University of Arkansas Office for Education Policy

Arkansas Teacher Turnover During the First Two Years of the COVID-19 Pandemic

In The View from the OEP on June 21, 2022 at 12:23 pm

This blog post and the research it summarizes was authored by Andrew Camp, Gema Zamarro & Josh McGee.

This summer, teachers across Arkansas will take a much-needed break to recharge after another difficult year. Since the start of the COVID-19 pandemic, educators have faced the unprecedented and unexpected challenges of remote learning, preventing the spread of COVID-19, and dealing with students’ absences due to quarantines and illness.

Throughout these unusual times, teachers have reported much higher levels of stress and a higher proportion of teachers report that they have considered leaving the profession. The increased stress that the pandemic has put on school personnel has sparked fears about a potential exodus of teachers and an increase in teacher shortages. Teachers are among the largest school-based contributors to students’ academic success and turnover is both costly for districts and harmful to students’ academic progress. As students face unprecedented learning losses, an increase in teacher turnover might also seriously hinder attempts at recovery.

In our prior research, we found that teachers were more likely to report that they have considered leaving their position or retiring if they are approaching retirement age, had to change instructional modes during the academic year, have higher COVID-19-related health concerns, or report high levels of job-related burnout. However, teachers who consider leaving may not ultimately leave.

So, are teachers leaving the classroom more than before the pandemic?

In our new report, we explore changes in Arkansan teachers’ mobility and attrition two years into the COVID-19 pandemic. Arkansas is an especially interesting context as districts in the state were required to offer in-person learning five days per week.

Like similar analyses from Washington and Massachusetts, we find stable turnover rates during the first year of the pandemic (2020-2021) but a 2.2 percentage point increase in teacher mobility and attrition in the second year (2021-2022).

Turnover, which we define as a teacher either moving schools or exiting the Arkansas teacher workforce entirely, did not change uniformly across the state. While all regions experienced increased turnover entering the pandemic’s second academic year (2021-22). Northwest, Southeast, and Southwest Arkansas saw the highest increases in teacher turnover. This is especially worrisome for districts in the Southeast and Southwest as many districts already experienced teacher shortages before the pandemic and tend to serve students from higher-poverty backgrounds. We also observed increased turnover due to transfers across schools in Northwest Arkansas during the first pandemic academic year.

Our prior research found that teachers approaching retirement age (55 or older) were more likely to report having considered leaving or retiring in a national survey from March 2021. When examining Arkansas, we find that the proportion of these teachers exiting the Arkansas teacher workforce increased by about 2 percentage points entering the 2020-21 academic year. Increased turnover among the most experienced teachers could have important consequences for Arkansas’s teacher quality, which may in turn impact student academic outcomes during the pandemic recovery.

We also study the retention and attrition patterns of beginning teachers who started their careers during the pandemic and likely envisioned a very different experience. While the proportion of these beginning teachers who exit the profession did not increase during the pandemic, we do observe a nearly 5 percentage point increase in the proportion of new teachers who move across schools or districts entering the 2021-22 academic year.

Increasing the diversity of the Arkansas teacher labor force is of great concern given the documented benefits that prior research has shown for racially minoritized students. For this reason, we also examine the retention of Black teachers during the pandemic.

Before the pandemic, the retention rate for Black teachers was 5 percentage points lower than teachers overall, although the gap between white and Black teachers was statistically insignificant once we controlled for teacher and school characteristics. During the pandemic, however, this gap has dramatically increased, with a larger number of  Black teachers now exiting the Arkansas teacher workforce.

We observe that during the pandemic retention of Black teachers in the same school decreased from about 75% in pre-pandemic academic years to 71% in the first pandemic year and 68% in the second pandemic year. This is in contrast with overall retention rates of about 80% pre-pandemic and during the first pandemic academic year and 78% during the second year of the pandemic. That represents a difference in retention rates for Black teachers of 10 percentage points in the second year of the pandemic.

We further analyze changes in teacher retention and exits using statistical logit and multinomial logit models. These models, which allow us to control for multiple factors and compare similar teachers who teach at comparable schools, confirm that as the pandemic has progressed, Black teachers have become more likely to exit the profession compared to white teachers. Black teachers became 3 percentage points less likely to remain in the same school than white teachers in the first pandemic year and 4 percentage points less likely in the second year. These results raise serious concerns about a potential reduction in the diversity of the Arkansas teacher labor force.

Our statistical models also reveal other changes in retention during COVID. Before the pandemic, teachers in more disadvantaged schools were retained less than teachers in advantaged schools. Compared to a school where none of the students qualify for the Free or Reduced-priced Lunch Program (FRL), a school where 50 percent of the students qualify can expect about a one percentage point decrease in the probability of teachers remaining at that school in the subsequent year. During the pandemic, however, this relationship has nearly tripled in magnitude. Now, a fifty-percentage point increase in the proportion of FRL students at a school corresponds to a 2.5-3 percentage point decrease in retention.

