ClearFactAll topics

Should schools implement frequent low-stakes testing as a primary teaching strategy to enhance student learning and long-term retention?

7 peer-reviewed studies for, 3 against

Key takeaways

Stronger evidence: For

Verdict

The debate over frequent low-stakes testing as a primary teaching strategy is among the most empirically active in educational psychology, and on balance the FOR side commands a decisive advantage in methodological rigor, scale, and convergent validity across decades of research. Roediger and Karpicke (Psychological Science, 2006) established the foundational testing effect with educationally relevant prose materials, demonstrating that students who were tested on material after a delay retained significantly more than those who merely restudied — even though the restudy group predicted better outcomes for themselves. Karpicke and Roediger (Science, 2008) extended this insight in a high-impact study showing that repeated studying after initial mastery contributed nothing to long-week retention, while repeated testing produced large positive effects; university students were systematically wrong about which strategy worked, making the case that deliberate pedagogical embedding of frequent retrieval practice is necessary and not simply something students will adopt on their own. The large-scale meta-analysis by Adesope, Trevisan, and Sundararajan (Review of Educational Research, 2017), synthesizing 272 effect sizes from 188 experiments, confirmed medium-to-large and robust effect sizes across age groups, formats, and educational contexts (g = 0.67 in classroom settings specifically), with secondary students — the exact population most frequently targeted by school-reform proposals — showing the largest gains. Crucially, McDaniel, Agarwal, Huelser, McDermott, and Roediger (Journal of Educational Psychology, 2011) moved the evidence into authentic 8th-grade science classrooms over a full semester, documenting 13–25% performance gains on summative unit examinations for quizzed versus unquizzed content, bridging the lab-to-classroom gap that critics typically exploit. The AGAINST side raises meaningful qualifications without overturning the core finding. Von der Embse et al. (Journal of Affective Disorders, 2018) established in a 30-year meta-analysis that test anxiety is negatively associated with academic performance and is most pronounced at the middle-grades level, raising legitimate welfare concerns about policies that normalize frequent testing for anxious students. The Frontiers in Education review (2019) correctly identifies that most classroom retrieval studies compare testing against passive restudy — a weak baseline — rather than against active alternatives like concept mapping or collaborative discussion, which inflates effect sizes and makes sweeping policy claims premature. Bangert-Drowns et al. (Journal of Educational Research, 1991) further demonstrated in an early classroom meta-analysis that test frequency gains are contingent on feedback quality and instructional alignment, not guaranteed by frequency alone. However, these AGAINST findings function as design principles and scope restrictions rather than refutations. The anxiety concern is addressed by the distinction between high-stakes summative testing and low-stakes retrieval practice; indeed, research by Agarwal and colleagues (cited extensively in the retrieval practice literature) finds that classroom retrieval programs can actually reduce end-of-year test anxiety by normalizing retrieval. The weak-baseline critique does not negate the positive effect against restudy, which reflects the most common default behavior students engage in outside of class. And the contingency findings of Bangert-Drowns simply specify best practices — provide feedback, use varied formats, align with curriculum — rather than argue against the strategy. The FOR side prevails because it presents a larger, more methodologically diverse, and more consistently replicated body of primary evidence, including both laboratory precision and authentic K-12 classroom implementation, with effect sizes that are educationally meaningful and robust across 50+ years of research.

Strongest study on each side

Ranked by citation count.

Strongest study for

The Critical Importance of Retrieval for Learning (Science, 2008)

Jeffrey D. Karpicke and Henry L. Roediger · 1,695 citations

Karpicke and Roediger published this landmark study in Science, examining four conditions in which students learned foreign-language vocabulary pairs with different combinations of continued studying and continued testing after initial correct recall. Repeated studying after mastery had essentially no effect on one-week delayed recall, whereas repeated testing produced a large positive effect. University students were systematically miscalibrated — they predicted that continued studying would be beneficial when it was not. These findings, published in one of the world's highest-impact journals, demonstrated that retrieval practice — not mere re-exposure — is the critical consolidating mechanism for durable learning.

