How to teach science
“We emphasize having fun to learn science,” says researcher Lin Zhang. “But did we tell students the full story of what science and math is?”
Thank you, paying subscribers, for supporting the kind of quality reporting, analysis and expert Q&As you can find every Friday right here at The Bell Ringer. Your membership contributions are allowing me to provide more long, in-depth research posts like this one to all subscribers (especially educators) for free. If you haven’t yet, consider becoming a member—it’s only $6 a month!
For as many billions as the US spends on STEM education—along with its adjacent industry, “getting students excited about STEM education”—we already know that American students don’t do very well in math. But something similar holds true for k-12 science. The number of students performing at “below basic,” the lowest score level on the National Assessment of Educational Progress (NAEP) grows the longer students stay in school: 27% of fourth graders and 33% of 8th graders scored “below basic,” and 41% of 12th graders performed “below basic.”
I’m surprised that k-12 science education doesn’t get more attention and scrutiny, considering how it’s so closely connected to our STEM and STEAM aspirations—billions of dollars that serve a national imperative, part economic and part national security, according to just about every expert, to funnel more students into STEM college majors and STEM careers.
Taking a closer look at what makes up science learning makes sense. First, experts argue, all students need basic science literacy in a world that relies, in areas from health to climate change, on scientific information. We live in a scientific century. “Building public trust in science should be our main goal…given the increasingly casual dismissal of scientific expertise among the broader public,” wrote John Rudolph in his book Why We Teach Science.
Then there are the more practical concerns. The fastest growing job sectors are in STEM fields, nearly double the rate of other fields. Nearly all of the top college majors where more than half of graduates make six figures or more are in STEM fields—more specifically, in engineering and computer science. Math, of course, is also a big part of being able to major in college STEM fields like medicine, engineering, and computer science. But what about science? Does science play a role in how many students can succeed in STEM? And are k-12 schools meeting expectations in science education?
Quality K-12 science is up against a lot of obstacles. A 2023 analysis reported that elementary schools barely cover science topics—according to the National Survey of Science and Mathematics Education, elementary students spend 18 minutes a day on science learning. (For reference, the average for reading is 89 minutes, 57 minutes for math.) Science teaching jobs are some of the hardest to fill; more than a third of middle and high schools don’t have a biology or life sciences teacher; for physics and engineering, half of schools have no one to teach them. High-quality materials to teach science are in short supply; many states haven’t adopted or even suggested materials for teachers to use. Existing materials are surprisingly lacking in the kind of background knowledge cognitive scientists say improve not only science learning but reading comprehension as well. A NAEP survey showed that only 21% of fourth graders have access to the kind of lab and materials needed to teach the science standards.
Teacher and curricular challenges notwithstanding, there are in addition big disagreements in the field on the most effective ways to teach science. For at least the last 20 to 30 years, says science education researcher Dr. Lin Zhang, it’s been assumed, and supported by the Next Generation Science Standards, that the most effective way to teach science is through inquiry and discovery learning, in which students are presented with exploratory projects in which they behave like amateur scientists, discovering natural phenomena, recording their observations and coming up with hypotheses. One of the key features of this style of learning, Zhang says, is a teacher that doesn’t provide answers for students, but instead allows them to discover their way to often novel solutions. For the last two decades, Zhang, an associate professor at Providence College in Providence, Rhode Island, has been trying to answer questions about whether that’s the most effective way to prepare students for a life of scientific literacy and a STEM career, if students so choose.
I recently had a fascinating conversation with Zhang, and what her research shows and doesn’t show about teaching science effectively. She has been part of an ongoing debate, along with researchers John Sweller, father of cognitive load theory, and Paul Kirschner, with another set of researchers on which instructional method research supports.
Their paper, “There is an evidence crisis in science educational policy,” was challenged by another group of researchers supporting inquiry-based approaches. Then Zhang, Kirschner, et al rebut the rebuttal here—and these three papers form the basis of the following interview with Zhang.