Finally, and in line with our prior research, having to change learning modes during the school year was associated with about a 4-percentage point reduction in teacher retention. This reduction seems to be evenly driven by teachers transferring to another school or exiting the Arkansas teacher workforce altogether.

What does this mean for Arkansas teachers, schools, and students?

The COVID-19 pandemic has been a challenging time for many but especially for teachers who had to adapt to new ways of instruction, challenging working conditions, and health concerns when returning to in-person learning. Early research documented high levels of teacher stress and burnout and an increase in the proportion who considered leaving their positions, raising concerns about a potential surge in teacher turnover and future teacher shortages.

While we do not observe the mass exodus of teachers that some warned about in popular media, the 2 percentage points increase in turnover we find raises concerns about potential instability in the Arkansas teacher labor force, given the relatively high pre-pandemic turnover rates in the state and documented shortages in school districts in the Lower Delta region (Southeast), Southwest, and Upper Delta regions (Northeast).

Our results show that teacher retention decreased during the pandemic, especially for schools in the Southeast and Southwest, as well as for schools with higher proportions of students eligible for FRL, which could put a strain on those schools already experiencing teacher shortages. In future research, we will further study the challenges facing these schools and the use of emergency teacher licenses, long-term subs, and teachers who are not fully certified to cover vacancies in these areas.

Our results also raise concerns about a potential reduction in the diversity of the Arkansas teacher workforce. We observe a significant decline in the retention of Black teachers during the pandemic, which could negatively affect students’ outcomes. Finding ways to support and better retain racially minoritized teachers in the state should be a high priority.

Finally, our results show that teachers who worked in a district that changed learning models during the 2020-2021 academic year were less likely to be retained at the same school the next academic year. Finding ways to facilitate a supportive work environment and adopting mitigation measures when needed could help reduce changes in learning modes and help retain teachers during this pandemic. In this respect, research indicates that schools with strong communication, targeted training, meaningful collaboration, fair expectations, and authentic recognition for their teachers were more successful at maintaining teachers’ sense of success.

Moving forward, it will be important to continue monitoring the pandemic’s effects on the Arkansas teacher labor force and find ways to better support teachers and schools, especially in those areas most affected by teacher shortages.

It’s Time to Address the “F”

In The View from the OEP on May 25, 2022 at 11:10 am

As report cards are being sent home this time of year, we thought it was a good time to talk about course failure.

Over the past 10 years, high school freshman course failure rates have declined, but more than 1 in 5 students still fail a course in 9th grade. We have shared our earlier research on the statistically significant relationship between high school freshmen GPAs and high school graduation rates and college enrollment rates. This led us to dig into the data, and find out which courses kids are failing, and which types of students are the most likely to fail.  We spent this spring semester analyzing ten different cohorts of Arkansas freshmen and from 2010-2019 to explore these three questions. You can read the whole paper or the shorter policy brief, but in this blog we will hit the highlights of what we found.

In 2018-19, the most recent year of data analyzed, we find that 22 percent of Arkansas high school freshman failed at least one course.

The demographic and programmatic characteristics of our Arkansas freshmen sample are below:

Male50.8%Free/Reduced Lunch (FRL)59.1%
White62.0%Gifted and Talented (GT)13.1%
Black19.2%English Language Learners (ELL)6.8%
Hispanic13.1%Special Education (SPED)11.6%
Other Races5.8%Total N35,180

And these groups of students have failure rates with respect to the average that look like this:

As you can see, freshmen demographic and programmatic groups with failure rates higher than average are Black, free-or-reduced lunch (FRL, our proxy for economically disadvantaged), English Language Learners (ELL), male, students receiving special education services (SPED), and Hispanic students. Black students have a failure rate almost 13 percentage points higher than the statewide average.

We then considered these failure rates might be different across the geographic regions of Arkansas. The failure rates by gender and race/ethnicity by region are are similar to statewide rates, but we find unexpected variation among the programmatic course failure averages by geographic region:

Overall, the Northwest region has the lowest failure rate at 18.7 percent, and the Southeast region has the highest failure rate at 28.5 percent. Among programmatic groups however, the FRL, GT, and ELL status course failure percentages are the highest in the Central region. Freshman students in the Central region that participate in the FRL program, or receive ELL or GT program services are more likely to fail at least one course than we would expect given the overall failure rate for the region and the statewide average.

Now that we’ve found Black students have the highest rate of course failures, the Northwest region has the lowest overall failure rate, the Southeast region has the highest failure rate, and programmatic groups are failing at the highest rates in the Central region, we need to investigate what courses Arkansas freshmen are failing. Below are the top ten most failed courses by Arkansas freshmen in 2018-19. We identify core content courses (mathematics, language arts, science, and social studies) with a check mark.

