Abstract
Prior research suggests that elementary school principals assign their strongest teachers to tested grades. As accountability frameworks have softened and principals’ experiences with them have matured, does the pattern still hold? We employ a convergent mixed-methods design to consider, at once, multiple explanations for how school leaders implement teacher assignments by combining data from surveys, interviews, and administrative records from North Carolina. Results reveal a reassignment pattern into second grade, with principals more likely to reassign teachers with lower scores on observation rubrics or value-added ratings to second grade than teachers with higher scores or ratings. Pushing beyond the literature that documents conventional notions of accountability-based staffing, we reveal a more nuanced story about how and why principals assign their teachers within schools.
Keywords
As accountability frameworks have softened and principals’ experiences with them have matured, we wondered whether the pattern still holds. In this study, we employ a convergent mixed-methods design that attempts to consider, at once, multiple explanations for how school leaders implement teacher assignments in elementary schools. The intent of mixed-methods designs is that the results of one method help develop or inform the second (Creswell & Plano Clark, 2018). Specifically, in this paper, we investigate school leaders’ practices to assign teachers and the factors that predict reassignments within elementary schools in North Carolina by combining survey data from elementary principals and assistant principals, interview data with a subset of those survey respondents, and administrative data from the population of North Carolina’s teachers from the state’s Department of Public Instruction.
As we will show, our results reveal a definitive reassignment pattern into Grade 2. Specifically, we find that nearly 6 in 10 elementary principals in the survey responded, “2nd grade” when asked, “If you had to place your least effective teacher in a single grade level, it would be . . .,” with response options ranging from kindergarten to Grade 5. Our analyses of statewide administrative data substantiate this finding, indicating that principals are more likely to reassign teachers with lower scores on classroom observation rubrics or test score value added (VA) ratings than teachers with higher scores or ratings to second grade. Follow-up interviews with a subsample of survey respondents reject any characterization of second grade as a dumping ground for less effective teachers; they describe it instead as a space for professional growth. In fact, emerging from the interview data is a repeated claim by principals that they make the move only when able to pair the less effective teacher with a high performing teacher in second grade.
Pushing beyond the literature that documents conventional notions of strategic staffing in elementary schools, our study investigates a potentially more complicated, nuanced story about how and why principals assign teachers to specific grades. In doing so, it begins to lay the groundwork for future studies that further test whether second grade is a “zone of instructional development,” which we define as an intentional space in which to develop less effective teachers into more effective ones. Focusing on the relationship between the effectiveness of teachers and principals’ decisions to place them, the specific research questions that the study seeks to answer are:
How often do elementary principals reassign teachers? When principals reassign teachers, to and from what grades are they most often moved?
Are more effective teachers, defined here as those who score higher on classroom observation metrics and/or value-added measures, more likely to teach in certain grades? If so, is there a second grade effect? and,
What factors, including the stock of current grade-level teachers, predict whether a teacher will be reassigned to another grade level and to second grade, in particular?
For this study, the term teacher assignment refers to decisions that principals make about how to allocate their key instructional resource—teachers—to grade levels within their school buildings. To date, while an extant set of studies has explored principals’ assignment practices, fewer have investigated reasons for the moves. Our purpose here is not to determine whether patterns in teacher assignments impact teacher turnover or student outcomes; that will be left for other research (Graham et al., 2023). Rather, combining data from a survey, interviews, and state administrative data, we dive deep into understanding why elementary school leaders assign (and reassign) their teachers. We argue that the discovery of a possible “2nd grade effect” has implications not only for its conceptual significance but also for the need for educators to track and protect against any potential learning deceleration during Grade 2.
Additionally, as we address in the discussion section, principals’ decisions to move teachers among grades within a school have implications for “teacher churn,” an increasingly common term used to capture students’ exposure to chronic levels of instructional instability, whether due to teacher turnover, within-school moves, or temporary assignments (e.g., use of long-term subs; see Holme et al., 2018; Redding & Henry, 2018; Ronfeldt et al., 2013). Some well-designed studies have found that minoritized students and students from families below the federal poverty line are more likely to experience teacher churn and that teacher churn has negative impacts on schools and the students they serve (Atteberry et al., 2017; Guin, 2004; Hanushek & Rivkin, 2013; Ost, 2014; Ronfeldt et al., 2013). Equity implications are significant to discuss as well, particularly now as teacher shortages and other cascading effects of the COVID-19 pandemic are disproportionately affecting students living in communities challenged by decades of discrimination and underfunding.
Literature Review
Teacher Assignment Within Schools
Teacher assignment studies in Pre-K–12 education are generally of two types: (a) those that examine the assignment of teachers to schools and (b) those that focus on the assignment of teachers within schools (Cohen-Vogel, Osborne-Lampkin, et al., 2013). Historically, most of the research has focused on the former, finding—with stubborn consistency—that the least experienced and least credentialed teachers are concentrated in schools that predominantly serve Black and Brown students and students living in households with lower-than-average incomes (Clotfelter et al., 2007; Lankford et al., 2002). The findings are especially alarming when coupled with the widely accepted view that teacher effects are the largest school-related influence on student performance 1 (Blazar, 2018; Rivkin et al., 2005; Rowan et al., 2002; Sanders & Horn, 1998; Sanders & River, 1996). The attention by researchers to school-to-school teacher assignment patterns and the resulting gaps in student outcomes has led to policy prescriptions—including selective teacher retention bonuses, strategic recruitment, and pay boosts for teachers in lower-performing schools—that have largely not been implemented at scale. While research to understand scaling continues (Coburn, 2003; Cohen-Vogel et al., 2015, 2016, 2024; Fullan, 2009), evidence of the impact of these policy prescriptions as they have been implemented to date is generally mixed (Ballou et al., 2016; Cowan & Goldhaber, 2018; Donaldson & Johnson, 2010; Pham et al., 2021; Protik et al., 2015), though recent findings have shown that there is both potential in and still much to learn about how we might best optimize the huge bucket of reforms that seek to disrupt the inequitable dispersion of teacher effectiveness among schools (Wyckoff, 2024). Less attention among scholars and policymakers has focused on the processes for and impacts of assigning and reassigning teachers to grades, subjects, and/or classes of students within schools, and especially how teacher effectiveness plays a role.
In 2011, that began to change. That year, two studies, drawing upon ideas from human resource management and personnel economics, documented a phenomenon in education that came to be known as strategic staffing. In the first, Cohen-Vogel (2011) described teacher assignment in Florida elementary schools, uncovering a pattern she called staffing to the test, in which principals weighed teacher preferences and evidence about teacher effectiveness in their staffing and reassignment decisions. Principals reported moving to untested grades teachers whose students did not perform well on the Florida Comprehensive Assessment Test; the pattern was stronger within schools that had earned a low score on the state’s school grading system. Cohen-Vogel attributed the finding to test-based accountability regimes that dominated federal and state reforms in K–12 education. Prior to this time, a few empirical studies in the tracking literature had documented demand- and supply-side pressures that influenced teacher assignment processes (Oakes, 2005; Watanabe, 2008). On the demand side, parents—especially those with time, information, and status—used voice and the threat of exit to press principals to make assignment decisions that favored their children. On the supply side, teachers stated their preferences for assignments to particular students (generally the higher-performing students), subjects (generally more advanced tracks), and grades (Oakes, 1992). With little leverage to reward teachers, principals often acquiesced, a practice that paired high-quality teachers with higher-performing students (Applebee et al., 2003; Donaldson, 2013; Kalogrides et al., 2013).
In the second 2011 study, Chingos and West (2011) followed the career paths of almost 25,000 fourth through eighth grade teachers in Florida for 7 years, beginning in the 2001–02 academic year. Like Cohen-Vogel (2011), they found evidence of strategic staffing, wherein effective teachers, especially those employed in schools receiving low ratings from the state’s school accountability system, were more likely to continue teaching in tested (or what the authors called “high-stakes”) grades and subjects (defined as reading or math in Grades 3–10 or science in Grades 5, 8, or 11) than less effective teachers.
Since then, teacher assignment studies have largely corroborated early results, providing additional empirical evidence that principals assign more effective and highly qualified teachers to tested grades and less effective and less experienced teachers to other grades (Cohen-Vogel et al., 2019; Dieterle et al., 2015; Fuller & Ladd, 2013; Grissom et al., 2017; Henry et al., 2022; Kraft et al., 2020). For example, Fuller and Ladd (2013) found that North Carolina elementary schools under No Child Left Behind (NCLB) were more likely to move lower quality teachers with relatively low Praxis exam scores and experience to untested grades and higher-scoring and more experienced teachers to tested ones (3–5). Kraft et al. (2020) found similar patterns of strategic staffing in one large district in North Carolina between 2002 and 2010; the practices were especially prevalent in schools with low ratings on the state accountability rubric (see also Brummet et al., 2017; Grissom et al., 2015; Henry et al., 2022).
