Chapter 4: Teacher Effectiveness

Other than students themselves, teachers are the most significant input in the public education system. A substantial body of research shows that having a highly effective teacher in a classroom is the most important within-school factor affecting student achievement, and the effect of a great teacher can help address or even overcome the negative impact of variables beyond the control of the school such as economic disadvantage and family factors.¹ And teacher salaries comprise the largest percentage of school budgets, making teachers the biggest economic driver in the system.

The national conversation on education policy has focused heavily on teacher quality. Traditionally, federal and state policies regarding teacher employment and compensation have been structured primarily around input measures — favoring seniority and advanced credentials. But in the last several years, the conversation has shifted to focus on measuring outcomes and structuring incentives for teachers more prominently around performance. States and districts across the country have made significant changes to the way in which teachers are evaluated and to thinking and practice around how that data is used in personnel decisions regarding compensation, teacher support, tenure, and dismissal. These changes have not been without controversy, particularly related to linking high-stakes decisions for teachers directly to student test scores, and the debate is ongoing.

In addition to the evaluation and performance management of those already in the classroom, attention has also focused on the teacher pipeline — examining the role of teacher preparation as well as on teacher quality. This attention has primarily come in the form of attempts at improving the measurement and communication of teacher prep program quality, and particularly in establishing links between programs and the effectiveness of the teachers they produce.

This chapter examines the makeup of the current teacher workforce in the United States, exploring trends and changes over time and what is known about how some of these demographic factors relate to measures of effectiveness. It discusses incentives like compensation, tenure, and pension policy — and how these systems are often structured in ways that disconnect job performance from critical personnel policies. It then describes recent shifts in policies around teacher effectiveness — primarily focusing on changes in teacher evaluation and measures of effectiveness and the associated implications for personnel policies. Finally, the chapter looks to the future of the teacher workforce — discussing issues related to the production of new teachers.

Who Is Teaching: The Current Teacher Workforce

Historical and current trends in demographic data highlight at least a couple of issues in the teacher workforce. First, there are considerable demographic differences between teachers and the students they serve — and these differences may increase as the student body becomes increasingly diverse. Second, schools now employ many more teachers with far less experience than they have in the past, raising questions around the attractiveness and stability of the teaching profession. Additionally, some of these demographics play a pivotal role in public school personnel policies. Experience and education credentials are the prime determinants in teacher compensation and other policies. These practices have a long history but are currently undergoing a shift as measures more directly related to job performance are increasingly brought to bear.

The teacher workforce in the United States comprises about 3.4 million teachers serving nearly 50 million students in public and charter schools. Over the last decade, the size of the teaching workforce remained relatively stable, ranging from 3.25 million to 3.4 million and growing at roughly the same pace as the student population. The demographic composition of the teaching workforce is predominantly female (76 percent) and white (82 percent), and these trends have changed little over time, though the composition of the student body has become increasingly diverse.

Figure 1. Demographics of the Teacher Workforce, School Years 2004 and 2012

Source: U.S. Department of Education, National Center for Education Statistics, Schools and Staffing Survey, available from: http://nces.ed.gov/surveys/sass/.

In the most recent decade, the most marked demographic shift among students has been a decrease in the percentage of white students and an uptick in the percent of Hispanic students. According to the most recent data available, one out of five students is Hispanic. In contrast, only one out of 12 teachers is Hispanic. Similarly, about one in six students in public schools is black, compared to one in 14 teachers, though this trend has been relatively stable over the decade.

Although the teacher workforce appeared to be aging between the late 1980s and the late 2000s, that trend may be largely over. Between the late 1980s and 2008, the modal (most common) age among teachers rose from 41 to 55. But between 2008 and 2012, that modal age dropped to 30. Further, the percentage of teachers aged 50 and older has been decreasing since 2008.² While age can be a proxy for teacher experience, it is not always an accurate representation of how long a teacher has been in the classroom. One reason is that a significant proportion of new teachers are older rather than recent college graduates; in the most recent federal data available, nearly one-third of new teachers were age 29 or older and one-tenth were over age 40.³

In terms of experience, over the long term the teacher workforce is trending toward less experience. In 1988, the modal years of experience among teachers was 15. By 2008, that modal experience level had shrunk to one year. That trend has softened somewhat since 2008 — the modal years of experience has climbed to around five. But there is still a higher proportion of less experienced teachers in the workforce compared with 20 years ago. The proportion of the teacher workforce with 10 or fewer years of experience has fluctuated from 37 percent in 1988 to nearly half in 2008 and 45 percent in 2012.

Analysis of the distribution of experience within the teacher workforce shows lower proportions of the least experienced teachers relative to those with the most experience, but it also shows a hollowing out in the middle with a dip in the proportion of teachers with 10 to 15 years of experience (Figure 2). During the five-year period between the 2007—08 and 2011—12 school years, the concentration of teachers in their first few years of teaching declined markedly, while the proportion of teachers with 10 or more years of experience grew. At the same time, the population of teachers in the middle, those with four to nine years of experience, remained stable (Figure 2).

These trends may reflect decreasing enrollment in teacher preparation programs over the past few years as well as high rates of attrition among teachers within their first five years.^{4 5} Turnover rates for new teachers have risen for the last two decades. Research through the 2008—09 school year found that just over 40 percent of new teachers leave the profession within five years, but more recent analysis suggests that trend may be slowing. Data through 2013 indicate that up to 70 percent of new teachers stay through the five year mark.⁶ In addition, minority teachers have higher rates of turnover than white teachers—likely contributing to the lack of racial diversity within the teaching workforce.^{7 8}

The most recent economic recession has certainly affected these trends. Numerous press reports have documented educator layoffs across states and districts. State and local policies regarding tenure often influence which teachers are retained when reductions are made, influencing overall attrition and retention trends. And economic conditions undoubtedly affect the choices of those considering entering the profession, as well as teachers who might consider leaving voluntarily.

Figure 2. Percent of Teachers by Years of Experience, School Years 2008 and 2012

Source: U.S. Department of Education, National Center for Education Statistics, Schools and Staffing Survey, available from: http://nces.ed.gov/surveys/sass/.

