Mobbing as a Predictor of Health Risk Behaviours and Seeking Specialist Support in Non-university Teachers

Introduction

Mobbing (bullying or harassment) at work (Leymann, 1996) has been defined as unethical and hostile behaviours systematically carried out by one or more individuals towards mainly one individual who is pushed into a helpless and defenceless position and held there by means of continued mobbing activities. The expressions bullying and harassment can be used synonymously with the concept of mobbing to refer to the same phenomenon (Einarsen et al., 2011). Sperry (2009) states that the terms bullying and mobbing are used more or less synonymously, although preferences vary geographically.

Mobbing in the workplace is considered an interpersonal workplace aggression performed by individuals to harm others with whom they work. This form of aggression is not necessarily related to discriminatory behaviours based on gender, race, or social group. The worker is exposed to negative actions from superiors or co-workers and finds it difficult to defend him/herself against these actions due to a real or perceived power imbalance between the victim and the perpetrator (Nielsen et al., 2021). Mobbing is a severe psychosocial stressor that involves long-lasting exposure to repeated negative behaviours at work, such as offending or ostracizing individuals, humiliating their honour and dignity by knowingly disseminating false information, making insulting remarks, performing intimidating behaviours, or taking actions that negatively affect an individual’s job (e.g., having his/her work sabotaged, having essential and relevant information withheld that affects his/her performance, etc.) (Conway et al., 2022). These actions must occur on a very frequent basis, at least once a week, and over a long period of time, lasting at least six months (Leymann, 1996).

Mobbing actions are characterized by an intentional use of power against another person that can result in harm to his/her physical, mental, or social development. Due to its high frequency and long duration, this maltreatment results in considerable psychological, psychosomatic, and social misery. It affects an individual’s mental health and it has negative consequences for the organization. Exposure to mobbing in the workplace is a more devastating problem for employees than all the other types of work stress combined, leading to psychosomatic and psychiatric problems (Ula et al., 2018), hopelessness, depression (Figueiredo-Ferraz et al., 2015; Miller et al., 2020), cardiovascular diseases (Romero et al., 2020), sleep problems (Nielsen et al., 2020), insomnia (Nielsen et al., 2021), and suicide risk (Alfano et al., 2021; Conway et al., 2022; Miller et al., 2020) in workers who experience it. Some studies have concluded that exposure to work-related violence and mobbing is associated with the prescription and use of anti-depressive (Rudkjoebing et al., 2021), anxiolytic (Messiaen et al., 2021), and psychotropic medication (Lallukka et al., 2012), as well as psychotherapy follow-ups (Messiaen et al., 2021). Moreover, victims may try to cope with mobbing by increasing the frequency of unhealthy behaviors such as alcohol intake (Aykut et al., 2016; Campo & Klijn, 2018; Giorgi, 2010; Giousmpasoglou et al., 2018; González & Delgado, 2008; Ula et al., 2018) and smoking (Hansen et al., 2021; Jacobsen et al., 2018; Lloyd, 2020).

The average prevalence of mobbing worldwide is estimated at 14.6% (Alfano et al., 2021), ranging from 11.3% to 18.1% depending on the measurement method used (Conway et al., 2022). Educational settings displayed prevalence levels that exceeded average mobbing rates, with values between 14% and 22% (Hodgins & McNamara, 2019). In addition, some studies have shown that teachers are frequently victims of mobbing (Scheeler et al., 2022; Zapf et al., 2020).

Taking into consideration the gender distribution of the victims, some studies have concluded that the prevalence of mobbing is higher in women than in men (Salin, 2021b; Zachariadou et al., 2018), with one-third of the victims being men and two-thirds being women in the majority of the samples (Zapf et al., 2020). In addition, the school as an organization can be a risk factor. Some studies have concluded that teachers working in high schools suffer more hostile behaviours than their colleagues teaching in elementary schools (J.-K. Chen & Astor, 2009).

Even though researchers on mobbing use both quantitative and qualitative research methods, there seems to be a preference for quantitative research. The most widely used questionnaires to assess mobbing are the Leymann Inventory of Psychological Terror (LIPT-45; Leymann, 1990) and the Negative Acts Questionnaire-Revised (NAQ-R; Einarsen et al., 2009). Validity studies carried out in different countries make it possible to affirm that both questionnaires are psychometrically robust enough for use and they have concluded that both the LIPT (see, Silva et al., 2021) and the NAQ-R (see, Erwandi et al., 2021) are suitable tools for assessing mobbing in different labour contexts.