Scholars who focus on within-school assignments have recently expanded their inquiries beyond strategic staffing in two fundamental ways. The first has to do with the frequency of teacher reassignments. From the limited evidence, they show that reassignments or what some have called “grade-switching” are not uncommon and may occur more often for beginning teachers. In 2017, for example, Atteberry et al. found that a quarter of the instructional staff in New York City changed grade-level assignments within their first 2 years of teaching, and more than half switched grades 3 or more times by their 15th year. Working with data from North Carolina, Ost (2014) observed that more than half of teachers switched grades at least once within their first 5 years of teaching, and most reassignments resulted in teacher moves to an adjacent grade.
Second, scholars have shown an increasing interest in assignments that “match” individual students and teachers. Ethnoracial matching, for example, has been studied for its possible impacts on student achievement (Bratsch-Hines et al., 2023; Egalite et al., 2015; Joshi et al., 2018; Redding, 2019), AP course-taking (Kettler & Hurst, 2017), college aspirations (Egalite et al., 2018), executive functioning (Gottfried et al., 2023), in- and out-of-school suspensions (Hughes et al., 2020; Lindsay et al., 2017; Shirrell et al., 2023), and educational attainment (Gershenson et al., 2022). Taken as a whole, this literature suggests that assignment to teachers who share students’ ethnoracial identities improves both academic and social-emotional outcomes for students of color (see also Easton-Brooks, 2019; Gershenson et al., 2021; McArthur & Muhammad, 2020), with benefits spanning early grades (Bratsch-Hines et al., 2023; Gottfried et al., 2023) through high school (Egalite et al., 2018). Another form of matching, equitable rostering, is also being studied (Springer et al., 2022; see also Aucejo et al., 2022). The approach involves intentionally changing the ways teachers are matched to students to ensure that individual students are not exposed to less effective teachers for more than 1 year in a row, based on strong and long-standing evidence that while students can recover from 1 year with an ineffective teachers, their ability to catch up to their peers after 2 or more years drops substantially and approaches zero; no outcome studies of equitable rostering have yet been published.
By combining approaches in a mixed-methods design, our study expands the extant literature surrounding teacher assignment within schools. We use different kinds of data to investigate new questions that arise along the way, and we also attend to micropatterns in the data, including specific grade-to-grade moves and the reservoir of teaching quality in teacher-receiving grades. In doing so, we test an emergent conceptual framework that extends beyond accountability-driven models of strategic assignment in education.
A Concept Model of Teacher Assignment: The Zone of Instructional Development
As education scholars have documented strategic assignment patterns that reflect an accountability logic—and decisions by principals to deploy their most effective teachers to tested grades, in particular—they have also worried about possible unintended negative consequences for the nation’s youngest students. Henry et al. (2022), for example, wrote that their results confirming earlier findings that elementary principals move lower-performing teachers into earlier grades “suggest that accountability-driven school reform can yield negative consequences for younger students that may undermine the success and sustainability of school turnaround efforts” (p. 1). In summarizing a set of studies documenting reassignment behaviors, the National Council on Teaching Quality (2018) similarly concluded that Many principals try to shift their strongest teachers to tested grades and subjects. This seems like a logical approach, but it often involves moving those less effective teachers into the earlier (untested) grades. Given this is when children learn how to read, this strategy can do real damage that is hard to undo in later grades. Moving stronger teachers to tested grades may boost school performance in that year, but the harm to students who do not learn to read as well as they might have otherwise become evident when they reach later grades. (p. 1)
Conceptually, the strategic assignment framework that reflects an accountability logic is grounded in economic theory. Insights from the field of human resource economics suggest that test-based accountability can lead to “perverse incentives” for schools as they seek to boost proficiency rates among students (Cohen-Vogel, 2011; Henry et al., 2022; Osborne-Lampkin & Cohen-Vogel, 2014). Emblematic of the theory, research has shown that district and school officials under accountability pressures may reduce time for recess and untested subjects (Center on Education Policy, 2005, 2008; Matthews, 2007; McMurrer, 2007, 2008); reclassify low-performing students as disabled or Limited English Proficient (Figlio & Getzer, 2002; Haney, 2000; Heilig & Darling-Hammond, 2008); and suspend lower-performing students or push them out altogether (Figlio, 2003; Heilig & Darling-Hammond, 2008). Claims that test-based accountability motivates other undesirable, unforeseen consequences, like less instructional attention to students who score above proficiency thresholds or far below them, have not been unilaterally confirmed, however (Cohen-Vogel & Rutledge, 2009; Ho, 2008; Krieg, 2008, 2011; Lauen & Gaddis, 2012; Midkiff & Cohen-Vogel, 2015). Applied to staffing, economic theory forecasts that principals are likely to use teacher effectiveness data to strategically assign higher-quality teachers to tested grades (in elementary schools, Grades 3–5) and to “hide” ineffective teachers in untested ones because performance in tested grades counts toward school performance designations—designations that often come with “carrot and stick” incentives (Henry et al., 2022).
But, what if the story of within-school teacher assignment is more complicated than that? There are good conceptual reasons (along with some early empirical signals) that accountability-driven strategic staffing may be incomplete in its ability to fully explain principals’ assignment decisions. Researchers who have interviewed school leaders have documented that principals understand the need for good teaching for the youngest school children, as they learn the foundational skills upon which their academic success will ultimately lie. Cohen-Vogel (2011), for example, found that some principals take care to place or retain top-performing teachers in kindergarten and first grade, noting that high-quality teachers are needed where building the fundamentals of reading and numeracy is critical (see also Dynarski et al., 2013). Recent policy movements, such as the proliferation of early literacy laws (e.g., the science of reading), will likely maintain or increase principal recognition of the critical developmental importance of the earliest grades (Cummings et al., 2021; Drake et al., 2023; Little et al., 2024).
As Figure 1 illustrates, our survey findings regarding second grade open the possibility of an extended conceptual framework—one in which principals may try to balance pressures from test-based accountability with efforts to invest in building students’ early foundational skills by targeting Grade 2 as a home for less effective teachers. If confirmed, understanding what goes on in second grade arguably takes on more interest. What if, instead of merely trying to post short-term successes on the accountability rubric, principals are also trying to do something longer term and potentially more meaningful? What if they are distributing their teacher resources in ways that promote their youngest students—those in kindergarten and first grade—to acquire robust foundational skills and concentrate instructional development efforts in Grade 2? In other words, do principals conceptualize second grade as a zone of instructional development, or a space to develop less effective teachers into more effective ones?
A developmental model of teacher assignment.
To find out, research is needed that will track teacher assignments over time not only between tested and untested grades but also from one grade to another and in relation to resourced opportunities for instructional development, like mentorship and instructional coaching, in Grade 2. We begin to meet that need here.
Research Methods
Study Setting
The study is set in North Carolina, primarily because it is where we work and live and because the primary data collection activities began there. The survey was originally designed around a larger purpose; a subset of the authors had fielded it as part of their work to understand principals’ early leadership competencies across the state. As we described above, the remainder of the work took root when we discovered a puzzling report by principals in the survey data of reassignments to second grade. An overlapping subset of authors was working at the time with state administrative data and the idea to combine these data sources and conduct interviews was born.
The policy context in North Carolina also made for a suitable setting. The state has long been known for its accountability system and the use of VA metrics in school and educator evaluation. Indeed, the person credited with pioneering and applying VA modeling to individual teachers was William Sanders, the director of the Value-Added Research and Assessment Center at the University of Tennessee and, later, the senior manager of Educational Value Added Assessment Services at SAS Institute© in North Carolina. Through his research, Sanders developed the EVAAS used today in North Carolina, Tennessee, and three other states. EVAAS measures students’ year-to-year progress against themselves to indicate whether and by how much their learning grew over time as they were exposed to different teachers.
Using the EVAAS system, North Carolina assigns students to one of three growth categories: did not meet expected growth, met expected growth, or exceeded expected growth. Growth levels for a school are based on the average amount of academic growth for all students in that school. Growth is also calculated for individual grades. If the students, on average, grow academically more than expected, then the growth level “exceeds” expected growth and so on. These growth data are used to rate districts and schools and, in some cases, pay educators. Today, the inclusion of growth data from EVAAS is no longer an official part of an individual teacher’s evaluation in North Carolina. The State Board of Education voted to end the use of growth data in teacher evaluations in April 2016. But the following year, in 2017, the state adopted a plan to tie principal pay to student growth. A North Carolina principal’s salary will increase 10% per annum if their school “meets growth” and 20% (or as much as $18,000 a year) if it “exceeds growth.” Apart from principal pay, student growth scores began being used in the calculation of publicly reported school performance grades in 2013 and continue today. School grades in North Carolina are calculated using a formula that includes 80% academic achievement, 2 essentially the proportion of a school’s students who meet a proficiency threshold, and 20% academic growth.
Compared with other states, North Carolina is one of only six states that uses an A through F school grading system. All told, states most commonly rate the performance of individual schools using Descriptive Ratings (13 states), Index Ratings (12 states) and Federal Tiers of Support (14 states; Education Commission of the States, 2024b). To calculate student growth, most states use Student Growth Percentiles (24 states), a metric that indicates how much on a scale from 1 to 99 a student has progressed academically compared to their peers with similar starting achievement levels. Other states calculate growth with value tables, growth-to-standard tabulations, and gain scores. Only eight, according to a report published last year by the Education Commission of the States (2024a), use VA. Among them, North Carolina’s weighting of student growth in its formula is the lowest and its percentages of D and F schools are the highest, despite posting scores on the National Assessment of Educational Progress (NAEP) that outperform most comparison states in the same years (North Carolina State Board of Education and Department of Public Instruction, 2023).
