Wordscores is a popular computational text analysis method with numerous applications in communication research. Wordscores claims to scale documents on specified dimensions without requiring researchers to read or even understand the language of the input text. We investigate whether Wordscores delivers this claim by scaling the Euromanifestos of 117 political parties across 23 countries on 4 salient dimensions of political conflict. We assess validity by comparing the Wordscores estimates to expert surveys and other judgmental measures, and by examining the Wordscores's estimates ability to predict party membership in the European Parliament groups. We find that the Wordscores estimates correlate poorly with expert and judgmental measures of party positions, while the latter outperform Wordscores in the predictive validity test. We conclude that Wordscores does not live up to its original claim of a "quick and easy" language blind method, and urge researchers to demonstrate the validity of the method in their domain of interest before any empirical analysis. Computational text analysis is a rapidly growing research field with many applications in political communication research. From using Twitter data to identify the political preferences of citizens (
In research on national identity, scholars have developed a wide variety of approaches to measure and better understand this ubiquitous yet complex concept. To date, most of these approaches have been theory-driven, while only a very few have been data-driven. In this article, we aim to contribute to the latter by introducing a new data-driven method that has not been applied yet—that of non-linear principal component analysis (NLPCA). In contrast to other commonly used methods such as factor analysis, NLPCA distinguishes itself by making relatively few assumptions about the data and by allowing for greater flexibility when discovering underlying dimensions of such a complex concept as national identity. Drawing on the 2013 ISSP National Identity module, our analysis focuses on the case of Germany, also taking into account Western and Eastern Germany. Running an NLPCA, we find four dimensions that cover the multidimensionality of national identity: nationalistic attitudes, national pride and attachment, cosmopolitan beliefs, and membership criteria defining national belonging. This article contributes to the empirical debate on measuring national identity by suggesting a new and flexible methodological approach that better grasps the concept’s complexity and which we believe can move empirical research on national identity forward in and beyond Germany.
Voting advice applications (VAAs) are online tools providing voting advice to their users. This voting advice is based on the match between the answers of the user and the answers of several political parties to a common questionnaire on political attitudes. To visualize this match, VAAs use a wide array of visualisations, most popular of which are the twodimensional political maps. These maps show the position of both the political parties and the user in the political landscape, allowing the user to understand both their own position and their relation to the political parties. To construct these maps, VAAs require scales that represent the main underlying dimensions of the political space. This makes the correct construction of these scales important if the VAA aims to provide accurate and helpful voting advice. This paper presents three criteria that assess if a VAA achieves this aim. To illustrate their usefulness, these three criteria-unidimensionality, reliability and quality-are used to assess the scales in the cross-national EUVox VAA, a VAA designed for the European Parliament elections of 2014. Using techniques from Mokken scaling analysis and categorical principal component analysis to capture the metrics, I find that most scales show low unidimensionality and reliability. Moreover, even while designers canand sometimes do-use certain techniques to improve their scales, these improvements are rarely enough to overcome all of the problems regarding unidimensionality, reliability and quality. This leaves certain problems for the designers of VAAs and designers of similar type online surveys.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.