We talked about how research supporting inquiry-based learning is often using a different kind of research than studies showing positive effects of explicit instruction, and how that should affect how we treat the results, as well as the big questions she’s trying to answer to improve science education for students.
The interview—in which I learned a ton, and hope you do, too—has been edited for length and clarity.
The Bell Ringer: How did you get interested in this work? What led you to start asking questions about how science is taught?
Lin Zhang: When I was in grad school, around 2006, educators were very much interested in inquiry-based science teaching. Later on, it became about STEM education—integrating the subjects science, math, engineering, and technology. Since then, they’ve added an arts integration, it’s called STEAM. A question I always have is: well, how do we integrate these quite different subjects? We have made a very “glory” name, and it sounds great to have everything combined. But what does integration look like in practice? In the classroom? For teachers?
For example, if you are using flour to create dough, you put some water and some flour in proportion to it to make the best bread, right? I was asking, how much do we need of each subject to make perfect dough? If you just put some water and some flour at random, and all of a sudden, you expect to have the best dough, in my opinion, that is unlikely to happen. I asked if we—in education society—have made enough effort in understanding the “integration” before we started adding more pieces to it.
Unfortunately, there was and still is minimal research on “integration.” But, there have been plenty of efforts made in expanding on “glory” terms. That’s how I got into this conversation—I’m asking for data from solid, rigorous studies that support these various highly-suggested “integrations” of science, technology, engineering, art and math teaching and learning. I can’t say I have found answers to the questions I asked.
Around 2000, in the education field, particularly in science ed, there was a push for inquiry-based teaching, a lot of grants, projects and centers were created for that. But when we looked at the data they presented, many of the studies are what we call program-based studies and they can hardly be used to support inquiry-based science teaching approaches.
The Bell Ringer: I’m not familiar with that term. What’s a program-based study?
Lin Zhang: It’s when a group creates a curriculum, project or program, and they implement the whole thing in school settings. There could be professional development embedded in the program, an enhanced learning environment created, and intensive content learning tools embedded. Then researchers do a pre- and post-evaluation to see if students made significant progress.
Many studies looking at the effectiveness of inquiry learning use this type of study. If you have a good PD focusing on content for teachers, it’s very likely you will have a learning gain. If you have a really good online portal that gives lots of support to students and emphasizes content well, there is a high possibility of productive learning. But this type of study would not allow you to say, with confidence, that learning is only caused by you doing inquiry-based teaching.
I’ve been in the field for 18 years and I haven’t seen much work examining the inquiry approaches by isolating inquiry procedures from all other instructional elements, or positive trends from this type of studies. In contrast, a lot of program-based studies are giving positive results when they only compare pre- and post- or compare their program as a whole with other programs. Without isolating inquiry procedures from other instructional elements, those results cannot be used to answer the questions about the effectiveness of inquiry-based approaches.
The Bell Ringer: What then would be the best way to study inquiry learning?
We need studies using randomly controlled trials to test instructional procedures. A comparison group and a controlled group that are identical to one another, through a random assignment, One group gets one thing or a procedure, the other group doesn’t. Then we are studying one element at a time isolating all other variables.
We have kept pressing inquiry-learning advocates, and asked: what do you mean by inquiry? Can you identify the feature instructional elements that researchers can test using randomly controlled trials? Unfortunately, there have been different terms and various descriptions. To the best of our understanding, there is a key element that features all inquiry teaching—to withhold answers and solutions from students. All the game-based, exploratory learning, inquiry-based learning, discovery learning—all those methods feature withholding answers and solutions from students.
Just to give you some context: it’s often suggested to teachers in inquiry classrooms to walk in and say, “I’m not going to tell you the answer. I’ll give you the materials, support and guidance, but not the answer. You are going to use the materials to find the answer.” That’s a key feature in an inquiry classroom. So how do we test this? We randomly assign students to two groups, in one group we give them procedures and withhold solutions, in the other we don’t withhold solutions, we tell them explicitly. Then we say, ok, which way do they learn better?
