2.4 Data & Methods
2.5.2 Multivariate Analyses
The multivariate analyses were carried out in two steps. First, the differences between natives and migrants with respect to their participation in tertiary and vocational educa- tion were compared between the federal states while holding demographic characteristics like sex and age constant. Second, the variations in the distribution of general secondary education described in section 2.5.1 also were held constant (compare TableA1).
Figure 2.2: Difference between native Germans and first generation migrants in the
probability to participate in vocational and tertiary education (percentage points)
Source: Microcensus 2008–2010; own analysis. Note: Representation of the results in TableA1
Figure 2.2 and 2.3 give an overview of the results of the two models. They show differences between natives and migrants in the participation in tertiary and vocational education in percentage points for each federal state.
Model 1 shows, as expected, that having a migrant background has a negative effect on the probability of participating in the tertiary and vocational education system in every federal state. This negative effect is statistically significant (p ≤ 0.05), except for peo- ple with a second generation migrant background living in Germany’s eastern states or for people with a first generation migrant background living in Schleswig-Holstein. The effect for people with a first generation migrant background is greater than for those with a second generation migrant background in every federal state, with the exception of Schleswig-Holstein. Furthermore, we can see that the level of the effect of having a migrant background on the probability to participate in tertiary and vocational education varies across federal states. In Bavaria, for example, the probability of participating in the tertiary and vocational education system is 30 percentage points lower for the first generation or 18 percentage points lower for the second generation than for the same age group without a migrant background. By contrast, in Schleswig-Holstein, this difference for the first generation amounts to only 15 percentage points, and in Germany’s eastern states to just four percentage points for the second generation. These results support the first hypothesis that the level of ethnic inequality varies between the federal states.
Figure 2.3: Difference between native Germans and second generation migrants in the probability to participate in vocational and tertiary education (percentage points)
Source: Microcensus 2008–2010; own analysis. Note: Representation of the results in TableA1
After holding the level of the secondary school degree constant in addition to the other covariates in Model 2, I found substantial changes compared to Model 1 regarding the size of the effect of having a migrant background. The level of inequality between natives and migrants decreases to a large amount, but it still persists in some federal states. The effect is statistically not significant for first generation migrants in Rhineland Palatinate, Schleswig-Holstein, Hesse, and Hamburg, and for second generation migrants in Hamburg, Berlin, Lower Saxony, and Germany’s eastern states. Thereby, after holding the general secondary school degree constant, the inequality between natives and migrants in Bavaria showed the largest decrease. This result is not surprising, since Bavaria not only is the federal state with the largest number of people with a migrant background having only a low level of general secondary school education, but also because it has the largest gap between natives and migrants with respect to the level of general secondary school education. In Model 2, the variation across federal states with respect to the level of inequality between natives and migrants remains but decreases compared to Model 1, and becomes statistically insignificant except for the difference between Schleswig-Holstein and Lower Saxony for second generation migrants.6 After holding the school leaving certificate constant, the inequality between native Germans and first generation migrants is still the
6
In order to test the statistical significance of the variation between the federal states I calculated a model including all federal states and interaction effects between having a migrant background and the federal
highest in Berlin and Germany’s eastern states, and the lowest in Hamburg and Schleswig- Holstein. The variation in the level of inequality between natives and migrants in all other federal states in the middle position is now very small. After holding the general secondary school degree constant, the inequality between natives and second generation migrants is still the highest in Schleswig-Holstein and the lowest in Lower Saxony and Germany’s eastern states.
The decrease in the degree of inequality, after holding constant the level of secondary school education, shows, that the rates of individuals with different levels of general sec- ondary education within a federal state can, to a large extent, explain the differences in the degree of inequality between the federal states.7
However, the differing level of general secondary school education between federal states can only partly explain the variations in the gap between natives and migrants between federal states because the level of general secondary school education also varies between the federal states with respect to their impact on the probability to participate in tertiary and vocational education. Figure 2.4, using the comparative perspective of Model 2, shows the effect of general secondary school education on the probability to participate in tertiary and vocational education. Having a low general secondary school education has, as expected, a negative effect on the probability to participate in tertiary and vocational education, and having a high general secondary school education increases the probability. Especially, the positive effect of Abitur varies statistically significant between the federal states. Whereas having Abitur compared to having Mittlere Reife increases the probability to participate in tertiary and vocational education by 29 percentage points in Berlin, this difference is only 8 percentage points in Baden-Wuerttemberg. Thus, the second hypothesis that the impact of the level of secondary school education varies between federal states can be approved.