Study Design
In the Spring of 2022, as part of a broader study on principal leadership in early elementary school, we fielded a survey with the population of elementary school principals in North Carolina. The survey focused on early grades leadership competencies and dispositions at a time when the state was increasingly locating publicly funded Pre-K classrooms inside elementary school buildings (Drake et al., 2023). While analyzing the survey data, we were struck by a finding that about 60% of sampled principals reported that they would move their least effective teacher to Grade 2 (if distributed evenly among grade levels, the percentage would be 20%). To investigate, we designed a convergent mixed-methods study to understand whether observed teacher moves captured by the state’s administrative data system confirmed the patterns and, through interviews, the reasons for them. Such approaches allow researchers to combine inductive and deductive thinking and offset limitations associated with exclusively quantitative and qualitative research through a complementary approach that maximizes the strengths of each type of data (Creswell & Plano Clark, 2018; Draucker et al., 2020; Erzberger & Kelle, 2003; Greene, 2007).
The quality of mixed-methods designs hinges on integration strategies, or the processes used by researchers to bring the qualitative and quantitative strands of study “into conversation with each other” (Guetterman et al., 2015; Plano Clark, 2019, p. 108). Purposeful interdependence in mixed-methods research involves linking the data collection or analysis procedures of one strand to the data collection or analysis procedures of the other strand (Bazeley, 2018). Various strategies are used for integration: (a) connecting—data from one strand inform sampling for the other strand; (b) building—data from one strand inform data collection for the other strand; and (c) merging—data from both strands are linked by analysis (Fetters et al., 2013). Following guidelines from the American Psychological Association (Levitt et al., 2018) and the National Institutes of Health Office of Behavioral and Social Sciences (2018), we worked intentionally to employ all three strategies. In connecting, we used data from our survey to develop a stratified random sample for our interviews; specifically, we drew an interview sample from survey respondents who indicated they would assign their least effective teachers to second grade and those who did not. In building, we used the second grade finding in the survey data to amass teacher-by-year variables from the state’s longitudinal administrative dataset to study observed teacher moves. Moreover, as interviews proceeded and new themes in those data emerged, we returned to the administrative data to develop a measure of the stock of teacher effectiveness in second grade. In merging, we developed an analytic coding scheme for the interviews that was informed by findings from the survey and allowed us to pick up information in the interviews that spoke to the reasons that second grade was targeted. As discussed in detail below, the design brings data together from our survey of North Carolina elementary school leaders, interview data with a subset of survey respondents, and teacher level data from the state’s Department of Public Instruction.
Principal Survey
Instrument and Measures
We developed the principal survey from a review of the literature, the collection of existing scales, and feedback from experts in early education and school leadership, including a range of university-based scholars, think tank researchers, and professional organization leaders representing elementary school principals. The survey gathered information on principals’ backgrounds and early education leadership practices as part of a larger study. Information about principals’ backgrounds included gender identity; race/ethnicity; highest education degree; principal certification program; principal preparation location (e.g., North Carolina, out-of-state); years of teaching experience; and years of administrator experience. Following Desimone and LeFloch (2004), we also administered cognitive interviews with four educational leadership faculty and practicing elementary principals to refine the instrument before administration.
For this study, we focused on survey items related to principals’ staffing practices in early grades. Items included, for example, “What percentage of your faculty are currently staffed in the grade that best suits their skill set?” and “How much weight do you typically assign the following factors (e.g., teacher preferences; observation scores) when making a decision about whether to move a teacher from a tested grade (3–5) into an untested grade (K–2)?” on a four-point Likert scale (Not a factor, minor factor, moderate factor, major factor). (Please see Supplemental Material [available in the online version of this article] for survey items.)
Survey Data Collection, Sample, and Analysis
Using Qualtrics, we administered the online survey over 2 months in the Spring of 2022. The survey took an average of 24 minutes for participants to complete and was sent to all lead and assistant public elementary school principals in North Carolina, though our analysis here only uses data from lead principals because they are the ones who received the questions about staffing practices. Working collaboratively with North Carolina’s State Education Agency, we acquired email contacts for the universe of principals in the state and included a letter of state endorsement of the survey in recruitment emails. At the end of data collection, we received completed surveys from 266 lead principals, representing about one in five public elementary schools in the state, or a response rate of 20%. Responding principals came from 80 of North Carolina’s 115 school districts.
Using data from the Common Core of Data, we used t-tests to compare schools with responding principals with schools with principals who did not respond and found our sample to be highly representative of schools in the state (see Supplemental Appendix Table A1 in the online version of the journal article). Of the 14 variables compared—including enrollment, Free and Reduced Priced Lunch (FRPL), and achievement proficiency rates, for example—we found only one statistically significant difference, where our sample of schools had a student population that was, on average, three percentage points less white than non-sample schools. Table 1 reports the background characteristics of responding lead principals. Nearly 70% were female, and about 30% were administrators of color. For 76% of respondents, the highest level of education was a master’s degree. North Carolina requires a Master’s of School Administration or principal add-on certificate for licensure, and nearly all the respondents obtained their certification through these pathways; further, 83% of respondents received their licensure from a North Carolina institution. In terms of administrator experience, nearly 88% of respondents had more than 3 years of experience as a school administrator.
Background Characteristics of Principal Participants and Comparison Group
Note. All data are provided by the North Carolina Department of Public Instruction.
We used basic descriptive methods to analyze the survey data, including descriptive statistics and generating graphical figures to display the relevant survey findings.
Principal Interviews
Sample
At the conclusion of the survey, we asked participants if they would be willing to participate in a follow-up interview. Among the 165 who agreed, we interviewed 27 school leaders using a stratified random sampling approach, ensuring we drew from respondents based on teaching and principal experience as well as whether they reported that they would assign their least effective teachers to second grade, among other factors. Among our participants, the average years of experience was 7.3 years, and over one-third reported they would assign their least effective teachers to second grade. (Demographic information on the qualitative interview sample is included in Table 1 to support findings’ interpretation.)
Instrument
We developed a semi-structured interview protocol with an eye toward mixed-methods integration. We aligned interview questions to survey responses, creating probes to capture more information than we could gather with the survey. For example, we asked: “In the confidential survey, we asked: ‘If you had to place your least effective teacher in a single grade level, [which would it be]?’ You said, [GRADE X]. Could you please tell me why?” Other items on the interview protocol included “Please describe the most common reasons for grade level reassignments while you’ve been an administrator at this school?,” “What measures of teacher effectiveness, if any, do you consider when making grade level assignment decisions?,” and “When you move teachers to different grade levels for effectiveness reasons, please describe any supports provided and follow-up procedures employed?” (Please see Supplemental Material [available in the online version of this article] for the interview protocol). The interviews lasted between 35 and 50 minutes. Each was audio recorded, and the recordings were transcribed. During the interviews, data collectors kept field notes, summarizing both emerging themes and areas for protocol revision. The notes were analyzed alongside the transcribed interview data. Throughout data collection, the research team met regularly to discuss emerging findings from the interviews and determine if modifications to the interview protocol were needed.
Coding and Analysis
We coded the transcribed data and fieldnotes in two rounds that were iteratively informed by findings and questions that arose from our analysis of survey and administrative data. In qualitative research, codes are simply research generated constructs—labels that ascribe meaning to data chunks or units of varying sizes (Vogt et al., 2014). They are, according to Saldaña (2021), “most often a word or short phrase that symbolically assigns a summative, salient, essence-capturing, and/or evocative attribute for a portion of language-based or visual data” (p. 4). We followed our coding activities with an analytic process for the purpose of categorization, pattern detection, and theory testing.
We began with a priori coding, with members of the research team applying three primary constructs: reassignment frequency; contexts for reassignment; and reasons for reassignment. Initially, members of our research team coded a common set of documents to establish inter-rater reliability (Coleman & Multon, 2018) and allow additional codes to emerge from the data. In the second round of coding, we applied new codes and added more descriptive subcodes to the data as we came to understand dimensions within each primary construct. Following Corbin and Strauss (2008), we met regularly to identify disconfirming evidence, refine the coding framework, and discuss emerging patterns and themes. Drilling deeper, we applied codes derived deductively from our emergent framework—for example, accountability-based staffing, defined as the ways principals respond to accountability pressures in their assignment decisions; and, foundational-based staffing, defined as ways principals consider childrens’ early foundational literacy and numeracy skills in their assignment decisions. By allowing other codes to emerge progressively, we identified additional constructs, including instructional development, coaching up, and coaching out.
