Perhaps the tritest advice in self-help books over the past two decades is that we should not be afraid to fail. Failure is held up as a motivator, a learning experience, and a badge of honor. And there are certainly a lot of examples of prominent successful people who overcame failure on their pathway to riches, fame, and glory.
The willingness to persist in the face of failure and to learn from mistakes and projects that did not succeed is indeed crucial. But, there are still a number of open questions about failure that are worth addressing. Do we generally believe that people learn from failures? Do people actually learn from failures? Why or why not?
These questions were explored in a fascinating paper in 2024 by Lauren Eskreis-Winkler, Kaitlin Woolley, Eda Erensoy, and Minhee Kim in the Journal of Experimental Psychology: General.
They start by exploring whether people overestimate the likelihood that someone who has failed will succeed on a subsequent attempt. In one study, they asked people to estimate the chance that someone who has taken a licensing exam to become a lawyer, nurse, or teacher and failed will pass on their next attempt. These questions are interesting, because the actual rates of success are known (35% for lawyers, 43% for nurses, and 58% for teachers). In each case participants significantly overestimated the likelihood that someone retesting would pass (58% for lawyers, 63% for nurses, and 67% for teachers).
Is this specific to failure? Perhaps people just assume that anyone retaking a test will do better. To explore this question, participants in another study in this series were told about a person who took a teacher licensing test and received a score of 219. They were then asked what score the teacher would receive when taking the test again. Half the participants were informed that 219 is a failing score on the license test, and half were not. When participants were told that 219 is a failing score, 70% predicted that the retest score would be higher, while only 56% predicted the score would be higher when they were not told that 219 was a failing score. Interestingly, this 56% is about the same as the actual percentage of teachers who pass the test after failing it (58%).
Another set of studies found that people pay less attention to information that would help them to succeed after a failure than we expect them to. In one study, participants had to learn the meaning of symbols they had never seen before. The study was set up so that participants were asked questions about what a particular symbol meant and chose from among two options for the meaning. This task is interesting, because the experimenters could choose the feedback they gave to the participant. Participants were told that they were incorrect on all three. Then, they had the chance to answer questions about these symbols again. Before the retest, participants could study information about the symbols. Participants were paid for their correct answers, so there was incentive for them to be correct. On average, participants chose to study information for 39% of the questions. The average score on the retest was 62%. Studying the symbols was a significant predictor of getting the answer correct on the retest.
A second group of participants in this study were told about the situation and were asked to predict both the percentage of questions participants would study as well as their ultimate level of performance. These predictors overestimated both the percentage of questions people would want to study (80%) as well as the percentage of correct answers they would give (81%).
So far, the results of these studies suggests that observers overestimate people’s likelihood of success following a failure, and this is due (at least in part) to the assumption that after someone fails they will change their behavior to do more things to help them succeed in the future.
Two additional studies from this paper shed additional light on the topic. One demonstrated that it is straightforward to de-bias people’s judgments. In one study, participants predicted the likelihood that a heart attack survivor would make positive changes in their behavior. Without any additional information, participants predicted that 62% of survivors would change their behavior (the actual percentage is 47%). A second group of participants was told that a survey found that few heart attack survivors are thinking about their health on a daily basis. This piece of information decreased people’s judgments of the percentage of survivors who would make a change in their behavior to 52%, which is close to the true value.
Finally, the researchers explored why this might matter. They suggested that people who overestimate the likelihood of success after a failure might be less willing to support programs that would help people to achieve their goals. Consistent with that suggestion, they asked participants in two subsequent studies how much they would want Congress to allocate funds to opioid rehabilitation or their willingness to sign a petition for Congress to allocate funds to opioid rehab. Some participants were just asked to respond to this question, while another group was told that 91% of people who enter a recovery program for opioids will relapse. The group told the failure rate for recovery programs were more likely to want government support for rehab programs and would be more willing to sign a petition.
This set of studies suggests that the typical advice we give about failure needs some revision. People can learn from failure, but they need more explicit instruction about how to do that, because they often will not do it on their own. When someone fails to achieve a goal important to them, it is important for them to recognize that without significant effort to change their behavior, they are likely to fail again in the future. People should not be afraid to try something for fear they should fail, but when they do fail, they have to work to ensure that they learn from that failure in order to succeed later.
