124 4 Chadi and Finnigan, 2019: 876; Dotti Sani and Quaranta, 2022: 132). Stacking the data of these three countries allows generalization across all countries similar to France, Germany and The Netherlands, particularly towards contexts that have encountered homonationalist rhetoric in debates on Islam (for instance, US, Canada, Belgium, Denmark, Austria, Switzerland), but less so where homonationalism is not prevalent (for instance Poland and Hungary). To enhance generalizability, I used population weights based on level of education, region, urbanization, migration background, and gender (see appendix and code for more information on the weighting process). I analyze and present the data using marginal means because I compare different subgroups (flankers, moderates) and wish to avoid confusing readers with different reference categories (Leeper et al., 2020). Marginal means also allow me to present the data in a way that most closely resembles how I gathered it – by presenting proportions of voters who expect politicians to be supportive of homosexuality, with confidence intervals of 95 percent. The analyses of the relation with perceived similarity also include linear models. I prepared the data using R-package “tidyr” (Wickham, 2020), analyzed it using marginal means with R-package “cregg” (Leeper and Barnfield, 2020) and “miceadds” (Robitzsch et al., 2021) and visualized it with R-package “ggplot2” (Wickham et al., 2020).
RkJQdWJsaXNoZXIy MjY0ODMw