Following Patton (2023) and Reyes et al. (2024), we organized the coded data in matrices and wrote a consensus analytic memo. As we looked for similarities and differences across cases, the memos helped us to identify patterns in the data, confront outliers in those patterns, and consider the themes and contradictions emerging in principals’ discourse surrounding teacher assignment. Throughout the memo-writing process, our research team continued to meet to discuss key findings and resolve inconsistencies (Corbin & Strauss, 2008). The process revealed an additional construct that sits alongside accountability-based strategic staffing and foundational staffing to explain the second grade effect: “the zone of instructional development.” As will be discussed in greater detail below, the new construct broadens the original explanations for teacher assignment patterns and centers principals’ efforts to target their efforts to develop striving teachers (or to coach them out).
State Administrative Data
Population/Sample
We complement our interview data with administrative data from the North Carolina Department of Public Instruction (NCDPI) to analyze the patterns in teacher assignment across grades in traditional elementary schools and to test the notion that less effective teachers are moved into second grade. As shown in Table 2, our primary analytic sample contains 70,374 teacher-year observations in Grades K to 5 in the 2015 to 2019 academic years. 3 Our unit of observation is at the teacher-year level due to the panel nature of our data—we observe a teacher only once in a given year but can observe the same teacher longitudinally across years if they continued to be present in the North Carolina data in Grades K to 5. We assign teacher grade levels based on the grade reported in the VA estimates in a year in Grades K to 5 only. Thus, a teacher who works outside of this grade range or who did not receive a VA score in a given school year would be missing from our sample as we cannot assign them a grade.
Summary Statistics for Three Teacher Samples
Note. Analytic sample was restricted to traditional elementary schools that include kindergarten and have no students above fifth grade. “First year observation” teachers were entering their second year of teaching and “First year VA” teachers were entering their third year of teaching. These represent the first time principals have official teacher observation scores and test score VA for teachers when considering grade assignments. The “All statewide data” column provides summary statistics for all available teachers and characteristics, regardless of their inclusion in the sample, from 2014 to 2019. EVAAS = Educational Value-Added Assessment System; VA = value added.
In terms of grade level, our focus here is on elementary grades as teacher assignment in later grades differ fundamentally in light of federal and state laws requiring teachers to be certified in the subjects they teach and the departmentalized nature of middle and high schools in which students are taught by numerous teachers during the school day (Cohen-Vogel et al., 2019). As such, our analytic sample is restricted to schools with traditional elementary grade ranges for North Carolina; to be included in the sample, teachers must have worked in a school that included kindergarten and had no students above fifth grade. We further restrict our sample to teachers working in traditional schools, omitting those settings classified by the state as alternative schools, special education schools, or others that focus on specific student populations as these schools are likely to have teacher assignment patterns that differ in practice from traditional public schools.
Table 2 also reports summary descriptive statistics for two subpopulations that we report on in supplementary analyses, including First Year Observation Teachers entering their second year of teaching (and having 1 year of teacher observation data) and First Year VA Teachers entering their third year of teaching (and having 1 year of VA growth data). Teachers entering their second and third years of teaching are of particular interest to our analysis because they represent the first time that principals have official teacher observation scores and test score VA scores for teachers, respectively, when considering grade assignments. Given our hypothesis that accountability pressures and foundational-based concerns about early reading fundamentals may drive the reassignment of less effective teachers to second grade, these analyses represent an opportunity to understand reassignment patterns when official performance data are newly available.
Measures
In Table 3, we summarize the measures used for our quantitative analysis. Below we detail each of these measures more fully.
Summary of Measures Used in the Analysis of State Administrative Data
Note. All data are provided by the North Carolina Department of Public Instruction. EVAAS = Educational Value-Added Assessment System. VA = value added.
Our data include teacher identifiers and information about what grade a teacher worked in for a given school year. We derive the grade a teacher worked in for a given year using the grade level of their state-reported VA score. Tracking changes in teachers’ grade levels over time thus requires that the teacher had reported VA scores from Grades K to 5 in multiple years. We also construct an indicator variable for whether a teacher (a) changed grades or (b) moved to second grade in a given year. We define a teacher as changing grades if their grade was observed in both the current and prior year in Grades K to 5, and their current year grade was not equal to the prior year grade. For teachers moving to second grade, teachers who worked in Grades K to 1 or 3 to 5 in the prior academic year and were reassigned to Grade 2 take the value of “1.” Teachers who worked in Grades K to 1 or 3 to 5 but who were not reassigned to Grade 2 take the value of “0.” Teachers who were already working in second grade in the prior school year (year t − 1) or first-year beginning teachers, whose grade assignment ipso facto cannot be a reassignment, are set to “missing” on this variable, as there is no possibility that they could be reassigned to second grade.
We complement these data with information on teacher race/ethnicity and sex, prior-year observation scores via the North Carolina Educator Evaluation System, 2-year-prior standardized English Language Arts (ELA) VA estimates of teacher effectiveness as calculated by the SAS Institute’s EVAAS, teacher years of experience, and a once lagged indicator for if a teacher held a National Board Certification.
We use two measures of teacher effectiveness available in the administrative data: scores assigned to teachers by principals or their designees during scheduled observations and teacher VA. We use both measures because teaching is a complex, multidimensional, context-sensitive, and dynamic enterprise and, as such, assessing teacher quality or effectiveness poses significant research challenges and is enhanced with the use of multiple measures (Darling-Hammond et al., 2012; Shulman, 1987; Youngs & Grissom, 2015). While both observation and VA scores have been shown to predict teacher performance (Bacher-Hicks et al., 2019; Kane et al., 2013), we treat them separately in our models because previous studies have shown that—despite state policy models that do so—combining them into a single index threatens validity. In fact, work by Martínez et al. (2016) affirms earlier studies that multiple measures are necessary for adequate coverage and validity in teacher evaluation, but that these measures need not (and in fact should not) be combined into summary indicators, but instead should be used in combination (Mehrens, 1989; Schmidt & Kaplan, 1971). Additionally, prior research shows that principals appear to rely on and use observation and VA measures differently. Some principals have demonstrated more formative uses of observation data, relative to VA uses. Goldring et al. (2015) found, for example, that VA measures were playing less of an exclusive role in principals’ decisions to hire teachers, renew their contracts, and assign them to classrooms as rigorous observation-focused evaluation systems developed. While we treat observation and VA scores here separately, future research might consider combining them into composite indicators that best reflect the state model (i.e., conjunctive, disjunctive [or complementary], and weighted [or compensatory]) in which the study of teacher reassignment is being conducted (Martínez et al., 2016).
Our measure of teacher observation scores ranges from 1 through 5 and represents the median score 4 earned by a teacher across all five North Carolina observation standards on which a teacher was assessed in a given year. Teachers with missing prior year observation scores are assigned a “missing” flag, allowing their inclusion in the regression analysis. The missingness may occur due to a lack of data or because a teacher previously worked in a grade outside of the K–5 range. Because teachers at all experience levels so rarely receive a median rating of “1” (0.1% of observations) we combine teachers rated as 1 or 2 into a single observation score. The 1-year lag (n − 1) on observation scores and National Board Certification captures the data available to principals when making grade assignment decisions in the current year.
In a VA model like EVAAS, if students outperform their predicted scores, the students’ teacher is credited with having an instructional effect above what was expected (i.e., the teacher “added value”). For our VA measure, we rely on EVAAS effectiveness categories, which take the value of “Above,” “Meets,” or “Below” expected growth. These categories summarize overall teacher performance and offer clear performance signals of effectiveness to principals who make grade assignment decisions. 5 Teachers with missing VA scores are assigned a “missing” flag so as to include them in the regression analysis. We use reading test score VA as it is available in all elementary grades starting in the 2014 school year, allowing us to examine teacher movement by EVAAS effectiveness across a wider set of grades than math VA would allow. In North Carolina, VA metrics from Spring standardized tests are typically not released until the subsequent fall. Thus, the twice lagged variable (n − 2) reflects the data available to principals when making rostering decisions in their current year. We observe substantial missingness in these VA metrics. Again, missingness could stem from a true lack of data or be due to teachers coming from grade levels outside of K through fifth.
As shown in Table 2, teachers in our sample 6 vary considerably in their observation scores and VA ratings. We see a modal observation score of 4, followed closely by observation scores of 3, and then relatively rarer scores of 5 or the combined 1 and 2 score. Likewise, the modal VA category is “meets expected growth,” with “above” and “below” expected growth being relatively infrequent. This distribution of teacher effectiveness suggests that principals are generating and/or receiving differentiating information about teachers at the top and bottom of the performance distribution that could conceivably be actionable when making grade assignment decisions.
We also measure school-level performance because prior research indicates strategic staffing may be more common in lower-performing schools (Cohen-Vogel, 2011; Henry et al., 2022). Specifically, our models measure the percentage of students rated as proficient on standardized tests at the school level in a given year. This measure is scaled to range from 0 to 100. We also generate two within-school second grade level flags, which indicate that a school had a second grade teacher in the year prior who was highly effective, defined as “Above” expected growth by the state or with a median observation standard score of 5 out of a possible 5 in a given grade level, who continued to work in second grade the following year. We refer to these cases as “highly effective peer teachers” and use this measure to test if the presence of a highly effective peer teacher in second grade modifies the relationship between an individual teacher’s effectiveness and the likelihood a teacher moves to the grade.
