An Experimental Investigation Into the Application of a Learning-From-Mistakes Approach Among Freshmen Students

Abstract

To achieve a more effective teaching method, an experimental study using a learning-from-mistakes approach was investigated. A novel approach was adopted from organizational learning literature involving two steps. A first step established the psychological safety notion in students, and the second step called for a student discussion of mistakes they had made. Two classes of freshmen university students studying basic accounting participated in this study. One class was assigned as the treatment group, whereas the other the control group. Students’ performance was repeatedly measured on three separate occasions: pretreatment test, midterm examination, and a final examination as the posttest. Results showed that students from the treatment group outperformed those in the control group on the latter two occasions, whereas both groups scored similarly in the pretreatment test. It is thus concluded that such a suggested approach may ultimately help students learn more effectively.

Keywords

learning from mistakes psychological safety constructivist learning theory

Searching for more effective methods of teaching is always a priority to higher education. In recent years, the Ministry of Education in Taiwan has encouraged universities to innovate teaching approaches in areas such as problem-solving, critical thinking, creativity, and communication for students. The traditional lecturing approach may not be the best way to accomplish the above objectives, especially for Taiwanese students. Taiwanese students tend to be passive and are afraid of asking questions in front of a crowd, fearing that they may be embarrassed. In this connection, it will be difficult for the instructor to know the progress of students in the course of traditional lecturing. Some researchers have tried to use games in lectures. Others tried to adopt devices such as a smartphone in collecting feedbacks. In this article, we propose the use of learning from mistakes approach.

Theoretical Backgrounds

Learning is not simply a passive reception of knowledge from teachers (Bada & Olusegun, 2015). According to constructivist learning theory, knowledge is actively constructed by students (Duffy & Jonassen, 1991). In other words, learning is a result of “mental construction” through fitting new information with what students have already known (Bada & Olusegun, 2015). Two mechanisms were at work during construction: assimilation and accommodation (Pritchard, 2009). Assimilation is the process of incorporating new experiences into existing mental construction, whereas accommodation requires the alternation of existing mental construction when new experiences contradict the existing model. Put differently, when students encounter new experiences, they will update their mental models to reflect the new information. At the same time, they need to reconcile it with previous ideas and experiences. Sometimes, they may change what they believed previously or discard the new information as irrelevant (Bada & Olusegun, 2015). Here, we perceive that students’ past experiences and their beliefs about them play a vital role. Because each student’s experience and belief are unique, it is quite likely that individuals are placed in different levels of understanding. The teacher’s role, according to Copley (1992), is to act as a facilitator to help students to make meaningful connections between prior knowledge, new knowledge, and the processes involved in learning. Under constructivism, a teacher should serve as a model in demonstrating to students the way to reflect on their evolving knowledge and to give direction when they encounter difficulties (Tam, 2000).

According to Lebow (1993), under the traditional instructional system, mistakes or errors serve three roles. First, they provide quantitative indicators of students’ performance. Second, they help to improve the quality of instruction. Third, they may indicate the need for remediation or additional instruction. Nonetheless, these errors or mistakes were not generally used to help students to understand why their learning strategies did not work (Clifford, 1984) and, in many cases, left students frustrated with their performance. Because of this, Lebow (1993) had proposed five principles derived from constructivist philosophy, which could help the instructional systems design. One of these principles is to strengthen learners’ tendency to engage in intentional learning processes by encouraging the strategic exploration of mistakes. Mistakes, in constructivist terms, can be seen as positive stimulants that create disequilibrium for self-reflection and conceptual restructuring. On the contrary, mistakes can serve as a signal for instructors to intervene in the learning process, to find out the missing links between prior and new knowledge, and to discover the past beliefs of students. By analyzing the real cause of mistakes, students may be able to discover incorrectly held assumptions or misunderstandings of certain topics. Their collective mistakes may also expose some important details in the course materials that are overlooked when they are learned for the first time. A thorough and personal analysis of mistakes can drive students to learn deeply, rather than just passively receiving information from their teachers. Mistakes may make them to modify their learning strategies and to adopt a review process as part of overall learning.

Kapur’s (2008) productive failure can be viewed as an example of the strategic exploration of mistakes. Productive failure includes two phases: problem-solving phase and instruction phase (Kapur & Bielczyz, 2011). First, problem-solving was done before the students learned the target concept. They need to use their prior knowledge to generate solutions that are likely to be incorrect. Second, they were taught about the correct solutions. DeCaro and Rittle-Johnson (2012) had found that adopting such an approach makes children learn mathematics better. Kapur (2016) provided some reasons for such an effect. First, starting with problem-solving may help students to activate and differentiate relevant prior knowledge. Second, the failures make students realize the limits of their prior knowledge (DeCaro & Rittle-Johnson, 2012). Third, by comparing and contrasting student-generated solutions and correct solutions during the instruction phase, students will be more likely to attend to the critical features of the new concept. However, we argue that these benefits apply not only to exploration before instruction, but also to situations wherever students made mistakes.