 Failure PercentageCore
Algebra I12.3
Spanish I9.2 
Physical Science9.1
Computer Business Applications8.9 
English 98.8
US History since 18908.4
Family and Consumer Sciences6.4 
World History since 14505.7
Oral Communications5.1 

Algebra I is the course with the highest failure rate for Arkansas freshmen, and this is consistently true statewide for all years examined. Spanish I is the highest failed non-core course for the 2018-19 group of freshmen, but other non-core courses like Computer Business Applications and Art are often in the highest failed non-core course spot.

Now, it might seem like we have addressed our questions, but not as mathematically confidently as we could. Econometrics has some handy mathematical tools that allow us to account for similarities and differences among the groups of students who share demographic and programmatic characteristics, who have similar prior academic achievement, who complete course work similarly their freshmen year, and who are in similar districts in Arkansas. This helps us find a fairer way to measure the effects of students across the state that aren’t skewed towards differences across regions and districts.

Using a logit regression analysis for the 2017-18 and 2018-19 school years to account for the differences of students across the state, we find statistically significant results below.

Initially, our Black student group had the highest course failure rate, but they are not necessarily most likely to fail a course their freshman year. After the regression that accounts and controls for differences and similarities,

  • White students are actually 1.5 percentage points more likely to fail a course their freshman year than Black students.
  • Economically disadvantaged students are almost 9 percentage points more likely to fail a course their freshman year compared to more advantaged students.
  • White economically disadvantaged students are 11.2 percentage points more likely to fail than White advantaged students.
  • Among economically disadvantaged students, White students are 4.5 percentage points more likely to fail a course their freshman year than Black students.

Interestingly, among our four program statuses, FRL, GT, ELL, and SPED, three of the four denote a less likely to fail association. Students not receiving special education services are 13.1 percentage points more likely to fail a course their freshman year than students receiving special education services. Students not in the GT program or ELL program are also more likely to fail compared to those who are in the programs. Participation in these three programs and services is associated with a lower likelihood of failure, whereas participation in the FRL program is associated with more likely to fail.

This points us to believe SPED, ELL, and GT students are receiving services and assistance that helps them succeed and pass their courses, but that FRL status students are not getting the support that they need. So, what can we do with this information?

In general, programs that have been found to be effective for helping failing or at-risk freshmen are professional learning communities (PLCs), reviewing student data that focuses on the most at-risk students (lower grades and higher absences), arranging Freshman success meetings, and forming intentional relationships with lower GPA students.

To get to the core of the problem the quickest, Arkansas district leaders should consider enacting a “no-zero” policy or minimum grading policies (Feldman, 2019). When a 0-100 scale is utilized in public schools, the weight of the failing 50 points disproportionately harms students of color, low-income students, and English Language Learners. Feldman (2019) reports schools that implement a minimum grading policy, assigning each letter grade the same amount of points, decreased student failures, reduced grade inflation, and reduced achievement gaps. While this policy recommendation may be uncomfortable for teachers as it challenges the norm and the standard zero grade as a punishment, it is necessary to help eliminate the possibility of grading bias.

Malecki and Demaray (2016) suggest social mentorship programs for FRL students from teachers and principals as a support to help them succeed. Economically disadvantaged students feel more welcomed and like they have a place at school when they have a direct relationship with a mentor.

Overall, since freshmen grades and GPAs are associated with future academic success, and we have concerns that our economically disadvantaged students might not be receiving the supports they need to succeed, we need immediate attention brought to each district on grading policies. Identifying and removing barriers to student success is a step towards helping all Arkansas students experience better academic and social outcomes.

If you are interested in the freshman failure rates for your district, shoot us an email at! We would be happy to identify which students in your district are failing and see if we can help you remove barriers to your students’ success. We are presenting this work at the ADE Summit in July, and we hope to see you there!

OEP is Seeking Summer Interns!

In The View from the OEP on April 27, 2022 at 11:45 am

The Office for Education Policy (OEP) at the University of Arkansas in Fayetteville, is seeking two interested, enthusiastic interns to work on special projects this summer.
Projects include developing policy briefs and/or Arkansas Education Reports (AERs) on relevant education policy topics, as well as participating in educational program evaluation.
OEP specializes in education research and policy to support lawmakers and educators in thoughtful decision-making for PreK-20 education in Arkansas.

Required Qualifications
• Strong writing skills
• Interest in education policy issues
Preferred Qualifications
• Prior research experience
• Experience working with large datasets and statistical software programs
• Interest in pursuing a degree in education policy, preferably the Ph.D. program offered through the Department of Education Reform

Location: Fayetteville, AR (possibility of remote work for qualified applicants)
Compensation: $3,500
Timeframe: Six weeks with a flexible timeline between June 1 and August 31
Housing: We do not provide housing; however, we can assist in finding temporary housing for the summer in Fayetteville, Arkansas.

How to apply: To be considered for this opportunity, please submit a letter of interest and resume by May 11, 2022 to Sarah McKenzie at