Analysis
We employ the following probit model to estimate the relationship between teacher characteristics and the likelihood of moving to second grade
where Yit represents the likelihood of teacher i in year t moving to second grade. A move to second grade is defined as a teacher working in Grades K–1 or 3–5 in our data in the previous year and switching to Grade 2 in the current year. Our independent variables of interest are measures of teacher effectiveness, like twice lagged test score VA (
To test whether having a strong potential mentor in the second grade would create greater incentives for principals to move a low performing teacher to second grade, we interact twice lagged test score value and once lagged observation scores with indicators for a highly effective peer teacher being present in second grade according to VA score (Xit) or observation score, (Yit) respectively. These separate variables take a value of 1 when the peer teacher was rated as having “above” expected growth in VA or had a median observation score of 5 across all available standards and said teacher was present in second grade in both year t − 1 and school year t.
Findings
How Frequently Are Elementary Teachers Assigned to New Grades? And, When Principals Reassign Teachers From Their Original Grade Placements, To and From What Grades Are They Most Often Moved?
Prior evidence has shown that, with few exceptions, 7 elementary school principals assign their teachers to a single grade level (rather than multiple grade levels) in a given year. Principals assign a teacher to a grade level when a teacher first arrives at the school (as a newly hired teacher or teacher transfer) and again in subsequent years as “reassignments.” Every year, principals decide to keep a teacher in the same grade level or make a grade-to-grade reassignment. No laws in North Carolina restrict principal assignment decisions, including the frequency with which they can move a single teacher between grades. Despite this freedom, there are good reasons to expect principals to use reassignments sparingly. Grade level changes come with disruptions; for example, teachers must learn (or re-learn) grade-level content standards and prepare new instructional materials and assessments.
In our survey, we asked North Carolina principals to indicate the percentage of teachers in their schools currently teaching in the grade that best suits their skills. In short, we wanted to know how well principals think they are doing when it comes to teacher-grade level matching. About 65% of principals reported that three-quarters or more of their faculty are well placed grade-level-wise. Approximately 30% reported that between 51% and 75% of their faculty members are teaching in the “right” grade, and about 5% think less than half are well matched to a grade level.
The longitudinal administrative data can tell us precisely how often principals move teachers from grade to grade. In any given year in our data, approximately 17% of the full teacher sample worked in a different elementary grade than they did in the prior year. For beginning teachers (entering their second year of teaching), the rate was between 20% and 23% in any given year of the sample. This level of reassignment—compounded across years—is significant. Of the 1,782 first year teachers in our data for 2014, for example, 42% experienced at least one grade change in the subsequent 5 years—a similar rate to the one documented by Ost (2014) who showed, in an earlier period in North Carolina, that about half of public school teachers in the state switched grades at least once within their first 5 years of teaching. Our inability here to differentiate moves to grade levels above fifth grade from teacher moves to other states or teacher attrition from the profession may explain our slightly lower values.
Table 4 provides descriptive information about the grade levels to and from which teachers are most often reassigned. Overall, the leaving columns in the table suggest that principals are less likely to reassign teachers away from kindergarten, compared with other grades. That is, kindergarten teachers appear to be the least likely to be reassigned to another elementary grade. This is true of the full sample as well as the two teacher subsamples. Specifically, only 12% of kindergarten teacher-year observations in the full sample were reassigned to another elementary grade level during the 5 year period for which data were analyzed, compared to between 16% and 21% of first through fifth grade teachers. The leaving columns also suggest that reassignments are more likely to move teachers away from fourth grade. For two of the three samples, fourth grade teachers appear to be the most likely to be reassigned to another elementary grade. For the third sample, fourth and fifth grade teachers are most likely to be reassigned. Specifically, a full 27% of beginning teachers with 1 year of observation data—nearly one in three—are likely to be moved out of fourth and fifth grades, compared to 18% to 23% of their K through Grade 3 counterparts. Overall, greater churn seems to characterize higher elementary grades.
Share of Teachers That Leave a Grade or Arrive to a Grade Due to Reassignment
Note. Percentages are based on (1) a subsample of “First Year Observation” teachers entering their second year, (2) a subsample of “First Year VA” teachers entering their third year, and (3) all teachers in the analytic sample. The subsamples represent principals’ first opportunities to access teacher observation scores and value-added measures when making grade reassignment decisions, respectively. Columns a, c, and e report the share of teachers that are moved out of the corresponding grade levels across all the years of our panel. Columns b, d, and f report the percentage of teachers that are moved into the corresponding grade levels across all years in our panel. A teacher is defined as reassigned if their grade was observed in both the current and prior year, and their current year grade was not equal to the prior year grade. VA = value added.
To which grades are teachers most often reassigned? Overall, the arriving columns in Table 4 indicate that teachers appear somewhat more likely to be moved into second and fourth grades than other grade levels. Beginning teachers with 1 year of observation data appear most likely to be moved into second grade, in particular. While 26% of them are reassigned to teach second grade, between 17% and 23% are reassigned to other elementary grade levels.
What Is the Relationship Between Teacher Effectiveness and Grade-Level Assignments Among Teachers?
Having established that grade-to-grade reassignments overall are not uncommon and that reassignment patterns seem to affect some grade levels more than others, we considered the relationship between teacher effectiveness and teachers’ grade-level assignments and whether there is evidence of accountability-focused staffing. As with prior studies, we find evidence that accountability logics inform principals’ assignment decisions. In short, we find that less effective teachers in tested grades (3–5), measured either with observation ratings or EVAAS scores, are significantly more likely to be reassigned to a non-tested grade (K–2) than their more effective counterparts.
The pattern is easy to spot among first year teachers, for whom moves are more likely. Figure 2 displays findings graphically. The percentages at the top of the figure represent the overall rates by which beginning teachers returning for their second year of teaching are reassigned to another elementary grade level. Median teacher observation scores by grade taught during the first year of teaching are arrayed along the X axis. Along the Y axis are the proportions of teachers who teach in each grade during their second year. The figure has six sets of histograms; each set of three represents one grade level, kindergarten through fifth grade. For example, the first set at the far left represents the population of teachers in North Carolina who taught in a kindergarten classroom during their second year of teaching. The first bar in the set represents those teachers who, during their first teaching year, scored the lowest (1 or 2) on the observation rubric. The second bar represents those who scored in the middle (3) and the third represents teachers who scored highest (4 or 5) on the rubric, meaning they had the highest classroom observation scores.
Grade level reassignments of beginning teachers by observation rating.
Five overall patterns emerge from the data displayed in Figure 2. First, paying attention to the light gray areas of the figure, we see that, in every grade (K–5), less effective first year teachers (with observation scores of 1 or 2) are more likely than their more effective counterparts to leave the purview of our data by year 2. These exits from our sample include leaving the public teaching force in North Carolina, and possibly the teaching profession altogether, as well as teachers who move to grades outside the kindergarten through Grade 5 span. Across all these sample attrition types, more than 25% of beginning teachers with low effectiveness ratings leave our sample before their second year. This is consistent with prior studies that show that less effective first year teachers are more likely to be coached out and/or leave teaching than their more effective counterparts (Boyd et al., 2008; Henry et al., 2011).
Second, we notice that teachers who remain teaching for a second year are more likely to experience a stable assignment—remaining in the same grade level—than they are to be reassigned to another grade for their second year of teaching (regardless of their effectiveness rating). Overall, 77% of teachers who had non-missing grade assignments in year 2 were assigned to the same grade as they were in year 1.
Third, we see that less effective first year teachers (with observation ratings of 1 or 2) are more likely than more effective teachers to be reassigned to another grade, regardless of the grade they teach in their first year. For example, 32% of first year fifth grade teachers with observation ratings of 1 or 2 were reassigned to a different grade the following year, while only 19% and 14% of teachers with ratings of 3 and 4 or 5, respectively, were reassigned. This pattern holds for every elementary grade level except second grade, where 20% of beginning teachers with observation scores of both 1 or 2 and 3 were reassigned.
Fourth, the proportion of less effective first year teachers who are reassigned to a different grade for a second year increases as grade levels increase. For example, in Grades K to 1 and 2, 19% and 20% of less effective teachers (with observation scores of 1 or 2), respectively, were reassigned to a different elementary grade, whereas less effective teachers working in Grade 3 to 5 were reassigned to a new grade at rates of between 29% and 42%. By fifth grade, less than 20% of beginning teachers with ratings of 1 or 2 remain teaching in fifth grade by their second year of teaching due to attrition or reassignment.
Fifth, we note that first year teachers who are reassigned for their second year of teaching are more likely to be moved to an adjacent grade than to non-adjacent grades. There is an important exception to the early grades pattern, however. The exception occurs in fourth and fifth grades—tested grades that are part of school accountability systems. Here, fourth grade teachers who earn a low observation rating (1 or 2) in their first year are more likely to be reassigned to second grade (non-adjacent) than third grade (adjacent) and not at all likely to be reassigned to fifth grade (adjacent). Similarly, fifth grade teachers are about as likely to be moved to second grade (non-adjacent) as they are to fourth grade (adjacent).
Is there a Second Grade Effect in the Teacher Effectiveness-Assignment Relationship?
