Behavioral insights for improving research integrity

This document considers efforts to enhance research integrity from a behavior change perspective. This perspective involves identifying the causes of undesirable behaviors and ways to shape these in an ethical and effective manner. To facilitate this way of thinking, we identify four broad strategies that can influence problematic research behaviors by targeting four broad categories of causes of these behaviors: shaping competencies, shaping incentives, shaping communities, and shaping opportunities (CICO). We use this framework to integrate existing literature on both the causes and proposed remedies of questionable research practices as well as research misconduct.According to the CICO framework, research integrity can be first facilitated by shapingcompetencies because research conduct depends in part on relevant knowledge, skills, and habits. For instance, improving research integrity training as well as equipping researchers to engage in self-regulation are promising ways to maintain integrity.Second, research integrity can be facilitated by shaping incentives because research conduct depends in part on the expected consequences of actions, both in terms external rewards and punishments as well as in terms of internal alignment with psychological goals, values, and identities. For instance, pre-registration and registered reports reshape incentives by disentangling publication likelihood from the content of findings. Reminding researchers of their integrity values might increase the weight of these internal incentives on choices. Third, research integrity can be facilitated by shaping communities because research conduct depends in part on the perceived behavior of pioneers, leaders, and majorities. For instance, the open science movement has sought to elevate the status of researchers who adhere to integrity principles and reduce the status of those who do not. Finally, research integrity can be facilitated by shaping opportunities because research conduct depends in part on which actions are made easier to consider and implement by the structure of situations. For instance, various defaults, reminders, feedback, and self-reflection prompts could be incorporated into research software or pre-registration and publication portals.These four strategies can be combined to form comprehensive interventions. Conduct of research is a complex behavior consisting of numerous decision points each with numerous action options. Such behaviors have many causes and therefore interventions targeting each one individually can have only a small effect. However, such small effects can accumulate over time and be combined to achieve larger results. We also conducted an online experiment (n = 634) to test if an incentive shaping strategy of increasing the psychological salience of internal values or external adverse consequences would alter participants’ evaluations and selection of low vs high integrity research behaviors in 10 vignettes from the Erasmus University Rotterdam’s Dilemma Game. We found that the consequences manipulation and to a lesser extent the values manipulation indeed increased Page 5 of 33preference for the high integrity option, but only in the first vignette where participants wrote about how the highlighted values or consequences applied to the situation described in the vignette. These results suggest that research integrity can be supported by interventions that induce researchers to think through, at the time of choosing between behavioral options, how the options at hand relate to specific adverse consequences that they wish to avoid and to specific personal values that they wish to live up to.