However, threats, aggression, and mobbing activities in the workplace are influenced by cultural and socioeconomic factors that produce differences in these actions, or in the way similar behaviours are developed, because cultural values have been shown to be a strong determinant of many organizational behaviours (Leng & Yazdanifard, 2014; Salin, 2021a; Salin et al., 2019). Due to these differences, it is possible that the most widely used questionnaires to assess mobbing, i.e., the LIPT-45 and NAQ-R, do not include some of the relevant and most frequent mobbing actions carried out by perpetrators in Mediterranean countries (e.g., Spain, France, Greece, Italy, or Portugal). Thus, in these countries, it would be more appropriate to assess mobbing actions with questionnaires that were developed with cultural contexts different from the Anglo-Saxon labour culture in mind to predict health risk behaviours and disorders of health related to mobbing. The Mobbing-UNIPSICO scale (Gil-Monte et al., 2006) aims to fill this gap by considering these cultural differences.

Overview of the Mobbing-UNIPSICO Scale

The Mobbing-UNIPSICO scale is based on the concept of mobbing developed by Leymann (1996). The scale contains 20 items designed to identify five categories of mobbing actions:

• Effects on the victim’s possibilities of communicating adequately (not having the possibility to communicate or meet with the boss, continuously interrupting the victim, indifference and verbal aggressions, yelling at or scolding the victim in the presence of others, being humiliated when asking questions or trying to participate in a conversation, receiving threats of being fired, receiving offensive telephone calls or written messages).
• Effects on the victim’s possibilities of maintaining social contacts (physical isolation, avoidance, rejection, social exclusion).
• Effects on the victim’s possibilities of maintaining his or her personal reputation (small errors unfairly exaggerated, attacks on opinions, rumours, ridicule, being humiliated in front of others, tasks sabotaged or intentionally distorted, victim’s job performance belittled regardless of what she/he does).
• Effects on the victim’s occupational situation (no tasks, lack of information necessary for work, uninteresting tasks, humiliating tasks, or tasks inferior to skills).
• Effects on the victim’s health (physical threats and sexual harassment).

The 20 items are evaluated on a frequency scale. Previous studies have concluded that this scale has good psychometric properties, and all the items loaded in one factor labelled Mobbing in a sample of employees working with disabled people (Gil-Monte et al., 2006). In addition, the scale has shown good internal consistency values, measured with Cronbach’s alpha (Figueiredo-Ferraz et al., 2015), and test-retest reliability (r = .82) (Figueiredo-Ferraz et al., 2015).

Moreover, the scale includes an additional item where the participants indicate the duration of the actions (“How long have you been experiencing these actions?”). This item is evaluated on a scale with seven options: (0) “I have been experiencing these actions for less than 6 months” to (6) “more than 10 years” (see Appendix). To identify a suspicious case of mobbing, two criteria used in the literature are combined (Leymann, 1996): at least one out of the list of 20 mobbing items evaluated on the frequency scale has to receive a rating equal to or higher than 3 (a few times a week); and the duration of the item has to get a rating equal to or higher than 1 (between 6 months and 1 year). Any positive cases should be confirmed in a psychological interview.

The Present Study

The purpose of this study was to investigate whether experiencing mobbing can predict different health risk behaviours, such as smoking, alcohol intake, increased use of medication as a consequence of psychological disorders at work, and the need to seek specialist support in non-university teachers. As a preliminary step, the one-dimensionality and invariance of the scale was analysed according to sex and the educational level at which they work.

Method

Participants

The sample included 9,350 non-university teachers from different levels of public education in the province of Valencia (Spain) (men, 2,704; women, 6,608; missing, 38). The data were collected in a non-random way from a population of about 35,000 subjects, with some small size variations between academic years depending on the personnel needs of the schools. The margin of error for this sample size is less than 1.25% (confidence level = 99%).

Mean age was 45.69 years (SD = 9.13, age range = 20-70 years). Regarding the type of contract, 98.70% of the sample (n = 9,226) were tenured teachers working as civil servants, and 0.60% (n = 52) were temporary teachers. The mean years of seniority in the profession of all the participants was 17.69 years (SD = 9.79, seniority range = 0.80-50 years). The sample distribution was: (a) 15.30% kindergarten teachers, (b) 35.50% primary school teachers, (c) 34.80% secondary school teachers, and (d) 14.40% trade-school teachers – i.e., children, 50.80% (a and b) and teenagers, 49.20% (c and d).

