Workplace bullying in a South African higher education institution: Academic and support staff experiences

‘bullying’


Introduction
Workplace bullying is a common occurrence in organisations (Hauge, Skogstad, & Einarsen, 2010) and has been reported in countries ranging from France, Canada and Sweden to the United Kingdom (UK) and the United States (Yamada, 2010); South Africa is no exception (Botha, 2019;Cunniff, & Mostert, 2012;Kalamdien & Lawrence, 2017). Workplace bullying is the act of one individual, organisation or even groups of individuals targeting another individual in the workplace through negative and aggressive behaviour that can cause degradation of the employee's mental and physical health (Madan, 2014(Madan, , p. 1742. The concept of workplace bullying is used conversely with the term 'mobbing' (Einarsen, Hoel, Zapf, & Cooper, 2011;Keashly & Neuman, 2010). In German-speaking countries, the term 'mobbing' is preferred, whereas in English-speaking countries, the term 'bullying' is preferred to describe the same phenomenon (Chirilă & Constantin, 2013). In this study, the terms 'mobbing' and 'workplace bullying' are used interchangeably to describe the same phenomenon.
Workplace bullying is identified through repeated negative acts at the workplace directed at a specific individual or groups of individuals who are unwanted and negative in nature (Langos, 2012, p. 285). Negative acts include snarly comments, unnecessary sarcasm, mockery, belittling (Langos, 2012), introduction of impossible target expectations, extreme forms of observation of work that may cause unnecessary stress, as well as physical harm (Kortjan & Von Solms, 2014). These negative acts must occur repeatedly over a set period (at least once a week for a period of Orientation: Workplace bullying is a common occurrence in organisations worldwide, and higher education institutions are no exception.
Research purpose: This study was conducted with the aim of determining the perceptions of workplace bullying amongst academic and support staff at a higher education institution in South Africa and to ascertain how they are affected by selected sociodemographic variables.
no less than 6 months) in order for it to be considered workplace bullying (Einarsen et al., 2011;Kortjan & Von Solms, 2014).
Workplace bullying has a negative impact on employees as well as organisations. Being a victim of bullying may cause serious physical and psychological health problems. The physical harm that may befall employees can range from heart or cardiovascular diseases to musculoskeletal diseases, including shaking, insomnia, blood pressure problems, heart palpitations and stomach problems (Kortjan & Von Solms, 2014;Langos, 2012;Miller et al., 2012). Psychological issues include lack of enthusiasm, low self-esteem, inability to concentrate, loss of motivation (Smit, 2014), depression, excessive fear, forgetfulness, panic attacks, insecurities, suicide and its thoughts (Momberg, 2011). People who are bullied to such an extent that they feel there is no possible way of escaping the bullying may eventually experience a psychosomatic breakdown (Bristow, 2016). On an organisational level, workplace bullying is negatively linked to organisational commitment, team cohesion, creativity, work climate, organisational performance, productivity, job satisfaction and employee engagement (Adera, 2017;Hoel, Glasø, Hetland, Cooper, & Einarsen, 2010;Livingstone, Stoilova, & Kelly, 2016;Madan, 2014). If not addressed effectively, bullying behaviour can escalate and become a problem for the organisation, which could lead not only to a reduction in productivity but ultimately to the collapse of the entire organisational structure -and in some instances, the closing down of organisations as a whole (Burke & Mouton, 2013;Kortjan & Von Solms, 2014;Law, Dollard, Tuckey, & Dormann, 2011;Livingstone et al., 2016).
Unfortunately, there is no specific legislation that deals with workplace bullying in South Africa (Smit, 2014, p. 229). Smit (2021) recently stated that: [T] (Ramjettan, 2020); the act became effective on 18 March 2022 (Stein, 2022). Workplace bullying is also listed and defined in the Code of Good Practice (Ramjettan, 2020). The goal of the Code of Good Practice is to set a framework for the interpretation and implementation of the Employment Equity Act (No. 55 of 1998) in regard to the elimination of violence and harassment in the workplace (Ramjettan, 2020; Republic of South Africa [RSA], 2020; Stein, 2022). Furthermore, it seeks to provide employees, employers and other parties with guidelines on the understanding and handling of violence and harassment in the workplace (Ramjettan, 2020;RSA, 2020). Finally, it aims to act as a guide to human resource policies and practices that are related to violence and harassment in the workplace (Ramjettan, 2020;RSA, 2020).
Workplace bullying at tertiary education institutions in South Africa has been grossly under-researched; few studies (Jacobs & Teise, 2019;Ngwane, 2018;Van Der Nest, Long, & Engelbrecht, 2018) have reported on the prevalence of the phenomenon amongst employees. This study was conducted to determine the perceptions of workplace bullying amongst academic and support staff at a higher education institution in South Africa. The study adds to the body of knowledge on workplace bullying in the tertiary education sector globally and in South Africa specifically. Furthermore, the results of the study provide management, human resource practitioners and industrial psychologists with insights into the phenomenon at tertiary education institutions.

