Vardaman 1

MaryAnne Vardaman

Economics 398

29 October 2012

Teacher’s Credentials or Value-added:

The True Determinant of Teacher Quality

Introduction

Many researchers have tried to predict teacher quality based on observable characteristics such as teacher’s highest education level, teaching experience, and possession of a teaching certificate. However, research has shown that this method of prediction is not very accurate. Therefore, researchers have proposed using a “value-added” method, similar to that used in production of goods, in order to measure the effect that the teacher has on her students throughout the year. This method can be a more accurate predictor of teacher quality, but it also has its issues. Therefore, I propose a longitudinal study of public school children and their teachers in order to compare some of the richer existing value-added models with “traditional” methods of measuring teacher quality that include some characteristics that are not always observable, like where the teacher attended college and the teacher’s SAT scores.

It makes sense that high-quality, effective teachers will have the greatest positive impacts on student achievement, but this has actually been a difficult point to prove. The difficulties in measuring teacher quality have arisen because not everyone agrees what it is that qualifies a teacher as “high-quality.” Obviously, there are some broad characteristics that people can generally agree that quality teachers possess including: knowledge of the subject they are teaching, possession of a “good education,” inclusion of all students, and an engaging demeanor. However, there is debate on what constitutes a “good” education for a teacher, how nurturing a teacher should be to her students, or whether the qualities that constitute a good teacher change as the child moves to higher grades.

Literature Review

Why does teacher quality matter? As Koedel and Betts demonstrate in their paper, “variation in teacher quality is an important contributor to student achievement.” [1] In the past, researchers like Hanushek have summed up previous research that shows teacher education, experience, and other readily observable measure of teacher qualifications are inaccurate predictors of teacher quality, and therefore, student achievement. [2] Naturally, researchers have been searching for an alternative way to measure teacher quality, so they have proposed using a value-added method. Ideally, this method would highlight the effective, high-quality teachers by showing the positive impacts they have on their students’ achievement.

There are some value-added models that are more commonly used than others, but these all have bias issues, due to the general simplicity of the models. Jesse Rothstein examines three of the most common models in greater depth, and he finds many potential sources of bias including: error in sampling, mean reversion (students who do poorly on year are more likely to do better the following year as they revert back to their mean score), nonrandom classroom assignment (students have the same peers in their classroom every year and teachers are assigned certain students deliberately), excessive dependence on sorting by observables (unobservable variables are virtually ignored in the method specifications), and omitted variable bias. [3] To help correct for these biases, it seems necessary to create more complex value-added models that can be tested before implementation.

Unfortunately, as work by Kirabo Jackson has shown, it is not currently possible to create a “one-size-fits-all” value-added model that can be applied across the board. He finds that value-added models of teacher performance for elementary school are not necessarily accurate for high school and that these models are more predictive of performance in certain subjects, but not in others. [4] These findings are supported by work from Koedel and Betts, who also demonstrate that value-added models must be limited in their application. [5]

“Track-level” bias is a relatively recent discovery in the investigation of value-added models. [6] This bias, discovered by Kirabo Jackson, arises when teachers are routinely assigned classes that lie on certain tracks or sequences of achievement, like honors Algebra. In these cases, the students exhibit higher test scores, not because of the teacher, but because of the class itself and the achievement track on which it is placed. He finds that there is greater variability within test scores of a certain subject within a certain grade for the school compared to achievement tracks within the school. Additionally, he finds that, “within the same course (but not the same track), some teachers consistently teach students who take more/less honors classes than the average student than other teachers.” [7] To account for this, Jackson creates a value-added model that controls for track-level achievement from the student perspective and the teacher perspective.