Strongest study against

Retrieval Practice in Classroom Settings: A Review of Applied Research (Frontiers in Education, 2019)

Bruna Fernanda Tolentino Moreira, Tatiana Salazar Silva Pinto et al. · 77 citations

This systematic review examined whether lab-based retrieval practice findings translate reliably to authentic classroom settings and concluded that evidence is insufficient to recommend retrieval practice as universally superior to all alternative active learning strategies. The authors identified a critical methodological problem: nearly all classroom studies compare retrieval practice against passive restudy — a known weak baseline — rather than against active alternatives such as concept mapping, collaborative discussion, or elaborative interrogation. Effect sizes in real classrooms are consistently smaller than laboratory estimates. The review explicitly cautioned that without comparisons against meaningful instructional alternatives, strong policy recommendations for frequent testing as a 'primary' strategy are premature.

Open the interactive page →

The evidence

For (7)

The Critical Importance of Retrieval for Learning (Science, 2008)

Jeffrey D. Karpicke and Henry L. Roediger · 1,695 citations

Karpicke and Roediger published this landmark study in Science, examining four conditions in which students learned foreign-language vocabulary pairs with different combinations of continued studying and continued testing after initial correct recall. Repeated studying after mastery had essentially no effect on one-week delayed recall, whereas repeated testing produced a large positive effect. University students were systematically miscalibrated — they predicted that continued studying would be beneficial when it was not. These findings, published in one of the world's highest-impact journals, demonstrated that retrieval practice — not mere re-exposure — is the critical consolidating mechanism for durable learning.

Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention (Psychological Science, 2006)

Roediger and Karpicke conducted two controlled experiments with undergraduates studying prose passages. Students either restudied material or took one or three free-recall tests; final retention was assessed at 5 minutes, 2 days, or 1 week. On immediate testing, repeated study slightly outperformed retrieval practice, but on both delayed tests, prior testing produced substantially greater retention than studying. The study was one of the first to demonstrate the 'testing effect' with educationally relevant materials, establishing that taking memory tests is a powerful means of improving long-term learning — not merely measuring it. The finding has been replicated hundreds of times and has a massive citation footprint.

Regarding Class Quizzes: A Meta-analytic Synthesis of Studies on the Relationship Between Frequent Low-Stakes Testing and Class Performance (Educational Psychology Review, 2021)

Lukas K. Sotola and Marcus Crede

Meta-analysis of 52 independent classroom samples (N=7,864) showing moderate positive association (d=0.42) between frequent low-stakes quizzes and academic performance. Effects stronger in psychology classes (d=0.47) and when quiz scores contributed to grades (d=0.51). Students taking frequent quizzes had 2.6x higher odds of passing.

Test-Enhanced Learning in a Middle School Science Classroom: The Effects of Quiz Frequency and Placement (Journal of Educational Psychology, 2011)

Mark A. McDaniel, Pooja K. Agarwal et al.

McDaniel, Agarwal, Huelser, McDermott, and Roediger conducted a novel experimental study in authentic 8th-grade science classrooms across a full semester. Students received multiple-choice quizzes with feedback on some content; other equivalent content was not quizzed. In Experiment 1, three quizzes spaced across a unit produced 13–25% gains in performance on summative unit examinations for quizzed versus unquizzed material. Experiments 2a and 2b demonstrated that post-lecture quizzing was more effective than pre-lecture quizzing. The study provided compelling evidence from a real K-12 classroom that frequent low-stakes quizzing causally improves long-term student achievement, not just laboratory recall.

Rethinking the Use of Tests: A Meta-Analysis of Practice Testing (Review of Educational Research, 2017)

Olusola O. Adesope, Dominic A. Trevisan et al.

Adesope, Trevisan, and Sundararajan synthesized 272 independent effect sizes from 188 experiments comparing practice testing against non-testing learning conditions. Practice tests were more beneficial than restudying and all other comparison conditions across the board: weighted mean effect size was g = 0.51 vs. restudy, g = 0.93 vs. filler/no-activity, and g = 0.67 specifically in classroom settings. Beneficial effects were robust regardless of methodological quality, age group, or subject matter. Secondary school students showed the largest benefits, and combining multiple-choice with short-answer formats yielded the strongest retrieval practice effect. The authors concluded that practice testing is among the most reliably effective learning interventions available.