Research purpose and objectives
The objective of this research study was to determine the perceptions of workplace bullying amongst academic and support staff at a higher education institution in South Africa and to ascertain how they are affected by the sociodemographic variables of gender, age, marital status, highest qualification, length of employment, nature of employment and employment contract.

Literature review Theoretical framework for workplace bullying
In order to investigate workplace bullying, the Leymann model (Leymann, 1996), the conflict escalation model of Glasl (1982) and the theoretical framework developed by Einarsen, Hoel, Zapf and Cooper (2004) were explored as the point of departure for the study and management of bullying at work. The Leymann model was created by Heinz Leymann in the late 1980s and early 1990s to understand the role that mobbing plays in the working environment (Leymann, 1996). Leymann (1996, p. 168) uses the term 'mobbing' in order to encompass all actions that are related to forms of bullying and/or harassment in the working environment. Mobbing is defined by Leymann (1996, p. 168) as social interaction that takes place between two individuals or groups of individuals, where one individual is attacked by another individual or a group of individuals over a period. The act of mobbing places the individuals who are being harassed (the victims) under immense pressure, making them feel helpless and extremely uncomfortable with the situation in which they find themselves (Leymann, 1996, p. 168). According to the Leymann model, organisational factors (not individual factors) such as weaknesses in leadership behaviour and work design; the victims' socially exposed position; the low morale of management and the workforce; and poor conflict management are determining factors of workplace bullying (Einarsen et al., 2004;Leymann, 1990Leymann, , 1996. The Leymann model (Leymann, 1990(Leymann, , 1996 outlines four stages of bullying that take place over time, and it is postulated that mobbing (bullying) can be viewed as an escalated conflict situation. The first stage is critical incidents. A triggering situation is most often a conflict; the source of the conflict may be unknown and the duration of this initial stage may be shorter than the others (Leymann, 1990(Leymann, , 1996. The second stage is mobbing (bullying) and stigmatising. In this stage, mobbing (bullying) activities may involve a number of behaviours that are targeted at an individual or group on a daily basis over an extended time. This stage is characterised by aggressive manipulation, and as a result, it often leads to a person being stigmatised and bullied (Leymann, 1990(Leymann, , 1996. The third stage is personnel management. Once the need arises for management to step in, the conflict situation becomes an official case. Unfortunately, it often happens that in this stage, management misjudges the situation as being the fault of the victim, thereby accepting the negative preconceived notions that have been projected by the bully's negative view of the victim. This most often results in a serious violation of rights and ultimately leads to the victim becoming marked and further stigmatised. Furthermore, colleagues and management have the tendency to create explanations based on personal characteristics rather than organisational and environmental factors (Leymann, 1990(Leymann, , 1996. The final stage is the expulsion of the victim from the organisation. The victim may develop serious illnesses and may seek medical or psychological help. There is the possibility of medical professionals not believing the victim and as a result, misdiagnosing the victim. The most common misdiagnosis is that of paranoia, character disturbance and manic depression (Leymann, 1990(Leymann, , 1996. According to Zapf and Gross (2001, p. 502), one of the strongest critiques that can be levelled against the Leymann model of workplace mobbing is that of specificity. The model focuses on organisational factors as antecedents to workplace bullying and by doing so, excludes important variables that need to be considered. For example, Einarsen, Raknes and Matthiesen (1994) found in their research within seven different organisational settings that working environment factors only account for 10% of the variance in the prevalence of workplace bullying; in no subsetting was this greater than 24%. The conflict escalation model of Glasl (1982) has been suggested as an appropriate model to explain how conflicts may escalate into bullying (Einarsen et al., 2004;Zapf & Gross, 2001). Conflict in the workplace occurs at two levels: conflict at the level of interest (i.e. there is a difference between parties over desired outcomes) and conflict at the level of behaviour (i.e. when parties seeking different outcomes, expressing their differences through gestures such as acting destructively) (Watson, 2017, p. 336). Conflict in the workplace is unavoidable, and in some instances, it is even fruitful, encouraging innovation, performance and learning (Einarsen et al., 2011, p. 26). Bullying is explained 'as a certain subset of conflicts' (Zapf & Gross, 2001, p. 499). Whilst conflicts consist of only one episode, can be resolved relatively quickly and are regarded as even positive, conflicts underlying bullying consist of a series of related conflict episodes, can last for a long time and may become extremely harmful for the victim (Zapf & Gross, 2001) as well as the organisation (Einarsen et al., 2011). Glasl's model distinguishes between three phases and nine stages of conflict escalation and also offers conflict de-escalation strategies for each phase. According to Zapf and Gross (2001), bullying is regarded as a kind of conflict and tending to occur between phases 2 and 3: • Phase 1: Rationality and control. In Phase 1 (stages 1-3), the parties who are engaged in conflict are still invested and interested in reasonable solutions to the problems they are facing (Zapf & Gross, 2001). During this phase, there is some degree of cooperation in order to deal mostly with impersonal issues. The parties are aware of the existence of conflict; however, they try to handle such conflict in a controlled and rational manner (Zapf & Gross, 2001). In Phase 1, the conflict de-escalation strategy to be used is moderation. Moderation takes place when both parties are still willing to resolve the conflict through mediation (talking through the conflict). The conflict may be resolved through self-help, help from friends and families, as well as professional moderation by a third party that is neutral and unbiased (George, 2007). • Phase 2: Severing the relationship. Phase 2 (stages 4-6) focuses on severing the relationship. During this phase, the origin of the conflict has more or less disappeared, and the root of tension has now become the relationship between the parties. Overt hostility, lack of respect and distrust start to evolve during this phase, and as a result, the parties no longer work together to solve the conflict but rather seek to exclude each other (Zapf & Gross, 2001). In Phase 2, the conflict de-escalation strategies to be used are process company (professional process support from within the company), sociotherapeutic process company (seeking help through external sociotherapeutic process support) and professional mediation (an objective third party decision-maker) (George, 2007). • Phase 3: Aggression and destruction. The final phase (stages 7-9) focuses on aggression and destruction. During the third phase, confrontations become seemingly more destructive than in the first two phases (Zapf & Gross, 2001). Because of the lack of willingness from either party to resolve the conflict, the 'other party' is seen as having no human dignity, and as a result, any possibility of positive outcome is blocked. The escalation has reached such a detrimental level that the parties involved are willing to risk everything, even their existence, in order to hurt or destroy the other (Zapf & Gross, 2001). However, Glasl argues that this phase would rarely be reached in an organisation (Zapf & Gross, 2001). In Phase 3, the conflict de-escalation strategies to be used are voluntary or mandatory participation in arbitrary awards and power intervention from seniors (Zapf & Gross, 2001). Einarsen et al. (2004) developed a theoretical framework to understand and manage bullying behaviour at work. According to the authors, bullying is a complex social phenomenon characterised by multicausality. It involves a variety of factors and can be explained at many levels (Einarsen et al., 2004). On an individual level, the personality of the perpetrator as well as the victim may contribute to both bullying behaviour and perceptions of being bullied. Individual factors play a role in the victim's emotional and behavioural reactions to the perceived treatment, as well as the extent to which the victim can cope with the treatment (Einarsen et al., 2004). On a dyadic level, the focus is on the power differential, as well as the dynamics between the parties (perpetrator and target), in the course of the conflict (Einarsen et al., 2004). On a social group level, 'scapegoating' processes in groups and organisations are used to explain bullying; groups transfer their frustration and aggression on to a suitable and less powerful group member (e.g. an outsider or a minority; Einarsen et al., 2004). On the organisational level, many factors may contribute to bullying at work, for example, the organisational culture, a high degree of cooperation combined with restricted control over one's own time and a strained and competitive atmosphere (Einarsen et al., 2004). On a societal level, the national culture and historical, legal and socio-economic factors may contribute to bullying behaviour. Furthermore, the level of stress of both a perpetrator and victim is enhanced by the following factors that characterise the working life in many countries: the high rate of change, increasing workloads and work hours and uncertainty regarding future employment. The level of aggression, the coping resources, the tolerance of organisations and their management of cases are influenced by societal factors (Einarsen et al., 2004). The framework developed by these authors suggests that organisational factors as well as an effective support system for victims are of utmost importance to moderate the perceptions and reactions of the victim (Einarsen et al., 2004).
From the above discussion, it is evident that more than one theoretical lens is required to help understand the workplacebullying phenomenon. The Leymann model suggests that organisational factors rather than individual factors contribute to workplace bullying and outline the stages of bullying. Glasl's conflict escalation model specifies how conflicts may escalate into bullying; the model presents the different phases and stages of conflict escalation and offers de-escalation strategies for each phase. The theoretical framework developed by Einarson et al. proposes that workplace bullying is a complex social phenomenon and involves a variety of factors that need to be considered in any bullying situation.

