Resumen

PEREZ-RAVE, Jorge Iván; MAYA-RESTREPO, María Alejandra  y  GONZALEZ-ECHAVARRIA, Favián. Do Internet search patterns predict election results? An analysis of the 2022 presidential election in Colombia using a reproducible descriptive analytics framework. Dyna rev.fac.nac.minas [online]. 2023, vol.90, n.225, pp.64-70.  Epub 02-Mar-2024. ISSN 0012-7353.  https://doi.org/10.15446/dyna.v90n225.104189.

Recent works have highlighted the relevance of Google Trends data for predicting electoral processes. However, the potential for the generalization of this evidence is still limited and there is still no consensus on the most appropriate method for performing analytics with data from this tool for this purpose. This study examines whether such data are useful for predicting presidential election results, taking as a case study the recent Colombian presidential election for 2022-2026, using a proposed reproducible framework of descriptive analytics in R. A sample of 1,020 observations was considered. All the analysis scenarios (considering Google, YouTube and both) were correct regarding the candidate who was finally elected by the citizens, and the best estimate was obtained by using data from one day before the election (less than 2% distance from the true result). This work is a cost-effective alternative for predicting presidential election results.

Palabras clave : electoral analytics; Google Trends; analytics; descriptive analytics; presidential elections.

        · resumen en Español     · texto en Inglés     · Inglés ( pdf )