When an employee causes harm at work, a manager can punish them, ignore the incident or try to repair the damage. Restorative approaches are designed to do the third. They bring affected people into the process, focus on accountability and future relationships, and give those involved a voice in deciding what repair should look like.
That sounds ethically preferable to a purely punitive response. New research suggests the consequences can nevertheless be more complicated than they first appear.
A study published in the Employee Responsibilities and Rights Journal on 29 September 2026 found that manager-facilitated restorative practices were associated with psychological responses that could pull employees in opposite ethical directions. Gratitude toward a manager was associated with less willingness to behave unethically on the manager’s behalf. But feeling an obligation to reciprocate favourable treatment and identifying strongly with the manager were associated with greater willingness to do so.
The result highlights an important tension in workplace ethics. A manager can respond to wrongdoing in a humane and constructive way while also strengthening a relationship that creates loyalty, indebtedness or identification. Those bonds are not unethical by themselves. The risk emerges when helping the manager begins to compete with wider ethical standards.
Studying employees after workplace wrongdoing
The researchers recruited full-time adult employees working in US organisations through Prolific. Participants were asked to recall an incident in which they had intentionally or unintentionally done something at work that caused harm to another person, several people or their organisation, and where their manager became aware of what had happened.
Data were collected in two waves approximately seven days apart. At the first time point, 350 employees completed the survey. Of these, 244 completed the second survey and passed the attention checks. The final two-wave sample was 40% female, 58% male and 2% non-binary. Most had substantial employment experience, with 64% reporting at least ten years of full-time work.
The researchers measured restorative justice practices initially through seven dimensions drawn from earlier theory. Exploratory factor analysis condensed these into three broader patterns. The first was a focus on participants’ needs, combining empowerment, flexibility and mutual accountability. The second was a focus on effective interaction, combining future orientation and transparency. The third captured inclusiveness.
These measures showed acceptable to strong internal consistency. The participants’ needs scale had a Cronbach’s alpha of 0.91, effective interaction 0.88 and inclusiveness 0.70.
The study then examined three possible psychological mechanisms: gratitude toward the manager, a desire to reciprocate the manager’s benevolent treatment, and identification with the manager. Moral identity was also measured to test whether seeing morality as central to one’s sense of self changed the relationship between these responses and unethical behaviour.
The outcome was unethical pro-supervisor behaviour. This did not mean observed misconduct. Participants reported how willing they would be to engage in actions such as misrepresenting the truth to make a supervisor look good or concealing information that could damage the supervisor. The six-item measure had a Cronbach’s alpha of 0.90.
Restorative practices strengthened different kinds of attachment
The measurement structure was checked in a separate sample of 218 full-time US employees. The seven-factor confirmatory model showed good fit, with a comparative fit index of 0.96, Tucker-Lewis index of 0.95 and root mean square error of approximation of 0.05.
The main path model also fit the data well. The chi-square statistic was 10.66 with eight degrees of freedom and a p-value of 0.22, while both the comparative fit index and Tucker-Lewis index were 0.99. The RMSEA was 0.04.
Managers who focused more strongly on participants’ needs had employees who reported greater gratitude toward them, with a standardised path coefficient of β = 0.26, and a stronger desire to reciprocate their treatment, β = 0.30. Both relationships were statistically significant at p < 0.01.
A focus on effective interaction was also positively associated with gratitude, β = 0.18, p < 0.01. Inclusiveness, meanwhile, was positively associated with identification with the manager, β = 0.18, p < 0.01.
These findings matter because restorative management did not appear to create one simple emotional response. Different features of the repair process were linked to different forms of connection with the manager. Giving people influence over the process, adapting to their needs and sharing accountability were particularly relevant to gratitude and reciprocity, while inclusiveness was linked to identification.
Reciprocity and identification carried an ethical risk
The next step produced the study’s central finding. Employees who felt a stronger desire to repay their manager were more willing to report unethical pro-supervisor behaviour, β = 0.22, p < 0.05. Identification with the manager showed a similar positive relationship, β = 0.21, p < 0.01.
In practical terms, benevolent treatment can create a sense of obligation. If an employee believes a manager gave them a fair opportunity to repair harm instead of simply punishing them, the employee may want to return the favour. Usually that impulse could support constructive behaviour. The problem is that reciprocity does not automatically stop at an ethical boundary. Loyalty can become a motive for protecting the supervisor even when protection requires dishonesty or concealment.
