Generative AI: New Framework of Using Large Language Models for Analysing Descriptive Qualitative Data

Development of Large Language Models (LLMs) offers new opportunities to analyse large qualitative data sets. This study proposes a framework that uses LLMs from five popular families (GPT, Mistral, Claude, DeepSeek and Llama) to automate data extraction process. The framework is tested on real estate listings from Luxembourg webscraped in 2025 (Jan) to demonstrate: (1) framework stability – ability to generate similar responses when provided with the same data set, but also of following the same pattern when faced with new questions, (2) framework calibration mechanism – by adding code sheets that clarify how to interpret specific situations, (3) mechanism of validating model behaviour over larger data set – by calculating correlations between partial responses generated by the framework and comparing them with the pattern generated on the test set.