Consistent with the federal definition of “highly qualified teachers” from NCLB, all states require a bachelor’s degree to qualify for a teacher license in the majority of fields (exceptions exist for certain career and technical education fields). As of 2012, over half of all teachers held a master’s degree or higher (Figure 3). And over the previous 10 years, that percentage grew, which is consistent with the substantial financial incentives many districts provide to teachers who earn graduate degrees.

Figure 3. Teacher Workforce by Highest Degree Earned, School Years 2004 to 2012

Source: U.S. Department of Education, National Center for Education Statistics, Schools and Staffing Survey, available from: http://nces.ed.gov/surveys/sass/.

While teachers are academically similar in that they must hold a bachelor’s degree, the academic caliber of those entering the profession tends to be relatively low. A recent analysis of Barron’s ranking categorizing colleges and universities into six categories by competitiveness — one indicator of academic caliber — found that about one-tenth of new teachers come from the top two tiers. On the other end of the spectrum, between 20 and 25 percent come from the lowest two categories of colleges or universities. These proportions have fluctuated since the late 1980s, indicating that there is not a clear trend in how the selectivity of teachers’ undergraduate institutions has changed.⁹ Other evidence based on SAT scores, however, indicates that the academic ability of teachers is beginning to rise.^{10 11}

Relating Teacher Job Performance and Compensation and Tenure

As with many professions, the rise of organized labor in the first half of the 20th century brought protections for teachers — most notably tenure policies designed to prevent unfair dismissal, uniform salary structures to help address issues of low or inconsistent pay for similar work, and pension benefits. Those structures created job security for teachers, who historically faced threats of dismissal for trivial personal or political reasons and inconsistent compensation for similar work, and established strong incentives in the system based primarily on longevity.

Despite significant changes in the labor market broadening opportunities for women and rising expectations for the role of teachers and schools, these structures and incentives have persisted almost unchanged to the present.

Nearly all (89 percent) school districts compensate teachers according to a uniform salary schedule that pays teachers based solely on years of experience and the level of degrees earned.

Figure 4. Salary Schedule Samples and Teacher Effectiveness Timeline

Source: Salary data from the National Council for Teacher Quality and other sources. ^{12 13}

Research links teacher experience to improved outcomes for students through the first few years, but the structure of most salary schedules does not align to the evidence. On average, first-year teachers are less effective than more experienced teachers, but they make significant gains in their first few years before their performance tends to level off. But most educator salary schedules do not align to this trajectory of teacher improvement.

Salary schedules in most districts tend to deliver small annual increases the first several years a teacher is in the classroom, rather than providing larger increases at the time the teacher is expected to show significant improvements.^{14 15} Further, although there is little evidence that holding an advanced degree has any measurable impact on improved student outcomes, a master’s degree is a significant determinant of compensation in many public schools.¹⁶

Similarly, pension plans for teachers are structured as if their final years are the most productive, only taking into account a teacher’s final average salary (usually the last three to five years) rather than earnings over an entire career. As a result, teachers accrue very little during the first couple of decades of teaching and instead accrue substantial pension wealth in their last years in the classroom as they near normal retirement.¹⁷

To qualify for a pension today, a teacher must teach a minimum number of years to meet what is called a “vesting requirement.” Today, most states require a minimum of five years for a teacher to qualify for a pension, and 17 states require 10 years. As a result, about half of all new teachers will not stay long enough to qualify for even a minimal pension.¹⁸

The traditional pension system favors the small percentage of teachers who stay for an entire career, and the vast majority of their pension wealth is accumulated in the last few years of service (usually after about 30 to 35 years).^{19 20 21} For all but a few teachers, pension wealth is modest; in many cases, it is nonexistent. Recent reforms have further whittled away at pensions by cutting benefits and imposing greater restrictions for new hires in an attempt to pay down liabilities accrued over years of inadequate funding and poor returns in the stock market.

Structuring a modest long-term benefit such that real value often does not begin to accrue until well after the period when many teachers leave the profession does little to provide an incentive for good teachers to stay longer. And the reality of comparatively low salaries and minimal retirement benefits in many school districts coupled with the fact that most teachers are not covered under Social Security has implications for stability and longevity in the teacher workforce overall. Among the 15 uncovered states, an average of 52 percent of teachers will not vest into their state retirement system, according to state pension plan withdrawal assumptions.

In addition to compensation structures that largely discount performance, tenure policies can separate teacher performance from employment decisions. And in some states, tenure policies allow ineffective teachers to remain in classrooms when schools must make layoff decisions — even in the case of serious misconduct or chronically poor performance. Tenure provides teachers protections aftera certain number of years on the job. Originating in the early 1900s to protect teachers from facing arbitrary dismissals, laws vary by state, but most commonly stipulate that teachers receive tenured status after three years.²⁵ Tenured teachers can still be fired, but dismissal requires due process. As such, the dismissal process can be slow, cumbersome, and costly.²⁶ In the majority of states, tenure is granted and held based on years of experience, not performance, though that is beginning to change.

Recognizing the importance of teacher quality and the disconnect between job performance and compensation and other personnel policies, policy-makers have started shifting focus to both the evaluation of teacher effectiveness and the application of that information to improving the overall quality of the teacher workforce. But most teachers are still paid solely based on years of experience and educational credentials.

Measuring Teacher Quality: Teacher Evaluation and Student Achievement

Beginning in the mid-2000s, stakeholders in the education field began to recognize and call out the problems with existing teacher effectiveness policies. TNTP (formerly The New Teacher Project) released The Widget Effect in 2009, documenting how 12 school districts in four states rated the great majority — between 94 and 99 percent — of teachers as “Satisfactory” (Figure 5). These statistics stood in sharp contrast to data on student achievement. For example, over a three-year period between 2006 and 2008, more than 98 percent of Denver Public Schools teachers received a “Satisfactory” rating in the district’s binary evaluation system. Yet, in that same period, no more than 10 percent of schools failing to meet minimum federal performance benchmarks (AYP) identified even one tenured teacher as “Unsatisfactory.”²⁷

The report concluded that in many districts, teachers were treated as “widgets” — interchangeable parts with no acknowledgement of variation in effectiveness. And because they weren’t differentiating teachers in a meaningful way during the evaluation process, districts did not — and in fact could not — consider teacher quality when making decisions around professional development, compensation, and dismissals.²⁸

Figure 5a. Teacher Evaluation Results: Satisfactory vs. Unsatisfactory, as Presented in TNTP’s The Widget Effect (2009)

