Early work-environmental indicators of bullying in an academic setting: a longitudinal study of staff in a medical university

ABSTRACT

The purpose of this study was to use a longitudinal approach to investigate the association between work-related psychosocial factors and workplace bullying in an academic setting. A questionnaire was sent out three times to about 4500 academic- and non-academic staff in a medical university. Those two group were analyzed together and separately. Logistic regression analyses were used to calculate odds ratios (OR). In the study, we found that factors such as lack of support, low control, inconsistent role demands, poor leadership, poor organizational climate, high sickness absence and high sickness presence can be important predictors of future bullying. We found that factors in an academic setting differ somewhat for academics and non-academic staff. This might indicate that type of job should be take into consideration in attempts to successfully prevent bullying at an early stage.

KEYWORDS:

• Academic setting
• work environment
• bullying
• longitudinal study
• university staff

Introduction

Workplace bullying has been an important area of research internationally for quite some time and is a field of global interest in which new topics are steadily emerging (Nielsen and Einarsen 2018). It has been suggested that workplace bullying is one of the most stressful phenomena (Pheko 2018), and several studies have demonstrated its detrimental effects on the mental health of targets (Conway et al. 2018). Given that bullying is a severe work-environmental factor, it can have a major impact not only on health but also on an individual’s work performance (Einarsen et al. 2011). A Swedish study has shown that production loss in academia was strongly associated with work-environment problems (Lohela-Karlsson, Nybergh, and Jensen 2018).

Bullying has been defined in many ways over the years. However, in a recent meta-analysis of bullying the authors conclude that workplace bullying is a unique and especially detrimental form of aggression in the workplace (Nielsen and Einarsen 2018). In studies of bullying it is often defined as situations in which an individual is harassed, offended or socially excluded, or in which their work is negatively affected. To be defined as bullying it must occur repeatedly and regularly (e.g. weekly) and over a period of time (e.g. about six months). Bullying can be seen as an escalating process, in the course of which the subject of the bullying ends up in an inferior position and becomes the target of systematic negative social acts (Einarsen et al. 2011).

Work environment and workplace bullying

A frequently used hypothesis about the causes of workplace bullying is the work-environment hypothesis (Einarsen, Raknes, and Matthiesen 1994; Leymann 1996). The work-environment hypothesis assumes that bullying is a consequence of a poor social environment in organizations. This hypothesis has received substantial empirical support (Bowling and Beehr 2006). A systematic review of work stressors from cross sectional studies showed that role conflict, workload, role ambiguity, job insecurity and cognitive demands were the most significant predictors of workplace bullying (Brande et al. 2016). This systematic review included several longitudinal studies. The results from these supported the cross-sectional findings showing that workload, job insecurity, and role conflict were the factors that increased the risk of workplace bullying (Brande et al. 2016). Previous research has also identified inadequate leadership as a precursor of bullying in the workplace (Leymann 1990, 1996), with managers themselves frequently being perpetrators (Rayner, Hoel, and Cooper 2002). Some studies have reported that those who perceive themselves as targets of bullying (compared to those who do not) consistently report poor leadership (Coyne et al. 2003; Coyne, Seigne, and Randall 2000).

Workplace bullying occurs in all kinds of organizations and professions, with academia being no exception. In fact, several of the studies report a quite high prevalence of bullying in academia compared to other types of organizations (Keashly and Jagatic 2011). After reviewing studies of bullying in academic settings, Keashly et al., concluded that bullying behaviors are an aspect of the academic landscape which not only damages the targets and bystanders but also the institution itself (Keashly and Jagatic 2011). Westhues (2004) describes academia as having a perfect climate for bullying with subjective performance measures and conflicting goals (Westhues 2004).

In an interview study among university staff, targets of bullying identified the three most common risk factors related to bullying were competition about status or duties, jealousy and the perpetrator being unsure of himself /herself (Björkqvist, Österman, and Hjelt-Back 1994). Studies have shown that negative behaviors found in the academic environment frequently involve threats to professional status and isolating and obstructive behavior (i.e. thwarting the target’s ability to obtain important objectives) (Nelson and Lambert 2001). However, there are different staff categories in an academic setting from academics such as researchers, lecturer to non-academics for examples administrators. In a study of Björkqvist et al., (1994) it was found that the highest rate of bullying was experienced by those whose work focused on administration, economy and service, whereas the lowest level was found among those whose activities instead focused on teaching and research.