Gender and workplace bullying
A large number of studies indicate ambiguous or conflicting results for the association between gender and workplace bullying (Salin & Hoel, 2013). According to Leo et al. (2014, p. 6059), the reason for this tendency might be because different studies conceptualise 'gender' and 'bullying' differently, and they might be using different measurement methods. In a study conducted by Botha (2019) investigating employees' perceptions and experiences of bullying in the workplace in a South African setting, the author found no significant relationship between gender and workplace bullying. In contradiction, a study conducted by Chan et al. (2019) investigating the prevalence of workplace bullying and its association with socio-economic factors and psychological distress amongst 5235 Malaysian employees revealed clear gender differences in terms of the prevalence of work bullying; a higher proportion of female employees reported being bullied in comparison with their male counterparts. Gender differences in the prevalence of workplace bullying were also confirmed by studies conducted by Moreno-Jiménez et al. (2008) and Salin (as cited in Jones, 2006). Moreno-Jiménez et al. (2008) conducted a study focusing on the prevalence of workplace bullying amongst Spanish employees and found that women experienced higher incidents of bullying than men in the workplace. Salin (as cited in Jones, 2006), in a study conducted amongst business professionals forming part of the Finnish association of graduates in economics and business administration, also found that women experienced more workplace bullying than men. According to Salin (2018, p. 24), the expectation that men should be independent and self-reliant might prevent them from seeking help and/ or reporting a bullying incident. This might be an explanation of the lower prevalence rate of workplace bullying amongst men (Namie & Namie, 2018).

