Introduction

School bullying is a common problem and has been of longstanding interest to parents, teachers, school administrators, and students [1]. Estimates of the prevalence of bullying vary. The largest review of school bullying appears to be one by Koyanagi and colleagues [2]. In a sample of 134,229 students from 48 countries aged 12–15, they found the overall prevalence of bullying victimization and related suicide attempts and were 30.4% and 10.7% respectively. A review by Espelage and De La Rue found that the worldwide prevalence for bullying ranged from 10% to 27% and they estimated that 30% of American youth suffer from being bullied. A review by Dake and colleagues found that the rate of bullying was highest in elementary schools, lower in middle schools and lowest in high schools [3, 4]. Gan et al. found that 55% of students reported having been bullied [5]. A study in middle schoolers found that 82.7% of respondents felt that bullying was a problem and that 49.1% had been the victim of a bully [6]. A study of Indian school children suggested 31.4% of children experienced bullying [7].

Although the general public probably thinks it understands when bullying occurs, one definition in the literature defines bullying as, “A student is being bullied or victimized when he or she is exposed repeatedly and over time to negative actions on the part of one or more students” [3]. Others would add that there is a power differential between the bully and the bullied. Some authors, however, have noted that a power differential does not always exist and the definition that students have for bullying can vary substantially from the definition used by researchers in the field [3]. Pergolizzi and colleagues suggest that bullying may be separated into five domains: physical, verbal, social exclusion, spreading rumors and cyberbullying [6]. Since cyberbullying is often anonymous, victims do not necessarily know the identity of their bully. This suggests that the proposed requirement for a power differential may need to be revised.

Those involved in bullying can be separated into three subtypes. There are bullies, there are those who have been bullied (victims), and there are bullies who themselves have suffered from bullying (bully-victims) [3, 8, 9]. Such bully-victims have been found to have a higher risk of various somatic and psychological complaints [8].

Although historically, bullying was often treated as one of the unpleasant realities of growing up, recently, some authors have argued that bullying represents a global public health crisis because bullying can have serious consequences [10]. A recent WHO report substantiates this concern [11]. Self-reported incidence of depression or anxiety has been found to be significantly correlated with having been bullied and symptoms worsen the more times a student suffers from being bullied [12]. A review by Rigby found that the consequences of being bullied result in a sense of low psychological well-being, poor social adjustment, psychological distress, and physical ill health [13].

A study in Arkansas found that bullying increased the risk of suicidal ideation and planning while a large meta-analysis reported that both traditional bullying and cyber-bullying were associated with an increased risk of deliberate self-harm [14, 15]. A study by Klomek et al. found that victims are at a higher risk for depression and suicidality and that bully-victims are at the highest risk [16].

According to the Centers for Disease Control and Prevention (CDC):

We know that bullying behavior and suicide-related behavior are closely related. This means youth who report any involvement with bullying behavior are more likely to report high levels of suicide-related behavior than youth who do not report any involvement with bullying behavior. … It is correct to say that involvement in bullying, along with other risk factors, increases the chance that a young person will engage in suicide-related behaviors [17].

Other studies suggest that bullying victims suffer a wide variety of somatic complaints including headache, sleep disorders, bed wetting, fatigue and similar ailments [4, 7]. In addition, bullying victims have been found to use medication for pain and exhibit psychological problems more frequently than those who have not been bullied [18]. There is some evidence to suggest that children with attention-deficit/hyperactivity disorder (ADHD) may suffer from being bullied due to an inability to regulate their emotions, and that episodes of being bullied increase inversely with the ability to regulate emotions [19].

A potential resource to evaluate school yard bullying is the periodic statewide youth survey carried out by various state government departments in most US states [20–23]. In some states, the report may focus on crime and safety issues in youth [22] while other reports focus more on issues related to school, family, substance use and abuse, and other broad, social issues [20, 21, 23]. Often, the youth survey includes questions related to bullying, and if so, the data can provide an estimate for how large the problem is in a given state [20, 21].

