Risk attitude and systematic bias in estimating and forecasting

This paper describes findings from the first phase of a research project investigating the modelling of risk and uncertainty in construction estimating and forecasting. The objectives of the project are to examine current methods and, experimentally, to explore techniques which offer the potential for the production of improved information from estimates and forecasts. Some of the causes of error and bias in both traditional deterministic and in probabilistic estimating and forecasting are described. The majority of the research in cognitive psychology which has led to the common assumption of errors and biases has been carried out with lay people thinking intuitively about problems. The present research is an attempt to test these biases in a domain-specific, non-intuitive context with individuals trained in that domain. The authors hypothesize that, if professional training has any value then they should find less evidence of bias than is the case in the general literature. Empirical results are reported and discussed. This work finds empirical support for only some of the biases commonly assumed to exist.