Panel data analysis in Political Science: Advantages and applications in electoral studies

Abstract

Political science has relied upon panel or longitudinal data incrementally in the last decades. In comparison with cross-section data, the inferences attainable through panel data are considerably more enlightening; however, its analysis should not be limited to the classical linear model of ordinary least-squares regression. Based on real electoral data and simulations, this research note demonstrates the greater robustness of the fixed and random effects models and the problems associated to the Arellano-Bond estimator, in the case of short panels, which are common in electoral studies.

Keywords

• panel, datos longitudinales, elecciones, simulación Monte Carlo.