James Montier, author of Behavioural Finance: Insights Into Irrational Minds And Markets, argues that a fear of uncertainty is largely responsible for our use of forecasts even when we know they aren't accurate. When we are uncertain, we are willing to latch onto even irrelevant data in order to attempt to quell that uncertainty. Montier refers to this as "anchoring", and uses the following example to illustrate how it affects us:
A question he asks professional fund managers is to write down the last four digits of their phone numbers, and then estimate how many doctors live in their capital city. Those with the last four digits above 7000 report on average 6762 doctors, while those whose last four digits are under 2000 arrive at an average estimate of 2270 doctors. Incredibly, the phone numbers of fund managers play a role in helping them answer a completely unrelated question!
This study is by no means unique. In this paper, Tversky and Kahneman asked subjects what percentage of African nations are part of the United Nations. Before they could answer, however, the authors had the subjects spin a rigged wheel of numbers that would land on either 10 or 65. As readers may by now predict, the answers of subjects were strongly influenced by the random number that came up on the wheel.
Montier strongly believes that we should ignore forecasts, and instead use value-based strategies based on trailing earnings or PE's. Furthermore, he argues that analysts should stick to analyzing, not forecasting. However, since this would require a radical re-think of the investment process, he doesn't see this as likely.