Why Even Great Teaching Strategies Can Backfire And What To Do About It

A famous study found that when preschool children were incentivized to draw they no longer chose to do so when they had free play time.

A famous study found that when preschool children were incentivized to draw they no longer chose to do so when they had free play time. (iStock)

Educators often look for classroom inspiration from instructional strategies that “work,” focusing on how many students improved based on a given strategy. While that’s important and helpful, focusing only on how a strategy works, without examining why it didn’t work for some learners, is a missed opportunity. Examining the conditions when a strategy is ineffective or unintentionally misleads students doesn’t necessarily mean teachers should abandon that strategy altogether, but it does help them plan ahead for how it might backfire.

“What seems to be a great way to learn for the teachers, the students, the instructional designers is often a great way to learn,” said Daniel Schwartz, dean of Stanford’s Graduate School of Education, at the Learning and the Brain conference in San Francisco. “But sometimes it’s a horrible way to learn.”

There are many examples in education of ideas implemented as though they were gospel backfiring because educators lost sight of the nuances. Rewards are a commonly misapplied tool in education, for example. Simple behavior theory predicts that rewards produce more of a desired behavior, while punishments yield less undesirable behavior. But a famous study by Mark Lepper, David Greene and Richard Nisbett found that misapplied rewards can have disastrous consequences for intrinsic motivation.

For their study, Lepper, Greene and Nisbett first observed a preschool classroom for baseline observations and found that drawing was one of the most popular activities. They wanted to test intrinsic versus extrinsic rewards, so they put out felt-tipped markers (a big treat) at the art table and told one group of students that if they chose drawing during free play time they would get a certificate with a gold seal on it. A second group was not told about the reward, but after making art they received one. The third group was neither told about the rewards nor received one. After a week or two, the researchers again put out the felt-tipped markers and observed from behind a one-way mirror what activities the children chose to play with on their own.

Children in the reward condition chose to draw much less during a three-hour play period than either of the other two conditions. What happened? “The [certificate] replaced the satisfaction of drawing,” Schwartz said. “When there was no more reward, the kids didn’t want to draw.” And, interestingly, when kids were being rewarded for their drawings, they produced less creative work.

Another example is the commonly believed notion that treating each case as unique is a good problem-solving strategy. But this, too, can be misapplied. “Sometimes you design instruction that leads students to inadvertently do the wrong thing,” Schwartz said.

In one study done with college undergraduates, physics students were learning about how magnets affect electric current. They were given three cases of how a magnet interacted with a lightbulb attached to a wire loop. In Case A, the magnet moved right and the lightbulb lit up. In Case B, the magnet moved up and the lightbulb did not light up. In Case C, the magnet was flipped and the light went on.

Students were asked to come up with one account that could explain all three cases. They were placed in two groups, one of which was asked to use the “Predict-Observe-Explain” (POE) strategy, common in science education. This is a difficult problem and only about 30 percent of the control group got the correct answer: the lightbulb lights up with a change to the x-vector of the magnetic field. However, only one student was able to get the right answer in the POE group.

The researchers found that when students used POE, they treated each case as separate and weren’t looking for patterns across the cases. Schwartz said another way the each-case-is-unique idea can go wrong is when students are doing problem sets. They often treat each problem separately, instead of thinking about how they relate.

This is an example of what Schwartz calls a “learning frailty,” or things students are likely to do and that teachers can predict and plan to circumvent. To do this, teachers often have to explicitly tell students what the frailty is and advise them not to give into it. “You have to address what you want them to do, but also what you don’t want them to do,” Schwartz said.


Schwartz wanted to know whether he could teach students to seek constructive feedback and to explore a space before prematurely settling on an idea, both strategies found to improve learning. He inserted an intervention into the setup of design thinking activities that 200 sixth-graders were doing in math, social studies and science. Students went through a design cycle where they were told to explore materials and ideas, generate solutions, create prototypes and reflect on the process.

One group was told that at each stage of the design process, they should seek constructive criticism on their idea. They were also told to avoid the learning frailty, “we like to hear what we have done well,” in favor of criticism that would help them improve. The other half were told that at each stage of the design process they should resist the temptation to settle on the first idea (the learning frailty), and instead to try multiple ideas before picking one.