Another strategic exploration of mistakes can be related to effective feedback. Juwah et al. (2004) had provided seven principles of good feedback practices in their conceptual model. One of their principles mentions that good feedback facilitates reflection in learning. Mistakes can force students to reflect on the reasons why they fell short of the goal and which part of their knowledge needs to be revised. Another principle is that feedback provides opportunities to close the gap between current and desired outcomes and should be done promptly. Mistakes undoubtedly indicated the current states of students and may expose the obstacles students are facing. They can also help instructors realign their teaching instructions in response to students’ needs without much delay. Consequently, students can restructure their understanding and build more powerful capabilities. A third principle argued that good feedback is not just information transmission from teacher to students, but should become a dialogue between the two parties around learning. Mistakes here can serve as a focal point for discussion and spark the interactions between teachers and students.

Learning-from-mistakes is also a popular topic in organizational learning, and it may be defined as the recognition that unexpected or undesired effects will have occurred. For example, in the car industry, learning-from-mistakes can be traced back to the 1960s when Toyota implemented its continuous improvement program on its assembly line. Rather than seeing a mistake or failure as threatening, the management hoped to exploit any and all potential benefits that could be learned from such mistakes (Rother, 2010). All of the mistakes or failures that occurred were to be sufficiently analyzed and understood for purposes of improvement. Transformation of analyzed failures into a positive contribution to performance has been a core theme in the Japanese total quality management approach (Cannon & Edmondson, 2001). Another example could be found in the aviation industry. It has been a tradition that all aviation accidents are investigated and a final report made public. Most people believe that learning-from-mistakes and failures resulting from accidents is crucial to avoiding similar future incidents (Kim & Rhee, 2017). In reality, the aviation industry’s safety record has been outstanding for many years. In 2015, the United States experienced only 0.0035 fatalities per million miles traveled by flights that occurred. Haunschild and Sullivan (2002) determined that airlines were far more likely to avoid future accidents after they learned from their own accident experiences.

Nonetheless, many researchers have agreed that learning-from-mistakes does not happen naturally (Harteis et al., 2008; Tjosvold et al., 2004). There are some psychological obstacles one needs to overcome before this may happen. For organizations, learning-from-mistakes is difficult to accept for some who view mistakes or failures as threats to an organization’s survival. Staw et al. (1981) theorized that such threats based on mistakes or failures would lead to certain forms of psychological anxiety and stress, which in turn would result in restricted information processing. Rather than exploring the possibility of acquiring new knowledge from mistakes in such circumstances, people would try to reduce the complexity of the situation by adherence to previously learned solutions. For teams, defensiveness, rather than openness to learning, may occur among members whenever mistakes or failures are found. Members may join together sometimes to make excuses for their behavior or attempt to shift the blame to other members (Tjosvold et al., 2004).

Similar reactions may also apply to individuals when they too are faced with their own mistakes. It is common for people to deny the mistakes they have made, or to attribute the causes to external factors such as the faults of others, insufficiency, or even common misfortune. The underlying mechanism is explained by Festinger’s (1957) theory of cognitive dissonance, which states that if a person encounters two cognitions that are in disagreement with each other (i.e., dissonance), a person will experience discomfort and such discomfort will motivate a person to reduce further dissonance. Most people view themselves as competent; nonetheless, a mistake made may contradict this cognition and thus change assessment possibilities. In such a case, a person may attempt to reevaluate the mistake by denying or minimizing its overall importance. However, if a mistake is covered up or ignored, an individual will have little motivation to learn from it.

A lot of other interrelated factors may also negate the potential benefits of using mistakes to learn. First, students assume different implicit theories of intelligence (De Castella & Byrne, 2015; Dweck & Leggett, 1988). Contrary to students who think that intelligence is malleable, students who presume that intelligence is fixed tend to exhibit a maladaptive motivational pattern when facing mistakes or failures (Lebow, 1993). The reason is that when they regard their intelligence is fixed, mistakes expose their limit, rather than represent opportunities to learn and improve. Second, constructivists purport that feelings, attitudes, and commitments of learners cannot be separated from the learning process (Lebow, 1993). Some individuals may not have a positive attitude toward mistakes. This may relate to their past experiences. For example, they frequently got punishments from teachers and parents whenever they made mistakes, and therefore, they tend to avoid mistakes or failures. Third, on the contrary, those who rarely experience failure may also be reluctant to learn from mistakes. Argyris (1991) had studied professionals and found that many of them never learned how to learn from failure. When he tried to teach consultants to learn, he discovered that they became defensive. This was not because they lack the abilities to learn, as many of the consultants had stellar academic records, but because those consultants implicitly tie up their self-esteem with effective performance. When they were told that their performance fell behind, they began to feel guilty and embarrassed. Such feelings make them defensive. If anyone tries to point out these consultants’ defensive behaviors, they will likely shift the blame to someone else.

