Teacher credentials and student achievement: Longitudinal analysis with student fixed effects

https://doi.org/10.1016/j.econedurev.2007.10.002Get rights and content

Abstract

We use a rich administrative dataset from North Carolina to explore questions related to the relationship between teacher characteristics and credentials on the one hand and student achievement on the other. Though the basic questions underlying this research are not new—and, indeed, have been explored in many papers over the years within the rubric of the “education production function”—the availability of data on all teachers and students in North Carolina over a 10-year period allows us to explore them in more detail than has been possible in previous studies. We conclude that a teacher's experience, test scores and regular licensure all have positive effects on student achievement, with larger effects for math than for reading. Taken together the various teacher credentials exhibit quite large effects on math achievement, whether compared to the effects of changes in class size or to the socio-economic characteristics of students.

Introduction

Education researchers and policy makers agree that teachers differ in terms of quality, and that quality matters for student achievement. Despite extensive research, however, debate still rages about whether measurable teacher credentials can reliably predict either teacher quality or student achievement. We shed new light on this issue by using rich administrative data from North Carolina to explore a range of questions related to the relationship between teacher characteristics and credentials on the one hand and student achievement on the other. The teacher credentials in which we are most interested are those that can be affected in one way or another by policy.

This paper builds on our previous cross-sectional research on teacher credentials and characteristics (Clotfelter, Ladd, & Vigdor, 2006), but differs in its use of longitudinal data. These data include all North Carolina students in grades 3, 4 and 5 in years 1995–2004 for whom we can identify their teachers of math or reading. The longitudinal aspect of the data allows us to include in our models student fixed effects, which provide powerful protection against the left-out variable bias that typically plagues research of this type. Such data also permit us to explore in some detail the mechanisms through which teacher credentials exert their impacts.1

Section snippets

Empirical framework

Although we focus here only on the findings related to teacher credentials, all the findings emerge from fully specified models of student achievement estimating using student level data. In recognition of education as a cumulative process, the standard starting point in the literature is a “value-added” model in which the learning that a student (denoted by i) brings to the classroom in year t is incorporated in the form of her achievement in the relevant subject in the previous year.2

The North Carolina data

The data we use for this study are derived from administrative records maintained by the North Carolina Education Research Data Center, housed at Duke University. Student information, including their standardized test scores, are derived from student test records, and the teacher data from a state-maintained archive of personal records. Particularly relevant to this study is that North Carolina has been testing all students in reading and math from grades three to eight since the early 1990s

Results by teacher credential

Once again, we emphasize that the results we report here all emerge from the full model that we have just described. The levels regressions are based on about 1.8 million observations for students in grades 3, 4 and 5. The gains regressions are based on about 1 million observations and represent gains for 4th and 5th graders alone.6

Interpreting the magnitudes

Each teacher brings to the classroom a bundle of personal characteristics and credentials. Hence, we illustrate the magnitudes of the estimated teacher effects by comparing teachers with different bundles of attributes. Consider for example a baseline teacher with the following relatively typical attributes listed in the first column of Table 6. The teacher has 10 years of experience, attended a competitive undergraduate college, and has a regular license, an average test score, and a graduate

Acknowledgments

This paper is a revised version of a paper initially prepared for a conference organized by the World Bank on “The Contribution of Economics to the Challenges Faced by Education,” Dijon, France, June 2006. The authors are grateful to the Spencer Foundation for funding for this project, the North Carolina Education Research Data Center for the data, and to Aaron Hedlund for excellent research assistance. Other research assistants, including Trip Stallings and Roger Aliaga-Diaz, contributed to

References (15)

  • E.E. Boardman et al.

    Using panel data to improve estimates of the determinants of educational achievement

    Sociology of Education

    (1979)
  • D. Boyd et al.

    How changes in entry requirements alter the teacher workforce and affect student achievement

    Education Finance and Policy

    (2006)
  • C.T. Clotfelter et al.

    Who teaches whom? Race and the distribution of novice teachers

    Economics of Education Review

    (2002)
  • C.T. Clotfelter et al.

    Teacher-student matching and the assessment of teacher effectiveness

    Journal of Human Resources

    (2006)
  • Clotfelter, C. T., Ladd, H. F., Vigdor, J. L., & Wheeler, J. (2006). High poverty schools and the distribution of...
  • Clotfelter, C. T., Ladd, H. F., & Vigdor, J. L. (2007). How and why teacher credentials matter for student achievement....
  • Goldhaber, D., & Anthony, E. (2005). Can teacher quality be effectively assessed? Updated version of an urban institute...
There are more references available in the full text version of this article.

Cited by (444)

View all citing articles on Scopus
View full text