Thursday, 25 June 2009

Unleash the crowd within

You can boost your quiz performance by unleashing the crowd within, a new study shows. The next time you're asked to estimate a historical date, for example, try doing the following: make your first estimate; then pause and assume your first guess was off the mark. Consider why, then use this new perspective to make a second estimate. Average your two estimates and, chances are, this newly calculated date will be more accurate than your original answer. The new approach is called "dialectical boot-strapping" and according to Stefan Herzog and Ralph Hertwig, it really works.

We've known since at least the time of Francis Galton that the averaged judgement of a group of independent individuals will nearly always outperform the judgement of a lone individual, no matter how expert he or she is. Galton, who was Darwin's cousin, showed this by averaging the guesses of 787 people as to the weight of an ox on show at a cattle exhibition in Plymouth in 1906. Remarkably, the crowd's averaged estimate was off by just one pound.

The wisdom of a crowd of independent-minded individuals emerges because the error in contrasting judgements is cancelled out. Imagine a jar of 100 beans. I estimate that there are 110 beans in the jar and you estimate their are 90. The errors in our judgements cancel out and together we are more accurate. Of course, real life isn't that neat, but the general principle holds, so long as our judgements are independent. If a group of individuals are not independently minded, perhaps because they're relying on the same faulty information, then collective wisdom will not emerge, because everyone's errors will all fall in the same direction.

Back to dialectical boot-strapping: Herzog and Hertwig asked 101 participants to estimate historical dates, such as the discovery of electricity. Crucially, half the participants used the dialectical boot-strapping technique. They made their first estimate, considered how it might be wrong, and then used this new perspective to make a new estimate. The other control participants simply made two best estimates.

The average of each dialectical boot-strapper's two guesses was, on average, 4.1 per cent more accurate than their initial estimate (72 per cent of them benefited by using this technique). By contrast, the average of each control participant's two estimates was, on average, just 0.3 per cent more accurate than their initial estimate.

"Part of the wisdom of the many resides in an individual mind," the researchers said. "Dialectical bootstrapping is a simple mental tool that fosters accuracy by leveraging people's capacity to construct conflicting realities."

ResearchBlogging.orgHerzog, S., & Hertwig, R. (2009). The Wisdom of Many in One Mind: Improving Individual Judgments With Dialectical Bootstrapping. Psychological Science, 20 (2), 231-237 DOI: 10.1111/j.1467-9280.2009.02271.x

Useful related links grabbed from the wisdom of the comments section (thanks people):

*Condorcet's jury theorem
*Two averaged guesses, without deliberate dialectical boot-strapping, can also be beneficial

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.


Anony Mouse said...

Interesting stuff. (Nitpick: it should be principle, not principal).

The Condorcet's Jury Theorem is closely related:

Unknown said...

Principal / principle corrected - thanks Michael! I'll also check out that link on the Jury Theorem.

Neuroskeptic said...

Ed Vul, of "voodoo fMRI correlations" fame, has also done some work on this.

Anonymous said...

So when was electricity discovered?

Post a Comment

Note: only a member of this blog may post a comment.