Chapter 4. Does ethnic heterogeneity of clubs affect member dropout? 93 Table 4.3 Predictor coefficients exponentiated into probabilities Model 1 Model 2 Model 3 Model 4 Model 5 Individual-level variables Age -0.18% -0.18% -0.19% -0.19% -0.18% Gender, ref: male Female +48.63% +48.46% +48.24% +48.28% +48.46% Income, ref: low Middle income -16.91% -16.75% -16.81% -16.81% -16.78% High income -20.79% -20.60% -20.74% -20.73% -20.69% Ethnicity, ref: Dutch Migrant background +37.36% +37.17% +3.57% +3.45% -5.95% Club-level variables Club size +0.04% +0.04% +0.03% +0.03% Income share, ref: low % middle income -0.44% -0.43% -0.39% -0.37% % high income -0.69% -0.68% -0.63% -0.61% Ingroup share -0.39% -0.47% -0.63% Outgroup fractionalization +0.13% +0.16% Cross-level interactions Migrant*ingroup share +0.18% Migrant*outgroup fractionalization -0.04% Note: Adjusted for membership duration (number of playing seasons) and time period (per playing season). differ between members with Dutch and migrant background. The reason for this is that I expected migrant background to have a mitigating effect on both effects due to a constrained opportunity structure and higher rates of interethnic contact (hypothesis 6). The cross-level interaction terms show that both effects are indeed weaker for members with migrant backgrounds. For ingroup share, this effect is -0.45% (-0.63 + 0.18) instead of -0.63% and for outgroup fractionalization, the effect is 0.14% (0.18 - 0.04) instead of 0.18%. This confirms the last hypothesis of this chapter.
RkJQdWJsaXNoZXIy MjY0ODMw