Measuring whether these interventions taught the students to use the strategy on their own was tricky because Schwartz and his team were interested in whether students would recognize the value in the strategy and choose to use it on their own when they weren’t explicitly told to do so. They needed a way to measure choice, not knowledge, so they chose a game format.

Screenshot from Schwartz' feedback game. Students could choose to either hear positive or negative feedback on their posters.
Screen shot from Schwartz’s feedback game. Students could choose to either hear positive or negative feedback on their posters. (Courtesy Dan Schwartz)

The seeking-criticism group played a game in which they are hired to make posters for booths at a school fair. The game offers various tools kids can use to create the posters, and then students present their first draft to a focus group of animals that provide feedback that includes praise as well as constructive criticism. Students read the feedback, make changes to the poster, and then see how many tickets they sold. Researchers were looking for how often students chose to hear more feedback from the focus group and made changes to their posters as part of their process.

“The more feedback you choose in this game, the more likely you do well on the California standardized tests,” Schwartz said. He also found that lower-achieving kids weren’t using this strategy before the intervention, but after the design thinking project they recognized its power and did use the strategy more. Kids who were already high achievers were already using this strategy, so it didn’t make much difference.

Similarly, Schwartz designed a game for the group that was asked to design in parallel instead of choosing the first idea they had. In the game, students are photographers with a variety of settings on their cameras. The game measured how many different camera settings students tried before settling on their final version. And, once again, kids who had not previously used the “exploring the space” strategy did improve.

“You want to teach students what to do and what to avoid. And acknowledge why you’d want to avoid it,” Schwartz said. Another common learning frailty is to do the thing that takes the least time. Teachers can try to circumvent the frailty by explaining why a better strategy, while more time-consuming, will pay off in the end.

Schwartz is wary of anyone who says teachers should never lecture, or never give rewards because it is “bad pedagogy.” “The key here is understanding that these instructional moves are good. You just have to figure out when,” Schwartz said. Rewards work well to incentivize something students don’t like to do, but educators have to be careful about unintentionally reinforcing the idea that whatever is being rewarded is work and therefore not fun.

Similarly, some educators argue that telling students information is wrong or anti-constructivist, but there is a time and a place for telling students information, a relatively efficient way to transfer knowledge. Schwartz and Bransford completed a study in 1998 showing that when college students analyzed contrasting data sets from classic psychology experiments and then read a text or listened to a lecture about why those experiments were important to the development of psychology, they were more prepared to understand and contextualize the new information. The students were then better able to grasp the outcomes of a similar set of data a week later, as compared to students who had summarized the information before the lecture. The analyze-and-lecture condition also predicted more accurately than students in a condition who analyzed the data twice.


Ultimately, Schwartz’s warning about unintended consequences of instruction is a rallying cry for teacher professionalism. “The science points out what’s necessary; the trick is making instruction where that component sits in an environment that’s sufficient for learning,” Schwartz said. For example, scientists can prove that overwhelming students’ cognitive load is bad. But reduce cognitive load too much and students are bored. That’s why teachers are so important; they are the investigators carefully taking note of how different students respond to strategies in the classroom, and are constantly tweaking ideas to improve them.

“The scientists can give you certain laws about learning, but they can’t put it together into instruction,” Schwartz said. They understand the neuroscience, not how to translate it into a classroom environment. That’s why Schwartz believes the most important thing for good instruction is for the teacher to be an “adaptive expert,” someone who is constantly reflecting, and learning from what he or she has tried in the past. Adaptive experts have growth mindsets about their teaching, whereas “routine experts” get good at one way and repeat it over and over.

“You develop a great deal of expertise by designing instruction and looking at the outcomes of the instruction,” Schwartz said. “You as the teacher need to think about this as a creative endeavor.” Observing how students interpret a lesson and thinking through what learning frailties may have led them in the wrong direction is one way to try to avoid unintended consequences of instruction.

This discussion of instruction misfiring may feel frustrating for educators looking for tried-and-true research-based strategies, but it also reaffirms the importance of educators’ expertise in the classroom. The one guideline Schwartz offers is that often when the rationale for an instructional strategy is to save time or be more efficient, the likelihood of an instructional backfire is high. Resorting to only telling students things, rewarding them for doing what you want them to do and oversimplifying are all ways this can happen.

