Evaluating Montessori EducationEvidence, Impact, and Future Directions

Chloë Marshall; Angeline S. Lillard, Megan J. Heise, Eve M. Richey, Xin Tong, Alyssa Hart et al. (+1)•Sources (2):Montessori education: a review of the evidence base Montessori Preschool Elevates and Equalizes Child Outcomes: A Longitudinal Study

OverviewBalancednova voice

Imagine walking into a classroom where the hum you hear isn't a teacher lecturing, but dozens of small, purposeful sounds: a child pouring beans from one pitcher to another, two older kids quietly coaching a younger one through a counting task, and a teacher kneeling to observe rather than direct. That's the core of Montessori. The classroom is a prepared environment—deliberately designed to support intellectual, physical, social, and emotional development—where the child chooses what to work on and for how long. Ages mix across multi-year bands, so four- and five-year-olds become natural tutors for three-year-olds, and the teacher acts more like a guide than a broadcaster. The day revolves around a long, uninterrupted work cycle—often close to three hours—where concentration is treated as something sacred. No grades, no gold stars, minimal interruptions. The materials do the heavy lifting; the child drives the learning. Those materials are not random. They're concrete, sequential, and self-correcting, so learning can move from hands to head. A classic example is the Pink Tower: ten cubes differing only in size, with no colors or numbers to cue you, so you have to attend to volume and order. The Sound Boxes pair cylinders that differ only by the grain inside, pushing discrimination by ear alone. That same concrete-to-abstract logic reaches into literacy and math. In language, children trace sandpaper letters to connect touch and sound, and they often learn to write before they read—hand control, letter formation, and spelling as component skills that later snap together. In math, quantities from zero to ten and the numerals that name them are introduced separately, then combined; tens, hundreds, and thousands arrive as actual bead quantities you can feel, before anyone asks you to manipulate symbols. There's a design principle threaded through all of this called "control of error"—materials that signal mistakes without an adult stepping in, letting the child revise and try again. And all of it plays out across two nested cycles: an internal cycle where a child selects and repeats a task until it's complete—or interestingly, until it isn't—and the longer, external work cycle that gives them the time and peace to go deep. Now, that's the promise. But does it work? Here's where the evidence gets both interesting and a bit fragile. Chloë Marshall, writing in "npj Science of Learning," pulled together the peer-reviewed evaluations and found that most are American, many are short-term, and the gold standard—true randomized trials—are rare. Fidelity complicates things further. "Montessori" can mean a lot of different things in practice, and the more the model is diluted, the harder it is to credit any effect to Montessori itself. Even with those caveats, some signals stand out. In several quasi-experimental and lottery-based studies, younger children in Montessori scored higher on early literacy—think letter-word identification and phonological decoding—and on early math. Measures of executive function and social understanding also nudged upward. In a striking high-fidelity preschool sample, gains were broader—reading, math, vocabulary, social problem-solving—and, crucially, how much time children actually spent with Montessori materials predicted how much they grew. That's a fidelity story in two beats: stick to the core design and the benefits appear more consistently; strip elements out or shorten the work cycle and those gains shrink or disappear. There are mixed notes, too. Two Head Start adaptations that diverged from the Montessori ideal didn't show immediate advantages, and one middle-school comparison found lower English scores after schools added conventional grading and non-Montessori features. The upshot from Marshall's review is not that Montessori is magic, but that when it's done well it can help across cognitive, social, and academic domains—and we need stronger causal tests to isolate what's doing the work. So let's talk about a study that leaned hard into that causal question. Angeline Lillard and colleagues used a natural experiment built into public school admissions: a district lottery for two high-fidelity Montessori magnet preschools in Hartford, Connecticut. Families applied, the state ran a computerized lottery, and some children got seats while others did not. That creates something close to random assignment. The team followed one hundred forty-one children—seventy assigned to Montessori and seventy-one to other public programs—from about age three to six, testing them four times across roughly three years. Attrition was low, and the analyses were "intent-to-treat," meaning children were analyzed by the group they were initially offered, even if they later switched schools. That matters because it keeps the lottery's randomness intact. These Montessori schools were not casual about fidelity. They were recognized by the Association Montessori Internationale, spanned eleven classrooms across two campuses, and were staffed by teachers trained by the Association Montessori Internationale. The control group was "business as usual" public preschool. Everyone was tested one-on-one, in the same order, to keep procedures tight across settings. If you care about causal inference, these are the moves you want: a fair assignment mechanism, high-quality implementation, and consistent measurement. What did they measure? Academic achievement through three widely used Woodcock-Johnson subtests—Letter-Word, Vocabulary, and Math—later combined into a single factor. Executive function with two tasks you might have seen in preschool research: Head-Toes-Knees-Shoulders, which taps inhibitory control and cognitive flexibility, and a NEPSY design copying task for working memory and planning. Social understanding through a set of well-validated Theory of Mind tasks from Wellman and Liu. Social problem-solving via Rubin's test. They also watched how children approached challenges with a puzzle task that let them choose an easy or impossible puzzle—mastery orientation in action—and asked how much they liked academic versus recreational tasks. Creativity came in through an Alternate Uses test: how many different ways can you use a paperclip? Under the hood, the team