable with this change?
If AI could instantly evaluate academic writing, should we replace the traditional peer review system with AI-driven evaluations?
If AI becomes so advanced that it can outperform professors in teaching, would you support replacing them with AI? Why or why not?
How would you go about analysing these open-ended responses?
For smaller datasets, you might consider directly using ChatGPT to summarise the responses. Just paste everyone’s answers into ChatGPT, ask it to summarise the results, and highlight any interesting patterns. This approach can be efficient and straightforward, providing a quick snapshot of the overall sentiment and key themes.
However, if you have a larger dataset or are aiming for a more systematic and reproducible approach, you can have ChatGPT assist you in creating an R Markdown script for analysis. This would allow for a more structured approach, ensuring consistency in the analysis process. Simply inform ChatGPT about your survey structure, the output you have, and the insights you’re interested in. ChatGPT can then guide you step-by-step, writing a script that not only performs the analysis but also explains each step, ensuring transparency and reproducibility in your research process.
In addition to thematic analysis, ChatGPT can also help you perform sentiment analysis on your qualitative data, which can provide useful insights into the general emotional tone of the responses. Given the prompts in the AI workshop, understanding the sentiment behind the responses could reveal how comfortable or uneasy participants feel about the prospect of AI taking over roles traditionally performed by humans.
ChatGPT presents a powerful tool for qualitative data analysis. It can adapt to your needs, whether you require quick summarisation or a more systematic approach. As we navigate the narrative landscapes of our research, ChatGPT can act as our guide, aiding us in the extraction and interpretation of meaningful insights from our data.
24.1.1 Prompt
I'm going to ask about 100 people tomorrow the following questions:
1. If AI could flawlessly grade student work and provide detailed feedback, eliminating the need for professors' involvement in assessment, would you be comfortable with this change?
2. If AI could instantly evaluate academic writing, should we replace the traditional peer review system with AI-driven evaluations?
3. If AI becomes so advanced that it can outperform professors in teaching, would you support replacing them with AI? Why or why not?
Here's what the results will look like (it will be a comma-delimited file):
--------------------------------------
PASTE A SAMPLE CSV FILE FROM QUALTRICS
--------------------------------------
What sort of questions can I ask about these results (assuming there will be data from another 97 people or so)?
----------------
CHATGPT RESPONSE
----------------
Great! Can you write code in R Markdown that will input this csv file and answer these questions and visualise the results?