Sickness absenteeism, presenteeism and bullying

Sickness absence is clearly multifactorial, but exposures in the workplace are certainly an important factor behind health and illness. Primary studies (Einarsen and Nielsen 2015) as well as meta-analyses (Verkuil, Atasay, and Molendijk 2015) suggest that exposure to bullying in the workplace increases the risk of mental and somatic health complaints prospectively. By extension, it seems reasonable to assume that bullying is a risk factor for sickness absence. One study which examined the frequency of bullying and its relationship with sickness absence found that the risk of long-term sickness absence was higher for those who reported being bullied most frequently (Ortega et al. 2011). A European study looking at the association between psychosocial work factors and sickness absence in 31 countries concluded that bullying had a stronger association with sickness absence than any other work-environmental factors (Niedhammer et al. 2012). Similar results were reported by Slany and colleagues (Slany et al. 2014). A concept closely related to sickness absence is sickness presenteeism, working while ill (Aronsson and Gustafsson 2005) is highly prevalent among workers in the European Union (Eurofound 2012). Although we do not know of any specific studies of presenteeism in an academic setting, we can assume that it will be equally high in this kind of environment. A relationship between bullying and sickness absence has been reported in several studies, sickness presenteeism, on the other has, has been studied less frequently.

Nevertheless, working while ill may compound the effects of the initial illness and result in negative job attitudes and withdrawal from work. This, in turn, can affect performance as well as the relationship with their colleagues in a negative way (Johns 2010). Those negative relations could in the worst cases results in conflicts or in the long run to harassments or bullying, especially if other individuals are dependent on the performance of the ill individual. Previous studies on bullying and presenteeism have reported that frequent (i.e. daily or weekly) exposure to workplace bullying was associated with higher reporting of sickness presenteeism in the preceding year (Conway et al. 2016). In another study it was found that negative relational experiences such as being exposed to harassement, discrimination and abuse were positively related to presenteeism (Miraglia and Johns 2016). Futhermore, across-sectional study has shown a quite strong relationship between high levels of bullying and presenteeim (Janssens et al. 2016).

Studies in academic settings tend to focus on students’ experience of bullying. There is therefore substantial potential for further research into the phenomenon of employee bullying in academia. A number of studies using a longitudinal design have been carried out in academic settings but not, to the best of our knowledge, looking at different staff categories.

In our study we include sickness absence and presenteeism as predictors rather than consequences. Since bullying has highly negative effects on both the individual, the organization and society at large, it is important to develop measures for prevention by the early identification of risk markers for bullying.

The purpose of the study

The overall aim was to use a longitudinal approach to investigate the association between work-related psychosocial factors and the reporting of being bullied in an academic setting. We also investigate whether those factors are similar for academic and non-academic staff.

The above-mentioned studies examined workplace bullying as a risk factor for sickness absenteeism and sickness presenteeism. The present study will reverse this and investigate the role of sickness absenteeism and presenteeism as potential risk factors for workplace bullying.

Method

Study population

An employee survey administered on three occasions using the same questionnaire was conducted at a medical faculty in a Swedish university.

Data collection

A questionnaire was sent out three times by web survey at 24-month intervals. Each survey lasted for three weeks. Two reminders were distributed during this time. Participation was voluntary and written informed consent was obtained from each employee. Three different measures were collected, for measure T1 N = 4 310, for measure T2 N = 4 168-and for the final measure T3 N = 4 661 questionnaires were sent. The response rate was: 69% at measure T1; 77% at measure T2; and 76% at measure T3. Table 1 presents the background factors for the different measures.

Table 1. Background factors at the different measures (T1, T2, T3).

