Allison Atteberry

Associate Professor of Education & Public Policy; Director, EdPolicyWorks

  • Ph.D., Stanford University, 2011
  • B.A., University of Chicago, 2005

Areas of Specialization

Teacher labor markets, education policy analysis, summer learning loss, production of social inequality, quantitative methods, causal inference

Allison Atteberry is an associate professor for the Education Policy Program within the School of Education and Human Development at UVA. She received her PhD in 2011 from the Stanford School of Education in educational policy analysis, with a minor in statistics. Atteberry conducts research on teacher- and school-level interventions designed to improve the quality of instruction experienced by historically underserved students. As a field, we are increasingly aware of how difficult it is to determine whether policies, practices, and interventions have the intended impacts, and so Atteberry approaches her work with a strong interest in what constitutes compelling evidence of causal effects in quantitative research.

In terms of methods, Atteberry teaches and uses both econometric and statistical approaches to education policy analysis. She has a particular interest in the estimation of education production functions in the context of value-added modeling, as well as randomized control trials, instrumental variables, regression discontinuity, propensity score matching, fixed effects, and difference-in-differences causal models. Atteberry also enjoys using hierarchical linear models given their unique suitability for asking sociological questions in nested settings (e.g., repeated observations nested within students, nested within schools, etc.).

Atteberry’s academic interests center on policies and interventions that are intended to help provide effective teachers to the students who need them most. This has led her to focus on the identification, selection, development, and retention of teachers who have measurable impacts on student achievement. Specific topics include teacher preparation, high quality professional development, mentoring and peer collaboration, efforts to use measures of effectiveness formatively to improve practice, policies that target district responses to teachers and schools based on measures of effectiveness, and incentives for the strongest teachers to work in hard-to-serve schools.

Selected Publications

Media Mentions


  • Program Area
  • Research Center
  • Current Projects

    1. A Research-Practice Partnership with Denver Public Schools on Supporting their Teacher Workforce. Professors Allison Atteberry and Mimi Engel from the CU Boulder School of Education are in the early stages of forming a long-term, research-practice partnership (RPP) with the Denver Public School District (DPS), called the Teacher Workforce Collaborative (TWC).  TWC connects CU professors with DPS’ Talent Management team. The focus of the RPP is closing Denver’s large and persistent achievement gaps. The mechanism for doing so—strengthening the District’s teacher workforce—is the focus of the Partnership. 

    2. The Causal Effects of Full- vs. Half-Day Pre-K: A Randomized Control Trial. Are children more successful in early schooling if they have access to full-day pre-k? Westminster Public School District is piloting a new full-day pre-k program to supplement its existing half-day offerings. Due to oversubscription to the pilot, we used a lottery to randomly assigned children to receive a full-day spot or an offer of half-day. As of 2019, all three cohorts have completed their pre-k year, and we have estimated the causal effects on student early literacy, special education referrals, and socio-emotional outcomes at the end of the year and fall of kindergarten (see EEPA publication here). We are also following study participants through third grade. We have a number of additional manuscripts under way, including studies of how time is used in these classrooms, how the provision of full-day pre-K affects family and home lives, and whether children who experience full day have different likelihoods of receiving special education designations. With colleagues: Daphna Bassok, Vivian Wong. 

    3. The Misattribution of Summers in Teacher Value-Added. In a 2020 ER article, I take up one methodological concern about teacher value-added model (VAM) based effect estimates: State tests are given only once annually in the spring. As a result, teachers’ estimated effects are based on their students’ test scores from the previous- to the current-spring, thus subsuming the summer before teachers even meet those students. Using a unique dataset with both fall and spring scores, I also estimated VAMs using current-fall to next-fall timing, which instead includes the summer after a given school year. Teachers’ VA-based scores from these two VAM timings—both of which purport to capture the same teacher’s effect in the same school year—turn out to be essentially unrelated to one another (ρ=0.13). This finding is concerning. There is no clear reason to prefer spring-to-spring over fall-to-fall time frames, and this choice would lead to an entirely different ranking of teachers.

    4. Trends in Student- and Teacher Outcomes during the Era of Denver ProComp. How has the distribution of student achievement and teachers changed since Denver began one of the first experiments with teacher merit pay in the U.S.? Denver Public Schools (DPS) started ProComp in 2006, and the system includes up to 10 different incentives for various forms of teaching effort and effectiveness. We explore whether highly effective teachers are more likely to remain in DPS since ProComp began. Have strong teachers been more likely to seek employment in DPS since the onset of the program? Has student achievement improved? We use interrupted time series methods to examine whether trends in outcomes are consistent with the roll out of this historic pay-for-performance system.  The manuscript is in press at Teachers College Record, and a link to the current draft is here.

    5. The Role of Summers in Achievement Disparities. What role does summer vacation play in the expansion of achievement gaps during K-12? We use a unique dataset with longitudinal records across K-12 for over half a million students to examine when during school-age years achievement gaps widen the most—during summers or between the first and last day of the school year. The article is in press at AERJ, and an Annenberg EdWorkingPaper version of the paper [No. 19-82] is available here

    6. Not Where You Start, But How Much You Grow: An Addendum to the Coleman Report . I remember learning, in my first year of graduate school, that the canonical 1966 Coleman Report established that only 10-20% of the variation in student achievement lies between schools. Schools, it seemed, were simply not a powerful lever to shape students’ outcomes—a takeaway that shook the field. Yet this “schools don’t matter” narrative is difficult to reconcile with the readily observable differences across schools, the formative nature of schooling, and a large body of subsequent evidence documenting aspects of schooling that can improve students’ trajectories. In an article in press at ER, I revisit the Coleman analysis, but instead of just decomposing the variance in students’ achievement levels within and between schools, I also decomposed the variance in students’ achievement growth rates. Coleman himself promoted this approach but did not have the necessary longitudinal data. Unlike achievement levels, most (over 70%) of the variation in student test score growth rates lies between schools. These findings are not definitive, as they touch on some key measurement issues. The article therefore calls for replication in other data settings. The results are nonetheless intriguing, challenging one of the dominant narratives about schools as weak influencers of student outcomes.