People Are Not Indiscriminately Optimistic

Although considerable evidence suggests that people’s predictions tend to be optimistically biased, the predictions are not entirely unreasonable. For example, predictions show high relative accuracy: correlations between predicted and actual outcomes are often substantial (e.g., in planning fallacy studies, correlations ranged from .4 to .7). People are also sensitive to their skills and circumstances. Most smokers recognize they are at higher risk than nonsmokers, but still believe they are less at risk than other smokers. Performance predictions reflect past performance.

Optimistic bias is greater under higher uncertainty. Early research found the strongest bias when objective odds were near 50-50, and reduced when odds were very high or low. People are sensitive to context and exhibit more optimism when accuracy is unlikely to be tested or consequences of error are minor. Most studies are conducted anonymously, allowing grandiose estimates without fear of consequences.

Despite this general sensitivity, optimistic bias is remarkably robust. In one study, participants rated their comparative risk for heart attacks, automobile injuries, and having high home radon levels. All three hazards showed small but significant optimistic biases (mean ratings of −0.29, −0.34, and −0.62 on a scale from −3 to +3). The bias was unaffected by whether the comparative-risk questions came before or after questions about risk factors. Merely having people think about their standing on relevant risk factors without clear feedback did not decrease or increase the bias. When participants rated their own risk factors on comparative scales, optimistic (risk-decreasing) biases emerged for most factors—such as blood cholesterol, dietary fat, vehicle weight, driving skill, basement openings, and soil density—while some factors unexpectedly showed risk-increasing biases (e.g., for exercise and driving speed). Interestingly, correlations between risk-factor ratings and comparative-risk judgments became more positive when the comparative-risk judgment followed the risk-factor questions, suggesting participants shifted their judgments to increase consistency. However, this did not change the mean level of bias because perceived standing on most risk factors was itself optimistically biased.

Other studies confirm that prompting people to focus on specific risk-factor sets can, in some cases, exacerbate optimistic bias. For example, when college students read a list of extreme risk-decreasing attributes for heart disease or alcoholism and then rated their own absolute risk, they became even more optimistic about weight problems (giving lower self-risk ratings) compared to a control group or to those who saw a list of risk-increasing attributes. Thus, while people are not indiscriminately optimistic, attempts to make risk factors salient can paradoxically strengthen rather than weaken the illusion of personal invulnerability.

Temporal Proximity

Predictions become more optimistic when outcomes are distant. Calderon (1993) found that analysts’ earnings forecasts became less biased as the forecast date approached the realization date. Gilovich, Kerr, and Medvec (1993) found that distal predictions were more optimistic than proximal predictions even without additional information. Shepperd et al. (1996) found that expectations became pessimistic just before receiving feedback.

Explanations include: strategic disappointment protection, accountability for accuracy, and anxiety interpretation (where anxiety reduces confidence).

Deliberative and Implemental Mind Sets

Optimistic biases are minimized before a decision (deliberative mind-set) and exaggerated after a decision (implemental mind-set). Armor and Taylor (2000) found that people deliberating pros and cons were less optimistic than those imagining they were about to complete the task. Deliberators considered multiple perspectives, while implementers generated specific plans.

These findings suggest that deliberating before a decision can produce less optimistic judgments than preparing to act on a commitment. They do not show that optimism is always lowest when error would be most costly, or that increased optimism after a decision necessarily improves performance.

The planning fallacy is one form of optimistic bias. People may expect a task to go smoothly, judge their own abilities too favorably, or underestimate obstacles to a goal. Confidence can help sustain effort, but a useful plan still needs to allow for delays, failure, and uncertainty.