To probe for possible accountability- and foundational skills-based logics in principals’ reassignment decisions, we included items in the survey designed to capture principals’ sense-making about the role teacher effectiveness may play in their assignment decisions. One item asked principals to answer, “If you had to place your least effective teacher in a single grade level, it would be . . .,” with response options that included kindergarten through fifth grade. As illustrated in Figure 3, nearly 60% of principals indicated a single grade: 2nd. When we analyzed whether principal survey responses to this item differed by school or principal characteristics, we found only one: Principals who have taken at least one child development course in their preparation program are 14 percentage points less likely to report that they would place their least effective teacher in Grades K, 1, or 2.
Principal reported assignment of their least effective teacher.
Intrigued by the overwhelming “2nd grade” response to the survey and understanding it was hypothetical, we returned to the state administrative data. Employing three teacher samples, the average marginal effects of our probit models tell a similar second grade story. Turning to Table 5, we see that teachers with low EVAAS and observation scores are more likely to be reassigned to second grade than observationally better teachers, net of teacher experience, board certification, gender, and race/ethnicity. 8 Overall, approximately 5% of teacher observations in our full teacher sample in the regression are reassigned to second grade in any given year between school years 2014 and 2019. In terms of EVAAS scores, the average marginal effects of our preferred full sample model in Table 5 Column 3 show that the percent probability of being reassigned to second grade is significantly higher for teachers rated as “below” expected growth relative to those rated “above.” The coefficient (0.0112 or 1.12 percentage points) is substantively large, representing ~20% of the overall 5% base level of teacher reassignments to second grade. The difference is not statistically significant for the beginning teacher subsample reported in Column 2.
Average Marginal Effects Predicting the Likelihood of Reassignment to Second Grade
Note. Standard errors in parentheses. Reports results for Model 1 for the analytic sample and subsamples of second and third years teachers. The interaction between highly effective teachers in second grade and teacher effectiveness measures are included in this model, but effects are only reported in Table 8, as interactions in non-linear models are more accurately interpreted as fixed values (Ai & Norton, 2003). Standard errors are clustered at the teacher level. We report the average marginal effect for independent variables on the likelihood of a teacher moving to Grade 2 in a given school year. Teacher demographic controls include race and sex. Coefficients for VA category, observation rating, and prior-year grade are interpretable relative to the listed “Omitted” level. Prior-year grade omits second grade as teachers in second grade in year t − 1 cannot move to second grade—thus we choose an omitted level of third grade. EVAAS = Educational Value-Added Assessment System; VA = value added.
p < .05. **p < .01. ***p < .001.
For teacher observation scores, we find a stronger pattern. Relative to teachers with a median observation score of 4, teachers who are rated 1 or 2 or 3 are significantly more likely to be reassigned to second grade. In our full sample reported in Column 3, relative to teachers with a median teacher observation rating of 4, being rated 1 or 2 in the prior school year is associated with a substantively large 6.16 percentage point increase in the likelihood of moving to second grade on average. This association is even larger in our samples of earlier career teachers in Columns 1 and 2. We also observe a similar overall pattern when comparing teachers rated as 3 in observation rubrics relative to those rated as 4, with coefficients being expectedly smaller in magnitude, but still large relative to the baseline likelihood of a teacher moving to second grade. The consistency of this finding across models and the relatively larger coefficients for low observation scores relative to low VA measures reflect prior research showing that principals tend to rely more heavily on scores from observation metrics than VA models in a host of school-level decisions (Donaldson & Johnson, 2010; Fox, 2016; Goldring et al., 2015).
Relatedly, Table 5 shows that less experienced teachers are significantly more likely than more experienced teachers to be moved to second grade, though there is no statistically significant relationship with National Board Certification. Lastly, we observe that principals are more likely to reassign teachers to second grade if a lower percentage of students in their schools are rated as proficient, with a one standard deviation decrease in proficiency rates being associated with a 0.364 percentage point increase in the likelihood of reassignment to second grade.
We expect that lower teacher effectiveness in the grades which contribute to school accountability measures (3, 4, and 5) will more strongly predict a move to second grade than lower effectiveness in nonaccountability grades (K and 1). We test this notion using the probit model’s average marginal effects reported in Table 6. 9 Column 1 of the table reports the same coefficients for the full sample as in Table 5, Column 3. In Table 6, Column 2, we run Model 1 on the subset of teachers who worked in the three grade levels for which test scores are used for accountability purposes (Grades 3–5) in elementary school in year 1. And, in Column 3 of Table 6, we run Model 1 on a subsample of teachers who worked in non-accountability grades (K–1). As we hypothesized, the association between teacher effectiveness and the likelihood that principals reassign teachers to second grade is stronger in accountability grades. For example, having a prior-year observation score of 3, relative to 4, is associated with a 3.38 percentage point increase in the likelihood of Grade 2 reassignment for teachers in accountability grades, a difference of about four times as large as the 0.847 percentage point difference observed in the model run on non-accountability grade teachers.
Average Marginal Effects Predicting the Likelihood of Reassignment to Second Grade for Three Grade-level Teacher Samples
Note. Standard errors in parentheses. Results for Model 1 for the analytic sample and subsamples of teachers working in accountability grades (3–5) and non-accountability grades (K–1). Standard errors are clustered at the teacher level. The interaction between highly effective teachers in second grade and teacher effectiveness measures are included in this model, but effects are only reported in Table 8, as interactions in non-linear models are more accurately interpreted at fixed values (Ai & Norton, 2003). We report the average marginal effect for independent variables on the likelihood of a teacher moving to second grade in a given school year. Teacher demographic controls include race and sex. Coefficients for value-added category, observation rating, and prior-year grade are interpretable relative to the listed “omitted” level. Prior-year grade omits second grade as second grade teachers in year t − 1 cannot move to second grade. The omitted is third grade. EVAAS = Educational Value-Added Assessment System.
p < .05. **p < .01. ***p < .001.
Referring back again to Figure 2, the pattern can be observed graphically for the first-year teacher subpopulation. Specifically, we can see that among first year teachers of tested grades who are reassigned for their second year, those with lower observation scores (1 or 2) are more likely than their more effective counterparts to be moved to second grade, in particular. Indeed, teachers with observation scores of 1 or 2 who taught in third, fourth, or fifth grade in year 1, are more likely to be moved to second grade over any other grade level. At the same time, the most effective fourth and fifth grade teachers (who scored a 4 or 5 on the observation rubric their first year) are more likely to be reassigned to any grade other than Grade 2. This suggests there is more selective movement to second grade overall and for teachers who start teaching in tested grades (3–5) relative to those who start in K and 1.
What Factors Do Principals Consider When Making a Decision About Whether and Where to Reassign a Teacher?
So, what factors do principals weigh when making assignment decisions and how might these factors explain the observed second grade pattern? Informed by prior research, we expected accountability pressures surrounding tested grades to be a driving factor. In North Carolina, school grades are calculated using average and growth scores on state tests in Grades 3 to 5; performance metrics in Grades K to 2 are not considered. Our survey of principals asked respondents how much weight they typically give to a set of factors when deciding whether to move a teacher from a tested grade (3–5) into an untested grade (K–2)?
As reported in Figure 4, over 60% of principals say they use teachers’ EVAAS scores as a “major” or “moderate” factor in these decisions but not substantively more than they report using teacher observation scores, teacher preferences, and prior experience in early grades. Indeed, principals—on average—report that they weigh two other factors more heavily and the last factor more significantly so: a desire to place their best teachers in classrooms with the greatest student need. We do not find the responses from principals who report that they would assign their least effective teacher to second grade to differ substantially from those who say they would assign their least effective teacher to another grade in terms of the weight they assign to the factors.
Weights that principals report giving to factors in decisions to reassign teachers from tested grades to untested grades.
So, while we find that second grade is the normative elementary grade for less effective teachers, we also have evidence that factors beyond those used to calculate school grades on the accountability rubric inform principal assignment decisions. With this in mind, we developed our interview guide to help build a more comprehensive understanding of principals’ reasoning. Findings from the qualitative interviews reveal that principals understand the value of grade-level stability for promoting teacher expertise in a particular grade level, avoiding the extra time involved in training teachers to deliver new content and develop instructional and assessment materials, and providing students with well-prepared teachers. Indeed, interviewed principals repeatedly said that reassignment (to any grade) is not a preferred strategy when they face a less effective teacher. They told interviewers that they are as likely to provide coaching in the stable grade or coaching out of teaching as they are to make grade-level shifts. In the words of one principal, for example, You really can’t hide a bad teacher anymore . . . I think really, the first and foremost thing to do is not necessarily change [the assignment], but to coach them through. To make sure they have great strategies and to provide the mentorship that they need. I think you’ve got to do the legwork no matter what grade they’re in. And then if they can’t make the change, I think you’ve got to document! I think that’s the priority. If you’ve got a less effective teacher and if they’re going to remain less effective, you’ve got three standardized tests that give you some hardcore data to say, “You probably need to find a different profession.” I don’t think a less effective teachers should be moved to another grade, I think they’ve got to be coached up or moved out.
