![Sample our Education journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5932/TOC-Subject-Sample-2021-Education.jpg"})
Abstract
Levels and Dronkers (Citation2006) showed that educational achievement differs between immigrant students from different regions of origin (Latin America, Northern Africa, and Western Asia). This follow-up paper establishes whether these differences in educational achievement between immigrant students from different regions of origin can be explained by school segregation, whether along ethnic or socioeconomic lines. Ethnic and socioeconomic school segregation have a negative influence on the scholastic achievement of all students, although the impact of socioeconomic school segregation is greater than that of ethnic school segregation. Ethnic school segregation affects the scholastic outcomes of native and immigrant students from some regions of origin more than those of immigrant students from other regions. The analysis shows that neither ethnic, nor socioeconomic, school segregation explains the lower mathematics achievement of immigrant students from Latin America, Northern Africa, and Western Asia.
Notes
1. An earlier version of this paper was presented at the spring meeting 2006 of the ISA Research Committee on Social Stratification and Mobility Intergenerational transmissions: Cultural, economic or social resources?, May 11th – 14th 2006, Nijmegen, The Netherlands. An earlier Dutch version was published in 2005 (Dronkers & Levels, Citation2005).
2. For instance, only the country of birth of the respondent is measured in the first wave of the European Social Survey (n.d.). This means that only the success of the first generation of immigrants can be estimated but not that of the far more important second generation (Dronkers & Wanner, Citation2006). Only the second wave of the ESS also contains the countries of birth of the parents.
3. Marks (Citation2005) was forced to analyze the role of education for immigrants without knowing their country of origin, because this information was not available in the PISA Citation2000 data. Although he found interesting differences between the educational achievement of immigrant students in various countries of destination, his analysis of the immigrant effects is unsatisfactory because he cannot control for the quite different effects of their countries of origin.
4. This preference is based on the observation that students in schools with better social composition attain higher educational outcomes.
5. The term “Matthew effect” refers to a text from the bible and is often used in sociology to indicate the phenomenon that the possession of resources tends to lead to further accumulation over time, while a lack of resources tends to lead to an even larger shortage in resources over time.
7. Students born outside the test country with parents both born in the test country are not coded as immigrant students.
8. The question for the students about the type of course (usually academic versus vocational) is quite imperfect.
9. We checked the consequences of deleting the migrants with a missing region (the most important group with irreparable missing values) for the differences between the correlations in the original file and the final file. These differences were not substantial enough for the research question of this article.
10. The −2*log-likelihood is 665816.
11. Another reason for this difference is that countries of destination might have immigrants from different countries within the same region.
12. The random effect of the first generation has a significant variance at the country level of 194 with a standard deviation of 87 in Model 2 with random effects for generations. This variance is significantly negative in relation to the intercept of the equation.
13. The random effect of the second generation has a nonsignificant variance at the country level of 93 with a standard deviation of 50 in Model 2 with random effects for generations. This variance is negative in relation to the intercept of the equation, but it is just not significant.
14. The first generation has a significant variance at the country level of 212 (with 94 as standard deviation) and the second generation has a significant variance at the country level of 117 (with 60 as standard deviation).
15. Levels and Dronkers (Citation2006) found only a few significant interactions between country of destination and individual immigrant characteristics, other than generation. For that reason we do not include them in these equations.
16. The random effects have a variance of 148 and 127, respectively, with a standard deviation of 68 and 60. The variance of first generation still correlates negatively with the intercept, but the variance of school average parental educational level does not correlate significantly with the intercept.
17. Variance of random effects at the country level of .083 with .045 as standard deviation.
18. In an analysis, not shown in this paper due to lack of space, we added a number of indicators of school climate to Model 7. This did not produce different results for the effects of ethnic and socioeconomic school segregation.
19. The effects of first-generation immigrant students and school averages of parental educational level still vary significantly between countries. The variance of the first-generation variable still correlates negatively with the intercept.