Think twice and then: combining or choosing in dialectical bootstrapping?

Individuals can partly recreate the "wisdom of crowds" within their own minds by combining nonredundant estimates they themselves have generated. Herzog and Hertwig (2009) showed that this accuracy gain could be boosted by urging people to actively think differently when generating a 2nd estimate ("dialectical bootstrapping"). Although the "crowd within" promises accuracy gains, it remains unclear whether and when people spontaneously reap those gains. What makes people combine their estimates rather than trying to identify the better one? This research found that participants were more likely to combine when they were instructed to actively contradict themselves. Furthermore, they were more likely to combine as the size of the disagreement between 1st and 2nd estimate grew. People thus acted as if they were hedging against the risk of making large errors. Finally, when people pursued a strategy other than combination, they were not able to outperform their crowd within.