Background factors total/included subjects	T1	T2	T3
Total number of respondents	2993	3213	3525
Age (Mean)	44.4	43.8	43.5
Men, n (%)	953 (32)	1093 (34)	1192 (43)
Women, n (%)	2040 (68)	2120 (66)	2333 (57)
Academics, n (%)	1779 (59)	1832 (57)	2026 (57)
Non-academic staff, n (%)	1214 (41)	1381 (43)	1499 (43)

Subjects

All employees at a medical faculty in Sweden were invited to participate in the employee survey. For the purpose of this study, only those who reported not being targets of bullying at baseline were included. The criteria for inclusion in the present study were (i) having worked for at least one year at the place of work, (ii) not having reported being bullied at baseline (iii) have responded to two of the three assessments. The number of participants were in T1 = 2993, T2 = 3213, and T = 3525. To optimize the number of available subjects in this study we combined subjects from measure 1 (T1) and measure 2 (T2) as baseline measures. Of those responding at T1, 1544 participants and from T2 840 fulfilled inclusion criteria’s and were thus included. Those participants were not bullied at baseline, they had worked at least one year and had responded to assessment a follow up assessment (Figure 1 and Table 1). After combining two baselines the total number of included subjects was n = 2.384. Table 2 presents the background factors for the included subjects.

Figure 1. A schematic representation of participants in the study, included in the logistic regression analyses.

Table 2. Background factors for the included participants.

Included participants	Number included	Number being bullied at T2
Total	2 384	95 (4.0)
Men, n (%)	816 (34)	33 (4.0)
Women, n (%)	1568 (66)	62 (4.0)
Academics, n (%) *	1 290 (55)	49 (3.8)
Non-academic staff, n (%) *	1071 (45)	46 (4.3)

*In some cases, information about profession was missing.

The questionnaire

Independent variables

Background variables used in the analyses are sex and age.

Sick leave was measured by a question taken from the Workability Index (Tuomi et al. 1994) ‘How many days during the past 12 months have you been absent from work due to illness (sick leave, care, treatment or study)?’ This self-reported sick leave measure has been shown to correlate acceptably well with register data (Voss et al. 2008).

Presenteeism was measured by the following question, ‘Over the last 12 months have you ever gone to work despite feeling that you really should have taken sick leave because of your state of health?’ The response format was (1) ‘No, never,’ (2)’ Yes, once,’ (3) ‘Yes, 2–5 times,’ (4) ‘Yes, more than 5 times.’ This question has been used in earlier research into sickness presenteeism (Aronsson, Gustafsson, and Dallner 2000).

Psychosocial work-environment factors were measured by the QPSNordic, an instrument for investigating psychological, social and organizational conditions in the workplace and developed by a group of Nordic researchers. The instrument has been psychometrically evaluated in the Nordic countries and in a variety of work settings. It has demonstrated satisfactory validity and reliability (Dallner et al. 2000).

The questionnaire covered a number of areas such as quantitative demands (4 questions), decision demands (3 questions), control of decision (5 questions), control of work pace (4 questions), empowering leadership (3 questions), fair leadership (3 questions), social climate (3 questions), innovative climate (3 questions), role clarity (3 questions), role conflict (3 questions), predictability of the work situation over the next two years (3 questions), support from colleagues (2 questions), support from the manager (3 questions), commitment to the organization (3 questions) and perception of mastery (4 questions). The questions have five response options from ‘very seldom or never’ to ‘very often or always’, using a 5-point Likert scale (value from 1 to 5).

The single item responses were treated according to the scale structure suggested by Dallner et al in the development of the QPSNordic scales (Dallner et al. 2000). These work index/scales were then used to examine their relationship to future bullying.

The dependent variable

Bullying included a question from the QPSNordic questionnaire (Dallner et al. 2000): ‘Have you been subjected to bullying in the last 6 months?’ with the response alternatives Yes / No. In the questionnaire, bullying was explained as follows

Bullying is a problem at some workplaces and for some employees. Bullying generally refers to reprehensible behavior on repeated occasions. However, in very serious cases (e.g. sexual violence), one occasion is sufficient. Bullying, however, should not be confused with temporary conflicts and problems in working relationships.

Statistical analysis

In order to investigate the association between psychosocial work-related factors and future bullying, a series of logistic regression models were performed. Before performing the statistical analyses, the work-environment factors were dichotomized into high and low respectively, using the median. Low was defined as all values below the median and high was defined as the median value as well as all values above. In this study, low will refer to a poor working environment and high to a good working environment.