used growth-curve modeling, a way to estimate not just where kids are but how fast they're changing, and they used full-information maximum likelihood to make principled use of the small amount of missing data. Here's the headline. Montessori kids' academic achievement grew faster over time. In the growth models, the Montessori group showed a steeper slope—about a tenth of a standard unit per time interval—than controls. By the final assessment around kindergarten exit, an analysis that accounted for where children started still showed a clear Montessori advantage. That's not a one-time bump; it's a shift in trajectory. In other words, the longer children were in high-fidelity Montessori, the more that gap opened. The story wasn't just test scores. On Theory of Mind, differences emerged by the third time point—roughly around age four and a half—and were still trending by the end. That suggests a classroom culture where children practice taking another's perspective, and mixed-age peer tutoring likely doesn't hurt. Mastery orientation told a similar story. By the later assessments, Montessori children were more likely to pick the hard puzzle and lean into challenge rather than flee it. And when the team simply asked what kids liked, Montessori children chose schoolish tasks more often. That's a small measure with a big implication: if you enjoy the work, you do more of it, and doing more of it is how you get good. Some domains didn't move. Creativity, at least as captured by alternative uses for everyday objects, didn't differ at the final time point. Social problem-solving scores were similar across groups. Executive function is a subtler case. Across the whole span, it wasn't consistently higher in Montessori, but at the age-four window, there was a measurable difference favoring Montessori that faded by kindergarten exit. That pattern fits what classroom observers often report: Montessori seems to scaffold inhibitory control and focused attention early, especially when the long work cycle is protected, but those advantages can wash out as children approach school age if comparison classrooms also emphasize routines and self-regulation. Now the part that caught almost everyone's eye. The income-achievement gap narrowed in Montessori. Across the whole sample, income and achievement were positively correlated, as they almost always are. But that link was stronger in the conventional programs than in Montessori. In numbers, the correlation was about zero point four six in the control group and roughly half that—zero point two three—in Montessori, and the difference wasn't just sampling noise. Zoom in on lower-income children and the pattern sharpens: by the final assessment, those in Montessori outperformed their lower-income peers in other programs, even after controlling for where they started. Within the Montessori group, differences by income were small or statistically weak by the end of preschool; within the control group, higher-income children stayed ahead. That's a concrete equity result: same city, same admissions process, different classroom ecology, and a gap that shrinks. There's another link that loosened in Montessori: the usual tight coupling between executive function and achievement. In the control programs, children with stronger executive function generally did better academically, which is exactly what you'd expect—better attention and working memory make school tasks easier. In Montessori, that link was flatter. Children with lower executive function performed closer to peers with higher executive function on achievement. One way to hear that is as a design effect. When materials provide built-in feedback—when errors are signals to try again, not reasons to stop—and when children can work at their own pace without interruptions, the advantage of being the most self-regulated kid in the room may matter less for learning core academic content. None of this makes the Hartford study a perfect test. A lottery isn't the same as flipping a coin in a lab; families self-select into application pools, and there were small wrinkles like sibling preferences. The Montessori programs were unusually high fidelity with recognition by the Association Montessori Internationale and teachers trained by the Association Montessori Internationale, which strengthens the internal test but limits how far we can generalize to "Montessori-ish" classrooms that mix in grades, worksheets, or shorter work cycles. And by the final time point, a handful of children had moved or crossed over. The team was careful—intent-to-treat analyses, sensitivity checks, measurement models to make sure the tests were tapping the same constructs over time—but generalizability is always the question waiting at the end. Step back, though, and a coherent picture comes into focus. Marshall's review told us that fidelity matters, and this lottery study is what that looks like on the ground: when the core conditions are in place—the prepared environment, mixed ages, long work cycles, self-correcting materials, minimal extrinsic rewards—children's academic growth speeds up, their social perspective-taking improves, they show more appetite for challenge, and they like schoolwork more. The big societal payoff sits in the equity results. A design that reduces the penalty of lower executive function and weakens the link between family income and achievement is not just a pedagogical win; it's a public health-scale idea. Where does that leave us? With a strong nudge toward bigger, careful tests. Lillard and colleagues argue for scaling studies that keep fidelity visible—monitor teacher training, track how much children actually use Montessori materials, and protect the work cycle—because fidelity was the hinge on which outcomes swung. And with an invitation to refine rather than romanticize. Practical life exercises seem to help fine motor control and attention in all kinds of classrooms. Systematic phonics, which Montessori already bakes into language work, is broadly supported beyond Montessori. Concrete math materials can deepen conceptual understanding before we fly to symbols. Those are elements you can evaluate piece by piece. The thread running through is simple to say and hard to build: a classroom ecology that respects concentration, hands-on sense-making, and intrinsic motivation. When that ecology is real, and not merely gestured at, the data suggest children learn more, more joyfully, and more equitably. That's not a myth of a method. It's a set of design decisions, tested in a lottery, that changed the slope of how children grew.