Against (3)

Retrieval Practice in Classroom Settings: A Review of Applied Research (Frontiers in Education, 2019)

Bruna Fernanda Tolentino Moreira, Tatiana Salazar Silva Pinto et al. · 77 citations

This systematic review examined whether lab-based retrieval practice findings translate reliably to authentic classroom settings and concluded that evidence is insufficient to recommend retrieval practice as universally superior to all alternative active learning strategies. The authors identified a critical methodological problem: nearly all classroom studies compare retrieval practice against passive restudy — a known weak baseline — rather than against active alternatives such as concept mapping, collaborative discussion, or elaborative interrogation. Effect sizes in real classrooms are consistently smaller than laboratory estimates. The review explicitly cautioned that without comparisons against meaningful instructional alternatives, strong policy recommendations for frequent testing as a 'primary' strategy are premature.

Test Anxiety Effects, Predictors, and Correlates: A 30-Year Meta-Analytic Review (Journal of Affective Disorders, 2018)

Nathaniel von der Embse, Dane Jester et al.

Von der Embse, Jester, Roy, and Post synthesized data from 238 studies spanning 30 years to examine the relationship between test anxiety and educational outcomes. Results showed test anxiety was significantly and negatively related to a wide range of outcomes including standardized test scores, university entrance exam performance, and GPA, with effects most pronounced at the middle-grades level. Between 15–22% of students exhibit high levels of test anxiety, and perceived test difficulty and high-consequence testing were strong predictors of elevated anxiety. Self-esteem, avoidant coping, and gender were also significant correlates. These findings raise substantive concerns about policies that make frequent testing a primary classroom strategy, particularly for the adolescent age groups most at risk.

Effects of Frequent Classroom Testing (Journal of Educational Research, 1991)

Robert L. Bangert-Drowns, James A. Kulik et al.

Bangert-Drowns, Kulik, and Kulik conducted an early meta-analysis of 40 classroom studies on the effects of frequent testing on student achievement. The overall mean effect size was modest (d = 0.54), but outcomes were highly heterogeneous, with several studies showing neutral or negative effects when feedback was absent or when testing displaced instructional time. Benefits were contingent on the presence of corrective feedback, appropriate test format, and alignment with instructional goals. The authors concluded that simply increasing test frequency without careful design — particularly without feedback — does not reliably produce learning gains, warning against simplistic policy prescriptions for universal frequent testing across school systems.

The debate

Lisa: The evidence for frequent low-stakes testing as a primary teaching strategy is, in my view, among the most solid in all of educational psychology. Roediger and Karpicke's 2006 Psychological Science study showed that students who took repeated free-recall tests on prose passages retained substantially more after one week than students who spent the same time restudying — and the restudy group were the ones who felt most confident. That's a foundational finding: testing doesn't just measure learning, it creates it.

James: I don't dispute the laboratory result, Lisa, but I think we need to be careful about the leap from lab to school policy. The 2019 Frontiers in Education review makes an important observation: nearly every classroom retrieval practice study compares testing against passive restudy — rereading, highlighting — which is one of the weakest learning strategies we know. We're not actually asking whether frequent quizzing beats concept mapping, or collaborative inquiry, or rich discussion. When we limit the comparison that narrowly, of course testing wins.

Lisa: That's a valid methodological point, but it's also a bit of a moving goalpost. Passive restudy is what students actually do when left to their own devices. Karpicke and Roediger's 2008 Science paper makes this painfully clear: students believed that continued studying after they'd already mastered vocabulary items would be helpful — but it had essentially zero effect on one-week recall. Repeated testing, by contrast, produced a large positive effect. If our question is 'what should replace passive rereading in schools,' the answer from the evidence is pretty unambiguous.

James: But there's a real welfare dimension here that the testing-effect literature tends to underweight. Von der Embse and colleagues published a 30-year meta-analysis in the Journal of Affective Disorders in 2018, covering 238 studies, and found a consistent negative relationship between test anxiety and academic performance — most pronounced at the middle-grades level. That's exactly the population we're discussing when we talk about implementing frequent testing as a primary K-12 strategy. Between 15 and 22 percent of students show high test anxiety. If we're normalizing assessment as the central learning mechanism, we may be systematically disadvantaging a large minority of vulnerable students.