Instrument

Mobbing was evaluated by the Mobbing-UNIPSICO scale (Figueiredo-Ferraz et al., 2015). This scale contains 20 items adapted from the results of some interviews carried out with mobbing victims. Items deal with: (1) effects on the victim’s opportunities to communicate adequately (5 items; e.g., “Your chances to communicate, talk, or meet with your boss are restricted”), (2) effects on the victim’s opportunities to maintain social contacts (2 items; e.g., “Being moved to isolate you from your colleagues”), (3) effects on the victim’s possibilities of maintaining his or her personal reputation (7 items; e.g., “Being ridiculed or humiliated in front of others”), (4) effects on the victim’s occupational situation (4 items; e.g., “Being left without any work to do, even if you initiate it”), and (5) effects on the victim’s health (2 items; e.g., “Being the victim of sexual harassment”) (see Appendix). The items are evaluated on a scale with five options: (0) never to (4) every day. All items load on one factor labelled Mobbing.

Alcohol and tobacco use were evaluated using one item each that alluded to the frequency with which these products had been used in recent weeks as a consequence of troubles at work: “As a consequence of troubles at work, has your daily alcohol use increased?”; “As a consequence of troubles at work, has your daily tobacco use increased?”. Medication use was evaluated using one item that alluded to the frequency with which medication had been used in recent weeks as a consequence of health disorders at work: “As a consequence of psychological or psychosomatic health disorders at work, have you had to take any medication to manage them?”. In the same way, professional support was evaluated with the item "Have you needed support from a specialist to overcome a personal crisis related to your work?”. Participants answered the items on all the subscales on a 5-point frequency scale ranging from never (0) to very frequently: every day (4).

Procedure

This study respected the fundamental principles of the Declaration of Helsinki (World Medical Association, 2013), taking into special consideration the confidentiality and non-discrimination of participants. All the public non-university schools in the province of Valencia were asked to participate. Teachers were informed of the purpose of the study. This study was carried out following the instructions of the Instituto Valenciano de Seguridad y Salud en el Trabajo (INVASSAT) (Government of the Valencian Region). Before starting the assessment, the ethical department of this institution was consulted and concluded that, because participation was voluntary and the teachers just had to answer an anonymous questionnaire, the research did not need to be checked by a bioethics committee. Data were collected by paper and pencil at the workplace between October 2015 and May 2020 by employees working in the INVASSAT. The INVASSAT employees went to all the schools and informed the director, union representative, and teachers at each school about the procedure. Then, each teacher filled out the questionnaire individually in the presence of an INVASSAT employee in order to address any doubts, and the completed questionnaires were given to the INVASSAT employee.

Data Analysis

As a preliminary step, a confirmatory factor analysis (CFA) was performed to test the one-dimensionality of the Mobbing-UNIPSICO scale, in line with previous literature (Gil-Monte et al., 2006) (Study 1). For this purpose, the WLSMV estimator (weighted least squares adjusted for mean and variance) was used due to the ordinal nature of the response scale (scale with five options: (0) never to (4) every day) and the non-normal nature of the data (Lloret et al., 2017). The fit indices that served as reference were the comparative fit index (CFI), the standardized root mean square residual (SRMR), and the root mean square error of approximation (RMSEA). Values above .90 for CFI and below .08 for RMSEA and SRMR can be considered indicators of good fit (Brown, 2015; Hu & Bentler, 1999; Raykov & Marcoulides, 2016). Reliability was assessed with the composite reliability (CR) index (Raykov, 2011) and Cronbach’ alpha. Next, the average variance extracted (AVE) was calculated to estimate the proportion of variance explained by each factor. Values equal to or greater than .70 for the CR and alpha and equal to or greater than .50 for the AVE are considered good (Raykov, 1997). The corrected item-total polychoric correlations were calculated as indicators of the corrected homogeneity indices for items with ordinal response scales (Raykov & Marcoulides, 2011).

Likewise, the measurement invariance across gender and educational stage was evaluated by calculating three nested invariance models: configural, metric, and scalar. To study invariance by educational stage, two groups were formed: teachers working with children up to 12 years of age (infant and primary), representing 50.8% of the sample, and teachers working with adolescents (secondary and trade school), representing 49.2% of the total sample. To assess the degree of invariance among the models, we followed the indications of Meade et al. (2008), who recommend the use of an approximative fit index such as CFI or RMSEA because the c² is greatly affected by sample sizes as large as the one in this study (Shi et al., 2018). Thus, evidence of measurement invariance was supported by the following changes in the between-model fit indices: reject if CFI change is lower than -.010, RMSEA change is greater than .010, and SRMR change is greater than .015 (Chen, 2007; Cheung & Rensvold, 2002).