It should be noted that learning-from-mistakes may be particularly difficult among Asian students. A 2018 Programme for International Student Assessment (PISA) survey carried out by the Organisation for Economic Co-operation and Development (OECD) has shown that students from several Asian countries display a high level of anxiety concerning what others think of them when they fail. Taiwan, Japan, and Singapore were among the countries with the highest levels of fear of failure. The percentages of students who expressed fear of failure in these countries were 89%, 77%, and 72% respectively, compared with the OCED average of 56%. Some researchers attribute this to the cultural factor. For example, societies such as Taiwan and Hong Kong are characterized by levels of low individualism (Auyeung & Sands, 1996) and high power distance (Hwang et al., 2005). In societies with low individualism, students tend to focus on watching passively in the classroom without obvious engagement. At the same time, students in a high-power-distance society tend to accept whatever the teacher says at face value (Sugahara & Boland, 2010). Whenever these two particular characteristics are put together, many Asian students may try to adhere closely to the teacher’s instructions and refrain from error. Furthermore, even if they make a mistake, students tend to cover it up; otherwise, they may be regarded as inattentive to the teacher’s instructions. Nonetheless, by doing so, they also miss out on the opportunity to receive clarification from the teacher in a timely manner (Cannon & Edmondson, 2001).

In addition, Asian students do not generally welcome mistakes of any kind, let alone their own. Castro and Rice (2003) found that Asian American college students are overwhelmingly concerned about mistakes and parental criticisms than European and African students. One of the reasons is that Asian parents are more focused on school achievements and demonstrated results. Okagaki and Frensch (1998) showed that Asian American parents are much more concerned about the grades compared with Latino or European American parents. In their study, Asian parents reported a significantly stronger negative reaction when their child gets “Bs” or “Cs” instead of “As” when given hypothetical scenarios. One explanation is that students’ behaviors will have definite effect on the reputation and honor of the entire family (Wang et al., 2007). It is well-known that the average Asian student tries hard to satisfy their parents’ expectations. It has also been found that Asian students are particularly concerned with socially oriented achievement (Stoeber et al., 2009), which is accompanied by a strong fear of personal failure (Bong et al., 2014). Such fear of failure or the resulting embarrassment, if discovered, makes students more likely to hide errors or to cover up mistakes that would otherwise provide ample opportunity for valuable learning opportunities to take place.

Overcoming Obstacles in Learning From Mistakes

To facilitate students to learn from their own mistakes, one should first assist them to overcome the psychological obstacles related to any admission of mistakes. When people admit to making a mistake, they may need to face some risks they are unaccustomed to. They may risk being looked down upon by teachers or by fellow students for being incompetent or may become overwhelmed with negative feelings based on pain, fear, and doubt. Quite a number of researchers would therefore agree that the creation of a psychologically safe state is necessary for people to encounter mistakes in both positive and productive ways (Cannon & Edmondson, 2001, 2005; Kahn, 1990; Kessel et al., 2012; Torralba et al., 2016). Psychological safety may be defined as a “feeling to be able to show and employ oneself without fear of negative consequences to self-image” (Kahn, 1990, p. 708). In other words, it is a situation where people can discuss their mistake without fear of ridicule or punishment (Tucker & Edmondson, 2003). Empirical evidence has been found that psychological safety is related to increased rates of medical error reporting and quality system improvement (Torralba et al., 2016). Academically, there is also evidence to support the positive relationship between psychological safety and academic performance (Soares & Lopes, 2020).

Two steps may help to create psychological safety among students. The first one is to change the overarching attitude toward mistakes (Cannon & Edmondson, 2005; Politis & Gabrielsson, 2009). Making a mistake should not be viewed as a sign of shame or weakness; rather, it should be viewed as an opportunity to learn and to improve (Cannon & Edmondson, 2005). In organizations, it is quite common for leaders to adopt a no-blame approach when discussing mistakes and failure so that members’ self-esteem will be preserved. Another purpose for such an approach is to reduce a person’s pain and dread as it is associated with a mistake. Similarly, in academia, students who have made mistakes may yet be regarded as incompetent and looked down on by others. To encourage students to have a positive attitude toward mistakes and to learn from them, it becomes rather necessary to remove such stigmas. For example, a mistake should not be labeled as “stupid” when in a discussion.

The second step involves changes in the applicable reward system. Many people in organizations tend to disassociate themselves from mistakes through a number of ways, such as cover-up or by blaming others (Tjosvold et al., 2004) as most organizations establish systems to reward success and to penalize failure. This is the same in many academic courses’ grading systems. Marks are deducted for making mistakes and those who make the least mistakes score the highest number of evaluation points. While it may be reasonable to evaluate a student’s learning outcomes in examinations, given such a traditional metric, it may also be desirable to promote student’s learning efforts in other ways such as through exercises. The purpose of exercise completion should be different from that of examination. Performing exercises should be treated as a tool for students to experiment with what they have already learned. In other words, it should be a more tolerant method of making mistakes. These mistakes should not be viewed as a lack of effort on the part of the student, but they may be viewed as the difference between the actual results and what students had previously expected. From these “learning” mistakes, students should be encouraged to both learn and improve. Thus, the grading system for engaging in exercises should not be the same as the marks-deduction system involved with examinations. Rather, it should be designed in a way that rewards real learning from mistakes.