Lisa: The anxiety concern is real, but it conflates two very different assessment regimes. The literature on test anxiety is overwhelmingly about high-stakes summative testing — GCSEs, entrance exams, standardized tests with consequences. Low-stakes retrieval practice is explicitly designed to remove those consequences. McDaniel, Agarwal, and colleagues ran a semester-long study in real 8th-grade science classes, published in the Journal of Educational Psychology in 2011, using multiple-choice quizzes that didn't count toward students' grades. Quizzed content showed 13 to 25 percent performance gains on actual unit exams. And separate work from Agarwal's group found that classroom retrieval practice programs actually reduce students' test anxiety over time — because the act of retrieval becomes normalized and familiar rather than threatening.

James: That's a fair distinction, and I'll grant that low-stakes framing changes the psychological context substantially. But let me raise a different concern: scale and contingency. Bangert-Drowns, Kulik, and Kulik published a meta-analysis in the Journal of Educational Research back in 1991 covering 40 real classroom studies of frequent testing. They found a mean effect size of d = 0.54 — positive, but modest — and crucially, outcomes were highly variable. Several studies showed neutral or negative effects when feedback was absent. If you take away the feedback loop, you lose much of the benefit. Implementing frequent testing as a primary school-wide strategy without guaranteeing high-quality feedback mechanisms is a recipe for disappointment.

Lisa: Bangert-Drowns is actually evidence on my side, not yours. A positive effect size of 0.54 across 40 real classroom studies from 1991 — before modern implementation science, before digital quiz tools, before we understood optimal spacing and formats — is a strong signal. And you're absolutely right that feedback matters: that's precisely what the 2017 meta-analysis by Adesope, Trevisan, and Sundararajan confirmed across 272 effect sizes and 188 experiments. Effect sizes with feedback were larger; classroom settings showed g = 0.67; secondary students showed the biggest gains. The message isn't 'quiz students and walk away.' It's 'embed low-stakes retrieval practice with immediate corrective feedback as a routine part of instruction.' That composite practice has a robustly supported evidence base.

James: I think we've actually narrowed the disagreement to something useful. My concern isn't with retrieval practice per se — it's with the word 'primary' in the question. When quizzing becomes the dominant mode of classroom activity, it can crowd out deeper forms of inquiry, creative synthesis, and the kind of slow thinking that builds conceptual understanding. The Frontiers review is right that we haven't adequately tested retrieval against those richer alternatives. I'd advocate for retrieval practice as one evidence-based tool in a diverse instructional toolkit — not the organizing principle around which everything else is structured.

Lisa: That's a reasonable pedagogical position, and I'd accept it as a design guideline rather than a refutation of the evidence. No serious proponent of retrieval practice argues that quizzing should replace all other forms of learning. The claim is that frequent low-stakes testing should be embedded throughout instruction — as a foundation that supports retention — rather than reserved for summative evaluation at the end of a unit. Four landmark papers, a 30-year meta-analytic record, and consistent classroom replications from middle school to university all point in the same direction. The science is clear enough that acting on it is the responsible choice.

Frequently asked questions

What is the testing effect?

The testing effect is the finding that recalling information from memory, such as by taking a quiz, strengthens long-term retention more than simply rereading or restudying it. Lab studies showed restudying can look slightly better right away, but prior testing produced substantially better recall days or a week later.

Does low-stakes quizzing work in real classrooms, not just labs?

Classroom meta-analyses covering hundreds of studies report medium positive effects of quizzing on achievement, and a full-semester study in 8th-grade science classes found clear exam gains for quizzed material. One review notes that classroom effect sizes tend to be smaller than lab estimates.

Does frequent testing increase student anxiety?

A 30-year meta-analysis found test anxiety is linked to lower academic outcomes and is most pronounced in middle grades, with high-consequence testing a strong predictor of anxiety. Supporters argue low-stakes retrieval practice differs from high-stakes testing, and some research suggests it may reduce test anxiety.

What makes frequent quizzing more effective?

Research points to corrective feedback, alignment with instructional goals, and appropriate formats as key factors; combining multiple-choice and short-answer questions showed the strongest effects in one meta-analysis. An early review found that testing without feedback, or testing that displaced instruction, did not reliably improve learning.

Related questions