A certain controversy exists when it comes to calculating the prevalence of psychological bullying in the workplace, given that the results are very uneven depending on the method chosen (León-Pérez et al., 2021). In general terms, the administering of questionnaires enables determination, through the use of two traditional methods, of whether a person has been subjected to bullying.

The first method consists of directly asking a person if they consider that they have been subjected to bullying during recent months, in such a way that the person themselves classifies themselves as bullied or not bullied, according to their experience. This is the question that was formulated by the interviewers for the European Working Conditions Survey (2021), whose results indicate that 5.9% of workers in the European Union answered affirmatively to the question of whether they had suffered abuse, bullying, or violence in their workplace during the last year. Obviously, the answers were partially biased by the level of sensitivity towards these issues and the sociocultural differences between interviewees. Therefore, it is not surprising that these studies report on broad differences in prevalence, by country and by sex. 

The second method consists of recording the frequency with which a person is the subject of negative behaviours, which, when suffered systematically and in a recurrent way over time, constitute a situation of psychological bullying. The frequencies obtained enable the drawing up of the so-called bullying behaviour inventories, which reveal that the figures for the prevalence of psychological bullying in the Spanish workplace context stand at around 15%. However, this method presents a problem, which is the criteria used; in other words, how one determines, firstly, the number of bullying behaviours and, secondly, the frequency with which a person must be exposed to said bullying to consider whether they are being bullied or not. In this case, it is obvious that the prevalence will depend on the criteria used, and therefore the results obtained could be somewhat arbitrary. If a low criterion is used, such as being exposed “weekly” to a negative behaviour (e.g. “being subject to excessive monitoring of my work”), then bullying cases will be much more numerous than if one opts for a high criterion, more difficult to meet, such as being exposed “daily” to four bullying behaviours (e.g., being ignored when one wants to establish communication with colleagues; being isolated from the work group; receiving groundless criticism in public; being the only person who must carry out tasks below their skills level and not receiving the information necessary to execute them, etc.). 

For this reason, in line with some proposals for alternative methods of detection of psychological bullying in the workplace that have emerged in recent years, the aim of the study by the Cármides Research Group is to offer a tool for detecting bullying that is more reliable and robust from a statistical viewpoint.

The following table contains four methods of estimation or detection of bullying: two traditional ones and two alternative ones.

The first step towards obtaining a less biased bullying detection tool consists of establishing classification criteria – or cutoff points – that are reliable and objective for a determined population; in other words, establishing the scores above which it can be considered that a person is a victim of bullying. 

Some researchers have opted for combining the alternative methods explained previously and applying advanced statistics that use Bayesian inference to establish bullying risk groups; in other words, applying a technique for data analysis that enables grouping of people into categories (i.e., latent class cluster analysis) through a pattern of responses to each question. In this way, it is possible to infer the probability that a person is classified in one group or another. Using this method, the only study conducted in Spain reported that 8% of individuals were classed as targets or victims of bullying in the workplace. Furthermore, this group reported a higher frequency of interpersonal conflicts in their workplace and poorer psychological well-being than for workers classified into other groups with a lower exposure to negative behaviours in the workplace (León-Pérez et al, 2014). 

However, given the analytical complexity of this alternative, researchers have preferred to focus on establishing cutoff points through ROC (Receiver Operating Characteristic) curve analysis. This statistical method determines the cutoff point – or score – on a scale that achieves better sensitivity (correct identification of people who present a determined characteristic; in the case of this study, suffering bullying in the workplace) and specificity (that there are no false positives; i.e., that individuals not exposed to bullying behaviours are not incorrectly identified as people suffering bullying). 

In other words, ROC curves indicate – based on the total score that a person obtains when responding to a questionnaire or inventory of exposure to bullying behaviours – from which score (or cutoff point) a person is being subjected to psychological bullying in the workplace. This focus is very useful for assessing the discriminatory capacity of a test and is based, mainly, on the combination of scores obtained on a scale that measures exposure to bullying behaviours through previously-validated health and well-being indicators. 

