![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
This paper explores the hypothesis that the electoral participation of Muslims varies according to two main types of opportunities, namely, institutional and discursive opportunities (DOs), characterising their country of residence. More specifically, we assess the impact of institutional opportunities (IOs) in terms of civic and cultural dimensions. We then add up the analysis of DOs in their quality of legitimating public debates over Muslims. We conduct our study by relating survey data to standardised contextual indicators of institutional and DOs in four European countries. The results show that both institutional and DOs have an impact, under control of a number of individual characteristics. However, while the civic and cultural dimensions of IOs have an equally important effect, we find that DOs are less crucial for Muslims’ electoral participation. A number of broader implications of these findings are discussed.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. Respondents were left the choice between the language of the host country and their native language.
2. The focus on elections allows us to take into account the same comparable voting events in each country, with no consideration of specific institutional features, such as, for example, the presence of direct democracy (voting on popular initiatives and referenda) in Switzerland.
3. The following newspapers have been used as a source for the coding: De Volkskrant, Trouw, NRC Handelsblad, De Telegraaf, and Het Parool in the Netherlands; Neue Zürcher Zeitung, Blick, Tagesanzeiger, Le Matin, and Le Temps in Switzerland; Bild, Süddeutsche Zeitung, Frankfurter Allgemeine Zeitung, Welt, and Tagesspiegel in Germany; Daily Mail, Daily Mirror, The Guardian, The Sun, and The Times in Britain; Het Laatste Nieuws, Le Soir, Gazet Van Antwerpen, La Dernière Heure, and De Standaard in Belgium.
4. These tests have yielded a strong consistency both with regard to the selection of claims and their description. The Cronbach alpha for selection bias (computed on a sample of 15 articles) is .91. The Cronbach alpha for description bias (computed on a sample of 4 articles) is, respectively, .97, .98, .98, and .98, for an average of 0.98 (see Cinalli and Giugni Citation2013b for further methodological remarks).
5. Marginal effects and predicted probabilities are based on the regression models and have variables set at their means. Since we have separate samples for each country which have then be pooled in a same data set, we estimate robust standard errors, hence allowing to relax the assumption that error terms are identically distributed (homoskedasticity). Ideally, we would have estimated cluster-robust standard errors so as to also relax the assumption that error terms are independently distributed (serial correlation). However, with only four clusters (countries) the treatment might be worse than the illness and we therefore avoid that. As a result, we pay more attention to the magnitude of the effects than to the significance level, as the standard errors in our regression models might be underestimated, leading to too small confidence intervals.
6. These figures are quite high. As is known, questions about voting are subject to responses dictated by social responsibility as people tend to overestimate their own participation so as to covey the image of good citizens. In our case, however, this is not a problem a far as we assume that this effect is constant across respondents.
7. It should be stressed that the relationship between associational membership and voting is subject to a strong risk of reverse causality. One is therefore only observing a correlation between these two variables rather than causality. In our case, however, this is not so problematic as we are only using associational membership as a control variable.
8. For the reasons mentioned earlier, beyond statistical significance, we are especially interested in the magnitude of the effects (as measured through odd ratios, marginal effects, and predicted probabilities).
9. The observed values here refer to the standardised opportunity scores.