Recent developments in the issue competition literature highlight that parties pay attention to how other parties discuss issues, rather than simply focusing on owning issues in which they are most competent. However, existing data remain limited in scope, restricted in the number of parties, focused on a single party type, confined to specific time periods, or failing to treat gender as a distinct issue. These limitations prevent a direct examination of how parties respond to their competitors in relation to gender issues and how the party system relates to gender issues. This paper introduces an original party-election level dataset measuring parties’ gender issue positions and emphasis across seventeen Western European countries from 1995 to 2023. Using the Comparative Manifesto Project corpus, I employ a supervised machine learning model based on the pre-trained transformer network BERT to classify sentences in party manifestos. Transformer models perform effectively on multilingual texts and are well-suited for classifying large volumes of political text. The classified sentences are then aggregated to produce party–election-level measures of gender issue position and emphasis.
