TLDR: This research introduces Q-Sat AI, a machine learning-based decision support system that provides objective, evidence-based recommendations for determining sample sizes in qualitative studies. By evaluating ten critical parameters like research scope and information power, the model, particularly the Decision Tree algorithm, accurately predicts optimal sample sizes, addressing the subjectivity of traditional ‘data saturation’ methods. A web-based application is available to assist researchers, reviewers, and advisors in standardizing sample size justification.
Qualitative research is invaluable for understanding human experiences, behaviors, and contexts in depth. Unlike quantitative studies that often focus on “what” and “how much,” qualitative methods delve into the “why” and “how,” providing rich, nuanced insights. However, a long-standing challenge in qualitative research has been determining the appropriate sample size. Traditionally, this has relied on the concept of “data saturation,” a subjective principle that can lead to inconsistencies and questions about methodological rigor.
The principle of data saturation suggests that researchers continue collecting data until no new information or themes emerge. While seemingly logical, this concept is often ambiguous and relies heavily on a researcher’s judgment. This subjectivity has led to a wide range of sample size recommendations across different qualitative designs, creating confusion and potentially compromising the credibility of studies.
Introducing Q-Sat AI: A Machine Learning Solution
To address this critical methodological gap, a new study introduces Q-Sat AI, a systematic model based on machine learning (ML) designed to bring objectivity and standardization to sample size determination in qualitative research. Developed by Hasan Tutar, Caner Erden, and Ümit Şentürk, this innovative approach aims to provide evidence-based guidance for qualitative researchers, journal reviewers, and thesis advisors.
The Q-Sat AI model utilizes an ensemble learning approach, drawing insights from a dataset derived from five fundamental qualitative research methods: Case Study, Grounded Theory, Phenomenology, Narrative Research, and Ethnographic Research. The researchers identified ten critical parameters that influence sample size, including research scope, information power, and researcher competence. These parameters are evaluated using an ordinal scale and serve as input features for the machine learning model.
How Q-Sat AI Works
After extensive data preprocessing and outlier removal, several machine learning algorithms were trained and compared. Algorithms like K-Nearest Neighbors (KNN), Gradient Boosting (GB), Random Forest (RF), XGBoost, and Decision Tree (DT) demonstrated high explanatory power, effectively modeling the complex, non-linear relationships involved in qualitative sampling decisions. The Decision Tree algorithm, in particular, showed the lowest Mean Absolute Error (MAE), indicating the most accurate predictions.
The model considers factors such as the depth of participants’ knowledge (information power), the diversity of the sample (homogeneity/heterogeneity), and the researcher’s ability to interpret data (researcher competence). For instance, if participants are experts in the research topic, the model might suggest a smaller sample size, reflecting the idea that high-quality information can compensate for a smaller quantity of participants. Conversely, a broader research scope or a highly diverse participant group would typically necessitate a larger sample.
Also Read:
- Boosting Random Forest Performance with Smart Feature Selection and Adaptive Tuning
- Intelligent AI System Enhances Medical Pre-Consultation with Proactive Inquiry
Key Insights and Practical Application
Feature importance analysis confirmed the vital roles of research design type and information power, providing quantitative validation for key theoretical assumptions in qualitative methodology. This means that the type of qualitative study (e.g., ethnography vs. phenomenology) significantly impacts the optimal sample size, and the depth of knowledge held by participants is a crucial factor.
The study concludes by proposing a conceptual framework for a web-based computational application, also named Q-Sat AI, designed to serve as a decision support system. This tool is already publicly accessible and operational, allowing users to input the ten key methodological parameters and receive an immediate, evidence-based sample size estimation. This system offers a transparent, quantifiable, and reproducible justification for sample size selection, moving beyond subjective claims of data saturation. You can explore the Q-Sat AI tool further by visiting its GitHub repository: Q-Sat AI GitHub Repository.
This model represents a significant step toward standardizing sample size justification, enhancing transparency, and strengthening the epistemological foundation of qualitative inquiry through evidence-based, systematic decision-making. Future work aims to expand the dataset, incorporate more qualitative designs, and utilize Natural Language Processing (NLP) to automate the scoring of metrics, further reducing subjectivity and enhancing the model’s generalizability.