The decision to use median values was based on the non-normal distribution of the variables. The use of median values as cut-offs has also been used in previous studies (Karasek and Theorell 1990). However, to avoid the risk of basing results only on the cut-off points, the psychosocial factors used were also analyzed as both a continuous variable and as categories. These results are not presented in this study but can be requested from the author.

The data sets were analyzed in two steps. In the first, bivariate logistic analysis was performed to examine the relationship between the psychosocial work-related factors, sick leave, sick presenteeism and future bullying. Logistic regression analyses were used to calculate odds ratios (OR) with 95 percent confidence intervals (CI). Both unadjusted OR and adjusted OR for age and sex are presented in the results. In the second step, a logistic regression model was performed which only included those independent variables with a statistically significant predictability, a 0.05 significance level. Following the initial analysis, we present a logistic regression model to determine which of the variables used in this investigation had the strongest relation with reports of future bullying. Both unadjusted (crude) OR and adjusted OR for age and sex are presented in the results as final models. All analyses were carried out using IBM SPSS Statistics v 23.0.

Results

Table 2 presents the results of the bivariate logistic regression for the relationship between future bullying and each of the psychosocial work factors investigated. The table shows the work factors associated with future bullying. In the analysis, ‘low’ indicates a poor working environment and ‘high’ a good one.

The findings, which include the total group of participants (Table 3), demonstrate that individuals who experienced a poor (low value) social climate at work had a more than three times higher OR of being the target of bullying than individuals who experienced the work climate as encouraging, supportive and comfortable (high value). Furthermore, those working in a poor innovative climate had a higher risk of being bullied.

Table 3. Unadjusted and adjusted odds ratios for future bullying N = 2384.

		Future bullying
Index Range 1–5	Categories	Unadjusted OR (95 % CI)	Adjusted (age and sex) OR (95% CI)
Quantitative demands	Low < 3.25 High >= 3.25	1.20 (0.80-1.80)	0.82 (0.54-1.24)
Decision latitude demands	Low < 3.30 High >= 3.30	1.11 (0.73-1.70)	0.87 (0.56-1.36)
Control decisions	Low < 3.50 High >= 3.50	2.05 (1.34-3.00)^a	2.00 (1.33-3.00)
Control work pace	Low < 4.25 High >= 4.25	1.44 (0.95-2.20)	1.14 (0.93-2.15)
Empowering leadership	Low < 3.33 High >=3.33	1.88 (1.23-2.86)	1.86 (1.21-2.85)
Fair leadership	Low < 4.00 High >= 4.00	2.17 (1.43-3.30)	2.16 (1.43-3.28)
Support from manager	Low < 4.00 High >= 4.00	1.73 (1.40-2.6)	1.70 (1.11–2.60)
Support from colleagues	Low < 4.00 High >= 4.00	1.98 (1.31-3.00)	1.95 (1.29–2.95)
Social climate	Low < 4.00 High >=4.00	3.13 (2.02-4.85)	3.13 (2.02-4.84)
Innovative climate	Low < 4.00 High >= 4.00	1.67 (1.04-2.67)	1.66 (1.04-2.66)
Predictability^b	Low < 3.33 High >=3.33	1.88 (1.23-2.88)	1.90 (1.24–2.93)
Perception of mastery	Low < 4.00 High >= 4.00	1.39 (0.92-2.10)	1.43 (0.94-2.16)
Incompatible roles at work (role conflicts)	Low < 2.33 High >=2.33	1.78 (1.13-2.8)	1.75 (1.11–2.76)
Role clarity	Low < 4.00 High >= 4.00	1.40 (0.92-2.12)	1.43 (0.94-2.17)
Organizational commitment	Low < 3.67 High >= 3.67	1.43 (0.92-2.20)	1.42 (0.91-2.21)
Sick leave (0 days reference)^b
	1- 24 days	1.60 (1-03-2.48)	1.45 (0.94-2.23)
	25- 365 days	2.25 (1.10-4.6)	2.96 (1.11-7.85)
Presenteeism (0 days reference)^b
	1- 10 days	1.97 (1.16-3.35)	2.03 (1.19-3.44)
	10+ days	5.3 (2.78-10.17)	5.5 (2.89-10.49)

^a Bold number indicate a significant level of 5% or less.