In the Cármides Research Group study, this analytical technique was applied to establish, according to the questionnaire of exposure to negative behaviours in the workplace called S-NAQ  (León-Pérez et al., 2019), three risk groups: a) low (without exposure to bullying behaviours); b) medium (at risk), and c) high (subject to or the “target” of bullying). Thus, as reflected in the chart below, the first result that should be highlighted is that, through application of this method, the prevalence of bullying in a sample of 5,000 Spanish workers is situated at around 11.2%, while 18.8% of them would be at medium risk and 70%, at low risk.

The negative consequences for the health and well-being of people exposed to bullying behaviours in the workplace are well documented in the scientific literature (Boudrias et al., 2021). For this reason, the capacity of the detection method used in the study for predicting the psychological well-being of workers was checked four months after classifying them in one of the three groups mentioned. In this regard, the probability of developing a generalised anxiety disorder– the most common psychological disorder in bullying situations – is nearly five times greater in people classified in the high-risk group than in those classified in either of the other two groups: medium risk or low risk (odds ratio = 4,466). Specifically, a screening test for possible anxiety disorder was applied four months after people were classified in the different risk groups. The data showed that, in the group of people with a high risk of suffering bullying in the workplace, 19.9% of them showed a broad symptomatology (high levels) of anxiety, which indicates the presence of a possible generalised anxiety disorder in comparison with the groups of people with a medium or low risk of bullying, in which the percentages with a high level of anxiety were reduced to 7.40% and 5.30%, respectively. In other words, as observed in the graph below, the higher the risk of suffering a situation of bullying, the greater the levels of anxiety and the higher the probability of developing a psychological disorder.

This study offers a tool that permits improvement of decision-making in relation to the occupational health of workers. It can be applied in diverse areas, by workplace health and safety or human resources professionals, with a view to detecting possible cases of bullying in a specific group, as well as by professionals at institutions entrusted with establishing the prevalence of bullying in a determined population and, even, on a judicial level, by clinical and forensic psychology professionals as evidence for elucidating reliably whether a person has been exposed or not to a bullying situation.

To produce this article, data were used originating from the questionnaire devised within the framework of the research project “Advanced statistical analysis for measuring and preventing bullying and cyberbullying in the workplace (PREVENT-US)”. This project, funded by the ”la Caixa” Foundation, is being executed by the Cármides Research Group: Social Behaviours and Health at the University of Seville, under the coordination of José María León Pérez. The study was conducted with a representative sample on a national scale of 5,000 workers, who filled out the questionnaire on two occasions, with a four-month separation between them (2,095 valid responses the second time around). With this information, in the second phase a computer app was produced that enables measurement of the frequency of exposure to psychological (cyber) bullying in the workplace and establishment of the degree of risk (low, medium, or high) for worker health.

BOUDRIAS, V., S. G. TREPANIER & D. SALIN (2021): “A systematic review of research on the longitudinal consequences of workplace bullying and the mechanisms involved”, Aggression and Violent Behavior, 56, 101508. 

LEÓN-PÉREZ, J. M., J. ESCARTÍN & G. GIORGI (2021): “The presence of workplace bullying and bullying worldwide”.  In: P. D’Cruz, E. Noronha, G. Notelaers and C. Rayner (eds.), Handbooks of Workplace Bullying, Emotional Abuse and Harassment, vol. 1: Concepts, Approaches and Methods (p. 55-86). Singapore: Springer.

LEÓN-PÉREZ, J. M., G. NOTELAERS, A. ARENAS, L. MUNDUATE & F. J. MEDINA (2014): “Identifying victims of workplace bullying by integrating traditional estimation approaches into a latent class cluster model ”, Journal of Interpersonal Violence, 29(7), 1155-1177.

LEÓN-PÉREZ, J. M., I. SÁNCHEZ-IGLESIAS, A. RODRÍGUEZ-MUÑOZ & G. NOTELAERS (2019): “Cutoff scores for workplace bullying: The Spanish Short-Negative Acts Questionnaire (S-NAQ)”,  Psicothema, 31(4), 482-490.