Chapter 4 found that people do not sufficiently perceive the importance of numerical variance information in capital allocation. This is important when business projects are dissimilar because people may fail to pay attention to the differing variance underlying NPV across different domains. However, there are also implications for high alignment scenarios. When projects are alignable, managers are likely to be able to use abstract metrics as well as intrinsic project features. Managers may use a metric such as NPV, the variance of which may suggest a lack of reliability, despite being able to use intrinsic project features. Therefore, they may miss the opportunity to use different and potentially more reliable measures.
Therefore, the evaluation of a non-alignable set of projects has many potential pitfalls. This situation is likely to occur in most hierarchical organisations, especially those that are highly diversified. As discussed in Chapter 3, a solution for managers who fail to aggregate the risk of multiple projects may be for them to concurrently evaluate projects as a portfolio. However, the solution to the evaluation of dissimilar projects in diversified organisations is likely to involve significantly more difficult structural changes in the organisation. For instance, this may mean divesting certain divisions of the organisation, as GE has done in the last few years (Scott, 2018).
Other solutions are also possible. For instance, organisations may develop a more normative use of metrics and take into account underlying uncertainties. However, this change may require substantially more statistical reasoning abilities than should be expected of managers without better decision-making guidelines. Another solution for managers is to seek evidence from similar projects from outside of the organisation. This may be useful because a diversified organisation may not have enough points of reference for a project proposal. It would also mean that substantial organisational restructuring such as divestment or training managers in statistical reasoning would not be required.
Evidence from similar projects may come in the form of an individual case study from another organisation or a research report that describes a statistical result. Case studies are especially important in managerial decision-making because they are used extensively in business school teaching materials. Therefore, managers are likely to seek case studies that may be used to inform their decisions. However, do they believe that a single case study is more useful than statistical data? The literature on anecdotal bias suggests that they might. Chapter 6 considers the influence of anecdotes on project allocation when they conflict with statistical evidence.
Previous work shows that people often do not give evidence appropriate weighting in their decisions (Griffin & Tversky, 1992). Statistical and anecdotal evidence often conflict because statistical estimates commonly refer to the mean value of a distribution, while individual cases may be sampled from either tail of the distribution. This comparison may produce conflicting information, especially if the distribution is skewed; therefore, it is important to appropriately weigh their influence when making a decision. In the same way that intrinsic project features conflicted with the abstract financial metrics in Chapter 4, anecdotal evidence conflicts with financial metrics of the target project in Chapter 6.
Chapter 6 considers how people deal with such conflicting information. That is, do they focus on one metric or use a trade-off? In the previous chapter, participants did not appear to predominantly use any one specific cue. The fact that those in the low alignment condition relied more on NPV compared with those in the high alignment condition means that the latter were still referring to intrinsic project features to some extent. Specifically, the influence of different measures may have been integrated in a type of trade-off. However, there was no clear way of determining this because the allocation measure was aggregated in the analysis. In Chapter 6, however, conditions are set up so that it is possible to determine whether participants were using anecdotes exclusively, partially, or not at all.
Griffin, D., & Tversky, A. (1992). The weighing of evidence and the determinants of confidence. Cognitive Psychology, 24(3), 411–435. https://doi.org/10/frw7xm
Scott, A. (2018, June 26). GE breakup leaves it with best and worst performers. Reuters. https://www.reuters.com/article/us-ge-divestiture-idUSKBN1JM0ZT