Figure 5b. Teacher Evaluation Results: Schools Not Meeting AYP

Source 5A and 5B: Graphics originally printed in TNTP’s The Widget Effect: Our National Failure to Acknowledge and Act of Differences in Teacher Effectiveness.²⁹

A great deal of attention has been paid to determining the characteristics of effective teachers. Research has yet to align distinct teaching practices with desirable student learning outcomes. As will be discussed later, many states and districts use teacher evaluation systems, particularly teacher observation, to try to determine which instructional moves lead to increased student learning, but research has yet to definitively identify a consistent and robust catalogue of specific practices that constitute effective teaching. Instead, most studies focus on measurable proxies for teacher effectiveness, such as certification, academic degrees, and years of experience.³⁰

As previously discussed, research links experience and effectiveness. And though that relationship is statistically stronger than many other measurable characteristics, including holding advanced degrees, test scores on licensing exams, and other input measures, experience still explains a relatively small proportion of teachers’ impact on student achievement.^{31 32 33 34 35} Given the lack of research-based support for effective teaching practices and weak links between other measurable inputs and teacher effectiveness, more recent focus has turned to outcomes-based measures — examining the impact of individual teachers on the achievement of the students in their charge.

Based in part on the influence of a handful of federal efforts, between 2009 and 2012, two-thirds of states adopted new teacher evaluation systems.³⁶ Under the 2009 American Recovery and Reinvestment Act, the U.S. Department of Education began to require states who received support from the State Fiscal Stabilization Fund to capture and report district-level evaluation data for teachers and principals. Subsequently, the competitive federal Race to the Top grant program encouraged states to implement rigorous teacher evaluation systems that factored in student achievement growth, which several Race to the Top recipients began doing. The most widespread federal effort to change the manner in which teachers are evaluated came through waivers from NCLB. As a condition of the waiver, the U.S. Department of Education required states to adopt rigorous teacher evaluation systems that

• Differentiate performance using at least three performance levels;
• Include data on student growth as a significant factor in measuring teacher performance;
• Use multiple measures in determining performance levels;
• Evaluate teachers and principals on a regular basis;
• Identify the need for professional development; and
• Are used to inform personnel decisions.³⁷

With 43 states (including the District of Columbia) currently operating under a waiver, recent years have seen substantial movement among state policymakers to develop systems — or improve existing ones — to evaluate the current performance of teachers in a fair and objective manner. Better evaluation systems are also designed to provide districts with the data they need to make performance management decisions — such as professional development, compensation, and tenure — around teachers and school leaders.

However, the development and implementation of these systems have not been without controversy. One salient theme is around the fairness of using student performance to evaluate teachers. Many factors both within and outside of school influence students’ academic performance — family factors, the impact of the overall school climate, previous academic background, and the influence of other school personnel among them. So evaluating teachers based on that performance in a way that truly isolates and reflects the work of the individual teacher presents a challenge. And the stakes for teachers can be very high, especially when employment or compensation decisions may be on the line. Another issue is determining what measures of student performance to use. For instance, performance on infrequent statewide assessments does not capture all dimensions of student learning and classroom outcomes.³⁸ Further, the majority of teachers work in grades or subject areas to which state assessments do not apply, which poses a challenge in defining valid and reliable measures of student achievement. According to a 2014 Brookings Institution study, only one-fifth of teachers in four urban school districts were eligible for evaluation based on student test scores.³⁹

States have attempted to address these concerns through the design of both the measures of student outcomes and through the manner in which student outcomes factor into teacher evaluations. While teacher evaluation systems vary across the country, no state or district has attempted to develop or implement a system that evaluates teachers solely based on student test performance. Instead, states are adopting policies to improve how districts evaluate and influence teacher performance using multiple measures. The most comprehensive evaluation models incorporate student learning outcomes while also capturing other dimensions of teacher quality, through both objective and subjective evaluation tools.

Measuring Student Achievement for Teacher Evaluation

Under teacher evaluation reforms, as of 2015, all but eight states have committed to using an objective measure of student achievement — such as performance on standardized assessments — as a part of teacher and principal evaluation systems.⁴⁰ However, given the challenges of fairly incorporating student test performance in evaluations, all states and districts engaged in these reforms must account for factors like the variation in student background and other external influences on performance.

Many states now require the use of value-added models in evaluations as a measure of the a teacher’s or school leader’s impact on student achievement, as determined by results on state standardized tests. Value-added models generally use statistical methods to isolate the impact of a factor of interest (like being taught by a particular teacher) on an outcome (like a test score) by mathematically removing the effects of other factors (like student family income or prior performance levels). In doing so, these research-based models are designed to assess teacher performance in a way that reduces bias favoring those teaching more educationally advantaged students.

Research shows value-added measures of student performance on assessments to be better predictors of teacher effectiveness than many other measurable characteristics that have traditionally been used to define incentive structures for teachers such as educational credentials or scores on licensing exams. A significant body of research supports the validity of well-designed value-added models — finding that value-added measures successfully identify the causal impact of a teacher on student outcomes. In other words, well-designed value-added models can differentiate whether a change in student scores over the course of a school year is because of the teacher or because of other factors. ⁴¹

A value-added measure based on a standardized test score is primarily predictive of the influence of teaching on standardized test scores; and value-added measures are found to be strong predictors in this capacity. And studies have found teachers’ value-added scores in one year to be partially predictive of impact on student performance in subsequent years. Their ability to measure influence on a broader definition of student achievement, however, is limited by the degree to which the testing data incorporated in the value-added model correlate with student achievement beyond the test score itself. In other words, if an Algebra I assessment is a poor measure of whether a student has mastered Algebra I concepts, then a value-added measure tied to that assessment would indicate how much the Algebra I teacher influenced student performance on the assessment, but not necessarily how much the Algebra I teacher influenced student mastery of Algebra I. This “Garbage In, Garbage Out” issue, however, points to a challenge in ensuring validity and reliability in assessments, not to a flaw in the concept of value-added measures as a measurement tool.