Finally, to study the predictive model a structural equation model (SEM) was specified for the Mobbing-UNIPSICO scale as a predictor of four health behaviours: tobacco use, alcohol intake, medication use, and professional support (Study 2). CFA analyses, corrected item-total polyserial correlations, measurement invariance, and the predictive model, were performed with Mplus 8.8 (Muthén & Muthén, 2017), whereas IBM SPSS 28 was used for the description of sociodemographic variables and item statistics of the Mobbing-UNIPSICO scale.

Results

Study 1. Dimensionality and Invariance of the Mobbing-UNIPSICO Scale

Table 1 shows descriptive statistics for the items and corrected item-scale correlations. All the items presented very good homogeneity indices, except for item 20, whose value is much lower than that of the other items (.309). This item had the highest kurtosis (Ku = 220.74) and the second most extreme skewness value (Sk = 13.71). In addition, Item 17 (Ku = 103.09, Sk = 9.56) and Item 14 (Ku = 87.22, Sk = 8.61) showed high values for kurtosis and skewness.

All coefficients were statistically significant (p < .001).

All factor loadings were statistically significant (p < .001), ranging between .608 (Item 1) and .908 (Item 11) (Figure 1). Table 1 shows the descriptive data of the Mobbing-UNIPSICO scale items and the item-total corrected polyserial correlations. All of them were statistically significant, ranging between .309 (item 20) and .684 (item 11), and with standard errors ranging between .004 (item 4) and .016 (item 20). The composite reliability index (CR = .975), Cronbach’s alpha (α = .93) and the average variance extracted (AVE = .709) showed very good values. The results confirmed the hypothesized model.

The one-dimensional CFA showed good fit to the data, χ²₍₁₇₀₎ = 5,443.66, p < .001, CFI = .962, SRMR = .052, RMSEA = .058, RMSEA 90% CI [.056, .059] (Table 2). Table 2 shows the results for the measurement invariance models. The fit indices of the estimated models confirm that scalar invariance was fulfilled by both gender and educational stage. Therefore, the comparison of the latent means of the subgroups was carried out. In the case of gender, the latent mean of the male group was set to 0, and the result of the comparison of the means showed that there were no differences between males and females on the mobbing scale scores (b = -0.014, z = -0.535, p = .593). For the educational stage, the reference group was teachers who taught children. The results of the comparisons of the means of the two groups showed that teachers who taught adolescents scored higher on the mobbing scale (b = 0.049, z = 1.957, p = .050).

Study 2. Predictive Model

To evaluate the predictive model, a SEM model was performed using the Mobbing variable as predictor and four dependent variables: alcohol intake, tobacco use, medication use, and professional support (see Figure 2). The model showed a very good fit, χ²₍₂₄₆₎ = 5,318.64, p < .001, CFI = .966, RMSEA = .047, RMSEA 90% CI [.046, .048]. All the coefficients of the model were statistically significant and in the expected direction (see Figure 2), so that the higher the score on the mobbing scale, the greater the search for specialist support (β = .337) and the greater the alcohol intake (β = .109), tobacco use (β = .142), and medication use (β = .300). In addition, the score on the Mobbing scale turned out to be the best predictor of the need for specialist support and the increase in alcohol intake. Moreover, the relationships between the variables related to health were all found to be statistically significant, especially the relationship between medication use and specialist support (Φ = .378). However, two of them were very close to 0 (between alcohol intake and medication use, Φ = .050; and between alcohol intake and specialist support, Φ = .028), but they are probably statistically significant due to the large sample size.

Discussion

The aim of this study was to explore whether experiencing mobbing can predict different health risk behaviours, such as smoking, alcohol intake, increased use of medication as a consequence of psychological disorders at work, and the need to seek specialist support in a sample of Spanish non-university teachers. The results of the previous factor analysis confirm the one-dimensionality of the scale, which also offers good reliability indicators. Likewise, the lower homogeneity, and the high both kurtosis and skewness values for item 20 may be due to the content of the item, which refers to sexual harassment, with more than 98% of the sample indicating that it has never happened to them. According to Çelik and Peker (2010), the least common mobbing behaviours in teachers working in high schools involved experiencing sexual harassment (verbal, visual, physical, etc.). Despite this, the relationship of this item with the total score of the scale is not negligible and it is necessary to evaluate the possibility of the existence of sexual harassment at work in order to evaluate mobbing in all its dimensions.

