High-Leverage Teaching Practices: What the Research Actually Supports

There's no shortage of claims about what makes teaching effective. Every professional development vendor has their framework. Every school improvement initiative has its research base. The result is that teachers are asked to implement multiple competing frameworks simultaneously, with the message that all of them are essential.

This is not how research works, and treating every instructional approach as equally evidence-based isn't honest. Some practices have substantially stronger research support than others. Here's what the evidence actually shows.

The Hattie Meta-Analysis (And Its Limits)

John Hattie's "Visible Learning" project synthesized over 800 meta-analyses on educational influences. It's the largest synthesis of educational research ever conducted, and its core finding — that effect sizes for most educational interventions are modest, and some are much stronger than others — is genuinely important.

The highest-effect practices in the synthesis include: collective teacher efficacy (teachers' shared belief that they can improve student learning), student self-assessment, feedback, formative assessment, and reciprocal teaching. The lowest effect practices include retention (holding students back a grade), which is actually slightly negative.

The limits: effect sizes are averages across contexts, and an average effect size doesn't mean a practice works in all contexts. The Hattie synthesis is a useful starting point, not a ranking to follow mechanically.

The Practices With Consistent Strong Evidence

Retrieval practice. The learning benefit of retrieving information from memory (through testing, recall practice, or flashcards) is one of the most replicated findings in cognitive psychology. Students who are regularly required to retrieve what they've learned retain it significantly better than students who re-read or review notes. This has been replicated across ages, subjects, and retention intervals.

Spaced practice. Distributing practice across time (rather than massed in a single session) produces better long-term retention. "Review it in September, then see it again in November, then again in February" produces more durable learning than "review it three times in September." This is why cumulative assessments and spiral curricula work.

Interleaving. Mixing different types of practice (rather than practicing one skill until mastered, then switching) produces better transfer and long-term retention than blocked practice, despite feeling harder and less efficient in the short term. Students who practice algebra and geometry problems mixed together perform better on later tests than those who practice them separately.

Explicit instruction with feedback. For procedural and skill-based content (reading decoding, math algorithms, writing mechanics), explicit teacher-led instruction with immediate corrective feedback consistently outperforms discovery-based approaches. Discovery-based learning has a place, but not for content that has a correct procedure students need to learn.

Formative assessment with instructional response. This is not just "give more quizzes." It's the process of regularly checking for understanding and adjusting instruction based on what you find. When teachers check for understanding, identify gaps, and respond instructionally, outcomes improve. The key is the response, not just the checking.

Practices With More Limited Evidence

Learning styles matching. The idea that students learn better when instruction is delivered in their preferred "modality" (visual, auditory, kinesthetic) is not supported by well-designed research. The learning styles hypothesis has been tested extensively and consistently fails to show that matching instruction to preference improves outcomes.

Brain-based learning claims. Many PD programs cite neuroscience to justify instructional practices. Some of these claims are legitimate; many are not. When a program says "brain research shows..." ask for the specific research. Often the link between neuroscience findings and classroom practice is more tenuous than the program suggests.

The Honesty About Complexity

Teaching is complex enough that single instructional strategies rarely dominate outcomes. The quality of the teacher-student relationship, the classroom climate, the coherence of the curriculum, and the specific content being taught all matter. Research can identify practices that improve outcomes on average; it can't tell you exactly which practices to use with which students on which days.

The useful question is not "what's the single best teaching method?" but "what does the research say about this specific practice with this type of content for this type of learner?" That's a more answerable question, and the answers are more useful.

The Practical Upshot

Invest in retrieval practice and spaced review. Use formative assessment and respond to it. Teach procedural skills explicitly. These three practices, applied consistently, have the strongest research support for improving student learning outcomes across a wide range of contexts. They don't require new materials or expensive PD. They require consistent implementation.