Ages mix across multi-year bands, so four- and five-year-olds become natural tutors for three-year-olds, and the teacher acts more like a guide than a broadcaster. The day revolves around a long, uninterrupted work cycle—often close to three hours—where concentration is treated as something sacred. No grades, no gold stars, minimal interruptions. The materials do the heavy lifting; the child drives the learning.

Those materials are not random. They're concrete, sequential, and self-correcting, so learning can move from hands to head. A classic example is the Pink Tower: ten cubes differing only in size, with no colors or numbers to cue you, so you have to attend to volume and order.

The Sound Boxes pair cylinders that differ only by the grain inside, pushing discrimination by ear alone. That same concrete-to-abstract logic reaches into literacy and math. In language, children trace sandpaper letters to connect touch and sound, and they often learn to write before they read—hand control, letter formation, and spelling as component skills that later snap together.

In math, quantities from zero to ten and the numerals that name them are introduced separately, then combined; tens, hundreds, and thousands arrive as actual bead quantities you can feel, before anyone asks you to manipulate symbols. There's a design principle threaded through all of this called "control of error"—materials that signal mistakes without an adult stepping in, letting the child revise and try again. And all of it plays out across two nested cycles: an internal cycle where a child selects and repeats a task until it's complete—or interestingly, until it isn't—and the longer, external work cycle that gives them the time and peace to go deep.

Now, that's the promise. But does it work? Here's where the evidence gets both interesting and a bit fragile.