  • “Instruction misfiring may feel frustrating for
    educators looking for tried-and-true research-based strategies, but it
    also reaffirms the importance of educators’ expertise in the classroom.” This is really a true conclusion, as the final result leads to the educators expertise in the classroom. If you want a best tuition experience for your wards then, check out Edventure.in for the best private tuitions in Mumbai & in India.

  • KC

    This article makes several important points about how to best implement teaching strategies in the K-12 classroom; the perspective of Dan Schwartz, as well as the use of various research studies to support his ideas, significantly enhances the quality of the article. It is important to recognize these “learning frailties” in the classroom in order to create an environment that is constructive and conducive to learning. I especially like Schwartz’s quote, “The science points out what’s necessary; the trick is making instruction where that component sits in an environment that is sufficient to learning.” Educational research is imperative to the success of a classroom environment. However, it is important that all levels of analyses that factor into a classroom are considered when implementing instructional methods. There are cognitive representations and processes, cognitive constructs, educational constructs, as well as specific attributes of a child (such as reading abilities and motivation), that all interact to shape a classroom environment. Trying to implement research-backed instructional methods at the classroom room level of analysis, while ignoring lower levels of analysis such as cognitive constructs, can hinder the ability to create a strong learning environment. Schwartz’s perspective is careful to consider all of these levels of analyses in the classroom, stressing that role of the teacher is to be able to adapt to the internal structure of a classroom. That being said, I do believe that there are additional components of research to be discussed in order to make aspects of your article stronger.

    First, you start this article by discussing research of Leppert et al. (1973) on how the use of misapplied incentives can backfire in a classroom. This is one of many, many research articles proposing that incentives can ultimately reduce task engagement after they end. However, recent research has found that although temporary incentives may decrease task engagement immediately after the task has ended, there are no long-term detrimental effects of these incentives on task engagement; in fact, post-incentive task-engagement will equal, or exceed baseline engagement levels, and the immediate decrease in engagement is rather due individual desire to take a “break” after performing tasks (Goswami and Urminsky, 2017). Therefore, although Leppert et al. (1973) presents evidence for one side of the incentive argument, this recent research of Goswami and Urminsky contradicts such studies. Including this research in your article could serve to strengthen your argument that the manner in which instructional strategies are applied is critical to the ultimate success of the strategy.

    Additionally, you also discussed the tendency of students to treat each problem separately, instead of thinking about how they relate, when completing problem sets. When reading this, I interpreted this as referring to difficulties in abstraction and transfer of material. The research of Shemwell et al. (including Schwartz) on the “Predict-Observe-Explain” strategy was used to support this claim in your article. However, I think incorporating additional examples of research on abstraction and transfer could strengthen this area of your article to emphasize this common “learning frailty,” as Schwartz would call it. For example, one of the best examples of research that demonstrates the difficulties of transfer during problem solving is that of Gick and Holyoak (1980, 1983) and the “classic radiation problem.” Here, participants are asked to propose a solution as to how to destroy a man’s inoperable tumor with radiation while also leaving healthy tissue intact. Results revealed that even if participants were told a story prior to being asked to solve the tumor problem that was analogous to the tumor problem (and also proposed a solution to it) but with different surface features, participants were still unable to recognize the identical structures in the two stories due to the surface differences. The background story only helped in solving the tumor problem is the participants were told to that the story might help them to solve the problem. This has important implications for student instruction. Your proposed method to circumvent abstraction difficulties is to have teachers “explicitly tell students what the frailty is and advise them to not give into it,” however I believe this is too simplistic of a solution. Research has proposed several alternatives to helping facilitate abstraction and transfer for students. For example, one solution to improve problem-solving performance is to incorporate explicit sub-goal labels to help students determine what is happening at each step of the problem-solving process, so that they understand important structural aspects of a new problem (Catrambone, 1996, 1998). Or, reducing the superficial, context-specific aspects of a situation when learning new material can be an effective strategy in promoting transfer, since interesting, but structurally irrelevant information, can distract students from learning important concepts (Garner, Gillingham, & White, 1989).