Age and workplace bullying
Previous research reported mixed results on the effect of age on workplace bullying. López-Cabarcos et al. (2017, p. 1023 state that there is uncertainty in the scientific community regarding the validity of age as a characteristic that makes individuals more susceptible to workplace bullying behaviour. Botha (2019) and Skuzińska et al. (2020) found no significant relationship between age and workplace bullying. Skuzińska et al. (2020) investigated the moderating role of individual (gender, age and education) and occupational (employment duration, workplace position, duration of negative behaviour and number of perpetrators) characteristics of victims and perpetrators of negative workplace behaviours in the relationship between workplace bullying and mental health. The study was conducted amongst administrative staff, police officers and hospital staff of a city in northern Poland (Skuzińska et al., 2020).
In contradiction, Ariza-Montes et al. (2017), Einarsen and Skogstad (1996), Hoel and Cooper (2000), Lange et al. (2019) and Cunniff and Mostert (2012) found that age affects the prevalence of workplace bullying. Einarsen and Skogstad (1996) explored the prevalence of harassment and workplace bullying in various Norwegian professions and organisations; the results indicated that older individuals experienced higher levels of bullying. In a study conducted by Hoel and Cooper (2000) amongst 5288 individuals from more than 70 different organisations in Manchester, England, the authors found that middle-aged and younger employees experienced higher rates of bullying. Concomitantly, a study by Cunniff and Mostert (2012) amongst 13 911 employed individuals across six different sectors of work (financial, mining, government, manufacturing, academic and call centres) in South Africa found that younger employees experienced higher rates of bullying in the workplace, whereas older employees experienced significantly lower instances of bullying. In a study by Lange et al. (2019) amongst 4143 employees working in all work sectors across Germany, it was found that older employees reported significantly fewer experiences of bullying. The study focused on prevalence estimates and the role of the perpetrator in the German workforce. Lange et al. (2019) concluded that there were significant differences between the three age groups (31-40 years; 41-50 years; 51-60 years) for both bullying by coworkers and bossing. The study reported that when focusing on supervisors or bosses as the perpetrators in the working environment, young employees were targeted more severely than their older counterparts (Lange et al., 2019

Marital status and workplace bullying
Previous studies revealed a significant relationship between workplace bullying and marital status. Yang and Zhou (2021), in their study on workplace bullying amongst nurses in China, found that marital status was a determining factor in the bullying process at work; married nurses were more likely to experience workplace bullying. Similarly, in a study by Yadav et al. (2020), focusing on the impact of workplace bullying on turnover intention, it was found that amongst Indian school teachers, marital status affects the likelihood of teachers in Indian schools experiencing workplace bullying. Furthermore, the results indicated that married women who have a longer work tenure tended to be least affected by workplace bullying (Yadav et al., 2020). Coherently, Giorgi et al. (2013) found that married individuals reported lower instances of workplace bullying than unmarried individuals. Giorgi et al. (2013) explored the organisational and personal determinants of workplace bullying amongst a sample of Japanese employees. Asakura et al. (Giorgi et al., 2013) argue that married adults might experience less workplace bullying than those who are single or divorced. The results of the study showed that marriage can be a source of support for such bullying experiences, whilst maintaining the integrity of the victim (Asakura et al., as cited in Giorgi et al., 2013).

Level of education and workplace bullying
Previous research indicated that level of education has an effect on the prevalence of workplace bullying. The study conducted by Moreno-Jiménez et al. (2008) showed that individuals with 'elementary education' reported higher levels of bullying than those with 'medium' or 'higher education'. According to Moreno-Jiménez et al. (2008), those with higher levels of education may experience less bullying, as education can offer security through supplying skills to manage conflict. Niedhammer et al. (2007) found that unskilled workers experienced higher incidences of bullying than those with qualifications or higher levels of education.
The study investigated the prevalence of workplace bullying amongst 3132 men and 4562 women of the general working population in the southeast of France. Cunniff and Mostert (2012) also found that South Africans with tertiary education reported less incidences of workplace bullying than those with only secondary education.