Chloë Marshall, writing in "npj Science of Learning," pulled together the peer-reviewed evaluations and found that most are American, many are short-term, and the gold standard—true randomized trials—are rare. Fidelity complicates things further. "Montessori" can mean a lot of different things in practice, and the more the model is diluted, the harder it is to credit any effect to Montessori itself.

Even with those caveats, some signals stand out. In several quasi-experimental and lottery-based studies, younger children in Montessori scored higher on early literacy—think letter-word identification and phonological decoding—and on early math. Measures of executive function and social understanding also nudged upward.

In a striking high-fidelity preschool sample, gains were broader—reading, math, vocabulary, social problem-solving—and, crucially, how much time children actually spent with Montessori materials predicted how much they grew. That's a fidelity story in two beats: stick to the core design and the benefits appear more consistently; strip elements out or shorten the work cycle and those gains shrink or disappear. There are mixed notes, too.

Two Head Start adaptations that diverged from the Montessori ideal didn't show immediate advantages, and one middle-school comparison found lower English scores after schools added conventional grading and non-Montessori features. The upshot from Marshall's review is not that Montessori is magic, but that when it's done well it can help across cognitive, social, and academic domains—and we need stronger causal tests to isolate what's doing the work.

So let's talk about a study that leaned hard into that causal question. Angeline Lillard and colleagues used a natural experiment built into public school admissions: a district lottery for two high-fidelity Montessori magnet preschools in Hartford, Connecticut. Families applied, the state ran a computerized lottery, and some children got seats while others did not.

That creates something close to random assignment. The team followed one hundred forty-one children—seventy assigned to Montessori and seventy-one to other public programs—from about age three to six, testing them four times across roughly three years. Attrition was low, and the analyses were "intent-to-treat," meaning children were analyzed by the group they were initially offered, even if they later switched schools. That matters because it keeps the lottery's randomness intact.

These Montessori schools were not casual about fidelity. They were recognized by the Association Montessori Internationale, spanned eleven classrooms across two campuses, and were staffed by teachers trained by the Association Montessori Internationale. The control group was "business as usual" public preschool.

Everyone was tested one-on-one, in the same order, to keep procedures tight across settings. If you care about causal inference, these are the moves you want: a fair assignment mechanism, high-quality implementation, and consistent measurement.

What did they measure? Academic achievement through three widely used Woodcock-Johnson subtests—Letter-Word, Vocabulary, and Math—later combined into a single factor. Executive function with two tasks you might have seen in preschool research: Head-Toes-Knees-Shoulders, which taps inhibitory control and cognitive flexibility, and a NEPSY design copying task for working memory and planning.

Social understanding through a set of well-validated Theory of Mind tasks from Wellman and Liu. Social problem-solving via Rubin's test. They also watched how children approached challenges with a puzzle task that let them choose an easy or impossible puzzle—mastery orientation in action—and asked how much they liked academic versus recreational tasks.

Creativity came in through an Alternate Uses test: how many different ways can you use a paperclip? Under the hood, the team used growth-curve modeling, a way to estimate not just where kids are but how fast they're changing, and they used full-information maximum likelihood to make principled use of the small amount of missing data.

Here's the headline. Montessori kids' academic achievement grew faster over time. In the growth models, the Montessori group showed a steeper slope—about a tenth of a standard unit per time interval—than controls.

By the final assessment around kindergarten exit, an analysis that accounted for where children started still showed a clear Montessori advantage. That's not a one-time bump; it's a shift in trajectory. In other words, the longer children were in high-fidelity Montessori, the more that gap opened.

The story wasn't just test scores. On Theory of Mind, differences emerged by the third time point—roughly around age four and a half—and were still trending by the end. That suggests a classroom culture where children practice taking another's perspective, and mixed-age peer tutoring likely doesn't hurt.

Mastery orientation told a similar story. By the later assessments, Montessori children were more likely to pick the hard puzzle and lean into challenge rather than flee it. And when the team simply asked what kids liked, Montessori children chose schoolish tasks more often.

That's a small measure with a big implication: if you enjoy the work, you do more of it, and doing more of it is how you get good.

