State standardized achievement tests were a central component of the federal government’s outcomes-based policies of the 2000s and 2010s. Policymakers and education leaders used tests to monitor the system’s progress toward performance goals and to inform decisions about accountability, including interventions in low-performing schools.
Since then, standardized tests and test-based accountability have come under serious criticism. And one of the most important questions is whether test scores measure something important that is predictive of later in life outcomes for students. After all, if standardized tests—the cornerstone of test-based accountability—are not measuring skills and knowledge connected to students’ longer term life outcomes, then the incentives created by outcomes-focused policies in the 2000s and 2010s were arguably misaligned with what most people would agree is one of the school system’s ultimate goals: preparing students for success after they leave K–12.
So, do test scores predict later success?
At first glance, the answer here seems clear: Students who do better on tests also tend to do better in college and work. Ample evidence suggests that test scores predict a range of student outcomes after high school. James J. Heckman, Jora Stixrud, and Sergio Urzua, for example, find that they are significantly correlated with educational attainment and labor market outcomes (e.g., employment, choice of occupation, work experience) and negatively correlated with risky behaviors (e.g., teenage pregnancy, smoking, illegal activity). More recently, Dajun Lin, Randall Lutter, and Christopher J. Ruhm find that cognitive skills at the end of high school are associated with rising labor market returns as people age. But, as the well-worn adage says, correlation is not causation.
The relationship between test scores and life outcomes will also reflect other, unmeasured student characteristics (e.g., diligence) and conditions (e.g., family characteristics, the systemic effects of racism and social inequality, adverse environmental exposures [such as lead]). The bottom line is that there is little doubt that a range of individual, social, and environmental factors affect both students’ opportunities and success in school and also later in life. The deeper question then is whether learning outcomes measured by tests tell us anything else? If a school improves student learning as measured by test scores, should we conclude that it is helping improve its students’ opportunities later in life? This is a key question for test-based accountability. And convincingly answering it is not easy. The underlying issue is that non-school factors that are unobserved to researchers, such as the degree to which students receive encouragement in the home, may influence both test scores and lead to better adult outcomes.
Researchers have generally relied on quasi-experimental methods to investigate the causal effects of interventions on learning (which is not observed) as measured by their effects on test scores (which are observed). The question is whether test scores capture the longer-run effects of changes in learning independent of other factors that also affect longer-run outcomes. Researchers have tried to tackle this problem by examining the causal effects of other educational inputs on both test scores and later-life outcomes. The reasoning for this approach is that if we find that these inputs have effects in the same direction on both tests and later life outcomes, then there is plausible evidence that test scores are measuring something that affects later outcomes.
Unsurprisingly, not all these types of studies reach the same conclusion. Some studies of the effects of school choice, for example, find that effects on test scores and later outcomes do not point in the same direction. However, a larger body of studies on teachers, peers, small class sizes, finance reform, and some school choice programs support the idea that there is a causal link between what test scores measure and life outcomes. Will Dobbie and Roland Fryer, Jr., for example, exploit oversubscribed charter schools to compare lottery winners and losers and find that attending a high-performing charter school increases test scores and college attendance and decreases the likelihood of risky behavior. Elsewhere, researchers have used the plausibly random timing of school finance reforms (often driven by lawsuits) to assess the effects of funding on short- and long-run student outcomes. Using this approach, Kirabo Jackson and colleagues, as well as Julien Lafortune, Jesse Rothstein, and Diane Whitmore Schanzenbach, find that changes in school spending affect test scores and adult outcomes. Consistent with these quasi-experimental results, Susan Dynarski, Joshua Hyman, and Diane Whitmore Schanzenbach leverage data from an actual experiment (the famous Tennessee Student-Teacher Achievement Ratio [STAR] class size project) and find that the experiment’s test score effects were an excellent predictor of postsecondary outcomes.
Widespread critiques of testing notwithstanding, these studies suggest that the measured outcomes on standardized tests do indeed capture aspects of student learning that matter for both assessing student progress in school and predicting student success after schooling ends. Research aside, this claim has some face validity: To the extent that your experiences later in life depend on literacy, numeracy, or some other specialized knowledge (e.g., chemistry), measures of your knowledge and skills in school should tell us something—again, not everything—about the opportunities you may have later in life.
Editor’s note: This is a modified excerpt from a report by the authors: Dan Goldhaber and Michael DeArmond, “What Does Empirical Research Say About Federal Policy From NCLB to ESSA?” CALDER Center American Institutes for Research (2023), in Looking Back to Look Forward: Quantitative and Qualitative Reviews of the Past 20 Years of K-12 Education Assessment and Accountability Policy, U.S. Chamber of Commerce Foundation (2023). For a more in depth discussion of the predictive power of tests and whether the relationship between tests and postsecondary outcomes is causal, see Dan Goldhaber and Umut Özek, How Much Should We Rely on Student Test Achievement as a Measure of Success? Education Researcher, 48(7): 479–483, https://doi.org/10.3102/0013189X19874061.
