Decision Bias in Project Selection: Experimental Evidence from the Knapsack Problem
54 Pages Posted: 14 Sep 2019 Last revised: 28 Feb 2023
Date Written: Feburary 26, 2023
Abstract
Selecting the most valuable projects given a finite budget constraint is a recurring decision challenge in all organizations. The optimization literature has long recognized the mathematical complexities of this knapsack problem. However, these complexities along with real-world data imperfections, such as how to fully determine the "values" of competing projects, have severely limited the adoption of optimization algorithms. Instead, decision makers employ mental heuristics. We explore the nature of these heuristics experimentally in a computer lab, and find them to be biased towards selecting too many small projects. We attribute this bias to a key structural characteristic of the decision makers' search process. Specifically, while they search for value-maximizing combinations of projects, they consistently keep their solutions within the feasible side of the budget boundary. They rarely generate infeasible solutions during their search, and then consider which projects to drop. We test two common strategies to debias decision makers: a problem framing that subtly nudges participants to search more in the infeasible solution space, and direct advice to participants to do so. We find that only the latter one reduces the small-project bias and improves resource allocation decisions.
Keywords: project selection, behavioural operations, knapsack problem
JEL Classification: A10
Suggested Citation: Suggested Citation