Beyond predicting assessment performance, some studies have identified merit in value-added measures ability to identify effective teachers, both in comparison with other evaluation methods and in long-term outcomes for students taught by teachers with high value-added scores. One study simulating a reduction in force with teacher layoffs based on three methodologies — principal evaluations, pure seniority, and results of value-added measures — found that the scenario utilizing value-added measures identified less effective teachers better than either principal evaluations or seniority. Seniority fared the worst, identifying teachers for layoff across the continuum of effectiveness. ⁴² And research ties significant long-term positive outcomes, such as greater rates of college enrollment and higher incomes, to students of elementary school teachers receiving high value-added scores relative to students taught by teachers with lower scores. ⁴³

Critics, including some researchers, have raised questions related to the complexity and validity of statistical methods used to calculate value-added scores. Value-added measures are complex statistical models incorporating multiple data points, which makes transparency about evaluation systems that include them challenging. And a poor value-added model design could attribute student outcomes (either positive or negative) to a teacher’s performance that are actually caused by other factors not measured in the model. ⁴⁴ Further, a recent study found weak to nonexistent relationships between the results of value-added measures of teacher performance based on standardized test scores and some other measures of instructional quality. ⁴⁵

These limitations of value-added measures — namely their sole dependence on standardized assessments and the limitations of statistical models more generally — mean that their best use is in combination with other measures of effectiveness. Plus, while value-added measures may help identify effective teachers, they cannot identify what teaching practices are linked to improved student performance, limiting their value in providing formative feedback to teachers that may lead to more effective teaching.  But in a role as one piece of a more comprehensive evaluation methodology, value-added measures can inject an objective measure that can amplify or validate information gleaned from more subjective methodologies. And research finds high levels of correlation between value-added measures based on test scores and high-quality, observation-based evaluation methodologies that focus specifically on instructional practice rather than outcomes. ^{46 47}

Concerns regarding the complexity and accuracy of value-added measures have fueled debate regarding the degree to which value-added measures should be included in teacher evaluations, particularly when evaluation results are tied to high-stakes employment decisions. But proponents of value-added measures point to their relative strength compared to other predictors of effectiveness, particularly when used as one component in an evaluation system combining multiple measures.

In addition to debate over their technical merit, one generally agreed-upon and significant drawback of value-added measures is that they can only apply to teachers in tested grades and subjects. To evaluate the majority of teachers who do not teach tested grades and subjects and cannot receive a value-added score, districts must use other measures of student growth. And they must consider the relative rigor of different methodologies used for purposes of evaluating teachers in different grades and subjects.

One type of alternative measure, Student Learning Objectives (SLOs), involves articulating specific learning goals for a course or subject area and measures to track progress toward those goals. ⁴⁸ Individual teachers and school leaders can develop SLOs and apply them to particular instructional contexts. As such, they can be adapted for a variety of grades and subjects, and they can incorporate district-wide assessments, end-of-course exams, or other measures of achievement, such as student performances for subjects, including art and music.

Classroom Observations and Student Surveys

In contrast to student growth and value-added measures that cannot be applied uniformly across subject areas and grade levels, classroom observations are a universally applicable measure that can provide insight into how all teachers are performing. Classroom observations also offer real-time feedback to teachers so they can make changes to their instructional practice during the school year.

Most states require that teachers undergo annual classroom observations. ⁴⁹ Districts have historically used locally-developed checklists and rubrics to observe teacher practice, but they are starting to move toward the use of higher-quality externally-developed rubrics.⁵⁰ These rubrics are designed to capture better depth and detail of classroom practice and provide ongoing, actionable feedback to teachers. A high-quality observation rubric will have a clear relationship to student achievement results. In other words, a teacher who scores high on the observation instrument should also have, on average, students with higher academic growth. Recent data from the Measures of Effective Teaching (MET) study, sponsored by the Bill & Melinda Gates Foundation, found that five research-based classroom observation rubrics were predictive of student achievement.⁵¹

These rubrics were found to be useful in identifying teachers who were most effective at improving student learning outcomes. That said, while the associations between the observation tools and student achievement were statistically significant, they were not particular strong. But the study also showed that the predictive power of these rubrics strengthened when classroom observation scores were combined with measures of student achievement gains and student surveys.

Policies on the use of classroom observation rubrics vary by state. Some states, including Rhode Island and North Carolina, have developed their own state rubrics, while others have a list of approved rubrics from which districts can choose. Some states allow districts greater autonomy to design their own rubrics or modify existing frameworks.⁵²

Although relatively less common, student surveys are another method districts use to measure teacher effectiveness. Students of all ages can take surveys that measure their perceptions of teacher quality, although the complexity of questions will vary by student grade level. Using data from 3,000 teachers, the Gates MET study found that survey results were broadly predictive of teacher impact on student learning. Another important finding of the study was that student surveys produced more reliable results from year to year than did objective measures of student achievement gains or classroom observations.⁵³ In other words, student survey results were more stable. However, some teachers are resistant to having student feedback integrated into the formal evaluation process, potentially limiting the likelihood of expanding the use of student surveys.⁵⁴

Teacher Evaluation System Implementation and Early Results

Since 2009, many states have succeeded in implementing certain teacher evaluation reforms (Figure 5). But even the best-designed teacher evaluation policies can only effect change if they are implemented well. Because states’ evaluation systems are still relatively new, fully assessing their impact on teacher effectiveness is difficult. But many states’ systems reflect promising practices.

For one, states are beginning to evaluate educators more frequently. As of 2015, 27 states required annual teacher evaluations, and 31 states required annual principal evaluations. Thirty-nine states incorporate objective measures of student achievement data into teacher evaluations, and of those, 17 report it is the most significant factor.⁵⁵ And most states now use at least three tiers to differentiate teachers (Figure 6).

Figure 6. State Policies for Teachers and Principals

Source: National Council for Teacher Quality, State of the States: Evaluating Teacher, Leading, and Learning (2015) ⁵⁶

One early indicator of a model’s impact is the extent to which it leads to differentiation of teachers by performance. Increased differentiation in teacher ratings is crucial for a district to identify its highest and lowest performers, provide teachers the support they need, and make performance-based personnel decisions.

Some early state data show greater differentiation than under prior systems.⁵⁷

Louisiana offers one such example: under its previous binary evaluation system, 98.5 percent of teachers received a rating of “Satisfactory.” After the state implemented the new four-tier Compass model in the 2013–14 school year, 43 percent received the highest rating of “Highly Effective,” while 8 percent were classified in the lower two tiers (6 percent in “Effective: Emerging” and 2 percent in “Ineffective”).