Some domains didn't move. Creativity, at least as captured by alternative uses for everyday objects, didn't differ at the final time point. Social problem-solving scores were similar across groups.

Executive function is a subtler case. Across the whole span, it wasn't consistently higher in Montessori, but at the age-four window, there was a measurable difference favoring Montessori that faded by kindergarten exit. That pattern fits what classroom observers often report: Montessori seems to scaffold inhibitory control and focused attention early, especially when the long work cycle is protected, but those advantages can wash out as children approach school age if comparison classrooms also emphasize routines and self-regulation.

Now the part that caught almost everyone's eye. The income-achievement gap narrowed in Montessori. Across the whole sample, income and achievement were positively correlated, as they almost always are.

But that link was stronger in the conventional programs than in Montessori. In numbers, the correlation was about zero point four six in the control group and roughly half that—zero point two three—in Montessori, and the difference wasn't just sampling noise. Zoom in on lower-income children and the pattern sharpens: by the final assessment, those in Montessori outperformed their lower-income peers in other programs, even after controlling for where they started.

Within the Montessori group, differences by income were small or statistically weak by the end of preschool; within the control group, higher-income children stayed ahead. That's a concrete equity result: same city, same admissions process, different classroom ecology, and a gap that shrinks.

There's another link that loosened in Montessori: the usual tight coupling between executive function and achievement. In the control programs, children with stronger executive function generally did better academically, which is exactly what you'd expect—better attention and working memory make school tasks easier. In Montessori, that link was flatter.

Children with lower executive function performed closer to peers with higher executive function on achievement. One way to hear that is as a design effect. When materials provide built-in feedback—when errors are signals to try again, not reasons to stop—and when children can work at their own pace without interruptions, the advantage of being the most self-regulated kid in the room may matter less for learning core academic content.

None of this makes the Hartford study a perfect test. A lottery isn't the same as flipping a coin in a lab; families self-select into application pools, and there were small wrinkles like sibling preferences. The Montessori programs were unusually high fidelity with recognition by the Association Montessori Internationale and teachers trained by the Association Montessori Internationale, which strengthens the internal test but limits how far we can generalize to "Montessori-ish" classrooms that mix in grades, worksheets, or shorter work cycles.

And by the final time point, a handful of children had moved or crossed over. The team was careful—intent-to-treat analyses, sensitivity checks, measurement models to make sure the tests were tapping the same constructs over time—but generalizability is always the question waiting at the end.

Step back, though, and a coherent picture comes into focus. Marshall's review told us that fidelity matters, and this lottery study is what that looks like on the ground: when the core conditions are in place—the prepared environment, mixed ages, long work cycles, self-correcting materials, minimal extrinsic rewards—children's academic growth speeds up, their social perspective-taking improves, they show more appetite for challenge, and they like schoolwork more. The big societal payoff sits in the equity results.

A design that reduces the penalty of lower executive function and weakens the link between family income and achievement is not just a pedagogical win; it's a public health-scale idea.

Where does that leave us? With a strong nudge toward bigger, careful tests. Lillard and colleagues argue for scaling studies that keep fidelity visible—monitor teacher training, track how much children actually use Montessori materials, and protect the work cycle—because fidelity was the hinge on which outcomes swung.

And with an invitation to refine rather than romanticize. Practical life exercises seem to help fine motor control and attention in all kinds of classrooms. Systematic phonics, which Montessori already bakes into language work, is broadly supported beyond Montessori.

Concrete math materials can deepen conceptual understanding before we fly to symbols. Those are elements you can evaluate piece by piece.

The thread running through is simple to say and hard to build: a classroom ecology that respects concentration, hands-on sense-making, and intrinsic motivation. When that ecology is real, and not merely gestured at, the data suggest children learn more, more joyfully, and more equitably. That's not a myth of a method.

It's a set of design decisions, tested in a lottery, that changed the slope of how children grew.