Do Some Students Learn Faster Than Others?

A deep dive into the research

Something striking about teaching is that some students seem to pick up what we're learning almost immediately while others learn more slowly. It seems like a truism that some students learn faster than others.

I'm going to tell a story about trying to understand research around whether some students learn faster than others. The research isn’t crystal clear and there’s plenty of contradicting evidence. It’s not enough to draw firm conclusions. Still, it's been enough to shift how I think about learning.

An Astonishing Regularity

A few months ago I stumbled across this study, "An Astonishing Regularity In Student Learning Rate." A group of researchers studied student learning through four different online learning modules. They learned two things. First, students started with a very wide range of initial knowledge. Second, students seemed to learn at very similar rates. Here is the key figure from the paper:

Each line represents a student’s learning trajectory. The y-intercept represents their knowledge at the start of the program, and the line shows their progress. The “astonishing regularity” is that the learning trajectories look almost identical.

This is only one study, looking at a few specific skills in an online practice program. And while three of the panels show learning curves that are more or less parallel, the fourth does show significant differences. My claim here isn’t that every student learns at exactly the same rate, it’s that those rates are more similar than they are different.

Before I started wrestling with this line of research my belief was that some students learn faster than others. Anecdotally I think that idea is a kind of "folk knowledge" in teaching. I don't remember anyone ever telling me in education school that some students learn faster than others, but the weight of my experience is that they do. It seems like many other teachers believe the same thing. This one paper doesn't get to end the conversation; more research would be needed to draw a firm conclusion. Still, it's provocative: maybe what seem like faster learners are just students who show up knowing more, and slower learners are students who show up learning less.

A Second Study

Here's a second study that tried to look at genetic correlates of intelligence vs learning speed. This type of research is messy. There's no "intelligence gene." Instead there are thousands of different genes that each have a small contribution toward things that researchers measure like IQ, and IQ is a flawed measure of what we value in school.¹ Still, I think a lot of people would expect a clear result: people who have more genes that are associated with intelligence are likely to learn faster. That’s not what happened: the study didn’t find statistically significant differences in learning speed. The authors point out that more research is needed and their sample size was probably too small. We could get nit-picky, they looked at nine different cognitive skills using a few different types of correlations and there are some places where you could pick out a small effect. Still, I find it interesting that with a sample of 131 people, nine different cognitive measures, and 100 days of learning and testing, they couldn’t find a clear pattern where some people learned faster than others. I can’t draw a conclusion from this study that everyone learns at exactly the same speed. Still, the pattern seems to be that people are more similar in learning speed than we are different.

Actual Classrooms

A reasonable response to the research I shared above is that it doesn't reflect real classrooms. One study was a very structured online practice program, the second was a tightly controlled lab study. Does this apply to real classrooms?

Graham Nuthall spent years studying classroom learning in New Zealand with an interesting methodology: using cameras and microphones, his team captured as many details about learning in a single classroom as they could over several months. Since the cameras and microphones were omnipresent, students and teachers got used to them and the team was able to capture learning in incredible detail. The team learned three major things that are relevant to my thesis. Nuthall published multiple papers on the topic, this one is a good overview of the methods.

First, they took a detailed inventory of what students knew at the beginning of the learning process. They asked teachers what they intended students to learn, then talked to students to see what they already knew. Students already knew, on average, about 50% of what the teachers intended to teach! One thing that is important about the previous two studies is that they very carefully controlled for prior knowledge. Nuthall's work suggests that there is a very large amount of variation in prior knowledge in actual classrooms.

Second, they observed the learning process very carefully to see when and how students interacted with different ideas, and then followed up at the end to see which things students learned. Using their observations, they could predict with about 80% accuracy which ideas students would learn. Without peering into students' brains that seems like a remarkable level of accuracy to me. The interesting part is, this prediction only used information about what students were paying attention to and engaging with in class. Their model didn't take into account any characteristics of the students besides what they observed directly. The implication here is that, when observed with this level of detail, they didn't find large differences in how quickly students learned.

These observations lead to an obvious question: if students learn at such similar speeds, why is there variability in learning at all? Why aren't all students in the exact same spot? Nuthall also observed that what each student learned was very different from what other students learned. Students paid attention to different ideas, processed ideas at different depths, and had surprisingly disparate experiences despite being in the same classroom. Most students learned things that were totally unique to them, ideas that no other student learned. The variation in student experiences within the same lesson, on top of the differences in prior knowledge, mean that a group of students can have a wide variation in what they know and can do. Even if learning happens at a similar speed between students when we zoom in to specific skills or pieces of knowledge, some students are learning multiple ideas at once because of what they pay attention to and engage with. Those differences in attention create differences in aggregate learning.

Conflicting Evidence

At this point in the story I have a rough thesis. The research is a bit thin, there isn't enough evidence to say "every student learns at the same rate." Instead, the evidence suggests "students seem to learn at similar rates when you account for prior knowledge. If I see some students learning much faster than others, maybe it's a good idea to dig in and try to figure out what ideas students are missing that cause them to learn more slowly."

But not all the research shows the same similarity in learning rates as the studies above. I found two categories of studies with different conclusions and I want to explore them here.

The first two are studies of learning fairly complex skills. In 1908, Edward Thorndike studied learning by having participants mentally multiply a pair of three-digit numbers, like 657 and 964, 96 times and observing to see how much people improved. He found significant differences: some people learned faster than others. The second study by Phillip Ackerman looked at learning in an air traffic control simulation, where participants acted as an air traffic controller communicating with planes in their airspace, giving headings to planes coming in to land, and giving flights permission to take off. Similar to Thorndike, the study found differences in learning rate.

