Prediction Markets dazzle audience at ESOMAR Congress 2014
Summary
ModifyTogether with Jon Puleston, VP of Innovation at Lightspeed-GMI, we showed the stunned ESOMAR audience how traditional questionnaires get even simple predictions completely wrong, all live with 10 interactive experiments.
Wiki article
Compare versions Edit
by Hubertus Hofkirchner -- Vienna, 12 Sep 2014
At this year’s annual congress of ESOMAR, the world-wide organisation of market research firms, Jon Puleston of GMI/Lightspeed and I presented an interactive session about “Predicting the Future” to about 250 market research professionals.
The audience’s enthusiastic feedback is an encouraging sign that the market research profession is increasingly interested in prediction markets, and hopefully in the new capabilities of Prediki’s second-generation system and method.
Predictive Biases
Our first aim was to demonstrate how badly survey questions about the future can go wrong if only asked with traditional methods. Just guess what even a crowd of very smart people will predict, on the average, when polled whether a coin flip will be heads or tails? A hint: the result of a traditional survey will not give the actual likelihood of 50/50. Not even close.
We packed the issues into a rapid series of small live experiments: question, prediction, question, prediction, question, prediction. With our 30-second pacing, the audience progressed at the typical speed we normally expect of a survey respondent. The stress emanating from the audience is a typical sign that people innately want more time to predict the right thing about the future. At the end, each listener’s prediction sheet was scored by their seat neighbour and everybody could see for themselves how one can be completely mistaken by unavoidable biases even with easy - not to mention realistically more complex – prediction questions.
Overcoming Bias
But how to solve this dilemma? To make our second point, we had 10 students perform several prediction market scenarios live on stage (thanks guys!), again under significant time-pressure. The audience could observe the good things happening once we employed various simulated prediction market settings, with or without information “nudges”, with single or multiple trading rounds providing an opportunity to reconsider, and most importantly, with or without monetary incentives depending on predictive performance.
Some participants will probably never believe in just a traditional survey for predictive questions ever again.
As they should not.
PS: For all those who missed the live session: Our paper is available online at the ESOMAR store for a small fee. And if you need a “nudge” to get it: not only was the session the congress audience’s absolute favourite by quite a margin, the paper itself was also nominated for two ESOMAR awards: “Best Paper Overall” and “Best Case History”.