Unfortunately, many youth surveys typically only report a tabulation of the results–e.g. in 2018, 31% of 6^th graders, 27% of 8^th graders, 19% of 10^th graders and 17% of 12^th graders reported being bullied [21]. Although this may give an estimate of the size of the problem, it does not provide any information on the consequences of bullying, the correlational factors, or the possible causes of bullying. In other words, although the results are tabulated, there may not be any actual analysis of the data. This can make it difficult for teachers and policy makers to create an effective means to mitigate bullying and its deleterious associations.

Although no one, single cause for bullying has been identified, the literature suggests a number of family, school, psychosocial, and other correlates [3, 4, 24, 25]. For example, if a child is brought up in a home with significantly disturbed family dynamics, and where problems are often solved with violence, such a child may be predisposed to see bullying as a means of protecting themselves and their interests [24].

Since the social, cultural, educational, and economic environment varies across all 50 US states, and since policies are often made at the state level, it then seems reasonable to undertake a more detailed analysis of bullying by means of a statewide youth survey. In this paper, we undertake such an analysis and use data from the 2018 Iowa Youth Survey to evaluate the relationship between bullying and feelings of sadness or hopelessness and suicide attempts [20].

Methods

Data for this study were obtained from the 2018 Iowa Youth Survey (IYS) [20]. The IYS is a broad questionnaire offered every two or three years since 1999 to private and public-school students in the 6th, 8th, and 11th grade across the state of Iowa. It is a joint effort between the Iowa Department of Education (IDE) and the Iowa Department of Public Health (IDPH) as well as several other state bodies.

The 2018 survey has 212 questions over 7 sections and collects data on demographics, student experiences, alcohol use, tobacco use, marijuana use, gambling, mental health, student beliefs and attitudes, peers, the school environment, family setting, and the community.

No personal identifiers are collected. School district and county information are redacted by IDPH making personal identification of respondents impossible. In the 2018 study, a total of 74,830 records were received via a survey completed by students using Survey Monkey® online. The IYS assumes that each record represents one student. After data cleaning, 4,379 records were removed leaving 70,451 validated responses available for analysis. The next IYS took place in 2021 with results available too late for inclusion in this paper. Data cleaning and tabulation of the data is performed by the Iowa Consortium for Substance Abuse Research and Evaluation at the University of Iowa. The data remains the property of the Department of Public Health and considered confidential.

The IYS has 7 Yes/No questions that inquire into mental health (Table 1) and 10 questions that inquire into the experience of being bullied. (Table 2) We divided the 10 types of bullying into four categories: cyber, identity-based, physical, and social. (Table 2). Cyber bullying consists of bullying online or through social media. Bullying about one’s race, religion, or sexual identity or gender orientation was considered to be identity bullying. Physical bullying involved physical violence, while social bullying was considered to be name calling, social exclusion or spreading falsehoods that did not take place online. This taxonomy differs from the 5 categories of bullying proposed by Pergolizzi but is more consistent with the questions asked in the IYS [6].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Mental health questions on IYS (n = 70,451).

https://doi.org/10.1371/journal.pone.0281106.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Types of bullying (n = 70,451).

https://doi.org/10.1371/journal.pone.0281106.t002

Of the 7 mental health questions, we chose 2 as dependent variables:

• During the past 12 months, did you ever feel so sad or hopeless almost every day for 2 weeks or more in a row that you stopped doing some usual activities? (Question B62)
• During the past 12 months, have you tried to kill yourself? (Question B65)

We chose these as the dependent variables for concision but also because sadness and hopelessness were thought to be the most common symptoms of mental illness in this population and are frequently related to depression [3, 8, 9]. Students who reported suicide attempts were assumed to have also thought about suicide and/or developed a suicide plan. All 10 bullying questions were included as independent variables.

Two questions (B60 and B61) inquired if a student was prescribed or was taking medicine to help them feel less angry, anxious, restless, nervous, or sad. Consequently, we controlled for psychotropic medication use. The IYS did not ask questions about specific medications prescribed or about the specific medical conditions.

Previous studies found that some bullies are also victims of bullying and refer to them as bully-victims, terminology we have adopted [3, 8, 9]. Thus, we also controlled for students who had engaged in bullying behavior and were themselves bullied.