In the following section, an experimental design will be delineated between two groups: The first is a treatment group that uses procedures that establish psychological safety parameters and encourages students to learn from mistakes. A second group, the control group, will be taught utilizing traditional methods of instruction. Our hypothesis is as follows:

Hypothesis 1 (H1): Students in the treatment group will outperform those in the control group, based upon their test scores.

Method

Experimental Design

The experiment used in this study employs two independent groups: A treatment group using a learning-from-mistakes approach and a control group using a traditional classroom instructional approach. The course chosen for this study is basic accounting, which is considered to be a compulsory course in the international college of a major private university in southern Taiwan. The textbook, language of instruction, and learning objectives were the same for all departments within the international college. The course was chosen as most questions have definite answers and, as such, mistakes or committed errors can be identified easily through a correction. Two departments had the same course taught by the same instructor. As a result, we randomly chose one department as the treatment group and the other as the control group. The treatment group turned out to be the International Tourism and Hospitality Department, whereas the control group was the International Business Administration Department. Students from these two departments were similar in many aspects. For example, they were not majoring in accounting, so there is little difference in their intrinsic motivation to participate in the course. Both departments had already started to adjust downward the possible admissions requirements for new intake students 2 years ago, and it was found that the traditional instructional approach used on new students was not as effective as the one previously used. Both classes were taught by the same instructor in English.

Materials

A pretest using 10 multiple choice questions was done in Week 3 of the semester after the first introductory chapter had been covered. Midterm examination in Week 9 and final examination in Week 17 were also used as performance evaluation.

Procedures

Treatment group

Learning from mistakes

To encourage students to learn from their own mistakes, for every exercise, the teacher instructed students to first attempt the given exercise with the knowledge they had previously obtained from the lecture. Students were allowed to form small groups for purposes of discussion. Rather than asking instructors how to proceed when they faced uncertainties, they were emboldened to attempt the problem on their own and with the assumptions and understanding they arrived at. After the students handed in their completed exercises, the instructor attempted to identify the type of mistakes made. In the next session, all of the mistakes were openly discussed and analyzed. Dialogues were encouraged between the teacher and students, as well as among students themselves. The students were then asked to correct the mistakes they had made and handed in the exercise again to get a final grade, a part of their course performance evaluation.

Psychological safety

After the pretest held on Week 3, the instructor took two steps to establish the psychological safety concerning mistakes for the students involved. The first step attempted to change students’ attitudes toward mistakes. A YouTube video (JukinVideoDotCom, 2015) was shown to students. It showed a 3-year-old boy attempting to follow the instructions of a teacher as part of a Taekwondo lesson. The little boy was too young to fully understand the teacher’s instructions concerning the breaking of a board with his foot, but he tried different things, for example, jumping on the board or stepping on the board. In the end, he successfully followed the order. The purpose of using this video corresponds to the implicit theories of intelligence. Students were persuaded that mistakes do not reflect the limit of their intelligence; rather, mistakes open the opportunity to learn and improve and are stepping-stones toward eventual success. Moreover, during the discussion of mistakes, no judgmental comments were attached to any mistake. For example, mistakes were never labeled as being bad or foolish. In other words, it is a de-stigmatization process concerning the self-perception of mistakes. The second step concerned the revision of the reward system. To alleviate students’ concern about their grades, the marking scheme was designed in such a way that a preliminary grade was given for the submitted exercises. We emphasized that this preliminary grade never becomes part of their assessment and is off-the-record. It attempts to inform students about the difference between the students’ current answers and the correct answers. Students were allowed to make corrections afterward; in this way, the next grade will become part of their assessment. The purpose is to provide students with the impression that mistakes convey information, not related to punishment.

The control group

Students in the control group were also given the same exercises to do in class and allowed to also form small groups to discuss. They were also given a preliminary grade for the first-time submission and then a final grade after correction took place. The only difference occurring was that they could ask the instructor questions when they attempted the exercises for the first time. Rather than discuss the mistakes that were made, the instructor explained the correct answers before they could make the necessary corrections.

Data Analysis

We used RStudio (Version 1.2.1335) to carry out the analysis. A total of 90 students participated in this study. Of them, 52 (57.8%) were female and 38 (42.2%) were male. In terms of nationality, 63 were Taiwanese students and 27 were international students. Fifty-five of them were in the treatment group and 35 were in the control group. Chi-square tests were carried out to determine whether there are significant differences in gender composition and nationality composition between the two groups. For gender composition, the difference was insignificant, χ²(1) = 1.99, p = .16. The difference was also insignificant for nationality, χ²(1) = 0.06, p = .81.