Length of employment and workplace bullying
Previous studies have shown that there is a significant relationship between the length of employment and workplace bullying. The study conducted by Moreno-Jiménez et al. (2008) found that workers employed for 16-42 years reported lower instances of workplace bullying than those employed for 8-15 years. In addition, those employed from 8 to 15 years reported lower instances of workplace bullying than those employed for 5-7 years (Moreno-Jiménez et al., 2008). Moreno-Jiménez et al. (2008, p. 104) noted that with the increase in time and experience at work, the risk of workplace bullying tended to decrease. Likewise, a study conducted by Owoyemi (2011) amongst 452 emergency service organisation employees in the UK found that individuals working in an emergency service organisation in the UK from 1 to 5 years reported higher incidences of bullying at work than those who have been employed longer. Awai et al. (2021) investigated the prevalence of workplace bullying and its associated factors amongst 178 hospital workers in Kuala Lumpur, Malaysia. The authors reported higher rates of bullying amongst individuals employed for 10 years or less than those employed for more than 10 years.

Type of employment contract and workplace bullying
Previous studies exploring the relationship between the type of employment contract and workplace bullying reported indecisive results (Moreno-Jiménez et al., 2008). Kivimäki et al. (2000) investigated workplace bullying and sickness absence amongst a cohort of 674 male and 4981 female hospital employees aged 19-63 years across Finnish organisations. The authors found no significant differences between the workplace bullying experienced by temporary and permanent employees. In contrast, Hoel and Cooper (2000) found that permanent employees experienced higher instances of workplace bullying than temporary workers. The results showed that temporary employees may be more at risk because of being seen as a 'potential disturbance' to the social setting of permanent employees. Temporary employees may be too afraid to defend themselves when the conflict arises with a permanent employee (Notelaers et al., 2010).
From the above discussion, it is evident that sociodemographic variables influence how workplace bullying is perceived and experienced in the working environment.

Research design and methodology
The study was conducted in a positivistic research paradigm, which is guided by a realist and objectivist ontology and an empiricist epistemology (Sarantakos, 2013). A quantitativebased survey design was used. A survey design describes trends, attitudes and opinions of a population and/or tests for associations amongst variables of a population (Creswell & Creswell, 2018, p. 147).

Research setting, population and sampling
The research was conducted in 2020 during the national lockdown period in South Africa. The research setting was limited to one higher education institution in South Africa. The target population of the study comprised all employees (academic and support staff) of the higher education institution. Therefore, employees with full-time, fixed-term and temporary employment contracts were targeted, as well as those with a postdoctoral fellowship. No one was excluded based on age, gender or ethnicity. A complete survey of the target population was undertaken, but as not all employees participated in the research, it resulted in a nonprobability sample. Therefore, convenience sampling was used to select the respondents. Convenience sampling refers to research participants who are available to the researcher by virtue of their accessibility (Creswell, 2014, p. 204).

Instrumentation and data collection
A web-based questionnaire was used to collect the data, which included two sections. Section A contained biographical questions, which included questions on gender, age, marital status, highest qualification, length of employment, nature of employment at the university and employment contract at the university. For Section B, the Negative Acts Questionnaire_Revised (NAQ-R) developed by Einarsen, Hoel and Notelaers (2009) for measuring perceived exposure to bullying and victimisation at work was used. The scale has excellent reliability and construct validity; Cronbach's alpha for the one-dimensional model, the two-dimensional model (person-related bullying and work-related bullying) and the three-dimensional model (person-related bullying, work-related bullying and physically intimidating bullying) exceeded 0.70 (Einarsen et al., 2009, p. 31

Analysis and reporting
The data collected were processed using the Statistical Package for the Social Sciences (SPSS) version 27. A confirmatory factor analysis was conducted to validate the factor structure of the workplace bullying construct and the reliability of Cronbach's alphas. Three goodness-of-model-fit indices were used to determine how well the measurement model fits the data of the sample. According to Hancock and Mueller (2010), it is a good practice to report multiple-fit indices, preferably from three broad classes. The following indices were reported: chi-square statistic divided by degrees of freedom (CMIN/DF), the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) with its 90% confidence intervals.
Cronbach's alpha coefficient was used to determine the internal consistency and reliability of the scale. Cronbach's alpha is the most common measure of scale reliability (Field, 2005, p. 667). Ideally, the Cronbach's alpha coefficient of a scale should be above 0.7 (DeVellis, as cited in Pallant, 2016, p. 101). Cronbach's alpha values are quite sensitive to the number of the items in a scale; therefore, a scale with fewer than 10 items might result in a low value. In this case, the mean inter-item correlation may be appropriate to report. Briggs and Cheek (as cited in Pallant, 2016, p. 101) recommend inter-item correlation values of 0.2-0.4 as an optimal range to be reported.
Descriptive statistics were interpreted and reported by mean and standard deviation. The independent-samples t-test was used to compare the mean scores of two independent groups on the continuous variable workplace bullying to determine whether there is statistical evidence that the population means differed significantly (Frost, 2020). Analyses of variance (ANOVAs) were used to determine whether the mean scores of two or more groups on the continuous variable workplace bullying differed significantly (see Pallant, 2016, p. 109).
Effect sizes were used to measure the strength of the relationship between two groups; the larger the effect size, the stronger the relationship between two variables in a population and vice versa (McLeod, 2019). Cohen's d-values were used as effect sizes to determine whether differences in means are important in practice. Cohen (1988) suggests that d = 0.2 be considered a small effect size, d = 0.5 a medium effect size and d = 0.8 a large effect size.
Spearman's rank-order correlation was used to determine the strength and direction of the linear relationship between two ranked variables (Pallant, 2016, p. 132). Cohen (Pallant, 2016, p. 137) suggests the following guidelines for interpretation: r = 0.10-0.29 is considered a weak (small) relationship, r = 0.30-0.49 a moderate (medium) relationship and r = 0.50-1.0 a strong (large) relationship.

