Avanços Metodológicos Emergentes para Análise de Painéis Longitudinais na Pesquisa em Marketing

Abstract The analysis of longitudinal and panel data has become essential for understanding the dynamics of consumer behavior and evaluating the temporal impact of marketing strategies. This article deepens the methodological discussion by integrating recent critiques of the traditional cross- lagged model, systematic comparisons between panel-based approaches for causal inference, and advancements in methods to address unmeasured confounding in longitudinal studies. Key models and estimators are discussed, including RI-CLPM (Random Intercept Cross-Lagged Panel Model), GCLM (Generalized Continuous-Lag Model), synthetic control methods, and proximal causal inference, highlighting their advantages, limitations, and practical applications. The article also explores how incorporating Artificial Intelligence and Machine Learning techniques enhances the analysis of large-scale temporal data, expanding the capacity for personalization, forecasting, and the measurement of digital brand equity. Finally, it proposes a research agenda that integrates statistical rigor, robust causal inference, and algorithmic innovation, aiming to guide both researchers and marketing professionals in making decisions based on dynamic and causal evidence. Keywords: Longitudinal panels, Causal inference, Artificial Intelligence, Longitudinal analysis, Proximal causal inference