For the midterm and final examinations, two different sets of data were collected for each examination. The first data set included the full examination scores. In other words, the scores included the marks from all the questions of each examination. We termed this data set the whole-test score. However, the examinations for the treatment group and the control group were not conducted simultaneously. The examinations took place in the same week, but they were several days apart. Therefore, the examination questions were not the same for both midterm and final. In this connection, we had another set of data by preselecting certain questions in the examinations so that they are more comparable in both groups. In the midterm exam, one out of a total of four questions was selected for this purpose. The question selected was about the application of the double entry system in basic accounting methods. Both classes received the same question, only the numbers used in both of the questions were different. For the final examination, two out of a total of five questions were used, and the questions pertained to the closing of the books. The questions were about the preparation of worksheets and closing the accounts and are central to basic accounting practices. Again, the questions were the same for both classes with only the numbers in the questions changed; the formatting remained constant. All the grades were changed to the base of 100 for ease of comparison. The scores extracted from these preselected questions formed the second data set. We termed this data set a specific-test score.

Other cautions were also made against the possible sharing of information. First, all the students were instructed not to share information with anyone. Second, students were requested to write answers in the space provided on the question papers, and all the papers were collected after the tests.

In the main analysis, as one set of data violates the assumption of sphericity, we used the multilevel linear model rather than mixed analysis of variance (ANOVA). The dependent variable was the scores of the students. The within-subject independent variable was the three test-occasions (pretreatment, midterm, and final) where the same students attempted. The between-subject independent variable was the two groups. For each data set, we built models of four levels: baseline, test-occasions, group, and interaction between test-occasion and groups (Test-occasion × Groups). The baseline model included no predictors except the intercept. The test-occasions added the three occasions to the baseline model. The group model further added the group factor (treatment and control) and the final model added the interaction term (Test-occasions × Group). If the model at the next level is significantly different from that at the previous level, it means that the additional level is significant. The most important part of this analysis is the final model, that is, whether the two groups scored differently across the three test-occasions.

Results

Our multilevel of two data sets generated similar results. First, for the data set of the whole-test score, there was a significant main effect of the test-occasions, χ²(2) = 51.03, p < .0001, which meant that the test scores were significantly different on the three test-occasions. The main effect of the group was not significant, χ²(1) = 1.98, p = .15. Most importantly, the Test-occasions × Group interaction was significant, χ²(2) = 20.43, p < .0001, which meant that the test scores were significantly different across the three test-occasions across between the treatment and control groups. The first significant contrast was found when the pretreatment test score was compared with the average examination score between the treatment and control group, b = 5.3, t(176) = 4.25, p < .0001, r = .31. When the final examination score was compared with the midterm examination score between the treatment and control group, the contrast was marginally significant, b = 3.8, t(176) = 1.75, p < .10, r = .13. Figure 1 demonstrated the trends. For the treatment group, the improvement of scores from pretreatment test to examinations is larger than that of the control group. Moreover, the treatment group continued to improve from midterm to final examination, whereas the control group showed some decline.

Figure 1.

Performances of treatment and control groups by using the whole-test score.

Similar results were also obtained for the data set of the specific-test score. There was a significant main effect of the test-occasions, χ²(2) = 27.04, p < .0001, which meant that the test scores were significantly different in three test-occasions. The main effect of the group was not significant, χ²(1) = 0.91, p = .33. Again, the Test-occasions × Group interaction was significant, χ²(2) = 6.44, p < .05, which meant that the test scores were significantly different across three test-occasions across between the treatment and control groups. A significant contrast was found when the pretreatment test score was compared with the average examination score between the treatment and control group, b = 4.7, t(176) = 2.53, p < .05, r = .19. When the final examination score was compared with the midterm examination score between the treatment and control group, however, the contrast was not significant, b = −.41, t(176) = −.13, p < .90, r = .01. Figure 2 demonstrated the trends. For the treatment group, the improvement of scores from pretreatment test to examinations is larger than that of the control group. Both groups continued to improve from midterm to final examinations. H1 was therefore supported by both data sets.

Figure 2.

Performances of treatment and control groups by using the specific-test score.

Further analyses have been done regarding the course evaluations of the two groups. Four related questions concerning the interactions and satisfaction of students about the course were selected. Nonetheless, a multivariate analysis of variance (MANOVA) test using Pillai’s trace showed that the differences in the four questions between two groups were insignificant, V = 0.32, F(4, 66) = 0.9, p = .46. Follow-up t tests and Wilcoxon’s ranked sum tests were carried out to test for the significance of difference for each question and all of the differences were found to be insignificant. Therefore, no significant difference was found concerning the course evaluations of the two groups.

Discussion

Results indicated that the class that was encouraged to use the learn-from-mistakes approach in a psychologically safe environment achieved better improvements than the traditional class. Originally, the two classes’ scores were similar according to the pretreatment test. Nonetheless, the treatment group later performed better than the control group had on the midterm examination. From Figures 1 and 2, we observed a similar pattern in both the whole-test scores and specific-test scores. For the whole-test scores, where the scores include all the questions in midterm and final examinations, the treatment group had a higher mean score than that of the control group in the midterm. In the final examinations, the treatment group continued to improve but the control group scored lower than the midterm. The only difference in the specific-test scores is that the control group also improved in the final examinations.

