Foreign-Born Teaching Assistants and Student Achievement: An Ordered Probit Analysis
Marvasti, Akbar, American Economist
Public debates about the impact of the foreign-born on various aspects of the life of native Americans has intensified in recent years because of increases in the number of foreign-born persons in the U.S. as well as changes in the composition of immigrants in terms of the country of origin. In the U.S., Canada, and several Western European countries, public policy has focused on the labor market, housing, and education. Education has been a sensitive area where concerns have been expressed with respect to the impact of the increasing number of immigrants on educational attainments or opportunities for the natives in high schools and colleges (crowding out), lowered standards in educational institutions because of language barriers faced by foreign born students, and the adverse effect of foreign-born teaching assistants (FBTA) on the scholastic achievement of native students. In a recent article, Borjas (2000) has investigated the impact of FBTA on the scholastic achievements of American undergraduate students. He uses OLS regression, which is inappropriate for ordered dependent variables, to confirm his hypothesis that FBTA have an adverse effect on the scholastic achievements of American undergraduates. Here, the maximum likelihood method is applied to the same data to arrive at an unbiased and efficient estimation of the grade function.
In recent years, a few other studies have dealt with the issue of FBTA performance in the classroom. For example, Watts and Lynch (1989) use a survey of 2800 economics students who were taught entirely by teaching assistants (TA) to estimate two functions--one uses post-test scores in a nationally normalized set of exams as the dependent variable, and the other uses student course grades. The authors properly employ statistical techniques of OLS and ordered probit, respectively, to test the hypothesis that FBTA have adverse effects on students' learning. While OLS results agree with the tested hypothesis, the coefficient of FBTA is not statistically significant in the ordered probit models. The authors also examine the drop rates in classes. The calculated dropout rates turn out to be very close between classes with FBTA and those with native TA, and no evidence in support of the "voting with their feet" argument is found.
In another recent study, Fleisher et al. (2002) investigate the effect of FBTA on the undergraduate students at Ohio State University. Contrary to the common expectation, their results show a significant positive effect for the FBTA in some cases. In this case, all TA, native as well as foreign-born, were subject to a teacher-training program during the study period at this school. Fleisher et al. also find the drop rate is actually lower for the FBTA than for the native TA. However, FBTA received lower ratings in the survey of student opinion. The authors attribute this apparent contradiction to more substantive content input by the FBTA, but perhaps a less desirable class environment due to a cultural gap. Fleisher et al. also examine the dropout patterns of the students. The authors argue that, although initial student perception of the FBTA is worse than perception of native TA, once the students learn about the teaching effectiveness of the FBTA, the drop rates fall in the FBTA classes and a higher re-enrollment is observed.
Some researchers are skeptical of student opinion surveys, questioning the validity of the student opinion survey instrument as a measure of teaching effectiveness (Marsh, 1984; and Marsh and Roche, 1997). It is also conceivable that student opinion regarding FBTA is biased. In fact, Watts and Lynch (1989) suggest that some native students may use the FBTA's English skill as a scapegoat to justify their poor performance in the class, which can bias the estimated coefficient. Becker and Watts (1999) question the existence of a strong correlation between student grades as a measure of learning and their ratings of the instructor. …