Approaches to displaying information to assist decisions under uncertainty

Abstract

The estimation of the costs of a product or project and the decisions based on these forecasts are subject to much uncertainty relating to factors like unknown future developments. This has been addressed repeatedly in research studies focusing on different aspects of uncertainty; unfortunately, this interest has not yet been adopted in practice. One reason can be found in the inadequate representation of uncertainty. This paper introduces an experiment which engages different approaches to displaying cost forecasting information to gauge the consideration of uncertainty in the subsequent decision-making process. Three different approaches of displaying cost-forecasting information including the uncertainty involved in the data were tested, namely a three point trend forecast, a bar chart and a FAN-diagram. Furthermore, the effects of using different levels of contextual information about the decision problem were examined. The results show that decision makers tend to simplify the level of uncertainty from a possible range of future outcomes to the limited form of a point estimate. Furthermore, the contextual information made the participants more aware of uncertainty. In addition, the fan-diagram prompted 75.0% of the participants to consider uncertainty even if they had not used this type of diagram before; it was therefore identified as the most suitable method of graphical information display for encouraging decision makers to consider the uncertainty in cost forecasting.

Information availability at the competitive bidding stage for service contracts

“…The empirical work focused on two experimental studies undertaken with a total of 72 cost engineers and bidding decision makers from practice. These studies focused on the different influences on the bidding decision-making process, including the approach of displaying the cost estimate [Kreye et al, 2012] and the influence of the existence of competition on the decision outcome and rationale [Kreye, 2011]. The participants were given a set of questionnaires which consisted of a pricing scenario and various questions connected to their decision-making process for this hypothetical example.…”

IMP: A decision aid for multiattribute evaluation using imprecise weight estimates

“…This would usually be thought a recklessly high value for statistical inference, but the purpose here is different: we wish to see the level of support for discrimination. Kreye et al [77] report that in giving forecasts experts and non-experts gave self-assessed confidence levels of about 40%, so p=0.5 may not be so unreasonable.…”

“…How results of imprecise analysis are shown is important if the analysis is to be useful [77]. The object of the screen design for this simple spreadsheet was that the user could readily see both the judgements made and their results so that exploratory interaction is made easy.…”

Setting tolerance limits for statistical project control using earned value management

“…Since the occurrence of a project management simulation model in [39], the question has been asked how to accurately represent the uncertainty present in the duration of activities in a real-life project [40]. Since the PERT-beta distributions have been found to fall short, different authors have suggested alternatives.…”