Identification creates a related risk. When employees begin to experience a supervisor’s successes and failures as personally meaningful, protecting the supervisor can become psychologically similar to protecting themselves or a valued group. The study suggests that this connection may help explain why a relationship built through an apparently ethical repair process can have an ethically undesirable downstream association.
Gratitude moved in the opposite direction
Not every positive feeling toward the manager increased the risk. The researchers had predicted that gratitude would be positively related to unethical pro-supervisor behaviour. Instead, the association ran in the opposite direction.
Greater gratitude toward the manager was associated with lower willingness to engage in unethical pro-supervisor behaviour, β = -0.20, p < 0.01.
This distinction is important. Gratitude and indebtedness can sound similar in everyday language, but they may operate differently. Gratitude can reinforce appreciation, interpersonal responsibility and a desire to maintain a relationship in a way that remains consistent with moral standards. Felt obligation to repay someone may place more direct pressure on the employee to provide a benefit in return.
The findings therefore do not support a simple conclusion that positive manager-employee relationships make unethical loyalty more likely. Instead, the type of psychological bond appears to matter.
Moral identity weakened some of the risky pathways
The researchers also examined whether moral identity acted as a boundary condition. Moral identity captures the extent to which characteristics such as fairness, honesty, compassion and helpfulness are central to how a person sees themselves.
Two significant interactions emerged. Stronger moral identity weakened the relationship between gratitude and unethical pro-supervisor behaviour, β = -0.16, p < 0.05, and weakened the relationship between desire to reciprocate and unethical pro-supervisor behaviour, β = -0.13, p < 0.05.
The moderated mediation analysis similarly indicated that the indirect effects of managers’ focus on participants’ needs and effective interaction depended partly on employees’ moral identity. Higher moral identity weakened positive indirect effects associated with unethical pro-supervisor behaviour and strengthened negative gratitude-based indirect effects.
This suggests that loyalty pressures do not operate in a vacuum. Employees who strongly organise their self-concept around moral traits may be better positioned to distinguish repaying a supervisor from doing whatever benefits that supervisor.
What restorative managers should take from the study
The research is not an argument against restorative justice at work. Restorative practices are intended to address harm, accountability and damaged relationships rather than simply impose punishment. The study instead identifies a possible side effect that managers and organisations may need to anticipate.
A repair process can make an employee grateful to a manager, indebted to them or more strongly identified with them. Those responses should not be treated as interchangeable. Organisations may benefit from making ethical boundaries explicit during and after restorative processes, particularly by separating appreciation for fair treatment from any expectation of personal loyalty to the manager who facilitated it.
That distinction could be reinforced through transparent procedures, independent reporting channels and clear norms that loyalty to a supervisor never overrides obligations to colleagues, the organisation, customers or the law.
Important limits on what the findings can show
The study cannot establish that restorative management causes unethical behaviour. Although data collection was split across two waves, participants reconstructed a past workplace episode from memory, and the core relationships are correlational.
The outcome also measured self-reported willingness to engage in unethical pro-supervisor behaviour rather than misconduct observed in a workplace. What people say they might do in a survey can differ from what they actually do when reputational, legal and organisational consequences are real.
The data came from a single source, and the sample was restricted to employees who had caused workplace harm and whose manager became aware of it. Those employees may differ systematically from workers who have not experienced that combination of wrongdoing, detection and managerial response.
Future longitudinal, experimental and multi-source research will be needed to establish causal direction and determine whether the same pathways predict observed behaviour. It would also be useful to test whether restorative systems facilitated by neutral third parties produce the same loyalty effects as repair led directly by a supervisor.
Even with these limitations, the study adds a useful complication to workplace ethics. Ethical management is not only about choosing a benevolent process over a punitive one. It is also about understanding the relationships that process creates. Repair can strengthen trust and connection, but organisations still need boundaries that prevent gratitude, identification or reciprocity from becoming a reason to cross an ethical line.
Source Information
Study: He, M.R., Butterfield, K.D., Tripp, T.M. et al. “Good Intentions, Bad Outcomes: Manager-Facilitated Moral Repair and Unethical Pro-Supervisor Behavior.”
Journal: Employee Responsibilities and Rights Journal
Published: 29 September 2026
Study design: Two-wave retrospective survey of full-time US employees, with 244 participants completing both waves and a separate 218-person sample used for confirmatory factor analysis.
Main analysis: Exploratory and confirmatory factor analysis, path analysis and moderated mediation examining restorative management practices, gratitude, reciprocity, manager identification, moral identity and unethical pro-supervisor behaviour.
DOI: 10.1007/s10672-026-09609-0
Research Today is a South African digital publication that makes credible research easier to understand.