These studies are interesting! Clearly not everyone learns at the exact same speed all the time. Still, I see a difference between these studies and the previous examples. Both three-digit by three-digit mental multiplication and air traffic control simulations are complex tasks. Each is composed of lots of different little skills. In both studies, participants didn’t get much instruction or time to focus on the component parts. They were tossed in, asked to practice and improve at a complex skill, and some learned faster than others. That’s a bit different than what school learning typically looks like.

The next two studies are in a different context. Christopher Zerr and colleagues studied how quickly people learned corresponding pairs of Lithuanian and English words. Some people learned faster than others. The team then ran a second study, replicating the original result and extending it to Chinese and English words, and associations between household items and locations in a circle. The result held up for Chinese and English words and, to a lesser extent, for the location pairs.

This is a clear challenge to my thesis. I can’t say that every student learns at the exact same speed; the evidence is definitely mixed. Still, one criticism I have of these studies, one that the authors point out, is that it is hard to control for prior knowledge when studying memory of words. Participants have different levels of vocabulary knowledge, and even if everyone knows the words that are being used they might have different associations with certain words that aid their learning. It’s interesting that the two language-focused parts of the study produced a more robust result than the location pairs.

I find it interesting that there doesn’t seem to be much research on this question. I tried to chase as many references as I could and this is all of the research I found that directly addresses whether some people learn faster than others. The first sentence in one of the Zerr studies above is “People vary in their ability to both learn and retain information,” yet there’s no citation for that claim. There’s not clear evidence one way or the other, and it seems like there is a common assumption of differences in learning speed that isn’t fully supported by the evidence.

Zooming Out

The studies I referenced above are all “zoomed in.” They’re focused on learning a single idea or practicing a specific skill. Zooming out, it’s clear some students do learn faster than others. Freddie DeBoer’s essay “Education Doesn’t Work” is a good summary of the research: some students perform better academically than others. Those gaps appear early on, and persist or widen as students move through their education.

I don’t think this contradicts my larger claim. In fact, it supports one of the key ideas from above. What students know has a huge influence on what they learn. The more they know, the more they learn. Through the effect of prior knowledge alone we would expect to see any gaps that appear early on to remain or get larger as time passes.

The thesis isn’t that every student always learns the same amount in every lesson. It’s that when we zoom in, isolate a specific skill, and control for prior knowledge, students learn at more similar rates than you might expect. Nuthall’s work helps to illuminate how that can be true yet students can have vastly different learning outcomes across long time periods. Each student is paying attention to different things. Some are soaking everything up around them; others engage much more shallowly, or miss opportunities to revisit ideas and build durable understanding. Learning isn’t zoomed in, it’s the constant accumulation of lots of complex experiences. When we zoom out, those differences get magnified.

Another area where a similar idea applies is in early reading instruction. There are lots of research reviews on phonics and reading instruction, here’s one by Anne Castles and colleagues that’s open access but I could easily cite others. One theme is programs that systematically introduce students to different phonemes and structure practice for those phonemes outperform programs that leave to chance which phonemes students learn and in what order. Importantly, gaps between students are narrower with systematic phonics. When instruction isolates and practices specific subskills that lead to effective reading, students learn at similar rates. These studies are large and it’s hard to compare to what I described above but the theme is similar. Learning to read is a complex skill that happens both in and out of schools. It’s not surprising that even with structured phonics there is a lot of variation in how quickly students learn to read. That’s the zoomed out view. But zooming in, being systematic and structured in a phonics program is a good way to minimize differences in prior knowledge and break reading down into small manageable skills. It makes sense that those practices narrow gaps between students who learn to read quickly and students who need more time.

Two Conclusions

What’s the point of all of this? I don’t want to make a strong conclusion like “every student learns at exactly the same speed” or “everyone can learn to the highest levels.” Those conclusions are not supported by the evidence, and they also sound ridiculous to most people who work in education. I think I can make two claims: differences in student learning speed are minimized when teachers account for prior knowledge of the topic they’re teaching, and differences in student learning speed are minimized when teachers break skills down into small, manageable pieces. So what does this mean for classroom teaching?

Every week, almost every day, I teach a concept in math class that some students pick up quickly and other students have a hard time with. One response is for me to say, “well, some students just learn faster than others.” Maybe I plan for more practice for some students, or I decide to move on because we don’t have unlimited time. A better response is to ask two questions. First, what do some students not know that’s holding them back? What concepts are they missing? What prior knowledge don’t they have? Rather than saying “ oh, some students just learn slower,” addressing prior knowledge is an actionable solution for me in the moment. Second, I can ask whether I can break a skill down into component parts. Are there smaller pieces that will be more accessible for students, that we can build back up into the whole? Again, that’s a clear action step I can take rather than throwing up my hands and saying that some students just learn faster than others.

Do some students learn faster than others? Yes. But to me, the evidence suggests much of that difference is about the knowledge students already have, and whether the learning process is structured and broken into logical pieces. I’m not saying that teachers can magically make every student learn at the same speed. Learning is messy, we can’t control everything, and even if we could there would still be differences between students. My point is that the differences in learning speed are probably smaller than you think, and if you see big differences there might be some simple ways to help the students who are falling behind.

Maybe this all seems like common sense. Good teaching accounts for prior knowledge and breaks skills down into manageable pieces. It doesn’t sound like a big deal when I phrase it like that. That’s a lot of teaching though, taking pretty simple ideas and applying them to the complexity that is real classrooms and students.

1. If you’re interested in learning more about this, The Genetic Lottery by Kathryn Paige Harden is a great place to start. Harden is a coauthor on the linked study.