The results also indicated that substantial progress of the treatment group had already been made between pretreatment test and midterm examination, compared with that of the control group. Progress in performance continued for the treatment group after midterm but to a lesser extent than it was before. It seemed that students could benefit quickly by adopting the new approach. In other words, compared with the traditional approach, the learning-from-mistakes approach in our study provided a more effective means of improving students’ performance.

Constructivists argued that knowledge is constructed by students (Duffy & Jonassen, 1991) and that their unique experiences and beliefs play a vital role. Our observation seems to support this view as different groups of students made a variety of mistakes in their exercises. Some students had difficulties with some aspects, whereas other students had difficulties with others. Our results also lend support to Lebow’s (1993) proposal about the use of strategic exploration of mistakes to facilitate learning as the treatment group in our study made better progress than the control group. Concerning Kapur’s (2008) productive failure, it seems that students learn better not only through problem-solving before instruction but also through reflecting and discussing mistakes made across the process of study. In other words, our study goes one step further.

Our results may also support the good feedback practices proposed by Juwah et al. (2004). Discussions and dialogues between teacher and students were carried out once mistakes were discovered. This provided a timely occasion for students to reflect on what they have learned and what was misunderstood. Through the discussions, the teacher may also be able to perceive students’ current level of understanding. Sometimes, the teacher may need to revise their invalid assumptions about the students and realign the instructions to the students’ needs. Probably, learning is somehow reinforced mutually between the teacher and students. Moreover, as students did exercises in groups, they may also learn from each other. Juwah et al. (2004) had argued that peer dialogue may add value to student learning. First, students sometimes are better able than teachers to explain to their peers in their own way. Second, peer discussion may also provide alternative perspectives on problems.

Another probable contribution to our results was the factor of psychological safety. Some people regard failures as threats (Staw et al., 1981), whereas others tie their self-esteem to failures (Argyris, 1991). These people normally become defensive when others point out their mistakes. Taiwanese students ranked first in fear of failure across all the countries in the 2018 PISA report done by OCED. As feelings and attitudes are integral parts of the learning process, according to constructivists (Lebow, 1993), we believe that Taiwanese students tend to be vulnerable to the negative feelings about failures and mistakes and become less motivated to learn from them. Targeting the fear of failures, we took two steps. The first one attempted to change the overarching attitude toward mistakes. This was done through instruction through the video that mistakes should be viewed as an opportunity to learn, rather than as a sign of shame or weakness. Moreover, no judgmental comments were given for any mistake made. The second step attempted to change the reward system so that mistakes were not associated with punishments.

We believe that these steps that attempt to establish psychological safety are conducive to learning from mistakes as evidence from previous research studies supports such a relationship. For example, Carmeli and Gittell (2009) have found that psychological safety is positively correlated with learning from mistakes. Siemsen et al. (2009) had demonstrated that psychological safety is an antecedent for knowledge sharing. Recently, a study also shows that psychological safety and academic performance are positively related (Soares & Lopes, 2020). The authors argue that this is because students who are high in psychological safety are more likely to admit errors and criticize past events. Such behaviors will have a positive impact on academic performance.

Practical Implications

Based on the results of this study, it may therefore be advisable to apply the learning-from-mistakes approach to other courses, especially those taught at the elementary levels of undergraduate coursework, because mistakes are easier to be identified. Examples may include basic economics or financial management. Politis and Gabrielsson (2009) had advocated that such a learning approach may be particularly suitable for courses that promote entrepreneurial behaviors. Instructors for such courses should encourage students to try out new things. If they fail, they should reflect on why the desired results were not achieved.

Nonetheless, not all of the courses can adopt the learning-from-mistakes approach effortlessly. Some courses may not have definite or objective answers; rather, they emphasize critical and logical analyses of the subject matter. For such courses, learning from weaknesses in analyses may substitute for the learning-from-mistakes approach. In the end, it is still possible for students to improve their analytical skills to solve problems.

In reality, businessmen start expressing their interests in students who can handle failures in a positive manner. James Dyson, the founder of the Dyson Company, had encountered innumerable failures before successfully applying the notion of cyclone technology to vacuum cleaners (Syed, 2016). Dyson believed that all those failures were steps toward his ultimate goal of creating an innovative and top-selling vacuum cleaner and he claimed that he would like to hire staff with positive attitudes toward failures. Richard Branson, the founder of the Virgin Group, did not regard failure as a bad thing; rather, he opined that “failure is a wonderful way of learning” (Wiener-Bronner, 2018).

