Heo, Nayoung (2017-04). A Comparison of Educational Outcomes between Multicultural Students and Non-Multicultural Students in South Korea: School-Level Analyses of 6th Grade Students. Doctoral Dissertation.
Thesis
I laid out the results of the multilevel ordered logit regression analyses of the school survey data that I collected in 2015 to compare the educational outcomes between the multicultural students (born to a foreign-born parent and a Korean-born parent) and the non-multicultural students (born to two Korean-born parents) in the same school settings. In detail, I examined the effect of being multicultural on the English language achievements of the 793 6th grade students in the 20 different elementary schools in Gyeongbuk Province of South Korea. The main hypothesis is that being multicultural leads to poorer English achievements. Most of the earlier Korean research based on small samples or interviews of multicultural students only suggested that the academic achievements of multicultural students on average were inferior to those of non-multicultural students. To test this, I predicted the log odds of English achievements of the students in the same classrooms using multilevel models with the dummy variable indicating whether the student is multicultural (yes=1; no=0), controlling for the other level-1 and level-2 variables. Next, I examined the interaction effects, i.e., the effect of a school-level variable on the slope of being multicultural at a time. The results are as follows: The "Multicultural" variable alone has a significant negative effect on the log odds of English achievements even after controlling for another independent variable at a time except the mother's educational attainment level. However, the negative effect of being multicultural becomes insignificant once the mother's education-attainment level and mother's information-sharing intensity and a school-level variable are additionally controlled. The three interactions that were hypothesized to have effects on the slope of the "Multicultural" variable turned out to be insignificant. The control variables, especially the mother's education level, mother's information-sharing intensity, school mean SES, and whether the school is multicultural-education-focused had significant effects in the full models. The results led me to conclude that being multicultural is not an automatic disadvantage as discussed in the previous literature.
ETD Chair
Poston, Dudley George T. and Gladys H. Abell Professor of Liberal Arts
