Convenient Overconfidence
We are more overconfident on tasks we don’t actually expect to perform, and when we don’t expect to have to explain our evaluation to others. On expecting to perform:
Participants made predictions about performance on tasks that they did or did not expect to complete. In three experiments, participants in task-unexpected conditions were unrealistically optimistic: They overestimated how well they would perform, often by a large margin, and their predictions were not correlated with their performance. By contrast, participants assigned to task-expected conditions made predictions that were not only less optimistic but strikingly accurate. Consistent with predictions from construal level theory, data from a fourth experiment suggest that it is the uncertainty associated with hypothetical tasks, and not a lack of cognitive processing, that frees people to make optimistic prediction errors. Unrealistic optimism, when it occurs, may be truly unrealistic; however, it may be less ubiquitous than has been previously suggested.
Accountability … [is] the expectation to explain, justify, and defend one’s self-evaluations (grades on an essay) to another person ("audience"). Experiment 1 showed that accountability curtails self-enhancement. Experiment 2 ruled out audience concreteness and status as explanations for this effect. Experiment 3 demonstrated that accountability-induced self-enhancement reduction is due to identifiability. Experiment 4 documented that identifiability decreases self-enhancement because of evaluation expectancy and an accompanying focus on one’s weaknesses.
It is almost as if we at some level realize that our overconfidence is unrealistic.
HH:
It looks like they tested that: "Experiment 2 ruled out audience concreteness and status as explanations for this effect."At least that's how I read that.
This is exactly Caplan's point. People are more irrational when it's cheap.
It's not that it's worth doing more cognitive work when the stakes are high, it's that people *actually try* to get it right. From an Evolutionary Psychology viewpoint, it's not surprising.
One of my favorite tricks is to ask myself "Would you bet on that at those odds?". The answer should never be obvious (if it is, shift your odds such that it isn't), so if you catch yourself thinking "WOAH! I never said anything about *betting*!", then you've made a mistake. The goal is to give odds such that you're willing to bet on either side if someone calls you out (unless conditioning on their willingness to bet changes the odds significantly).