The characteristics of the teaching profession could explain the high kurtosis and skewness values obtained for Item 17 and Item 14. Item 17 evaluates the effects of being moved to some room to isolate him/her from colleagues on the victim’s possibilities of maintaining social contacts. However, in the workplace, the teaching activities impede these actions because teachers spend most of their time inside the classroom, carrying out activities with students individually and communicating with parents or guardians (McShane, 2022; OECD, 2020). Item 14 evaluates the effects of receiving offensive telephone calls or written messages on the victim’s opportunities to communicate. These actions may not be used frequently because they provide evidence of aggressive behaviours (e.g., the telephone call can be heard by a witness who can identify the perpetrator, and the written messages can be physical evidence in a judicial process). In Spain, aggressions toward teachers with the status of civil servants working in a public school are considered an “attack on authority” (article 550 and article 553 of the Spanish Penal Code), and they could be punished by a prison sentence of up to three years. In this study, 98.70% of the sample were tenured teachers working as civil servants. By contrast, the most frequent mobbing actions were related to the victim’s possibilities of maintaining his or her personal reputation (Item 3 and Item 4), the victim’s possibilities of maintaining social contacts (Items 6), and the victim’s occupational situation, assigning absurd or senseless tasks (Item 1). These types of mobbing actions are common in educational institutions (Tosten et al., 2018).

Likewise, the Mobbing-UNIPSICO Scale has shown invariance by gender and educational stage. The results obtained when comparing the groups indicate that there are no differences between men and women in terms of perceived mobbing, but there are differences depending on the educational stage of the teacher. In this case, the results indicate that teachers who work with adolescents perceive more mobbing at work. This result is similar to the results obtained in previous studies that have concluded that teachers working in high schools experience more hostile behaviours than their colleagues in elementary schools (J.-K. Chen & Astor, 2009). Non-appreciation and criticism of performance and spreading rumours are the most common behaviours reported by teachers working in high schools (Çelik & Peker, 2010).

The predictive validity of the Mobbing-UNIPSICO Scale has been verified with the validity model because the total score on the mobbing questionnaire predicts the increase in both alcohol and tobacco consumption, a greater use of medication, and the need to seek support from a specialist to overcome some personal crisis related to work. On the whole, these results support the predictive validity of the Mobbing-UNIPSICO Scale, given that specialist support (Messiaen et al., 2021; Rudkjoebing et al., 2021), alcohol intake (Aykut et al., 2016; Campo & Klijn, 2018; Giorgi, 2010; Giousmpasoglou et al., 2018; Ula et al., 2018), tobacco use (Hansen et al., 2021; Jacobsen et al., 2018; Lloyd, 2020), and medication use (Lallukka et al., 2012; Messiaen et al., 2021; Rudkjoebing et al., 2021) have been identified as relevant consequences of mobbing activities. Moreover, the model shows that the consumption of alcohol and tobacco are positively correlated, whereas the search for specialist support is more related to the increase in the use of medication.

Finally, taking the size of the sample into consideration, our results can be generalized to other Spanish samples of non-university teachers in other places and support the external validity of the scale.

Limitations and Future Directions

In interpreting our findings, several limitations should be noted. First, some significant results could be influenced by the large sample size (e.g., relationship between mobbing and alcohol intake and between mobbing and tobacco use). Second, it should also be noted that answers to the questionnaire were subjective and, therefore, may be biased. Third, another limitation of the study is that the score on the Mobbing variable was calculated as a mean of all the scores on the scale, and it included all the individuals in the sample, both those who met the mobbing criteria and those who did not. Fourth, the study only focuses on teachers, which restricts the generalizability of the results. Therefore, we suggest that other organizational contexts be studied in future research.

Conclusion and Implications

On the whole, the results of our study indicate that the Mobbing-UNIPSICO Scale possesses adequate psychometric properties for the study of mobbing in Spanish teachers. This study is relevant because it provides evidence supporting the predictive validity of an alternative mobbing measure, taking into consideration some of the relevant and most frequent mobbing actions carried out by the perpetrators in a cultural context different from the Anglo-Saxon labour culture. In Mediterranean countries it would be more appropriate to assess mobbing actions with questionnaires developed by taking into consideration cultural contexts different from the Anglo-Saxon context. The Mobbing-UNIPSICO scale is designed to fill this gap.