Limitations and Future Studies

One limitation of our study is whether psychological safety was effectively established in the treatment group is uncertain as it was not measured. Given the high fear of failure usually present among Taiwanese students, as surveyed in the 2018 PISA report by OCED, we believe that our attempts in the experiment may have some effects in the treatment group. Moreover, several studies have found that learning-from-mistakes cannot achieve desired outcomes without the presence of psychological safety (Cannon & Edmondson, 2001, 2005; Hirak et al., 2012; Torralba et al., 2016; Tucker & Edmondson, 2003). Nonetheless, future studies are needed to verify whether psychological safety provides an additional effect on learning-from-mistakes. For example, investigations can be done using three groups: control group, learning-from-mistakes group, and learning-from-mistakes plus psychological safety group. Alternatively, the measuring of changes in the level of psychological safety through a timeline can also help.

Another point that is worthy of further investigation is how long the learning-from-mistakes approach can take effect. Our results indicated that substantial progress was made between pretreatment and midterm examination, whether using the whole-test score or specific-test score, and relatively small improvements continued until the final examination. It will also be fruitful to see whether progress is started when the new approach starts or when the attitudes toward mistakes change.

The third limitation is that in this study the students’ evaluations concerning interaction and satisfaction by the two groups were not statistically different although the treatment group’s performance was better in the midterm and final examinations. Future studies may need to address this issue.

Conclusion

Searching for more effective ways to teach students has become imperative for some universities and faculty, which would like to improve their students’ skills, such as problem-solving and critical thinking. This study showed that a learning-from-mistakes approach, combined with psychological safety measures, could be one of the possible solutions to achieving more effective learning. When one does not view mistakes as a threat to self-esteem, but instead views them as an opportunity for learning, one will start to reap the benefits of learning-from-mistakes.

Nonetheless, future studies are still needed to determine whether this innovative approach can be adopted in other areas. Moreover, one needs to put some effort into studying how each element in such an approach works and interacts with one another.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Wing Shing Lee

References

Argyris

(1991). Teaching smart people how to learn. Harvard Business Review, 69(3), 99–109.

Auyeung

Sands

(1996). A cross cultural study of the learning style of accounting students. Accounting & Finance, 36(2), 261–274.

Bada

S. O.

Olusegun

(2015). Constructivism learning theory: A paradigm for teaching and learning. Journal of Research & Method in Education, 5(6), 66–70.

Bong

Hwang

Noh

Kim

S-I.

(2014). Perfectionism and motivation of adolescents in academic contexts. Journal of Educational Psychology, 106(3), 711–729.

Cannon

M. D.

Edmondson

A. C.

(2001). Confronting failure: Antecedents and consequences of shared beliefs about failure in organizational work groups. Journal of Organizational Behavior: The International Journal of Industrial, Occupational and Organizational Psychology and Behavior, 22(2), 161–177.

Cannon

M. D.

Edmondson

A. C.

(2005). Failing to learn and learning to fail (intelligently): How great organizations put failure to work to innovate and improve. Long Range Planning, 38(3), 299–319.

Carmeli

Gittell

J. H.

(2009). High-quality relationships, psychological safety, and learning from failures in work organizations. Journal of Organizational Behavior: The International Journal of Industrial, Occupational and Organizational Psychology and Behavior, 30(6), 709–729.

Castro

J. R.

Rice

K. G.

(2003). Perfectionism and ethnicity: Implications for depressive symptoms and self-reported academic achievement. Cultural Diversity and Ethnic Minority Psychology, 9(1), 64–78.

Clifford

M. M.

(1984). Thoughts on a theory of constructive failure. Educational Psychologist, 19(2), 108–120.

10.

Copley

(1992). The integration of teacher education and technology: A constructivist model. In Carey

Willis

C. R. D.

Willis

(Eds.), Technology and teacher education (p. 681). Association for the Advancement of Computers in Education.

11.

DeCaro

M. S.

Rittle-Johnson

(2012). Exploring mathematics problems prepares children to learn from instruction. Journal of Experimental Child Psychology, 113(4), 552–568.

12.

De Castella

Byrne

(2015). My intelligence may be more malleable than yours: The revised implicit theories of intelligence (self-theory) scale is a better predictor of achievement, motivation, and student disengagement. European Journal of Psychology of Education, 30(3), 245–267.

13.

Duffy

T. M.

Jonassen

D. H.

(1991). Constructivism: New implications for instructional technology? Educational Technology, 31(5), 7–12.

14.

Dweck

C. S.

Leggett

E. L.

(1988). A social-cognitive approach to motivation and personality. Psychological Review, 95(2), 256–273.

15.

Festinger

(1957). A theory of cognitive dissonance (Vol. 2). Stanford University Press.

16.

Harteis

Bauer

Gruber

(2008). The culture of learning from mistakes: How employees handle mistakes in everyday work. International Journal of Educational Research, 47(4), 223–231.

17.

Haunschild

P. R.

Sullivan

B. N.

(2002). Learning from complexity: Effects of prior accidents and incidents on airlines’ learning. Administrative Science Quarterly, 47(4), 609–643.

18.

Hirak

Peng

A. C.

Carmeli

Schaubroeck

J. M.

(2012). Linking leader inclusiveness to work unit performance: The importance of psychological safety and learning from failures. The Leadership Quarterly, 23(1), 107–117.

19.

Hwang

N.-C. R.

