Gender codification of 412 national (general) and European Parliament elections in six European countries (2003-2021)

This dataset has been produced by applying the Manifesto Gender Analysis (MGA) codebook to 412 national (general) and European Parliament elections in the six countries participating in the UNTWIST project (Denmark, Germany, Hungary, Spain, Switzerland, and the UK) from 2003 to 2021.

 The Manifesto Gender Analysis coding procedure, developed by WP4 of the UNTWIST consortium, aims to analyse gender-related content in party manifestos. It relies on existing manifestos collected by MARPOR and EM projects from 2003-2021 in six national contexts: Denmark, Germany, Hungary, Spain, Switzerland, and the United Kingdom. The process involves splitting manifestos into quasi-sentences, coding them based on a scheme inspired by previous projects and feminist typology, and completing an expert survey. This method ensures comprehensive analysis and potential scalability through computational methods. 

The coding procedure involves a series of essential steps, divided in two main activities: the classification of manifestos’ quasi-sentences, and the completion of a survey dedicated to more general concepts which can be gauged by evaluating the content of the entire documents. In the latter case, then, the unit of measure of each coder consists in the manifesto document, whereas in the former the units of measure are quasi-sentences - i.e., arguments denoting a verbal expression of a political idea or issue. Coders are instructed to split sentences containing multiple arguments into quasi-sentences and ensure that each quasi-sentence encapsulates a single political idea or issue. 

Once the manifestos are split into said units, coders classify the arguments following the MGA coding scheme. The coding scheme (MGA) consists of 5 domains and 25 coding categories, covering various aspects of gender-related issues. Each domain includes an "other" category for relevant statements that do not fit precisely into the defined categories. Apart from coding categories related to specific themes, the coding scheme then includes additional dimensions. The classification process consists of seven steps: (1) assessing whether the quasi-sentence addresses gender-related issues, (2) defining both the domain and coding category, (3) determining whether the quasi-sentence refers to a specific recipient or group based on gender and/or sexual orientation, (4) evaluating intersectionality, (5) assigning the sentiment or connotation, (6) determining if it's related to a goal, issue, or policy, and (7) characterising the policy if applicable.

After completing the classification of the quasi-sentences in a given manifesto, coders fill in a survey for each manifesto document. The surveys provide information that cannot be directly inferred from the quasi-sentences, focusing on the gender ontology of a manifesto, the degree to which a manifesto entails a binary conception of sexes, the extent to which a manifesto promotes a patriarchal conception of the society, and how much a manifesto promotes heterosexuality as the only normal and socially acceptable sexual orientation of individuals. While the last four characteristics are gauged relying on quasi-interval measures (scales ranging from 0 to 10), the first one, gender ontology, consists in a categorical variable which distinguishes between manifestos with an essentialist ontology – gender and sex are the same and inseparable –, a constructivist ontology – biological sex is mediated through social construction of femininity and masculinity –, and other or undefined ontologies.