Teachers do become more effective with a few years of teaching experience, but some declines in effectiveness appear in the second decade after a teacher has begun teaching, particularly on low-stakes tests. These and other findings with respect to the correlates of teacher effectiveness are obtained from estimations using value-added models that control for student characteristics as well as school and (where appropriate teacher) fixed effects that estimate teacher effectiveness in reading and math for Florida students in fourth through eighth grades for eight school years, 2001–02 through 2008–09.

In recent years, it has become the conventional wisdom that teachers vary substantially in their effectiveness at lifting student classroom achievement, as measured by their performance on standardized tests.

Despite that variability, it has been difficult for scholars to identify types of training that correlate well with teacher effectiveness. Teacher classroom performance is correlated neither with the type of certification a teacher has earned, nor with the acquisition of an advanced degree, nor with the selectivity of the university a teacher attended. Only on-the-job training that comes with each year of experience in the classroom has been regularly identified as a correlate of teacher effectiveness.

Support for this research was received from the William Simon Foundation and the Searle Freedom Trust. The data were generously provided by the Florida Department of Education. We are grateful to staff members Tammy Duncan and Jeff Sellers for their assistance in this regard as well as to Antonio Wendland for administrative assistance and to Ashley Inman for research support. This is a revised version of a paper originally prepared for the Conference on “Merit Pay: Will it Work? Is it Politically Viable?” sponsored by Harvard’s Program on Education Policy and Governance, Taubman Center on State and Local Government, Harvard’s Kennedy School, 2010. Helpful comments were provided by Brian Gill, Brian Jacob, Steven Rivkin, Martin West, and conference participants at Harvard University.

Neither holding a college major in education nor acquiring a master’s degree is correlated with elementary and middle school teaching effectiveness, regardless of the university at which the degree was earned.

The emerging consensus depends upon a limited number of studies, however, so it is worth continuing to scrutinize available information to see whether findings can be replicated as well as to explore certain lacunae in the literature. Due to data limitations, most prior studies of pre-service training, for example, have relied upon crude indicators of the type of training a teacher has received—whether or not the teacher is certified, or whether or not the teacher has attended a selective university. Important effects produced by specific university training programs could be masked by lumping pre-service training into such broad categories. Similarly, estimates of the impact of acquiring a master’s degree have usually been unable to adjust for the fact that only some teachers pursue an advanced degree. If those pursuing advanced training do so in order to make up for teaching deficits, prior research may have under-estimated the benefits from that training. Also, the impact of specific masters’ degree programs has not previously been estimated.

Prior estimations of on-the-job training (years of experience) also suffer from certain limitations. Most do not take into account the fact that more experienced teachers are a selected population. And those that do adjust for that fact (by conditioning on teacher fixed effects) have been unable to track the effects of experience of individual teachers over eight years of employment. Nor have they distinguished between effectiveness at teaching to high-stakes and low-stakes tests. In short, the effects of on-the-job training over the teaching life cycle have yet to be precisely estimated.

Finally, the literature has yielded inconstant findings with respect to certification by the National Board of Professional Teaching Standards (NBPTS). Some studies show positive impacts, while others show little impact. Further research on this question is clearly merited, especially since the program is growing in size.4

In this paper we provide additional information on the correlates of teacher effectiveness, as estimated by the additional amount of learning taking place by students in their classroom (adjusted for student background characteristics). In addition to confirming propositions conventionally believed to be true, we find that effectiveness does not vary by the specific university one has attended. We also show that the value of additional on-the-job training decays after several years of teaching and turns negative after a decade or more of teaching, particularly when measured by low-stakes tests. Finally, we show that teachers certified by NBPTS are more effective than those not so certified, but the certification process yields no value-added in teacher effectiveness, except perhaps for teachers of middle school math.

These findings come from data made available by the Florida Department of Education from its K–20 Education Data Warehouse (EDW). These data include test score performance in reading and math for students in grades four through eight for the years 2002 to 2009. (All years are identified by the year in which the school year ends, the point in time when standardized tests are administered.) Administrative data on student characteristics, teacher characteristics, school characteristics, and the specific teacher each student had each year were also made available.

In recent years, research on the value of pre-service, in-service, and on-the job training has begun to accumulate rapidly, as many school districts and states have made available test score information to members of the scholarly community.

Nearly every state requires teachers to have earned both a bachelor’s degree and taken a certain number of courses in the field of education in order to receive a teaching license from the state. The practice has become so pervasive that the 2002 federal law, No Child Left Behind, 5

required that by 2006 all teachers in schools that received compensatory education funding be “highly qualified,” defined as “a bachelor’s degree as well as a state teaching license and demonstrated competence in the academic subject(s) he or she teaches” (Murnane and Steele 2007, p. 24). The number of required courses for certification is usually around 30 courses in the subject matter and for the specific grade (elementary, middle, or high school) being taught, about the same number that is required for a college major in the subject. School districts, in emergency situations, are able to hire temporary teachers without certification, and in some states teachers may be hired provided they begin a set of courses leading toward what is known as alternative certification (Hess et al. 2004; Nadler and Peterson 2009).

Prior econometric research has generally failed to detect positive impacts of pre-service teacher preparation programs on student learning. In their review of the literature, Murnane and Steele (2007, p. 24) conclude that “in general, empirical studies find little or no difference in average effectiveness between those teachers who are traditionally licensed and those who enter the profession through alternative routes.” In recent studies of high quality, Clotfelter et al. (2006) found no benefits from certification in North Carolina and Kane et al. (2006) found no differences in the effectiveness of certified, non-certified, Teach for America (TFA), or teachers recruited through a special New York City initiative.

In Florida alternative training programs have been developed for teachers (Llaudet 2006, pp. 38–39; Moe 2006), and the state’s department of education (2009) found that teachers who received their training in such programs were no less effective than traditionally certified teachers. However, the methodology employed in this study makes a number of highly restrictive assumptions. In another Florida study, using data similar to ours but for an earlier time period, Harris and Sass (2008) report no correlations between majoring in education and6

classroom effectiveness. They do find that math training may be associated with effectiveness in teaching high school math, but they caution the reader to interpret these results cautiously because of the sample size limitations they faced (Harris and Sass 2008, p. 55).

As mentioned above, it is possible that a particular teacher preparation program could be particularly effective, despite these negative findings about certification programs in general. If one could identify each degree program separately, one might find some to be effective and others to be ineffective.

Full paper