People systematically underestimate how often things go wrong in the world. Researchers call this the “failure gap.” This huge project Journal of Personality and Social Psychology.
We rely on our perception of what events are common when forming opinions, making decisions, and supporting policies. Previous research has shown that these perceptions are highly biased. Much literature focuses on optimism. For example, people tend to believe that good outcomes are more likely to occur and bad outcomes are less likely to occur, especially if those outcomes affect them. But does this trend extend beyond our personal lives to broader social issues such as crime, health, and economic ruin?
Lauren Eskreis-Winkler and colleagues investigated this broader issue, asking whether people misjudge how often failure occurs in different areas of life. They suggested that the problem may not be simple optimism, but the way information is shared. Failures are less discussed than successes because of the discomfort, embarrassment, and social costs associated with communicating them. As a result, people are exposed to incomplete information, which can lead to systematically distorted impressions of reality.
The team ran a large-scale, multi-study research program involving approximately 3000 participants, combining controlled online experiments, analysis of real-world data, and field research to understand how people perceive failure and how that perception can be changed. In an early series of studies, we asked participants from platforms such as Amazon Mechanical Turk and Prolific to estimate how often different types of failures occur in more than 30 domains, including national issues (e.g., crime, health care), global issues (e.g., poverty, pollution), and everyday personal experiences (e.g., broken relationships, returned products).
Participants sometimes estimated multiple items within a single domain, and other times focused on one type of disorder. These estimates were compared with real-world statistics extracted from official data sources. The researchers also framed the questions differently, sometimes asking directly about failures and sometimes about successes, to ensure that the bias was specific to failures rather than general misjudgments.
To investigate why these misconceptions arise, researchers looked at how often failure and success are discussed in widely available information sources. They conducted a large-scale search of approximately 2.4 million news articles using databases such as Nexis Uni and systematically compared the frequency with which failures and successes were mentioned across domains that participants had previously rated. They extended this approach to other forms of shared information, such as social media and online consumer reviews, to test whether this pattern holds across traditional news.
The researchers also designed an experiment in which participants were exposed to a carefully selected information environment. For example, a series of reviews and headlines that underestimate or accurately reflect the true rate of whether exposure to biased information shapes people’s beliefs cannot be directly tested.
Subsequent research looked beyond perception to investigate when bias disappears and what its consequences are. Participants were asked to estimate failure in situations where sharing negative experiences has recently become socially acceptable, such as reporting sexual misconduct following the #MeToo movement, allowing the researchers to test whether reduced stigma increases awareness.
Finally, we investigated how misconception correction affects real-world attitudes and decision-making through a series of field and online experiments. These included samples of voters, educators, and workplace managers who were provided with accurate statistics on failure rates and asked to make judgments about policies such as criminal penalties, school discipline, workplace bias, and parental leave.
A robust and consistent pattern emerged across the study results. In other words, people significantly underestimated how often the disorder would occur. This is true across national, international and personal domains, as well as in specific contexts such as sports, education, and drug efficacy. Even when the true answer was clear from the structure of the situation, such as in competitive sports that require a balance between winning and losing, participants still underestimated the failure rate. On average, failures occur far more often in reality than people believed, indicating a widespread and systemic gap between perception and reality.
The researchers also found strong evidence that this gap is related to how information is shared. News coverage, social media, and online reviews consistently downplayed the failure compared to what actually happened. When people were experimentally exposed to an information environment that downplayed failure, their estimates became even more inaccurate.
Conversely, if the information they encountered accurately reflected real-world failure rates, the gap narrowed. In situations where discussing failure is more normalized (such as public conversations about sexual misconduct), the usual pattern may be weakened or even reversed, suggesting that visibility and openness play an important role in shaping perceptions.
Importantly, correcting these misconceptions had important downstream effects on attitudes and decision-making. Once participants learned the true prevalence of failure, they became less supportive of harsh punitive measures, such as harsh discipline and mass incarceration, and more supportive of policy changes aimed at addressing the underlying problems. In workplace and policy contexts, increased awareness of failure rates has also reduced stigma and encouraged more supportive practices, such as extended parental leave.
Taken together, these findings show that not only do people misjudge failure, but that these misjudgments can shape important social and institutional decisions.
One limitation is that the failure gap may be context- and culture-dependent. Since most participants were from Western-educated populations, it remains unclear whether the same pattern generalizes globally.
The study, “The Failure Gap,” was authored by Lauren Eskreis-Winkler, Kaitlin Woolley, Minhee Kim, and Eliana Polimeni.
