The modeling of change over time in learning, cognition, and behavior involve complex statistical models to resolve short-term ups and downs in performance from long-term skill building. As a quantitative methodologist, Ethan McCormick’s research focuses on the development of timeseries and longitudinal models for understanding human development at both time scales. His primary goals are 1) to build models which test specific and meaningful theoretical hypotheses, 2) link short- and long-term longitudinal models in novel ways, and 3) promote dissemination of complex statistical results in understandable terms for applied researchers, policy makers, and the broader public.
My plans for the Fellowship
Learning proceeds in fits and starts, periods of confusion followed by eureka moments, and partially interrupted by holidays and school absences. Evaluation, by contrast, often uses one-off, high-stakes tests where children must perform under pressure. This mismatch between natural skill learning and common methods for evaluating those skills weakens the validity of these tests and can disadvantage children from less-privileged backgrounds. During my fellowship, I will develop statistical models for continuous tracking of math learning and validate continuous metrics of ability against traditional one-off testing.
My first projects will focus on developing statistical models which accommodate two aspects of tracking short- and long-term math ability development: 1) distinguishing between practice-related improvements (e.g., memorization/recall) on given sets of math problems and generalizable learning. The second is to link timeseries models for weekly problems sets (i.e., dense, short-term data) with longitudinal growth models which capture long-term changes across educational years. I will then take the results of these first projects and the compare model-based estimates with the results of end-of-year testing, and explore the role of contextual (e.g., SES) and personal (e.g., immigrant background, minority status) factors to explore whether minoritized students are especially disadvantaged by the current testing paradigm.
How will my work change children’s and youth’s lives?
The educational landscape is changing rapidly, with transitions to remote/hybrid learning, changing demographics, and the increased incorporation of technology into the classroom. While these changes present challenges for traditional education models, they also provide the opportunity for data-rich evaluation of new approaches and heuristics for measuring educational success. During my fellowship, I will generate new insights for the possibility of measuring academic achievement continuously, rather than the traditional model of one-off end-of-year testing to evaluate proficiency. If these models prove successful, this will impact the lives of children and their families in numerous ways. Lessening reliance on end-of-year testing will remove a large source of pressure and stress for students and families as ability can be continuously evaluated throughout the year, minimizing the impact of any one mistake/challenge. Furthermore, this approach has the potential to mitigate some disadvantages that certain students face, including lower socioeconomic and minority status. These factors often introduce precarity for students which can exacerbate the chances for underestimating their ability when their broader challenges align with testing periods. The larger aim of my research for society is to evaluate whether continuous measures of ability show reduced bias towards these disadvantaged populations.