Out of controlled studies from emerging literature, very strong data suggest students learn better in explicit instruction conditions.
A program-based study tests the program as a whole, with everything in it. But a controlled study is looking at just one thing or only one instructional procedure.
The data from controlled studies simply indicate that students tend to learn better from explicit instruction than from discovery. Thus, this leads to a very critical question: what does this mean to our science education community that highly promotes inquiry learning that features discovery and exploration?
Let me give you another example. Our team is examining the “integration” of explicit instruction and exploration, using randomly controlled trials. We randomly assign students into two groups. You have one group of students do the exploration first and then do the lecture, and have the other group of students do the same lecture first and then the exploration. The only difference between the two groups is the sequence. Using this method, there are no other elements that can be colluded and after replications we can say things with confidence.
The Bell Ringer: In my experience, some people tend to react very negatively to explicit instruction and direct instruction.
People have misunderstandings on explicit instruction, and think that in explicit instruction students do not get to ask questions. That is a misconception.
**Explicit instruction simply means teachers don’t have students wonder or invent solutions when they don’t have built solutions in their mind yet.
Instead, teachers tell students what the solution is. They help students develop an understanding of the solution, then use the solution as an example to solve new problems.
The Bell Ringer: Based on your understanding, students don’t perform as well in classrooms relying on inquiry-based learning. If data doesn’t support it, then why are people so attached to it?
It is a great question and in my opinion, many factors contribute to it.
One could be that people confuse academic learning with daily experience. People think we learn many daily things by doing and it must be true for everything we learn in our life. But the kinds of knowledge we are talking about are different. To do things like listening, we often do not need to be taught. But there is certain knowledge that requires training and schooling, such as writing. It is also often the case for learning science and math.
I would say the subject matters are different—some of the knowledge can be obtained simply by having experience with it. But when the knowledge is very abstract, simply experiencing it is not going to be sufficient. Intensive training is often required, and such training often involves explicit instruction. It is unfortunate that people equate different types of knowledge the same and treat them uniformly in teaching and learning.
Another important factor contributing to the continuation of inquiry-based teaching relates to how educational research data are used for policy-making. I would say our Evidence Crisis article has done a good job in identifying disconnections between scientific data and educational policy-making.
The Bell Ringer: A lot of people are interested in using science ed for attracting more kids to science and STEM careers. It’s my understanding that one purpose of the inquiry-based science class is getting more students interested in and excited by science. Is there research on which instructional approaches tend to get more kids interested in science?
In one of the papers I’m working on with some other scholars right now, we are using 2015 PISA data and looking at how different instructional approaches are associated with student enjoyment of science. Inquiry learning is associated with high enjoyment, interest, and confidence—but so is explicit instruction. So it’s not quite right to say providing explicit guidance necessarily leads to a negative attitude toward science. There are also data supporting that explicit guidance reduces students anxiety during explorations.
With attitudes, it’s hard to draw conclusions.
It is also hard to predict if people will end up choosing STEM or not. For example, gathering data from interviews, we found that whether students pursue STEM is a very complex decision-making process. It might have something to do with their past experience, their conversations with siblings and parents, and their value of the field. There are various factors contributing to their career choices and plus, everybody has a different life journey. It’s hard to say.
I would like to frame the question differently: successful STEM professionals are often prepared with high-quality content knowledge, and explicit instruction is essential to the high-quality content knowledge acquisition, on which even the inquiry learning advocates agree. I refer to our recent debate.
However, explicit instruction has been very much deemphasized and traded off by games and play.
In our k-12 system, we teach kids to play in order to learn something, science, math, you name it—but when they go to college that’s not the case anymore. The k-12 curriculum often emphasizes playing and having fun, then all of a sudden, when they get to college or enter professional careers, students are facing a totally different thing, and there are not always games and playing.
My question is: how do we prepare our kids in k-12 so they are ready?
Did we tell them the full story of what science and math is? These two subjects are hard. It requires intensive learning and training, attention and repetition. But we didn’t tell them that, we often emphasized having fun, “it is going to be fun and you will love it!”