Ethical considerations
The following ethical considerations were adhered to when conducting and reporting the research: maintaining professional integrity, obtaining informed consent, voluntary participation and protecting the privacy of the respondents by ensuring the anonymity and confidentiality of the research setting, as well as during data collection, analysis and reporting, as recommended by Babbie and Mouton (2011) and Sarantakos (2013). Approval to conduct the research was obtained from the Basic and Social Sciences Research Ethics Committee of the university under investigation (reference number NWU-00900-19-A7).

Confirmatory factor analysis results of the workplacebullying measurement model
Confirmatory factor analysis was conducted to test the structure and relations between the latent variables that underlay the data. As the three-factor solution of the Negative Acts questionnaire had been theoretically and empirically supported (Einarsen et al., 2009), the measurement model was tested. The measurement model included seven items on work-related bullying, 14 items on person-related bullying and three items on physically intimidating bullying. The results of the confirmatory factor analysis showed that all factor loadings were statistically significant at the 0.05 level. Factor loadings ranged from 0.540 to 0.869 for work-related bullying, from 0.422 to 0.841 for person-related bullying and from 0.401 to 0.672 for physically intimidating bullying. The standardised regression coefficients were interpreted as factor loadings. Field (2005) suggests that a factor loading of an absolute value of more than 0.3 is considered important. Confirmatory factor analyses showed that the three-factor solution of the workplace bullying scale did not fit well to the data of the sample (see Table 2). A CMIN/DF value of 4.628 was found for the measurement model, indicating an acceptable model data fit, as suggested by Mueller (1996) and Paswan (as cited in Shadfar & Malekmohammadi, 2013). A CFI of 0.801 was found for the measurement model, indicating not a good model data fit as it is lower than 0.9 as suggested by Hair, Anderson, Tatham and Black (2010) and Mueller (1996). The measurement model obtained an RMSEA value of 0.114 with a 90% confidence interval of 0.107 (low) and 0.120 (high), indicating not a good model data fit. Hu and Bentler (as cited in Brown & Moore, 2012) recommend a value equal to or smaller than 0.06 as a cut-off value for a good fit. According to Blunch (2008), models with RMSEA values of 0.10 and larger should not be accepted.

Exploratory factor analysis of workplace bullying
As the measurement model analysed indicated 'not a good fit', an exploratory factor analysis was conducted on the 24 Likert-type scale items measuring perceived exposure to bullying and victimisation at work. Principal component analysis and oblimin rotation were used. The Kaiser-Meyer-Olkin test returned a value of 0.942, indicating that the sample size was adequate for factor analysis. The p-value of Bartlett's test of sphericity obtained a value smaller than 0.05, indicating that the correlation between items was sufficient for factor analysis (Field, 2005). Four factors, namely exclusion, hostility, managerial misconduct and humiliation and belittlement, were extracted through Kaiser's criteria (Field, 2005), which explained 66.41 of the total variance.
The Cronbach's alpha coefficient for all four factors (exclusion: α = 0.928; hostility: α = 0.765; managerial misconduct: α = 0.787; humiliation and belittlement: α = 0.920) measured above 0.7, showing high reliability and internal consistency, as suggested by Field (2009). Eight statements loaded on exclusion (ranging from 0.427 to 0.869), four on hostility (ranging from 0.525 to 0.792), four on managerial misconduct (ranging from 0.475 to 0.841) and eight on humiliation and belittlement (ranging from 0.505 to 0.705). All items loaded satisfactorily with a factor loading of above 0.4.
A confirmatory factor analysis was conducted to test the structure and relations between the latent variables that underlay the data of the four-factor workplace bullying model; the results are presented in Figure 1.
All factor loadings were statistically significant at the 0.05 level. Factor loadings ranged from 0.622 to 0.914 for exclusion, from 0.555 to 0.788 for hostility, from 0.662 to 0.748 for managerial misconduct and from 0.709 to 0.814 for humiliation and belittlement. Table 2 indicates the goodness-of-model-fit indices of the four-factor model. All the indices showed an acceptable model fit; a CMIN/DF value of 3.218, a CFI of 0.880 and an RMSEA value of 0.089 with a 90% confidence interval of 0.082 (low) and 0.096 (high) were obtained as opposed to the values reported for the three-factor model.