^b Data is missing in some of the cases.

Individuals who experienced poor leadership also had a higher risk of being bullied. The results show that employees who experienced unfair leadership or lack of empowering leadership (low values) were around twice as likely to be the subject of future bullying. An incompatible role at work (high role conflicts) also predicted a higher likelihood of being bullied, as did a lack of support from managers (low value). High reported sick leave (25 days or more) and high presenteeism were associated with the probability of being a target of future bullying. Additional factors were low control over decisions about one’s work and low predictability of the individual work situation over the next two years.

In the analysis presented in Table 4, the two professional groups were separated. It was shown that they differ somewhat in terms of work-environmental factors. However, a number of factors associated with a significant high risk were common to both groups. These were lack of empowering leadership, unfair leadership, lack of support from colleagues, poor social climate and presenteeism of 10+ days.

Table 4. Adjusted odds ratios for future bullying N = 2361. Academics and non-academic staff.

	Future bullying Adjusted (age and sex) OR (95% CI)
Index Range 1–5	Academics N = 1290	Non-academics  N = 1071
Quantitative demands	0.78 (0.40-1.51)	1.68 (0.88-3.11)
Decision latitude demands	0.83 (0.42-1.67)	1.38 (0.76-2.52)
Control decisions	2.31 (1.30-4.10)^a	1.73 (0.90- 3.48)
Control work pace	1.25 (0.70-2.23)	1.67 (0.89-3.09)
Empowering leadership	1.84 (1.03-3.30)	1.94 (1.01-3.71)
Fair leadership	2.20 (1.24-3.93)	2.13 (1.17-3.87)
Support from manager	1.44 (0.80-2.58)	2.13 (1.13–4.00)
Support from colleagues	2.09 (1.17-3.70)	1.91 (1.06–3.48)
Social climate	3.22 (1.75-5.93)	3.07 (1.63-5.76)
Innovative climate	1.85 (0.97-3.53)	1.41 (0.70- 2.81)
Predictability^b	2.26 (1.23-4.10)	1.50 (0.79–2.84)
Perception of mastery	1.34 (0.76-2.38)	1.55 (0.84-2.85)
Incompatible roles at work (role conflicts)	2.00 (1.0-3.40)	1.77 (0.94–3.32)
Role clarity	1.22 (0.67-2.20)	1.68 (0.92-3.05)
Organizational commitment	1.58 (0.85-2.84)	1.22 (0.65-2.29)
Sick leave (0 days reference)^b
1-24 days	1.45 (0.33-6.34)	1. 57 (0.81-3.03)
25-365 days	1.71 (0.93-3.18)	2.48 (1.04-5.95)
Presenteeism (0 days reference)^b
1-10 days	2.00 (0.91-4.43)	2.17 (1.01-4.62)
10+ days	5.94 (2.65-15.67)	6.0 (2.13-16.97)

^a Bold number indicate a significant level of 5% or less.

^b Data is missing in some of the cases.

There were also differences between the professional groups. Lack of support from managers; sick leave of more than 25 days; and presenteeism of less than 10 days were the additional factors with high OR for the non-academic staff. Academics, on the other hand, had high OR for low predictability, high role conflict and low control over decisions.

The multivariable analyses in Table 5 show the final models (Step 2) in which all statistically significant factors from tables 3 and 4 were included. In the final models, the same factors predicted a higher likelihood of future bullying, namely poor social climate and high presenteeism.

Table 5. The final model, multi-adjusted odds ratios (OR) with 95% confidence intervals (95 percent CI) for future experience of bullying. Total population and the two groups (academics and non-academic staff) separately.

