Confiabilidad en los estudios sobre fake news: datasets y métricas

Abstract

Many empirical studies on disinformation use fake news datasets and real news datasets to compare them. Such studies often appeal to reliability (or confidence, trustworthiness, credibility, accuracy) as a factor to characterize and evaluate the message, medium, source and perception of the audience. We selected the 50 most cited articles in WoS and Scopus (2017-2022) on fake news and reliability and investigated where the researchers extracted their samples. That is, we propose a meta-research aimed at evaluating how reliable such samples are. We extract and analyze the criteria, indices, metrics, resources and databases that researchers declare to justify their selection of both authentic and false news.