Merely adopting a multi-tiered system may not result in schools and districts effectively distinguishing among high- and low-performing teachers. When Delaware rolled out its new evaluation system in the 2012–13 school year, 99 percent of teachers were categorized as “Highly Effective” or “Effective.” Only 1 percent of teachers received the third rating of “Needs Improvement,” and 0 percent of teachers received the lowest rating of “Ineffective.”⁵⁸ Delaware may be an example of a state facing challenges in using all its evaluation tiers to meaningfully differentiate teachers.

Political challenges and other education priorities have also led to slower progress toward implementing additional evaluation policies, particularly the use of student growth measures based on state assessments in evaluating teachers.^{59 60} States have responded to public backlash by decreasing the weighting of value-added scores or other student growth measures in teacher evaluation ratings. In Ohio, for instance, the state legislature voted in 2014 to decrease the weighting for student growth from 50 to 42.5 percent of a teacher’s evaluation.⁶¹ In New Jersey, the governor mandated that student performance on statewide exams account for 10 percent — a significant decrease from the previously adopted 30 percent — of a teacher’s evaluation in the 2014–15 school year.⁶² Indiana, Hawaii, and Maryland have also decreased the weighting of test-based student growth in teacher evaluations.^{63 64 65}

Other states have outlined vague parameters of how student growth should be used when evaluating teachers but left the specific methodology or weighting up to local school districts. For example, Massachusetts has not stipulated how much weight student growth on state assessments will have when districts and schools evaluate teachers. Colorado, while mandating that 50 percent of a teacher’s evaluation be based on student growth, allows districts to choose their own student learning measures — including state assessment results — and decide how to weight them.⁶⁶ In addition to adjustments to how much student growth factors into evaluations, some states, including Georgia and Connecticut, have opted to delay full implementation of their evaluation systems while they transition to the more rigorous Common Core standards.^{67 68}

The degree of flexibility states give to school districts may also lead to inconsistent implementation of reforms. As of 2015, only 9 states had adopted a single statewide teacher evaluation system, and another 12 had a state model from which districts could opt out.⁶⁹ Other states offer guidelines or criteria for districts but give them greater flexibility to design their own evaluation systems. Districts may, for instance, choose which measures to include in evaluation systems or decide the specific weighting of each measure in determining overall ratings. But while allowing districts to control evaluation gives them greater autonomy, it may potentially lead to issues of consistency and quality control. For instance, among six districts in Indiana, student academic growth contributes anywhere between 0 and 40 percent of a teacher’s final rating. This degree of flexibility may explain why the distribution of teacher evaluation ratings varies drastically within Indiana. Two of those six school districts offer one example: During the 2013–14 school year, only 4 percent of teachers in South Bend were rated “Highly Effective” compared to 15 percent in Fort Wayne.⁷⁰ Among the states that provide school districts flexibility to design their own systems, fewer than half require that the state review and approve district models.⁷¹

What Comes Next: Efforts to Use Teacher Performance Data to Improve Quality

Robust teacher evaluation systems can only lead to improved teacher quality if schools use the resulting data effectively. Schools and districts can use teacher evaluation data to inform meaningful decisions around on-the-job support as well as compensation and employment.

Professional Development

At least 40 states require professional development for teachers in some form or fashion, and significant resources are devoted to providing professional learning opportunities. Notably, the federal No Child Left Behind Act (NCLB) requires that low-performing schools set aside 10 percent of their Title I funding allocation specifically for professional development, and billions of dollars allocated under Title II expressly support professional development.⁷²

Though nearly all schools deliver some form of professional development, there is little evidence that common types of professional development efforts, such as workshop-type training, have positive impacts on teacher quality or student outcomes. Rather, research on professional development finds that those with the most measured impact on student learning focus on specific content knowledge, are sustained over a long period of time, and are collaborative in that they create networked learning communities that can provide ongoing support. These features appear in analysis of professional learning in other countries that are high achieving on international measures (like PISA and TIMSS).⁷³ YYet professional development opportunities for teachers in the United States have tended not to reflect many of these research-supported features, which has negatively affected teacher performance growth. A recent study found that despite large investments in professional development, most teachers in the United States do not appear to improve substantially from year to year, even though many have not yet mastered critical skills.⁷⁴

Additionally, there is growing evidence that professional development targeted at individual teachers based on their unique needs can be effective. For example, teachers in Cincinnati Public Schools improved their math instruction after receiving targeted feedback based on classroom observations.⁷⁵

These findings highlight the role of performance evaluation for purposes of improving instruction and not solely for purposes related to high-stakes employment decisions such as compensation, retention, and tenure. A study by the Brookings Institution found that classroom observations in particular have the potential to provide formative feedback in real time to teachers that helps them improve their practice, whereas feedback from state achievement tests is often too delayed and vague to produce improvement in teaching.⁷⁶

In concert with reforms to teacher evaluation systems across the country, states increasingly require that identification of and plans to meet professional development needs at the teacher level be part of the evaluation process. As of 2015, 25 states had policies to ensure that schools design and deliver professional development based on teachers’ individual evaluation results.⁷⁷

Differential Pay: Performance-Based Compensation and Financial Incentives

As of 2012, the vast majority of districts still had a salary schedule in place, and only 11 percent of districts used pay incentives to reward teachers for excellent performance, though the practice is increasing.⁷⁸ And as states develop more effective ways to measure teacher quality, districts should be better able to incorporate performance data when making compensation decisions.

Federal policy has helped encourage a shift away from strict lockstep salary schedule systems. In addition to ESEA waiver requirements that districts incorporate evaluation data into personnel decisions (though compensation isn’t specified), the Teacher Incentive Fund (TIF), a federal grant program that has allocated over $2 billion since its inception, began supporting state and district efforts to implement performance-based compensation in 2006.⁷⁹ Grantees from across the country have included Miami-Dade County Public Schools, Mississippi, Pittsburgh Public Schools, Seattle Public Schools, and Washington, D.C. Public Schools (see Case Study: Lessons From Washington, D.C.’s IMPACT System).

In addition to encouraging performance-based pay, another key purpose of TIF was to increase the number of high-quality teachers in challenging environments such as high-needs schools or hard-to-staff subjects through targeted compensation.