When principals do make reassignments, they report that they do so primarily in response to enrollment bubbles or bursts in particular grades. When enrollment pressures arise and reassignments are needed, the principals we interviewed report taking a host of factors into account, just as the larger principal sample does on the survey. They describe a holistic approach. In terms of teacher performance, they report considering what they glean from walkthroughs, classroom observation scores, and student growth (EVAAS).
In the interviews, principals’ responses indicate that they feel a tension related to their reassignment decisions between the tested grades and early foundational skills in K and 1. One said she would place her least effective teachers in second grade because Kindergarten and 1st grade to me are the most critical grade levels so I definitely need my strongest teachers there. Third grade is when expectations really start to elevate and there’s a lot of testing in 3rd grade and your content becomes more difficult to understand . . .. If the teacher doesn’t understand the content, how can they teach it to a student? . . . 2nd grade is missing the foundational “have to have this,” and even though in K and [Grade] 1 the concepts aren’t tricky, it’s understanding a child’s brain and how they learn and how very different it is at that age than the upper grades. So, 2nd grade is what I found to be best for least effective teachers—the least damaging, if I can say it that way.
Another remarked, I think the early years are very formative and you need a strong teacher to understand development. Because if they have those structures, then they [the children] are going to be okay in the future . . . You have to be effective as a 1st grade teacher or Kindergarten teacher. So that’s quite a dilemma. And, 3rd grade is a testing grade. So, 2nd might be safest. I think especially since LETRS [NC’s new reading program], you know, and a lot of us are actually going through the [professional development] program with the teachers, and we’re realizing how crucial it is to support students reading in their print concepts early on and their number knowledge, because if they don’t have these foundational skills, they’re just struggling.
In interviews, participants denied that second grade is a stagnant space for less effective teachers. Instead, many principals spoke in detail about reassigning less effective teachers so as to deploy resources to further develop them. For example, the interviewed principals spoke about efforts they made to focus the time of their instructional coaches on less effective teachers, and to pair them up with strong teachers who can serve as peer mentors and share resources. “The biggest thing, whenever you move someone,” according to one principal participant, “is to make sure that you’re moving them to a space where they can be comfortable, and they feel supported. We moved a 4th grade teacher down to second grade, and she was overwhelmed from the start . . . She was teaching second graders like they were 4th graders, and that just doesn’t work.” Echoing the sentiment that factors that principals consider in their reassignment decisions include the stock of the receiving grade teachers, a principal from another part of the state remarked: When we move for effectiveness reasons, we’re making sure that we’re going to move them to a grade level that has a strong person on that team that’s going to partner up with them and work together with lesson planning and implementation and classroom management. We are also going to make sure we have those extra admin walks in the classroom . . . We also have district level ELA and math and science specialists that we will get to come out and work with those teachers to provide that extra professional development, with the, you know, that curriculum piece.
From another participant, we heard: “Because of our enrollment, we’ve had to shift a lot of teachers across grade levels . . . I want to place them with a team of teachers that are strong to help guide them. That way, they would have a partner—somebody else besides the leadership—to come in to give feedback. We also have academic coaches to go in to assist teachers.”
All of this made us wonder if principals target second grade as a grade to which to reassign less effective teachers to help develop them into more effective ones. Instead of trying to post short-term successes on the accountability rubric, as strategic staffing posits, perhaps principals are taking a longer, potentially more meaningful view of second grade as a zone for instructional development?
To answer, we returned to the longitudinal administrative data. First, if as we are positing that there is something unique about second grade that makes it appealing for principals to place teachers there who need development, then we should ensure that we are not seeing the same pattern when analyzing the movement to other grades. To test this, we ran robustness checks, estimating 10 the likelihood of (a) moving to Grade 2, (b) moving to Grade 1, and (c) moving to kindergarten from test-based accountability grades (3–5). Overall, as displayed in Table 7, the results are supportive of a second grade effect. Findings demonstrate that the relationships between VA category and the movement of teachers previously in Grades 3 through 5 are significant for second grade only, with a “below” expected growth score associated with a 2.56 percentage point increase in the likelihood of moving to second grade, and null effects for kindergarten and first grades. Low observation scores in Grades 3 through 5 predict movement to all untested grades; however, the size of the coefficients for second grade are much larger. Interpreting the statistic, we see that an observation score of 1 or 2 is associated with a 12.2 percentage point increase in the likelihood of moving to Grade 2 compared to 2.17 and 3.89 percentage points for first grade and kindergarten, respectively. Also confirmation is that moves to second grade from tested grades is significantly less likely in schools with higher overall proficiency rates whereas there is no such association for moves to first grade or kindergarten.
Robustness Check on Movement to Untested Grades for Teachers in Accountability Grades.
Note. Standard errors are in parentheses. Results for a version of Model 1 with both Highly Effective Peer Observation and VA variables and their interactions were removed due to missingness within interaction cells. This modified model is run on the same subsample of teachers working in test-based accountability grades (3–5). Each model predicts the likelihood of moving to a different nonaccountability grade. Standard errors are clustered at the teacher level. We report the average marginal effect for independent variables on the likelihood of a teacher moving to a given non-accountability grade in a given school year. Teacher demographic controls include race and sex. Coefficients for value-added category, observation rating, and prior-year grade are interpretable relative to the listed “omitted” level.
p < .05. **p < .01. ***p < .001.
Second, while the North Carolina administrative do not provide a way for us to analyze the amount of time or attention instructional coaches or district curriculum specialists spend with teachers by grade level, we are able to use the data to test whether principals reassign less effective teachers to second grade only when a highly effective teacher is there to mentor them. Contrary to principals’ claims, our analysis in Table 5 shows that having at least one highly effective peer teacher (with an observation rating of 5 or an EVAAS score of “above” expected growth) among the current stock of second grade educators does not significantly increase the likelihood of other teachers—regardless of their effectiveness—being reassigned to second grade. Rather, analysis suggests that the presence of a teacher who earned top scores on the evaluation rubric or a high EVAAS teacher being present in second grade decreased the probability that a teacher is reassigned to second grade on average, with all other covariates held constant.
We ran additional analyses to analyze principals’ assertions that they work to ensure that less effective teachers, in particular, are moved to second grade only when effective teachers are there to mentor them. Specifically, we wanted to understand whether the interacted relationship between the presence of a highly effective peer teacher in second grade and the likelihood that other teachers will be reassigned to second grade is stronger when the teachers being reassigned are less effective (see Table 8). Running post-estimation conditional marginal effects on the interaction between the presence of highly effective peer teachers in second grade and individual teacher effectiveness, however, we found that the average marginal effect that the presence of a highly effective peer teacher in second grade has on the likelihood of other teachers moving to second is not significant when the reassigned teachers have lower observation scores. Additionally, as shown in Table 8 Column 2, we found a negative relationship when the reassigned teachers’ efficacy was measured by EVAAS scores. Specifically, the presence of a highly effective peer teacher actually decreased the average likelihood that a teacher with an EVAAS score of “Below,” signaling that her students in year one underperformed their predicted scores, would be reassigned to second grade. In fact, the presence of a highly effective peer teacher in Grade 2 seems to decrease the likelihood of teachers anywhere on the EVAAS effectiveness distribution from being moved to second grade. This pattern does not align with a hypothesis that principals are more likely to reassign less effective teachers to second grade when there is a highly effective peer teacher present who can serve as an informal “coach” and, therefore, where instructional development is possible.
Post-Estimation Conditional Marginal Effects: Interacting Teacher Effectiveness With the Presence of a Highly Effective Peer Teacher in Second Grade
Note. Standard errors are in parentheses. Table reports the marginal effects of the interaction run in Model 1, which was reported as average marginal effects for the full sample of teachers in Table 5, Column 3. Coefficients represent the interaction effect of having a highly effective teacher present in the second grade and various levels of teacher effectiveness across both observation scores and test score value added.
p < .05. **p < .01. ***p < .001.
Discussion
Results from our convergent mixed-methods study show that about one in every five returning elementary teachers in North Carolina is reassigned to a new grade annually, with the reassignment rate even higher for beginning teachers who return to the classroom after their first full year of teaching. Principals’ decisions to reassign a sizable proportion of teachers to new grade levels each year have consequences for teachers and students alike. These moves, referred to by some researchers as “churn,” translate into the need for additional investments in time and resources to retrain returning teachers in terms of grade-level and age-appropriate content standards, child development, effective engagement, and classroom management strategies. They also mean that students in reassigned teacher classrooms are exposed to inexperienced grade-level instruction (Blazar, 2015).
Among teachers who are reassigned, we continue to see some evidence for the kind of accountability-driven strategic assignment that was first detected in the No Child Left Behind era. Like earlier studies conducted in Florida, North Carolina, and Michigan (with value table, VA, and student growth percentile calculation approaches, respectively), our results show that elementary principals in North Carolina reassign less effective teachers at higher rates than more effective teachers and that larger proportions of less effective beginning and continuing teachers are reassigned out of tested grades, and fourth and fifth grades, in particular.
