Analysing data
Data analysis in research uses relevant techniques to evaluate data to find meaning. For quantitative data, this can involve statistical techniques. For qualitative data, various techniques are used to identify patterns and trends.
Responsible use
- Do appraise output of genAI tools for potential biases.
- Do record all use of genAI tools for transparency of data analysis.
- Do be specific with prompts.
- Do include a plan to ensure privacy and anonymity if genAI is used in ethics applications and participant consent forms.
Cautions
- Don’t share confidential or sensitive data into genAI tools.
- Don’t upload copyrighted or licensed information into genAI tools.
Case examples
Click on the plus icons below to explore exemplar and cautionary hypothetical scenarios for analysing data.
Exemplar - Case example 1
Fahdi has collected a large set of quantitative data for a project he is working on. He now needs to analyse and interpret the data using statistical modelling. In order to do this, Fahdi is using some advanced Excel functionalities, including complex formulae. Fahdi uses genAI to provide him with step-by-step instructions on how to do some of the technical skills he is unsure of, and the genAI tool is also able to provide him with the formulae that he needs. Fahdi always keeps a record of his prompts and the output from the genAI, in case he needs to disclose this in the future.
Cautionary - Case example 2
Gina has completed her data collection for her project. She has a mixed methods approach, using both measurements and interviews, and needs to conduct both statistical analysis and textual analysis. She has been trained on using both Stata and R for statistical analysis, but has decided to use an open, commercial genAI tool for her text-based analysis as she feels this will give deeper meaning to her data and will be more time effective. She uploads her interview transcripts into the genAI and prompts it to identify the emergent themes.
Gina has breached Deakin’s guidelines to support the use of generative AI for researchers because she has breached her human subjects’ privacy by uploading their transcripts into the open genAI tool. She has also breached confidentiality and has failed to disclose her use of genAI.
Cautionary - Case example 3
Elyse’s field work involved interviewing early-career healthcare workers. Her field work generated over 50 hours of interview data, which she has tirelessly manually transcribed. She now has hundreds of pages of data to analyse. Elyse knows that she can’t get open, commercial genAI to do this work for her, but she is curious nonetheless to see what interpretations a genAI tool can offer. Elyse copy-pastes the full-text of one of her interview transcriptions into ChatGPT and asked the genAI to analyse the data. She quickly sees that the output is helpful in identifying a list of codes that she can use for her own analysis. Elyse proceeds to analyse her data in NVivo, using codes that she took from the ChatGPT output.
However, Elyse may be in breach of Deakin’s guidelines to support the use of generative AI for researchers; Elyse did not disclose to her participants when they consented to participating in the research that their data would be shared with a genAI tool, so she does not have permission to share this data with ChatGPT. Moreover Elyse failed to deidentify the information that she shared, and the interview transcripts may have included personal or sensitive information.
In future, Elyse should include any planned use of participant data with genAI in the consent form so that participants can make an informed decision about participating in her study. The consent form should also include specifics on how participant data will be deidentified and other steps she plans on taking to protect participant privacy and anonymity.