Completeness in Twitter datasets A critical review on Twitter research methodologies

Abstract

This thesis addresses how the limitations of Twitter research influences the datasets, and this is discussed in the literature. Twitter is used for a range of topics, from tracking countries online to predicting epidemics. This paper researches how the limitations of using Twitter as a data source is discussed, and how completeness influences the research. By analyzing the methodology used in a study of the Dutch Twittersphere as a case study, several limitations are revealed. The current practices of conducting Twitter research are analyzed, and an overview of the current discussion about limitations is presented. This is followed by an introduction of completeness in Twitter research. Completeness is discussed on three levels, the literal definition, disclosure of methodology, and addressing limitations in research. The influence of these three levels on the replicability of research is discussed, and consequently a proposal for a model is presented.