Research has shown that combining forecasts improves forecasts accuracy - but people misappreciate its value.

The value of combining

Armstrong (2001) reanalyzed 30 studies conducted between 1960 and 2000. He found that, across the studies, combining reduced forecast error from 3 to 24 percent, with a mean improvement of 12.5 percent. Under ideal conditions, the combined forecasts were sometimes even more accurate than their most accurate components. He concluded that, to improve forecast accuracy, combined forecasts should be derived from methods that differ substantially and draw on different sources of information.

Misappreciation of combining

Regardless the evidence, people don't appreciate the value of combining. Larrick and Soll (2006), in a clever series of experiments, showed that among highly intelligent MBA students at an elite institution, most did not understand that the error of the average judgment is almost always smaller than the error of the average person in a group. Even more surprising to them was that the average judgment is sometimes better than the best judgment.

Ignoring scientific evidence

The finding that people resist evidence that contradicts their beliefs is well-known. Reporting on findings from four case studies, Hogarth (2006) showed that people are reluctant in accepting that simple models can be superior to more complicated ones. When making complex decisions, people believe complex models or information are necessary to deal with the situation.

References

• Armstrong (2001), "Combining forecasts", Principles of Forecasting: A Handbook for Researchers and Practitioners, J. Scott Armstrong (ed.), Norwell, MA: Kluwer Academic Publishers.
• Hogarth, R. M. (2006), "On ignoring scientific evidence: The bumpy road to enlightenment", Working Paper.
• Larrick, R. P. & Soll, J. B. (2006), "Intuitions About Combining Opinions: Misappreciation of the Averaging Principle", Management Science, 52, 111-127.