103 3 In Model 4 of Figure 3 we add together ethnic and religious background characteristics. The effect of ethnicity is severely diminished compared to Model 2, whereas the effect of religion remains roughly the same compared to Model 3. This suggests that religion partially mediates the effect of ethnicity. In other words, religion explains most of the effect of ethnicity we see in Model 2 of Figure 3, while ethnicity does not explain the effect of religion we see in Model 3 of Figure 3. In Figure 4 we analyse whether issue preferences explain voting for DENK. Besides the religion and ethnicity variables, we added eight economic and cultural issue positions of respondents. The only statistically significant issue positions are those on climate change, immigration and LGB rights. The more a respondent believes that immigrants are an asset to the Netherlands, the more likely they are to vote for DENK. The more a respondent believes in the need to combat climate change and supports same-sex adoption, the less likely they are to vote for DENK and vice versa. Further analyses, reported in Appendix 4, show that the impact of climate change views on DENK voting is driven mostly by respondents with a Moroccan background, while the impact of views on LGB rights are mostly driven by their Turkish counterparts. Appendix 4 also reveals that respondents with a Surinamese background and those without a migration background are most likely to reveal a positive relationship between immigration views and voting for DENK, while this is much less likely amongst respondents with a Turkish and Moroccan background. Most importantly, adding the eight issues to the model only reduced the coefficient of Muslim religion by 16%, from 3.57 to 3.01. Likewise, the coefficients of Turkish, Moroccan or Surinamese descent scores are only reduced a little, by 0.25, 0.35 and 0.10 points, respectively. The impact of a Moroccan and Surinamese background lost its significance after adding the variables, though only a small reduction in the effect size was necessary to make this happen. The impact of identifying as Hindu remained the same. Thus, there is little evidence to support the claim that issue positions substantially mediate the relationship between ethnic or religious background characteristics and voting for DENK.
RkJQdWJsaXNoZXIy MjY0ODMw