Effect of sociodemographic variables on workplace bullying
Effect of gender, nature of employment, nature of employment contract and marital status on workplace bullying: Table 3 presents the results of the independent sample t-tests for gender, nature of employment and nature of employment contract.
The results of the t-tests indicated no significant differences between the means of male and female respondents, as well as for respondents with full-time and fixed-term, temporary or postdoctoral fellowship employment contracts for all the factors of workplace bullying. The results of the t-test revealed no significant differences between the means of academic and support staff respondents for hostility and managerial misconduct; the p-values measured higher than 0.05. However, the p-values of exclusion and humiliation and belittlement both measured 0.01, indicating a significant difference in the mean scores of academic (exclusion: M = 1.81; humiliation and belittlement: M = 1.55) and support staff (exclusion: M = 2.08; humiliation and belittlement: M = 1.77) respondents. The effect sizes showed a small effect; the d-values for both factors were 0.28. The results of the ANOVA test showed no significant differences between the means of the different categories for marital status for all the factors of workplace bullying.
Correlation between age, highest qualification and years working at the university and workplace bullying: Table 4 presents the results of Spearman's rank-order correlation test between ordinal sociodemographic variables and workplace bullying.
Small negative correlations were found between highest qualification and hostility (p = 0.01, r = −0.146), as well as highest qualification and humiliation and belittlement (p = 0.02, r = −0.135). Small positive correlations were found between the number of years working at the university and managerial misconduct (p = 0.014, r = 0.147) and the number of years working at the university and humiliation and belittlement (p = 0.038, r = 0.124).
Correlation between workplace bullying factors: Table 5 outlines the results of Spearman's rank-order correlation test between exclusion, hostility, managerial misconduct and humiliation and belittlement.
Medium to large positive correlations were found between exclusion and hostility (p = 0.00; r = 0.472), exclusion and managerial misconduct (p = 0.00; r = 0.621) and exclusion and humiliation and belittlement (p = 0.00; r = 0.811). Medium to large positive correlations were observed between hostility and managerial misconduct (p = 0.00; r = 0.428) and hostility and humiliation and belittlement (p = 0.00; r = 0.585). A large positive correlation was found between managerial misconduct and humiliation and belittlement (p = 0.00; r = 6.33).