Though increasing in popularity, differentiated pay policies of either type (performance pay or incentives targeted to high-needs areas) are not widespread. According to the most recent data available, only seven states have policies that factor in teacher evaluations in salary determinations, and nine allow for performance-based bonuses to individual teachers (Figure 7). Compensation decisions tied to higher-stakes evaluations remain controversial, particularly among teacher unions.⁸⁰ More states, however, allow for incentives tied to recruitment and retention, with 24 states having policies supporting financial incentives for shortage areas, high-needs schools, or both. However, survey data indicate that such incentives may not be a common practice at the district level. In 2012, 14 percent of districts reported providing additional compensation to either reward or retain teachers in fields of shortage, and only 6 percent used financial incentives to reward or retain teachers in harder-to-staff locations.⁸¹

Figure 7. State Policy on Differential Pay for Teachers

Source: National Council for Teacher Quality, State of the States: Evaluating Teacher, Leading, and Learning (2015) ⁵⁶

Because of the relative rarity of true performance-based teacher compensation systems, research on outcomes of differentiated pay is scarce, and the results are mixed. In addition, compensation research experts note that research designs have been limited by the ways performance pay programs are designed and implemented in practice. Therefore, making definitive claims about the outcomes of such programs remains a challenging task.⁸³ Some studies find no link between financial incentives for student achievement and higher test scores, while others see higher achievement for students in systems with performance bonuses.⁸⁴ A large-scale 2014 study on TIF presented findings on early implementation from 153 districts. Fewer than half the districts carried out all required components of the grant, potentially indicating implementation challenges with pay-for-performance systems.⁸⁵ The relative impact of the programs evaluated depended on program design elements such as the amount of compensation provided, the link between individual efforts and rewards, and whether the funding is perceived as temporary or permanent (see sidebar on Washington D.C.’s IMPACT system). As the practice becomes more common, additional research on the impact on both student performance and other potential effects related to new teacher recruitment and retention of effective teachers is warranted.

Dismissals, Tenure, and Transfers

Under recent state changes to teacher evaluation, policies emphasizing performance in personnel decisions regarding dismissal, tenure, and transfer decisions have gained traction. In particular, 29 states specifically stipulate poor performance as grounds for dismissal.⁸⁶ But the dismissal process continues to be a challenge when it comes to tenured teachers, who have the right to multiple appeals processes in most states.⁸⁷ To address these challenges, state policymakers are increasingly reforming teacher tenure systems. For instance, 16 states now require that tenure decisions factor into performance on the front end, and seven states allow districts to revoke tenure in the case of consistently poor performance.⁸⁸

In some states, tenure systems are being challenged prominently in court — as evidenced by the polarizing 2014 Vergara v. California ruling, which initially found teacher tenure laws in California unconstitutional.  A unanimous state appellate panel reversed the trial court’s decsion, and a petition for review in the state supreme court is expected. ⁸⁹. Yet making large-scale changes to long-standing teacher tenure policies is controversial.

Policies related to tenure and seniority can also influence teacher transfers and layoffs in the event that a school needs to eliminate staff positions. LIFO, or “last-in, first-out,” rules refer to the process of newer teachers facing layoffs first — even if they are more effective in the classroom than their colleagues with greater seniority. And tenured teachers are often entitled to vacancies in other schools regardless of their performance. Effectively, these policies limit decision making of school principals regarding staffing because they are constrained by central district office placement decisions based on seniority. One potential solution is implementing mutual consent policies under which the teacher seeking a new placement and the principal of the receiving school agree to the transfer. Rather than granting the district’s central office sole control, these policies are designed to decentralize staffing decisions to the school level.⁹⁰ Colorado is the only state that has passed legislation to require mutual-consent hiring statewide.⁹¹ Large districts, including Chicago, Washington, D.C., and Baltimore, have also implemented such policies.⁹²

The Role of Teacher Training and Recruitment in Building an Effective Workforce

Developing policies to measure and improve the effectiveness of teachers who are part of the current workforce approaches teacher quality from one angle. But system-level change in teacher quality also requires understanding and influencing the quality of new entrants into the profession.

Entering a public school classroom is a multistep process. In all states, teachers are required to meet state licensing requirements to be employed in most public schools.⁹³ Although specific standards vary by state, the basic process universally requires a bachelor’s degree, passage of state licensing exams, and completion of an accredited teacher preparation program. Federal policymakers, as well as some state ones, are beginning to push for higher-quality teacher preparation programs proven to produce graduates who can have a positive impact on student achievement.

Teacher Preparation Program Structures

There are two broad types of teacher preparation programs. Traditional teacher preparation programs are typically offered by colleges and universities, run concurrently with an undergraduate or graduate degree program, and include instruction both in content area and teaching pedagogy. Most programs also include a student teaching experience during which teacher candidates are paired with classroom teachers to gain experience in a real-world setting. Student teaching requirements vary significantly in length among states and programs. In addition, standards for selecting mentor teachers and other factors that could affect the quality of the student teaching experience vary widely.

The other category of teacher preparation programs is alternative certification programs, which emerged in the 1980s. Alternative programs generally offer a non-degree option for eligible candidates, often of shorter duration than a traditional program and providing training concurrent with paid employment by a school district as a teacher of record. The structure of alternative programs varies from state to state, as do eligible providers. Federal data from the 2009–10 academic year indicate that 12 percent of individuals enrolled in teacher preparation programs were in alternative programs.⁹⁴

The current evidence base of teacher quality across certification pathways does not point to one route as being more effective than the other. In fact, the variation in effectiveness within pathways is far greater than the average differences between pathways.⁹⁵ In other words, individual programs of both types produce teachers who are more or less effective, and there is little evidence that either type of teacher preparation program is systematically better able to train an effective teacher. Many states have designed alternative certification programs to be very similar to traditional ones; the primary difference is that candidates are already in the classroom while they complete coursework. And among programs within the same category, there is often a great deal of variation in selection criteria, coursework, and practice.⁹⁶ A large-scale study of 20 school districts, sponsored by the U.S. Department of Education, did not find significant differences between traditionally and alternatively certified teachers in their ability to improve student performance.⁹⁷ Certain alternative programs produce promising results; evaluations of Teach For America corps members, for instance, show a positive impact on math outcomes for elementary and secondary students.⁹⁸ And a separate study in North Carolina found that Teach For America teachers outperformed other teachers in multiple subjects and grade levels and in other measures of performance assessed against state teaching standards.⁹⁹ However, it is difficult to extrapolate Teach For America’s results to  alternative certification programs generally, given both the competitive nature of the program’s selection process and the unique nature of its training.

