University of KwaZulu-Natal, School of Mathematics, Statistics and Computer Science, Westville Campus, Durban, KwaZulu-Natal, South Africa.
PLoS One. 2019 Mar 20;14(3):e0213973. doi: 10.1371/journal.pone.0213973. eCollection 2019.
This paper uses regression adjustment and entropy matching to achieve an appropriate balance between the two groups. Having achieved this balance any difference in the overall performance between the two groups can then be attributed directly to the new language policy that has been implemented. Our results indicate, after matching, that a significant difference in performance between the two groups occurs with Black African Zulu home language speakers in particular not performing as well as students from the other race and language groups. The data came from the University of KwaZulu-Natal and the study has been approved by the University Ethics committee and Research committee.
Using a weighted mean of marks to measure performance, the ATT estimates that result indicate that students in the treated group would all perform significantly better had they been allowed instead to enrol in the non-treated group. Furthermore, Zulu home language speakers, who are not actually forced to take this course, are significantly underperforming whether they have chosen to take this course in Zulu or not. Surprisingly, their underperformance is worse in the treated group. Because one would expect them to be scoring a higher mark for Zulu in this treated group, forfeiting the chance to take another course in their chosen area of study is clearly affecting the type of mark they could be getting for their other courses had they chosen to remain in the non-treated group. With English being the medium of instruction at this university, should the university not also consider introducing a compulsory course in English for these Zulu home language speakers? Females are doing better than males in both groups. The effect being stronger in the treated group suggests that females appear to cope better with the learning of a new language. Significant college effects are also being observed, suggesting that this new language rule possibly needs to be adjusted for the college a student wants to study in.
本文采用回归调整和熵匹配来实现两组之间的适当平衡。在达到这种平衡后，两组之间的整体表现差异可以直接归因于实施的新语言政策。我们的结果表明，在匹配后，特别是对于祖鲁语为母语的黑人祖鲁族学生，两组之间的表现存在显著差异，他们的表现不如其他种族和语言群体的学生。数据来自夸祖鲁-纳塔尔大学，该研究已获得大学伦理委员会和研究委员会的批准。
使用加权平均分来衡量表现，ATT 估计结果表明，如果允许 treated 组的学生选择加入非 treated 组，他们的表现会显著提高。此外，即使没有被迫修读这门课程，祖鲁语为母语的学生的表现明显较差，无论他们是否选择以祖鲁语修读这门课程。令人惊讶的是，他们在 treated 组中的表现更差。因为人们会期望他们在这个 treated 组中祖鲁语的分数会更高，放弃在所选研究领域修读另一门课程的机会显然会影响他们在其他课程中可能获得的分数，假设他们选择留在非 treated 组。由于这所大学的教学语言是英语，大学是否也应该考虑为这些祖鲁语为母语的学生开设一门英语必修课？在这两组中，女性的表现都优于男性。在 treated 组中，这种效应更强，表明女性似乎更能适应新语言的学习。还观察到显著的学院效应，这表明这一新的语言规则可能需要根据学生希望就读的学院进行调整。