	Total N = 2179^b	Academics N = 1195^b	Non-academic N = 1065^b
Index	OR (95% CI)	OR (95 % CI)	OR (95 % CI)
Control decisions	1.37 (0.83-2.25)	1.46 (0.76–2.82)	Not included
Empowering leadership	1.23 (0.68-2.22)	1.12 (0.55–2.29)	1.14 (0.50-2.68)
Fair leadership	1.54 (0.90- 2.65)	1.20 (0.58–2.46)	1.30 (0.63–2.68)
Support from manager	0.70 (0.38-1.30)	Not included	1.03 (0.42-2.52)
Support from colleagues	1.21 (0.74-2.00)	1.17 (0.60-2.29)	1.24 (0.64-2.41)
Social climate	2.46 (1.42-4.30)^a	2. 24 (1.06–4.70)	2.17 (1.06-4.45)
Innovative climate	0.71 (0.40-1.30)	Not included	Not included
Predictability	0.67 (0.42-1.10)	0.60 (0.32-1.13)	Not included
Incompatible roles at work (role conflicts)	0.95 (0.56-1.60)	0.97 (0.46-2.06)	Not included
Sick leave (0 days reference value)
1–24 days	1.25 (0.80-1.97)	Not included	1.27 (0.66-2.43)
25–365 days	1.50 (0.48 -4.72)	Not included	1.80 (0.51-6.36)
Presenteeism (0 days reference value)
1–10 days	1.44 (0.82–2.57)	1.41 (0.62–3.20)	1.77 (0.81–3.85)
10+ days	3.93 (1.88 – 8.25)^a	3.96 (1.44–10.9)	3.83 (1.26-11.65)

^a Bold number indicates a significant level of 5% or less.

^b In the analysis only the cases that have data on all of the indexes are included.

Discussion

The overall aim of the study was to investigate bullying in an academic setting at a medical faculty among both academics and non-academic staff. The analyses were performed for the total group as well as separately for academics and non-academic staff.

The main contribution of this study is to shed further light into bullying in an academic setting. We were able to study the relationship between work environment and bullying longitudinally and use several psychosocial work-environment factors simultaneously. We were also able to investigate whether the factors predicting future bullying were the same for different professional groups.

In this study, the longitudinal analyses of the relationship between working environment and bullying were performed in two steps. In the first step, a logistic bivariate regression analysis was performed to examine the relationship between the various factors separately and future bullying for the whole group and separately for the two professional groups. Poor social and innovative climate (low values); unfair leadership; lack of empowering leadership; incompatible work roles and lack of support from superiors and colleagues were the factors most strongly related to future bullying in both groups, as well as long-term sick leave (≥25 days) and high presenteeism (≥5 days).

This confirms results from prior studies showing that type of leadership has a strong relationship to occurrence of bullying (Feijó 2019). Poor and sometimes destructive leadership has proved to be associated not only with psychological distress but also with a direct risk of cardiovascular disease (Nyberg et al. 2009). A central role of leadership is to give social support and encouragement in order to reduce stress (Sosik and Godshalk 2000; van der Doef, Maes, and Diekstra 2000). When stress in the workplace and interpersonal tensions are not dealt with, it can eventually lead to workplace bullying (Skogstad, Matthiesen, and Einarsen 2007).

In the present study, the leadership behavior that demonstrated the strongest association with future bullying was unfair leadership. Unfair leadership and unfairness have also been found to be strongly related to bullying in previous studies. Fair treatment and justice have proved to be important to commitment, motivation and employee health and are related to the concept of trust (Elovainio, Kivimaki, and Vahtera 2002; Stoetzer et al. 2014).

Furthermore, our study also shows that lack of social support, especially from colleagues, resulted in increased bullying. For the non-academic staff, lack of support from managers also increased bullying. A previous study found that bullying targets received significantly less support than others from colleagues and superiors (Zapf, Dormann, and Frese 1996). Earlier studies have also reported that low support is associated with a higher risk of mental illness such as depression (Stoetzer et al. 2009; Zapf, Dormann, and Frese 1996). In accordance with the previous discussion, it is the lack of support related to increased stress or the inability to cope with stress which can result in behaviors that can create bullying behaviors.