Teacher Preparation Program Reform

As part of the effort to improve teacher quality, many states are attempting to improve teacher preparation programs. A 2014 report from the National Council on Teacher Quality (NCTQ) found that 33 states had passed significant laws or regulations to improve teacher preparation between 2011 and 2013. State changes included implementing higher admission standards for teacher preparation programs, improving ways to test content knowledge, and developing higher-quality student teaching experiences.¹⁰⁰ A separate report from NCTQ focused on states’ progress toward aligning teacher preparation programs to new Common Core academic standards. The report found that five states — Indiana, New York, North Carolina, Rhode Island, and Texas — were ahead of the rest of the country in preparing aspiring teachers for the new academic standards.¹⁰¹

Historically, preparation programs have done little to track and report academic outcomes of students for the teachers they produce. As of 2014, only 10 states routinely collected data linking teacher preparation programs and student achievement outcomes, and among those, only four established related performance standards for programs.¹⁰²

Louisiana, North Carolina, and Tennessee (see Case Study: Teacher Preparation Reform — Lessons From Tennessee) are leaders in establishing data systems that tie student achievement outcomes to teacher preparation programs. While these states have made great strides in developing data linkages and reporting outcomes, they currently do not use this data to make program approval decisions. Linking program approval to student outcomes would represent a significant shift away from most current state policies, which typically focus on program input measures such as faculty qualifications and the length of student teaching. But linking teacher preparation programs to student learning outcomes, and using this data to inform approval processes, could ultimately help improve program quality. As it stands, existing approval and reporting policies make it difficult for schools, potential teachers, and the public to differentiate among programs for quality; 25 states have never identified a single low-performing teacher preparation program.¹⁰³

Recognizing these challenges, the U.S. Department of Education proposed a set of regulations designed to improve teacher preparation in 2014. Under these changes, states would develop new reporting systems to report on the annual performance of teacher preparation programs based on comprehensive data, including employment outcomes, employer feedback, and student learning outcomes data. These changes would allow for increased transparency around employment and student learning outcomes for graduates of all teacher preparation programs, including alternative certification ones, within each state.¹⁰⁴

There is also a burgeoning movement for preparation programs to hold themselves accountable for the outcomes of their graduates. In higher education, colleges and universities must be accredited to qualify for federal financial aid dollars, such as Pell grants. Additionally, specific programs within colleges seek out specialized accreditation. In most states, teacher preparation programs choose whether they want to seek out this specialized accreditation, while other states require programs to pursue accreditation to receive the state’s permission to license new teachers.

Accreditation is often criticized as setting only a minimal bar, and few programs ever lose their accreditation status. But that paradigm seems to be shifting in education, as two long-standing educator preparation accreditation bodies are currently merging and establishing new, more rigorous standards and protocols for accrediting colleges of education. The new standards encompass five areas, including:

• Content and pedagogical knowledge;
• Clinical partnerships and practice;
• Candidate quality, recruitment, and selectivity;
• Program impact; and
• Provider quality assurance and continuous improvement.¹⁰⁵

There are still questions about whether CAEP can enforce stricter standards over the long run — it is, after all, still reliant on a self-policing model where the accreditor has little incentive to hold programs accountable — but it shows signs of promise. Once enacted, CAEP’s standards will require programs to slowly raise the admission requirements for new teachers and will require programs to use data to evaluate the impact their graduates have on student learning outcomes.

Beyond accreditation processes, the NCTQ is developing an “inspectorate model” for states to use in program approval process. Under the model, states would employ independent inspectors to evaluate teacher prep programs against established standards of practice through comprehensive site visits.¹⁰⁶

Entering the Profession: Teacher Licensure

After completing a teacher preparation program, teacher candidates in all states are typically required to pass state licensure exams before teaching in a public school.¹⁰⁷ Licensure exams, which cover both the content area of licensure and teaching pedagogy, are designed to act as a screening point for teachers — ensuring a minimum competency in the content of the subject to be taught and knowledge regarding the practice of teaching. There is some evidence that candidate performance on these exams is predictive of classroom effectiveness. A study of teacher testing in North Carolina found a positive relationship between teacher performance on licensure exams and student learning gains.¹⁰⁸ However, other studies have failed to find a significant relationship between teacher effectiveness and performance on licensure exams.¹⁰⁹ Some of that inconsistency may be linked to the quality of the exams, which vary by state. Research supports the general notion that teachers with strong content knowledge in their field are more effective.¹¹⁰

Though not yet reflected in the research linking assessment performance to teacher effectiveness, the use of performance-based assessments for teacher candidates is rapidly gaining traction, with the edTPA instrument, developed by Stanford University and partners, at the forefront. Through performance-based assessments, candidates submit a portfolio of work, including lesson plans, video clips of actual teaching, and other materials, for evaluation. Field testing of edTPA ended in 2013, and teacher preparation programs in 33 states use it in some capacity.¹¹¹ Some states are beginning to require some form of performance-based assessment as a condition of licensure.

Upon successful completion of an accredited teacher preparation program and passage of all required licensing exams, candidates apply for a teaching license. State licensure structures vary widely. Some states grant limited or provisional credentials to new teachers with the opportunity to earn a standard professional license after a period of employment, in some cases requiring evidence of effectiveness. Other states grant full licensure from the outset. Most states require that fully certified teachers apply for renewal periodically, often tied to requirements for professional development. In a few states (Delaware, Louisiana, Rhode Island, Tennessee), license renewal is tied to evaluations, but this practice is uncommon.¹¹²

District Recruitment and Selection Policies

District recruitment and selection policies may also play a role in influencing teacher workforce quality. Growing evidence from the private sector indicates that prioritizing human capital can have a significant impact on organizational performance.¹¹³

Yet school districts’ hiring practices remain relatively outdated. For example, districts may only recruit new teachers from local teacher training programs rather than nationally — limiting their ability to screen a greater number of applicants for top talent. Hiring processes in school districts also tend to be highly centralized, limiting the ability of school leaders to choose the best educators for their schools.

