Distress stood out
Employees classified as both targets and perpetrators of workplace bullying reported the highest psychological distress among the four groups examined. In the direct comparison with perpetrator-only employees, distress was also the clearest adjusted correlate of dual-role membership. That is an association, not proof of cause: because the analysis was cross-sectional, it cannot show whether distress came before bullying involvement or followed it.
Of 2,359 people in the main analysis, 149, or 6.32%, were classified as target-perpetrators. There were 51, or 2.16%, perpetrator-only employees, 576, or 24.42%, target-only employees, and 1,583, or 67.10%, who were not involved. Dual-role participants therefore outnumbered perpetrator-only participants, although both groups were much smaller than the not-involved group.
A group that does not fit one simple profile
The categories came from a behavioral rule. Employees were assigned target or perpetrator status when the sum of observed responses on a four-item scale, divided by four, reached at least 2, or when any single observed item scored 4 or more.
Compared directly with perpetrator-only employees, target-perpetrators reported more distress. They also had lower agreeableness and conscientiousness and higher neuroticism, while the two groups did not differ significantly in age, physical symptoms, alcohol or tobacco consumption, or openness to experience.
Across the wider set of comparisons, chronic psychological conditions were more common than expected among target-perpetrators than among target-only or not-involved participants. Cramér's V, a measure of the strength of association between categories, was 0.11 versus target-only participants and 0.15 versus those not involved.
The adjusted comparison
To examine what separated the two perpetration-related groups, the researchers used hierarchical binary logistic regression, a method that estimates the odds of one category rather than another. Target-perpetrator membership was coded as 1 and perpetrator-only membership as 0. The analysis included separate health and personality models.
In the adjusted health model, higher psychological distress was associated with greater odds of target-perpetrator rather than perpetrator-only membership. The odds ratio was 1.68, with a 95% confidence interval from 1.11 to 2.53. The other health measures tested, including physical symptoms, chronic psychological condition, alcohol use and tobacco use, were not statistically significant.
The adjusted personality model found a positive association between neuroticism and dual-role membership. The odds ratio was 1.52, with a 95% confidence interval from 1.03 to 2.25. Agreeableness, extraversion, conscientiousness and openness were not statistically significant in that model.
That personality result was less secure than the distress finding. A further bootstrap analysis, which repeatedly resamples the data to test the stability of an estimate, kept distress as a statistically significant correlate: the coefficient was 0.52, with p = 0.014 and a 95% confidence interval from 0.15 to 1.11. Neuroticism remained positive but was not statistically significant, with a coefficient of 0.42, p = 0.052, and a confidence interval from 0.004 to 0.96.
A less settled picture under a stricter rule
The pattern changed when the researchers required the reported behavior to occur weekly. Among 2,357 classifiable participants, 184 were target-only, 28 were perpetrator-only, 22 were target-perpetrators and 2,123 were not involved. With the perpetration-related groups much smaller, distress was no longer statistically significant: the odds ratio was 1.79, with a 95% confidence interval from 0.94 to 3.40. Physical symptoms and tobacco consumption were associated with role membership under this rule, while no focal personality trait was significant.
Why the result needs caution
The study was a secondary, cross-sectional analysis of Wave 1 data from a broader three-wave longitudinal study. It began with 3,663 respondents and retained 2,359 after exclusions for insufficient work hours, missing consent, duplicate responses or undetermined workplace-bullying role status.
Missing answers were another source of uncertainty. Little's MCAR test, which checks whether missingness is completely at random, was significant: the test statistic was 842.53 with 546 degrees of freedom, and p was below 0.001. Missingness was associated with openness to experience and chronic psychological-condition status.
The personality measure itself also had a measurement caveat. Internal consistency, a check of whether items in a scale hang together, was low for extraversion, with alpha = 0.52, and neuroticism, with alpha = 0.51. That makes the neuroticism association harder to interpret, especially because it was not statistically significant in the bootstrap analysis.
Overall, the analysis presents employees in both roles as a mixed group rather than one defined by a universal health or personality profile. Psychological distress was the most consistent adjusted correlate in the primary analysis, while the personality pattern was less stable and the stricter classification changed which associations reached significance. The cross-sectional design leaves the direction of the relationship unresolved.
Paper data and sources
Original title: Not all workplace bullying perpetrators are alike: health, personality, and occupational profiles of perpetrator-only and target-perpetrator employees
Authors: Sun J, Özer G, Escartín J
Journal/Repository: Frontiers in psychology
Status: Peer-reviewed
First online: 2026-10-08
DOI: Not available
Original paper · Full text