[Taxonomy of Public-Health research based on data-analysis complexity]

Public health requires operational classifications that organize studies according to the complexity of data analysis, not just by design or purpose. The objective of this paper was to construct a classification of levels of public health research based on the analytical treatment of data using Grounded Theory.

A theoretical-constructive qualitative research was conducted between January 1 and April 30, 2025, with iterative searching in PubMed, Scopus, Web of Science, SciELO, and the Virtual Health Library, theoretical sampling, and documentary analysis of thirty-five traceable sources. Open, axial, and selective coding was applied with constant comparison and coder consensus until theoretical saturation.

Six levels emerged, defined by the type of analysis: exploratory (without statistics), descriptive (univariate), relational (bivariate), explanatory (multivariate), predictive (modeling), and applicative (improvement methodologies). Each level was operationally delimited (purpose, unit of analysis, techniques, and example) for reproducible use in evaluation and planning.

The proposed classification is coherent, verifiable, and replicable; it places data analysis as a structuring criterion and facilitates alignment between objectives, design, and techniques, providing an operational framework for researchers, reviewers, and decision-makers in public health. Its empirical validation in applied contexts is recommended.