Another factor that were related to be a target of future bullying were lack of control but in this study it was only applied to the academics. Our study included two different dimensions of control, namely low control over work pace and control over decision-making. We found that, among academics, lack of control over decision-making about their own work was a significant risk factor for becoming a target of bullying. Other studies have also found lack of control to be a major risk factor. The experience of not having control over the situation can be seen as a potential stressor. Using skills/professional knowledge is also a form of control. Employees can deal with a lack of control if there is good support from colleagues or managers and if they know they will get help in dealing with unexpected problems (Karasek and Theorell 1990).

In this study, sickness absence and sickness presenteeism were seen as risk factors for bullying rather than consequences of bullying. Earlier research has revealed bullying to be a risk factor for both sickness absence (Nielsen, Indregard, and Overland 2016) and sickness presenteeism (Conway et al. 2016). However, our results show that high sickness absence and high sickness presenteeism can also be strong risk factors for future bullying. This indicates that both sickness absenteeism and presenteeism can be seen as both a consequence of and a risk factor for bullying. Nevertheless, experiencing illness, the choice between absenteeism or presenteeism would seem to be a particularly personal decision. However, working while ill may not only the performance be affected but also the job attitude being more negative as well as of withdrawal from work. All this may affect the relationship with colleagues in a negative way to the extent that the person experience bullying or harassment (Johns 2010). More research into the association between presenteeism and bullying is needed to be able to further understand this association.

In an academic setting there are different types of work tasks and roles academics and non-academics. The groups may be different not only concerning their work tasks but in Sweden at least, insecure, short-term work contracts which create job insecurity are more common among the academics than non-academics. Previous studies have reported that job insecurity can be a risk factor for bullying (Zapf and Einarson 2005). Therefore, the group academics and non-academics were analyzed separately to investigate whether the factors predicting bullying were the same or not for the two groups. The separate analysis of academics and non-academic staff revealed some differences. The factors that differed between the groups were high role conflict, low predictability and low control decision for the researchers and lack of support from managers, presenteeism of less than 10 days and long-term sick leave for the non-academic staff. Some of the difference could be explained by the two groups having different employment contracts and by the academics also having a variety of academic roles. It is likely that all of this can affect the outcomes.

In the final model (step 2), the results show that additionally to presenteeism being a strong predictor of bullying, poor social climate was the factor that demonstrated the strongest association in both groups. Previous studies have also indicated that social climate is one of the most important factors in predicting bullying (Einarsen, Raknes, and Matthiesen 1994; Vartia 1996). The importance of the organizational social climate has also been identified in more recent studies, which found the number of interpersonal conflicts to be particularly important. The more conflicts there are, the greater the likelihood that they will lead to workplace bullying (Baillien et al. 2009). Bullying may thus be the ultimate consequence of a generally poor social climate as Vartia (1996) points out it is easier to bully someone you do not care for or feel connected to (Vartia 1996).

Methodological considerations

Bullying and the preconditions for bullying have been studied for decades (Samnani and Singh 2012). A common problem in research into bullying is that most of the studies are cross-sectional, i.e. they ask questions about bullying when it already exists as a problem. This may result in the over-reporting of negative work factors, which in turn will incorrectly show an increased risk. The strength of the present study is that we have longitudinal prospective data. Only those who were not exposed to bullying at enrollment were used in the study. Since the measures were collected at intervals of about two years, it allows us to detect indicators of bullying early in the process, at a stage at which interventions to prevent victimization could be implemented.

One should always be cautious when attempting to draw general conclusions. This study can primarily tell us something about employees in one type of work setting.

Conclusions

Bullying in the workplace is often highlighted in the public debate because it is a serious health and safety problem. In our study, we found that factors such as lack of support, low control, inconsistent role demands, poor leadership, poor organizational climate, high sickness absence and high sickness presence and can be important predictors of future bullying. We found that factors in an academic setting differ somewhat for academics and non-academic staff. This might indicate that type of job should be take into consideration in attempts to successfully prevent bullying at an early stage.

As this study and others show, bullying is a serious workplace issue in academic settings (Shelley et al. 2017) and there is an urgent need to manage the problem at an organizational level (Hauge, Skogstad, and Einarsen 2010). It is an ethical and moral obligation of managers and supervisors to protect employees from an unfriendly environment (Pheko 2018). The findings of the present study can hopefully help to prevent bullying at an early stage among a variety of professional roles in academic settings.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.