A Second Grade Effect
New to the literature, however, we also find that elementary principals in North Carolina appear to be reassigning less effective teachers disproportionately into Grade 2. This possible “2nd grade effect,” as we refer to it, shows up in all three datasets analyzed in the study—data from the survey, administrative records, and principal interviews. It manifests in the survey data when the vast majority of principals report that they would reassign their least effective teacher to second grade out of a possible six grade choices. It shows up in the administrative records when principals grade-retain (do not reassign) less effective beginning first and second grade teachers at higher rates than they do less effective beginning teachers of all other elementary grades. It shows up in the administrative records again when larger proportions of less effective elementary teachers overall are more likely than their more effective counterparts—measured by either observation scores or VA—to be moved to second grade, and when beginning teachers with observation scores of 1 or 2 who taught in fourth or fifth grade during their first year are more likely to be moved to second grade over any other grade level. The pattern holds for more experienced teachers as well, where reassignments to second grade are significantly more likely for teachers with lower EVAAS and observation scores. Finally, we observe it in the interview data when principals appear to argue that a less effective teacher may do the least “damage” in second grade.
The second grade effect appears to be attributable, at least in part, to a balance that principals attempt to strike. From their own accounts in our interviews, principals seem to be deploying effective teachers in tested grades as well as kindergarten and first grade, where foundational reading skills—phonemic awareness, sight word recognition, and decoding—are primarily emphasized.
Zone of Professional Development
Our interviews with principals also reject any characterization of second grade as a dumping ground for less effective teachers, and in doing so suggest instead that they see it as a zone for professional learning and development. Despite the claim, however, we are unable with the North Carolina administrative data to confirm principal reports that they are more likely to make reassignments to second grade when a high performing peer teacher is already teaching there. Specifically, our analysis shows that reassignments of less effective teachers into second grade do not appear to occur more often when a highly effective peer teacher is on the second grade team. So, do principals target second grade as a zone for instructional development? The evidence is inconclusive.
We say it is inconclusive because limitations associated with our study prohibit a more thorough investigation at this time. Specifically, we are unable to measure all possible sources of instructional support that a principal may deploy within this zone. While our administrative data permit us to estimate the proximity to a highly effective peer teacher (and, therefore, a potential source of professional support), they do not provide information on exposure to instructional coaches or others who provide teacher feedback and mentorship. Put differently, we cannot evaluate principal claims that they deploy instructional coaches to second grade classrooms or to the classrooms of grade-reassigned teachers, more generally. Studies show that instructional coaches can significantly improve classroom practice, especially when part of a coherent system that links evaluation with instructional improvement aligned to ambitious standards (Hochberg & Desimone, 2010; Woulfin & Rigby, 2017). They help teachers with day-to-day instructional challenges, support the implementation of instructional materials, and work to help teachers build collaborative practices together. Recently released data from the federal School Pulse Panel, a study collecting information on the impact of the COVID-19 pandemic from a national sample of public schools, show that approximately 6 in 10 American public schools have instructional coaches, that there are more coaches dedicated to reading instruction than to math, that schools in the South are more likely to have instructional coaches than schools in other regions, and that the number of instructional coaching positions has increased over the last several years to help with COVID-related learning recovery (National Center for Education Statistics, 2024). We could not, however, find federal or state level administrative data sources that measure where instructional coaches within schools spend their time and could not, as such, test whether coaches are deployed to the second grade zone when ineffective teachers are assigned there. To help fully interrogate whether second grade is a zone of professional development, future research might be able to leverage data curated inside of third-party technology platforms used by some school districts across the country to direct and guide instructional coaching (Glover et al., 2019).
In addition to data limitations that omit our ability to consider coaching supports in second grade, state administrative records in North Carolina do not permit us to track where principals spend their own time. Principals, especially those who have cultivated their skills as instructional leaders, can themselves provide mentoring and support to developing teachers. Education researchers who have ongoing studies tracking where elementary principals go throughout the school day could analyze the proportion of time they spend in Grade 2 (relative to other grades), and especially in relation to the current stock (number and effectiveness) of second grade teachers.
Of course, other explanations, beyond the zone of professional learning explanation offered by the North Carolina principals we interviewed, could exist for why principals in the state appear to be reassigning less effective teachers to second grade. It is possible, for example, that looping, wherein principals assign teachers in ways that have them following students from one grade level to the next, instructing the same group of students for at least two school years (Cistone et al., 2004), might be in part responsible. Looping is a practice that has been found in some studies to improve teacher effectiveness and student achievement and decrease absences, truancy, and suspensions (Hill & Jones, 2018; Hwang et al., 2021; Wedenoja et al., 2022). Unfortunately, state administrative records in North Carolina do not allow us to reliably test for whether principal looping practices might explain some of the observed elementary assignment decisions. As such, while looping is rare, with recent studies estimating that the practice affects less than 2% of teachers (Wedenoja et al., 2022), we could not test whether the practice contributes to teacher reassignments overall and to grade-to-grade level reassignment patterns specifically.
These questions and others, then, should be the subject of future research as should questions related to whether the second grade patterns hold in other states with different policy contexts and accountability regimes. Within states, it is also important to begin to ask how the teacher reassignment patterns we describe here in relation to second grade vary by school size, school geography, and the family income of students served by the school.
Critically important, too, are studies that examine the potential practice implications of a second grade effect. For example, what are the downstream effects of moving less effective teachers to second grade? Specifically, is a move to second grade typically reciprocated by hiring a new teacher, by a moving a second grade teacher into a tested grade, or some more complex and yet to be understood reorganization? We encourage future studies designed with these questions in mind from the start.
Finally, future research should consider the implications of a second grade effect for student learning. What do principals and other school leaders need to do to set up a system for tracking and protecting against any potential learning deceleration during second grade, for example? Implications may also include an unintentional stigmatization of second grade teaching (should teachers catch on that less effective teachers appear to be disproportionately moved to second grade), on the one hand, and a chance to deliver more and better opportunities for instructional development more efficiently, on the other. From a research perspective, implications may entail a focus on whether reassignment to second grade is actually accompanied by an increased developmental focus and, in turn, improvements in teaching practice and, ultimately, teaching effectiveness.
In the context of a teacher assignment literature that has largely focused on the interplay of accountability policies and assignments, our study and its findings reveal a potentially more complicated, nuanced story about how and why elementary principals assign teachers to specific grades and begin to lay the groundwork for future inquiry into whether second grade is and should be a zone of instructional development for teachers.
Supplemental Material
sj-pdf-1-epa-10.3102_01623737251348202 – Supplemental material for Is Second Grade a Zone of Instructional Development for Teachers? Rethinking Strategic Staffing With a Mixed-Methods Study of Elementary Principals’ Assignment Decisions
Supplemental material, sj-pdf-1-epa-10.3102_01623737251348202 for Is Second Grade a Zone of Instructional Development for Teachers? Rethinking Strategic Staffing With a Mixed-Methods Study of Elementary Principals’ Assignment Decisions by Lora Cohen-Vogel, Christopher D. Brooks, Michael Little, Timothy A. Drake, Thurston Domina, Matthew G. Springer, Austin Gragson and Victor Cadilla in Educational Evaluation and Policy Analysis
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research reported here was supported in part by grants from the Belk Foundation of North Carolina and from the Institute of Education Sciences, U.S. Department of Education through grant #R305A21035. The views expressed are those of the authors and do not represent the views of the Belk Foundation or the U.S. Department of Education.
ORCID iDs
Notes
Authors
LORA COHEN-VOGEL, PhD, is the Frank A. Daniels Distinguished Professor of Public Policy and Education at the University of North Carolina at Chapel Hill. She helps to lead research-practice partnerships that use the science of improvement to raise schooling outcomes for traditionally underserved students.
CHRISTOPHER D. BROOKS, PhD, is a researcher at the American Institutes for Research. His research focuses on promoting equal K–12 educational opportunities through the equitable distribution of educational resources, such as school funding and effective teaching.
MICHAEL LITTLE, PhD, is associate professor of educational evaluation and policy analysis at North Carolina State University. His work focuses on improving young children’s learning and development outcomes by building coherent, aligned systems of support from pre-K through elementary school.
TIMOTHY A. DRAKE, PhD, is associate professor of education leadership and policy at North Carolina State University. His research focuses on the policies, practices, and environments that shape school leader effectiveness.
THURSTON DOMINA, PhD, is the Robert W. Eaves ‘28 Distinguished Professor of educational policy and organizational leadership and associate dean for academic affairs at the University of North Carolina at Chapel Hill. His scholarship explores the relationship between education and social inequality.
MATTHEW G. SPRINGER, PhD, is managing partner at Basis Policy Research and an adjunct professor in the School of Data Science and Society at the University of North Carolina at Chapel Hill. His research studies educational innovations and policy developments to improve the educator workforce and student educational opportunities.
AUSTIN GRAGSON, MEd, is a doctoral candidate at North Carolina State University and an associate researcher at the American Institutes for Research. His research focuses on early childhood education policy and understanding the differences in access and costs to the system.
VICTOR CADILLA, PhD, is a post-doctoral researcher at North Carolina State University. His research focuses on the ways educational policies expand and limit the educational opportunities of K–12 students.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