Discussion
This study was conducted to determine the perceptions of workplace bullying amongst academic and support staff at a higher education institution in South Africa and to ascertain how they are affected by the sociodemographic variables of gender, age, marital status, highest qualification, length of employment, nature of employment and employment contract.
Confirmatory factor analyses showed that the three-factor structure of the workplace bullying measurement model did not fit well with the data of the sample. Only the CMIN/DF goodness-of-model-fit indices indicated an acceptable fit. Therefore, an exploratory factor analysis was conducted on the Likert-type scale items of the workplace bullying scale; four factors (exclusion, hostility, managerial misconduct and humiliation and belittlement) were extracted through Kaiser's criteria (Field, 2005). The Cronbach's alpha coefficient for all four factors showed high reliability and internal consistency. All three goodness-of-model-fit indices (CMIN/DF, CFI and RMSEA) showed an acceptable model fit. The mean scores for all four factors ranged from 'Never' (1) to 'Now and then' (2). The mean scores showed that on average, the respondents were subjected to negative behaviours related to exclusion (M = 1.95), managerial misconduct (M = 1.79) and humiliation and belittlement (M = 1.66) more often than hostility (M = 1.23). Davenport (2014, p. 16) states that workplace bullying is increasingly experienced amongst educators who lack the knowledge and skill to be able to cope with or handle the phenomenon. Concomitantly, Gilman (2015) argues that the longer the phenomenon is left unattended, the worse it will become and the more difficult it will be to counteract workplace bullying. If bullies believe that they can get away with their behaviour, workplace bullying can become even more prevalent. In this regard, Woodrow and Guest (2017) emphasise the necessity of proper prevention and intervention measures being put in place to protect employees against the workplace bullying phenomenon. Through the use of intervention measures as both proactive and reactive to workplace bullying, Catley et al. (2013) believe that employers can play a pre-emptive role in the prevention of and intervention in workplace bullying.
Regarding the effect of the sociodemographic variables on workplace bullying, the results of the statistical tests revealed the following: The results of the t-tests showed no significant differences between the mean scores of male and female respondents, as well as for respondents with full-time and fixed-term, temporary or postdoctoral fellowship employment contracts for all the workplace bullying factors. Furthermore, the results of the ANOVA test indicated no significant differences between the mean scores of the different marital status categories for all workplace bullying factors. The results of this study confirmed those of the study conducted by Botha (2019) and Kivimäki et al. (2000). Botha (2019) found no significant differences between the workplace bullying experiences of male and female respondents. Kivimäki et al. (2000) found no significant differences between the workplace bullying experiences of temporary and permanent employees. However, the results of this study contradicted those of previous studies, which found that gender (Chan et al., 2019;Moreno-Jiménez et al., 2008;Salin, as cited in Jones, 2006), marital status (Giorgi et al., 2013;Yadav et al., 2020;Yang & Zhou, 2021) and employment contract (Hoel & Cooper, 2000;Moreno-Jiménez et al., 2008;Notelaers et al., 2010) influence the experiences of workplace bullying.
The t-tests showed that the mean scores of academic and support staff respondents differed significantly in terms of exclusion and humiliation and belittlement. It can be deduced that the support staff respondents were more subjected to exclusion and humiliation and belittlement than the academic staff respondents. Westbrook et al. (2021) as well as Kakumba. Wamala and Wanyama (2014) found that amongst management and other employees, administrative staff reported being bullied more regularly.
Small negative correlations were found between highest qualification and humiliation and belittlement and hostility. This indicates that the more qualified respondents were, the less they were subjected to hostility and humiliation and belittlement. This result is supported by studies conducted by Moreno-Jiménez et al. (2008), Niedhammer et al. (2007) and Cunniff and Mostert (2012), who found that employees with higher levels of education may experience less bullying.
Small positive correlations were found between the number of years working at the university and managerial misconduct and humiliation and belittlement, indicating that the longer the respondents had worked at the university, the more they had *, correlation is significant at the 0.05 level (two-tailed). **, correlation is significant at the 0.01 level (two-tailed).
(1) Small effect: r = 0.1, (2) medium effect: r = 0.3 and (3) large effect: r > 0.5 http://www.sajhrm.co.za Open Access been exposed to managerial misconduct and humiliation and belittlement. This result is in contrast with the those of the studies conducted by Moreno-Jiménez et al. (2008), Owoyemi (2011) and Awai et al. (2021), who found that workers with longer tenure reported lower instances of workplace bullying.
Medium to large positive correlations of between 0.428 and 0.811 were found between the four workplace bullying factors. This suggests that the more respondents experienced being excluded, the more they were subjected to hostility, managerial misconduct and humiliation and belittlement. Furthermore, the more hostile the working environment was perceived to be by the respondents, the more likely they were subjected to exclusion, managerial misconduct and humiliation and belittlement. Finally, the more the respondents experienced managerial misconduct, the more they were subjected to hostility, exclusion and humiliation and belittlement.

Limitations
The research was conducted in the early stages of the national lockdown period in South Africa in 2020. This may have affected the participation of the respondents in the webbased survey, as people still had to adapt to the variety of challenges posed by the lockdown and might not have been willing to participate in the research. Furthermore, the sample of the study only included employees of one higher education institution in South Africa, and therefore the results cannot be generalised to the tertiary education sector of South Africa.

Recommendations
Bullying in the working environment has become an increasingly detrimental phenomenon that can also be observed in higher education institutions. Similar studies could be conducted at other higher education institutions in South Africa to create a better understanding of the experiences of workplace bullying in the higher education context in the country. Further research on the psychometric properties of the four-factor workplace bullying model could be conducted to determine whether it is applicable in other contexts as well. Qualitative investigations might provide further in-depth insights into the phenomenon. Managers, human resource departments and human resource practitioners can implement the following practical recommendations in an attempt to combat bullying in organisations: • Establishing an anti-bullying culture by adopting a clear zero-tolerance stance against bullying behaviour. • Implementing and operationalising an anti-bullying policy; such a policy should entail clear definitions of the acts that are associated with workplace bullying and the procedures to follow to report and handle incidents. • Creating awareness of workplace bullying and procedures to follow in the case of incidents. • Education and training on policies, procedures and practices to follow in reporting bullying incidents.
• Effectively addressing incidents of conflict and bullying in the working environment; in this regard, the conflict de-escalation strategies suggested by Glasl (1982) can be followed.

Conclusion
The main purpose of this study was to determine the perceptions of workplace bullying amongst academic and support staff at a higher education institution in South Africa and to ascertain how they are affected by selected sociodemographic variables. From the quantitative results, it is evident that workplace bullying was experienced, to a certain extent, at the higher education institution under investigation. The respondents experienced negative behaviours related to exclusion the most, followed by managerial misconduct, humiliation and belittlement and hostility the least. The results revealed that nature of employment, highest qualification and length of employment played a role in the experiences of workplace bullying at the higher education institution under investigation. In order to understand and address bullying in working environments, it is of utmost importance that the variety of factors (individual, personal, power differentials, organisational and societal) that might contribute to bulling behaviour are considered. The study contributed by producing a four-factor workplace bullying model that can be further investigated in different contexts.

Funding information
This research work received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Data availability
The data that support the findings of this study are available on request from the author (D.B.).