AUTHOR=Doerflinger Johannes T. , Martiny-Huenger Torsten , Gollwitzer Peter M. TITLE=Exploring the determinants of reinvestment decisions: Sense of personal responsibility, preferences, and loss framing JOURNAL=Frontiers in Psychology VOLUME=13 YEAR=2023 URL=https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2022.1025181 DOI=10.3389/fpsyg.2022.1025181 ISSN=1664-1078 ABSTRACT=

Two potentially costly errors are common in sequential investment decisions: sticking too long to a failing course of action (escalation of commitment), and abandoning a successful course of action prematurely. Past research has mostly focused on escalation of commitment, and identified three critical determinants: personal responsibility, preferences for prior decisions, and decision framing. We demonstrate in three studies using an incentivized poker inspired task that these determinants of escalation reliably lead decision makers to keep investing even when real money is on the line. We observed in Experiments 1, 2 and 3 that reinvestments were more likely when decision makers were personally responsible for prior decisions. This likelihood was also increased when the decision makers had indicated a preference for initial investments (Experiments 2 and 3), and when outcomes were framed in terms of losses as compared to gains (Experiment 3). Both types of decision errors – escalation of commitment and prematurely abandoning a course of action – could be traced to the same set of determinants. Being personally responsible for prior decisions, having a preference for the initial investment, and loss framing did increase escalation, whereas lacking personal responsibility, having no preference for the initial investment, and gain framing increased the likelihood of prematurely opting out. Finally, personal responsibility had a negative effect on decision quality, as decision-makers were still more likely to reinvest when they were personally responsible for prior decisions, than when prior decisions were assigned optimally by an algorithm (Experiments 2 and 3).