Métodos cuantitativos para estudiar a las élites: Aplicaciones prácticas, sesgos y potencialidades
Métodos cuantitativos para estudiar a las élites: Aplicaciones prácticas, sesgos y potencialidades
Authors
González-Bustamante, Bastián
Authors
Date
2022
Datos de publicación:
Revista Chilena de Derecho y Ciencia Política, Vol.13, Nº 2, 12-44, 2022
Keywords
Élites políticas - Élites económicas - Análisis de supervivencia - Emparejamiento - Chile
Collections
Abstract
Este trabajo metodológico discute aplicaciones prácticas, ventajas y desventajas del análisis histórico de eventos o de supervivencia en comparación con estrategias econométricas clásicas en ciencias sociales como las re-gresiones logísticas binarias. En consecuencia, se analiza cuándo el enfoque de supervivencia es conveniente en comparación con modelos logísticos y los riesgos de trabajar con datos observacionales y estimaciones potencialmente sesgadas. Para responder ambas preguntas se realizan dos demostraciones econométricas con técnicas de propensión y emparejamiento para analizar el efecto de la trayectoria y vínculos empresariales en el acceso y permanencia en posiciones políticas relevantes. Para esto se utiliza la Encuesta de Élites en Chile (1990-2010) y un conjunto de datos de ministros que desempeñaron su cargo entre 1990 y 2014. Los hallazgos evidencian que la trayectoria empresarial no tiene impacto significativo en el acceso a posiciones relevantes, sin embargo, si está asociada con un menor riesgo de ser removido del gabinete. Las demostraciones econométricas muestran que el análisis de supervivencia es útil para fenómenos que implican intervalos de tiempo, mientras que las regresiones logísticas resultan pertinentes para estudiar el acceso a posiciones siempre que el muestreo sea adecuado. También se evidencian diferencias significativas entre modelos sin ajustes o placebos y modelos con mejores especificaciones o estimados después de la aplicación del algoritmo de emparejamiento.
This methodological paper discusses the practical applications, ad-van¬tages and disadvantages of historical event or survival analysis, compared to clas¬sic econometric strategies in social sciences such as binary logistic regressions. Accordingly, it analyses when the survival approach is preferable to lo¬gistic models, and the risks of working with observational data and po-tentially biased estimates. To answer these two questions, we carried out two econometric demonstrations using propensity score matching techniques, in order to analyse the effect of entrepreneurial background and ties on access to and tenure in important political positions. The data were obtained from the Chilean Elite Survey (1990¬-2010) and a data set of ministers who held office between 1990 and 2014. The findings show that business background had no significant impact on access to important positions; however, it is asso-ciated with a lower risk of being removed from the cabinet. The econometric demonstrations show that survival analysis is helpful for phenomena involving time intervals, while logistic regressions are appropriate for studying access to positions (as long as the sampling is adequate). Significant differences are also evident between unfitted models, or placebos, and models with better specifications or fitted after applying the matching algorithm.
This methodological paper discusses the practical applications, ad-van¬tages and disadvantages of historical event or survival analysis, compared to clas¬sic econometric strategies in social sciences such as binary logistic regressions. Accordingly, it analyses when the survival approach is preferable to lo¬gistic models, and the risks of working with observational data and po-tentially biased estimates. To answer these two questions, we carried out two econometric demonstrations using propensity score matching techniques, in order to analyse the effect of entrepreneurial background and ties on access to and tenure in important political positions. The data were obtained from the Chilean Elite Survey (1990¬-2010) and a data set of ministers who held office between 1990 and 2014. The findings show that business background had no significant impact on access to important positions; however, it is asso-ciated with a lower risk of being removed from the cabinet. The econometric demonstrations show that survival analysis is helpful for phenomena involving time intervals, while logistic regressions are appropriate for studying access to positions (as long as the sampling is adequate). Significant differences are also evident between unfitted models, or placebos, and models with better specifications or fitted after applying the matching algorithm.