  • Sydney Cadogan

    As an undergraduate psychology and education student, I found this article particularly thought provoking as it addressed the fine line issue of praise and rewards- something recently discussed in class. The Lepper, Greene, and Nisbett study does a great job of highlighting how extrinsic rewards have the potential to backfire with their experiment revolving around markers and certificates. The major issue with telling student’s upfront that they will receive a reward is that it creates a misattribution of motivation. Instead of playing with the markers for fun, they play with the makers to earn the shiny, tangible certificate. Once that reward is obtained, in the eyes of the young student, there is no longer a point in choosing to play with the markers. It is also important to note that the students who play with the makers, once promised a reward, substantially decreased the creative quality of their drawings and appeared to only draw to complete a drawing, and in turn, get a reward. This “backfire” occurs when students initially liked the activity before being promised an end reward and when there were in fact expecting to receive the reward upon completion. On the idea of taking shortcuts, if a student is promised a reward upon completion of a certain number of books for example, students who do not find reading the most enjoyable pastime will just read picture books as opposed to suggested reading list books in order to simply reach that minimum number.

    I can imagine that as a teacher, it is especially difficult to avoid falling into the trap of promising rewards or desired outcomes to students. “Work” is something every student dreads at times, and being promised a sticker might allow encourage a student to complete an assignment in the short run, but at the cost of quality and presentation of work. Since kids will not continue to work if tangible rewards stop or run out, it is plausible to argue that consistent, verbal/written affirmation and praise to students is a much more beneficial to superficial and tangible rewards. Instead of sticker, a spoken “job well done” or a sincere note of praise written on a student’s paper will have much more pronounced long-term effects on a student’s desire to learn in school.

    The challenge of encouraging teachers to become “adaptive experts” within the classroom is something else that could pose a future issue moving forward. It appears to be that some teachers become too accustomed to going through the motions of a certain classroom dynamic that they will neglect to see any fallacies within those strategies. The contrary is also true—a teacher might hear that extrinsic rewards or certain peer learning strategies can be ineffective and thus eliminate them completely from the classroom. There is also evidence that suggests that extrinsic rewards could have short term benefits (Goswami and Urminsky, 2017). My suggestion for moving teachers in the direction of becoming these creative investigators is to incorporate this into teacher evaluations. One such method of evaluation, the CLASS method, specifically addresses the student-teacher interactions that promote student learning and how they affect classroom climate for students. Adding a component within these evaluations that specifically addresses flexibility and how effectively a teacher can revise a standard lesson plan to a targeted classroom will further provide teachers with the assurance that it is ok to tweak strategies and step back and further analyze exactly how students are interpreting specific teaching strategies.

  • JackM

    This article’s concept of “learning frailties” illustrates the importance of metacognition in education, a large part of which is making sure that students know proper learning strategies and can apply them. Gunstone’s research into metacogntion in education has served as a basis for showing the effectiveness of these strategies, and there’s been more research done on planning as part of metacognition (Anderson, 2005) to show that students require these techniques to learn properly and without them, students and teachers both feel like they have failed at some point in the process.

    This article also provides good insight of a persistent problem in education, the problem of extrinsic rewards in the classroom which impede intrinsic motivation. The Nisbett marker study is a landmark in demonstrating this effect and newer research on self-determination theory by its leaders E.L. Deci and Richard Ryan show how crucial internal motivation when completing a task as opposed to influence from external factors. It is important to remember that while this research has been very conclusive, putting this knowledge into action will require a great deal of effort from both administrators and educators, who are often set in the ways of extrinsic motivation which was famously written about by Alfie Kohn. These teachers cannot simply have their stickers or treats taken away; they need to learn how to emphasize effort and promote students’ unique curiosities while providing good feedback, which is not an easy undertaking.

    The importance of feedback and how it improved standardized test scores in Schwartz’ study was very informing, but also brings a conflict when it comes to grades. Many people think of grades as extrinsic motivators which pressure students, but they are also a very basic form of feedback which is much more simple for a teacher to implement that written feedback for every assignment, so we have to find a way to keep providing feedback while lessening the amount of extrinsic motivators teachers are using.

  • itforce

    nice article admin good information you mentioned in your post we like it keep post


Katrina Schwartz

Katrina Schwartz is a journalist based in San Francisco. She's worked at KPCC public radio in LA and has reported on air and online for KQED since 2010. She's a staff writer for KQED's education blog MindShift.

Sponsored by

Become a KQED sponsor