None of the researchers believe that value-added methods alone should be used to determine teacher effectiveness. As Hanushek and Rivkin illustrate, because of the great potential for bias, it is necessary to combine these models with other (subjective) methods of evaluating teacher effectiveness and teacher compensation. [8] Otherwise, the high-stakes situations can create incentives for teachers to help students cheat on tests or only teach to those students who have the best possibility of doing well on the test. Hanushek and Rivkin also cite random occurrences prior to and during the tests that can introduce measurement bias. [9] In these cases, the presence of something distracting (like construction) during the test can result in lower test scores, despite student and teacher preparation. It is impossible to control for such randomness, so the value-added model must not be weighted too heavily in teacher evaluations. Also, Kane and Staiger believe that if test scores are aggregated from several years, this would help reduce the bias these random events introduce into the data. [10] Basically, teachers would be evaluated on their average “performance,” rather than by their students’ achievements each year.

There is no easy answer on how to measure teacher effectiveness. A teacher’s impact on a student may not be immediately observable and there are various factors influencing a student’s achievement besides their teacher. Considering all of the factors and all of the potentials for bias, it seems impossible to create a value-added model that on its own will accurately reveal a teacher’s quality. Therefore, it is necessary to make the most complex, accurate models possible and then hope to explain the rest of the variation by subjective means.

Works Cited

Hanushek, Eric A. The Impact of Differential Expenditures on School Performance, 1989. JSTOR. Web. 29 Oct. 2012.

Hanushek, Eric A., and Steven G. Rivkin. "Generalizations about using Value-Added Measures of Teacher Quality." American Economic Review 100.2 (2010): 267-71. EconLit. Web. 23 Oct. 2012.

Jackson, C. Kirabo. (2012) "Teacher Quality at the High-School Level: The Importance of Accounting for Tracks" NBER Working Paper No. 17722. 25 Oct. 2012.

Kane, Thomas J., and Douglas O. Staiger. "The Promise and Pitfalls of using Imprecise School Accountability Measures." Journal of Economic Perspectives 16.4 (2002): 91-114. EconLit. Web. 24 Oct. 2012.

Koedel, Cory, and Julian R. Betts. "Does Student Sorting Invalidate Value-Added Models of Teacher Effectiveness? an Extended Analysis of the Rothstein Critique." Education Finance and Policy 6.1 (2011): 18-42. EconLit. Web. 23 Oct. 2012.

Koedel, Cory, and Julian Betts. Re-Examining the Role of Teacher Quality in the Educational Production Function., 2007. EconLit. Web. 5 Oct. 2012.

Rothstein, Jesse. "Teacher Quality in Educational Production: Tracking, Decay, and Student Achievement." Quarterly Journal of Economics 125.1 (2010): 175-214. EconLit. Web. 5 Oct. 2012

[1] Koedel, Cory, and Julian Betts. Re-Examining the Role of Teacher Quality in the Educational Production Function., 2007. EconLit. Web. 5 Oct. 2012.

[2] Hanushek, Eric A. The Impact of Differential Expenditures on School Performance, 1989. JSTOR. Web. 29 Oct. 2012.

[3] Rothstein, Jesse. "Teacher Quality in Educational Production: Tracking, Decay, and Student Achievement." Quarterly Journal of Economics 125.1 (2010): 175-214. EconLit. Web. 5 Oct. 2012

[4] Jackson, C. Kirabo. (2012) "Teacher Quality at the High-School Level: The Importance of Accounting for Tracks" NBER Working Paper No. 17722. 25 Oct. 2012.

[5] Koedel, Cory, and Julian R. Betts. "Does Student Sorting Invalidate Value-Added Models of Teacher Effectiveness? an Extended Analysis of the Rothstein Critique." Education Finance and Policy 6.1 (2011): 18-42. EconLit. Web. 23 Oct. 2012.

[6] Jackson, C. Kirabo et al

[7] Jackson, C. Kirabo, p.11

[8] Hanushek, Eric A., and Steven G. Rivkin. "Generalizations about using Value-Added Measures of Teacher Quality." American Economic Review 100.2 (2010): 267-71. EconLit. Web. 23 Oct. 2012.

[9] Hanushek and Rivkin et al

[10] Kane, Thomas J., and Douglas O. Staiger. "The Promise and Pitfalls of using Imprecise School Accountability Measures." Journal of Economic Perspectives 16.4 (2002): 91-114. EconLit. Web. 24 Oct. 2012.