The aim of this project is to develop a contextual understanding of polling errors and their triggers. Unlike most previous studies, we take a cross-election comparative perspective and put the theoretical focus on characteristics of the electoral contest which may encourage polling errors.