دانلود رایگان مقاله اندازه کلاس و اثرات معلم در آموزش عالی

Abstract

Using student evaluations of their instructor as an outcome measure, we estimate and compare class size and teacher effects for higher education, with an emphasis on determining whether a comprehensive class size reduction policy that draws on the hiring of new teachers is likely to improve educational outcomes. We find that first time teachers perform significantly worse than their peers, and we find substantial class size effects. Hence higher education institutions face a tradeoff if they wish to increase admission. This tradeoff implies that as class size increases, at first the negative class size effect is smaller than that of introducing a first time teacher. However, beyond a certain level, the class size effect dominates and it is better to create a new class with a first time teacher.1

6. Conclusion

We study the tradeoff between smaller class sizes and teacher effects in the production function for higher education. While reducing class size has a positive effect, we argue that a comprehensive class size reduction policy has to be coupled with an expansion in the number of teachers. If this is the case, then the relative quality of marginal teachers is critical for the success of such a policy. In order to explore whether the quality gap between infra-marginal and marginal teachers dominates the class size effect, we estimate class size and first time teacher effects. Our findings show that a negative class effect does exist, and its’ impact when class size is reduced can be offset by the negative impact of a first time teacher. Hanushek and Rivkin (2010) survey various studies and argue that the effect of a ten student reduction in class size is between 0.10 and 0.30 standard deviations of the dependent variable. In comparison, we predict that said impact is roughly 0.10 standard deviations, a relatively small class effect. At the same time, Rockoff (2004) finds that a one standard deviation increase in teacher quality raises learning outcomes in 0.24 7 0.187 × 42.5/18.7 − 0.72 standard deviations. We find that a first time teacher lowers outcomes in roughly 0.41 standard deviations, and that first time teachers that are not invited to teach again lower them in 0.7 standard deviations. These results imply that it is Pareto optimal to break up a class, giving half the students to a first time teacher, for class sizes above 85 students. This is the case because the students who keep the same teacher gain a 0.425 standard deviation increase in satisfaction on average due to lower class size, while the students who are assigned to the new teacher have a 0.01 standard deviation increase in satisfaction. Since this is precisely the rule in place at FACEAPUC, one could judge this institution’s hiring policies to be on average getting this decision right. However, the effect of low quality first time teacher, defined as individuals who are not invited to teach again, is so large (−0.71 SDs) that it is Pareto optimal to break up a class only when class size is greater than 140 students. This suggests that the ability to detect low quality first time teachers is important, and while on average FACEAPUC seems to be doing this correctly, we are unable to determine whether the marginal hire is of high or low quality. These results highlight that finding methodologies to identify poor quality first time teachers seems like a relevant area of future research. Regarding the external validity of our results, to begin we find results that relate well to those in the literature. But many results are dependent on criteria used by administrators, characteristics of the student pool, of the teacher pool one has available, and on the characteristics of full time (or experienced) professors. The stopping rule we described is key in determining the margin at which we are measuring results. However, much of the discussion is relevant to any education institution and in particular to any higher education institution.