Under common “transfer and excess” rules, school leaders are expected to hire senior teachers from other schools that want to transfer in or teachers whose positions at other schools have been eliminated — even if these teachers are not the most qualified or the right fit. Another consequence of transfer and excess rules is that many districts must wait until these placements are complete before hiring new teachers.¹¹⁴ Delayed hiring that does not take place until right before the school year begins, or even after the school year has started, leads to a smaller pool of competitive talent for schools to consider.¹¹⁵

Such policies and approaches to hiring limit the ability of school leaders to make the best choices for schools’ specific needs from the broadest pool of talent possible and may prevent talented teachers from securing positions in a timely fashion or at all.

The Impact of School Leaders

Studies show that nearly 60 percent of schools’ impact on student achievement is attributable to principal and teacher effectiveness together. Broken down, teachers account for 33 percent and principals account for 25 percent of a school’s total impact on student achievement.¹¹⁶ Teachers’ impact on student achievement receives a significant amount of attention, while there is much less focus on school leaders’ roles in the equation. Research shows that highly effective principals raise the achievement of a typical student in their schools by between two and seven months of learning in a single school year and that ineffective principals lower achievement by the same amount.³ And principal quality affects teacher turnover and specifically turnover among effective teachers. Teachers who leave schools with the most-successful principals are much more likely to have been among the less-effective teachers in their school than teachers leaving schools run by less-successful principals.¹¹⁸

But recruitment, selection, and retention of high-quality candidates to principal positions proves a challenging task for many districts and schools. Urban districts in particular often have low ratios of applicants to open principal positions, limiting selectivity.¹¹⁹ And principal turnover is rampant, particularly in schools serving significant high-needs populations. Nationwide, principal turnover ranges from 15 to 30 percent on average, with the highest turnover rates occurring among schools serving the highest-need students.¹²⁰ And fully half of new principals leave their positions by the third year.¹²¹ The complexity of the job may be a factor. A recent national survey reported that 69 percent of principals indicated their responsibilities had changed in the past five years, and 75 percent said their job had become too complex, suggesting that the field of school leadership must evolve as the education landscape changes.¹²²

School Leadership Pipelines

Similar to the traditional routes for teacher preparation, historically principals have been primarily trained through university-based training programs. And similar to criticisms of teacher preparation, principal preparation programs are often cited as being poorly aligned with what educators need to succeed in the school setting. For example, little emphasis is placed on clinical experiences that prepare prospective principals for the job’s daily duties. A study of traditional principal preparation programs found that clinical experience tends to be squeezed in while candidates work full time and generally occurs in the school where the candidate is employed. For the most part, candidates in the study described their clinical experience as something to be gotten out of the way, not as a valuable learning opportunity.¹²³ Not only that, but university-based preparation programs where the vast majority of principals are trained have long been under intense scrutiny for lacking rigor and not adequately equipping principals for the multifaceted role of effective instructional leader, among other things.³

In recent years, alternative approaches to developing school leaders have emerged — including efforts to improve existing routes and accountability for both programs and principals, to develop new pipelines for potential leaders, and to provide alternative training programs for aspiring school leaders.

At the state policy level, 46 states have adopted leadership standards for principals, and many have begun aligning them to all components of a school leader’s career continuum.¹²⁵ In addition, recently more than half of states, through a mix of state and federal initiatives, have passed laws to strengthen the evaluation process for principals.³ AAnd under the federal Race to the Top grant, states such as Louisiana and Rhode Island implemented new requirements for principal preparation programs. Louisiana’s principal preparation programs must show proof of collaborating with schools and districts to ensure leaders can better serve the diverse needs of individual schools. And in Rhode Island, programs are required to offer multiple field experiences and a more intense internship to better prepare principals.¹²⁷ Other states are working to build principal pipeline and development capacity. For instance, under Race to the Top, Florida developed an intensive, job-embedded leadership pipeline for aspiring turnaround principals and assistant principals that aims to provide professional development to encourage and enable qualified, motivated teachers to become successful principals.¹²⁸

Beyond state policy efforts, numerous efforts to create new types of principal preparation programs have emerged across the country. Programs vary in focus — ranging from efforts to identify potential leaders very early in their teaching careers and develop leadership skills with an eye toward longer-term potential to intensive residency-based programs that place aspiring principals in embedded-training programs in schools with the goal of moving to a full-time principal position immediately upon completion. Some programs focus on serving the charter sector, turnaround schools, or districts schools specifically; many are based in specific cities or regions.¹²⁹ Other programs are national in scale, but some geographic regions remain uncovered. Together, these programs produce only several hundred principals annually — just a small fraction of the total need. National data suggests that on average 23 percent of principals leave their positions annually, and between 11 and 15 percent leave the profession entirely. Based on these data, schools must hire over 10,000 brand new principals each year.¹³⁰

Because many of these programs are relatively new, there is little rigorous evidence on their effectiveness in producing principals who improve student outcomes. The New Leaders Aspiring Principals program, a national program that prepares principals through a residency-based program, is one of the few programs that has been the subject of a rigorous, independent evaluation. The study found that students who attended schools led by New Leaders principals experienced slightly larger achievement gains on average than did similar students in schools led by non-New Leaders principals.¹³¹ While these results are promising, more research and data collection on both new and existing programs is needed to inform policy efforts to improve principal preparation.

Conclusion

It is clear that most states and districts are in some stage of transition in tying data on teacher effectiveness to critical decisions regarding compensation, tenure, hiring, and dismissal. Though practices like incorporating student achievement measures into the teacher evaluation process are becoming widespread, the application of that information to personnel policies remains highly variable across the country. For instance, more and more states allow for districts to factor performance into decisions regarding tenure and dismissal, but relatively few states require performance to enter into these decisions. And few districts factor performance information into compensation structures.

The linchpin in systemically improving policies tied to effectiveness for both the current teacher workforce and for the preparation of future teachers is information. The quality, fairness, and reliability of teacher effectiveness measures will make or break any policy aimed at identifying and rewarding highly effective teachers, exiting low performers, and preparing and supporting new teachers or those struggling somewhere in the middle. To that end, the role of research and advocacy as new approaches mature is critical to shaping the policies that will drive teacher quality.