Do we prepare them enough to make them successful at it?
The Bell Ringer: It makes sense to take a closer look at whether inquiry-based learning helps students learn more science, and under what circumstances—it’s a legitimate question. There’s kind of an assumption being made among the science education community.
Yes, this is one of the things we have argued the back-and-forth in the debate with lead scholars in science education: there’s only one type of study overwhelmingly supporting inquiry-based learning, and that’s the program-based study we talked about earlier. If we look at large data—TIMSS and PISA, it’s consistently and repeatedly reported the more you involve students in inquiry, the lower their science achievement is.
An important point to make here is that it has been a long time since 1996, since this kind of inquiry-based learning has been promoted in science educational standards. If there ever is anything from inquiry, after three decades, now, we should be able to see it . Yet, it’s been negative across countries over years.
In one of the papers we wrote, the Evidence Crisis paper—there is one section that summarizes all these correlational studies up to 2020. Correlational studies examine trends from large populations over time.
Quite differently, in program-based studies, you as a researcher designed something and went ahead and implemented it in classrooms. And then you came back and said, “The thing I designed was excellent.” Is it questionable how objective these conclusions can be?
Similar to the objectivity of correlational studies, In a randomly controlled trial, you’re not testing somebody’s work, you're testing an instructional procedure or element. So the questions I raised are: if we create policy based solely on studies from self-created programs, shouldn’t we be concerned? Especially if the policy doesn’t include results from the other types of studies that are more objective?
The Bell Ringer: It’s my understanding in science education that there are other factors contributing to students’ low achievement as well, not just instructional strategies—teacher shortages, lack of high-quality teaching materials.
Agreed. Education is very complex, and influenced by many factors. The things you mentioned can contribute to students’ low achievement too. In my state, Rhode Island, High Quality Instructional Materials (HQIMs) have been built around NGSS (Next Generation Science Standards). They are newly developed. I’m not surprised that a lot of money has been spent on designing these materials. From my observations, they have been distributed to schools, and it's highly recommended for schools to use them.
I’d say that it’s helpful if the curriculum itself provides solid content coverage. But if there was nothing then teachers have to make it up, lesson by lesson. Remember, lesson designing is not the only goal that teachers are supposed to hit—think of how busy teachers are—it’s hard. You need years of experience to build lessons.
High-quality instructional materials are carefully designed in advance and are ready to go. I say it’s helpful in that sense. But, given that I have concerns with promoting explorations and withholding answers to deemphasize explicit instruction in science teaching, that NGSS is built to promote such instructional approaches, and that these HQIMs were built around NGSS, as they themselves claimed, my question is: are these HQIMs designed using the best scientistic data, such as randomly controlled trials and correlational studies? Data from these two types of research do not uniformly support what the NGSS suggests.
Because some of these issues get intensified in the field we observed, we feel we have the responsibility to step in. We cannot ignore them at this point. It’s why we got into the debate.
Some of my students benefit from inquiry and some benefit from explicit instruction, so my classroom incorporates both. I suppose that makes things harder to study and quantify, but I gotta do what is good for my students!
This was a terrific interview. Perhaps I have a unique perspective. I taught earth science to middle schoolers, for six years, back in the 1970s. I desired for every student to be curious about his environment; where we came from, where are we going. Of course, for many students, science was just another class. I made it as interesting and as relevant as I could.
I like the idea of discovery from inquiry. But very little material can be learned, considering the time it takes to do such inquiry. So, I did mostly explicit teaching. Still, I didn't just drone out data. I made the instruction as relevant as I could to students' experience. I encouraged questions and speculative observations (hypothesis).
Perhaps my best instruction to them was that few things are totally absolute. Theories evolve, and I would give examples. Science is a continual endeavor, not something that is settled. And I taught that it's hazardous to presume to be correct without absolute proof; and there is very little absolute proof. As much as anything, science is about having an open mind, and I think most of my students learned that from me, if they didn't already know.