Lui

Tong

M. Y. J. W.

(2005). An empirical test of cooperative learning in a passive learning environment. Issues in Accounting Education, 20(2), 151–165.

20.

JukinVideoDotCom. (2015, August 10). Little boy trying to break board in taekwondo—The New Karate Kid. YouTube. www.youtube.com/watch?v=pLx77bG1GSQ&t=0s&index=7&list=PLORwTFZaxLboASHh5pxSS6ohPNEhBavRc

21.

Juwah

Macfarlane-Dick

Matthew

Nicol

Ross

Smith

(2004). Enhancing student learning through effective formative feedback. https://www.heacademy.ac.uk/sites/default/files/resources/id353_senlef_guide.pdf

22.

Kahn

W. A.

(1990). Psychological conditions of personal engagement and disengagement at work. Academy of Management Journal, 33(4), 692–724.

23.

Kapur

(2008). Productive failure. Cognition and Instruction, 26(3), 379–424.

24.

Kapur

(2016). Examining productive failure, productive success, unproductive failure, and unproductive success in learning. Educational Psychologist, 51(2), 289–299.

25.

Kapur

Bielczyz

(2011, July 20–23). Classroom-based experiments in productive failure [Paper presentation]. Proceedings of the Annual Meeting of the Cognitive Science Society, Boston, MA, United States.

26.

Kessel

Kratzer

Schultz

(2012). Psychological safety, knowledge sharing, and creative performance in healthcare teams. Creativity and Innovation Management, 21(2), 147–157.

27.

Kim

Rhee

(2017). How airlines learn from airline accidents: An empirical study of how attributed errors and performance feedback affect learning from failure. Journal of Air Transport Management, 58, 135–143.

28.

Lebow

(1993). Constructivist values for instructional systems design: Five principles toward a new mindset. Educational Technology Research and Development, 41(3), 4–16.

29.

Okagaki

Frensch

P. A.

(1998). Parenting and children’s school achievement: A multiethnic perspective. American Educational Research Journal, 35(1), 123–144.

30.

Politis

Gabrielsson

(2009). Entrepreneurs’ attitudes towards failure: An experiential learning approach. International Journal of Entrepreneurial Behavior & Research, 15(4), 364–383.

31.

Pritchard

(2009). Learning theories and learning styles in the classroom. Routledge.

32.

Rother

(2010). Toyota Kata: Managing people for improvement, adaptiveness, and superior results. McGraw-Hill.

33.

Siemsen

Roth

A. V.

Balasubramanian

Anand

(2009). The influence of psychological safety and confidence in knowledge on employee knowledge sharing. Manufacturing & Service Operations Management, 11(3), 429–447.

34.

Soares

A. E.

Lopes

M. P.

(2020). Are your students safe to learn? The role of lecturer’s authentic leadership in the creation of psychologically safe environments and their impact on academic performance. Active Learning in Higher Education, 21, 65–78.

35.

Staw

B. M.

Sandelands

L. E.

Dutton

J. E.

(1981). Threat rigidity effects in organizational behavior: A multilevel analysis. Administrative Science Quarterly, 26, 501–524.

36.

Stoeber

Feast

A. R.

Hayward

J. A.

(2009). Self-oriented and socially prescribed perfectionism: Differential relationships with intrinsic and extrinsic motivation and test anxiety. Personality and Individual Differences, 47(5), 423–428.

37.

Sugahara

Boland

(2010). The role of cultural factors in the learning style preferences of accounting students: A comparative study between Japan and Australia. Accounting Education: An International Journal, 19(3), 235–255.

38.

Syed

(2016). Black box thinking. John Murray.

39.

Tam

(2000). Constructivism, instructional design, and technology: Implications for transforming distance learning. Educational Technology & Society, 3(2), 50–60.

40.

Tjosvold

Z. Y.

Hui

(2004). Team learning from mistakes: The contribution of cooperative goals and problem-solving. Journal of Management Studies, 41(7), 1223–1245.

41.

Torralba

K. D.

Loo

L. K.

Byrne

J. M.

Baz

Cannon

G. W.

Keitz

S. A.

. . .Kashner

T. M.

(2016). Does psychological safety impact the clinical learning environment for resident physicians? Results from the VA’s Learners’ Perceptions Survey. Journal of Graduate Medical Education, 8(5), 699–707.

42.

Tucker

A. L.

Edmondson

A. C.

(2003). Why hospitals don’t learn from failures: Organizational and psychological dynamics that inhibit system change. California Management Review, 45(2), 55–72.

43.

Wang

K. T.

Slaney

R. B.

Rice

K. G.

(2007). Perfectionism in Chinese university students from Taiwan: A study of psychological well-being and achievement motivation. Personality and Individual Differences, 42(7), 1279–1290.

44.

Wiener-Bronner

(2018, May 11). How Richard Branson went from high-school dropout to billionaire entrepreneur. CNN Business. https://money.cnn.com/2018/04/26/news/companies/richard-branson-rebound/index.html