As suggested by previous research, the consequences of being bullied may be more severe according to the number of experiences of being bullied a student has undergone [12]. The question stem for each bullying question asks how many times a respondent has experienced bullying and offers respondents a choice of six responses: 0 times, 1 time, 2 times, 3–5 times, 6–10 times, ≥ 11 times. In an attempt to evaluate if greater experience of bullying was correlated in our data, while not providing an overwhelming amount of data, we limited reporting each type of bullying to 0 times (referent), 1 time, 3–5 times and ≥ 11 times.

All variables were included in a multivariable logistic regression analysis to assess significant relationships between independent variables and the presence of mental health concerns. Descriptive statistics examined include 95% confidence intervals calculated for all demographic groups and bullying instances as well as an estimated odds ratio indicating the strength of association between independent variables and mental health concerns. We treated the entire data set as a single cohort and did not attempt to undertake separate analyses for girls or boys, or grade level. Adjusted models included the following co-variables: being prescribed a psychotropic medication; taking a psychotropic medication; behaving as a bully. Correlations were considered significant if the 95% CI did not overlap 1.

Results

A total of 70,451 validated responses were received to the most recent (2018) Iowa Youth Survey. The demographics of respondents are shown in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Demographics (n = 70,451).

https://doi.org/10.1371/journal.pone.0281106.t003

Data analysis

Unadjusted Odds Ratios (OR) for feelings of sadness or hopelessness and for suicide attempts after experiencing all 10 types of bullying are shown in Table 4. ORs and 95% Confidence Intervals are shown for a single experience, 3–5 experiences, and 11 or more experiences of being bullied. Adjusted ORs (aORs) are shown in Tables 5 (adjusted for psychotropic medication use) and 6 (adjusted for being a bully-victim).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Unadjusted odds ratios of undergoing type of bullying and feelings of sadness or hopelessness and suicide attempt (Referent is 0 in each case).

https://doi.org/10.1371/journal.pone.0281106.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Adjusted odds ratios of undergoing type of bullying and feelings of sadness or hopelessness and suicide attempt adjusted for psychotropic use.

(Referent is 0 in each case).

https://doi.org/10.1371/journal.pone.0281106.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Adjusted odds ratios of feelings of sadness or hopelessness and suicide attempt adjusted for bully-victims (Referent is 0 in each case).

https://doi.org/10.1371/journal.pone.0281106.t006

Discussion

Our hope is that results such as ours may be helpful in the creation revision, and implementation of school anti-bullying policies and programs. The oldest and most widely cited is the program developed in Norway by Olweus [26–28]. The core components of the program included activities at the school level (school conference day, improved schoolyard supervision), the classroom level (class meetings and rules against bullying), and at the individual level (talking with bullies and victims, talking with parents). The program was found to have a marked effect to decrease bullying as well as other forms of antisocial behavior. The Olewus program has been adopted widely, including in the United States [29, 30]. and has since been updated to include information about cyberbullying and sexual or gender-based bullying. [31] Our results would support making these updates.

Other anti-bullying programs and attendant evaluations were reported by Evans and colleagues and Albayrak and co-workers [32, 33]. The Evans paper was a systematic review of 32 studies that examined 24 bullying interventions [32]. Overall, the results of these programs were mixed. Reviews of anti-bullying programs outside the United States, which involved fairly homogeneous student populations compared to the United States, reported more positive results than those focused on programs in the United States. In a study performed in Turkey, Albayrak and colleagues found that an anti-bullying program based on two theoretical models was effective in preventing bullying [33].

In a recent meta-analysis, Gaffney and colleagues reviewed 100 studies of anti-bullying programs [34]. In general, programs reduced bullying by 15–20%, but the authors noted that study populations were heterogeneous, as were study designs leading to differences in effect sizes.

Descriptions of anti-bullying programs in the literature generally do not provide much guidance as to the content that should be included when developing an such programs. Our literature review suggests areas, some of which are not well-discussed in the literature, where our results may have a useful influence as discussed below.

Although some literature suggests that the negative correlations from bullying increase positively with increased experiences of being bullied, this does not align with our results [13, 16]. Program design may thus not necessarily have to focus on the frequency of bullying.

Based on our results and our literature review, the highest priorities for designing anti-bullying interventions should be sexual and gender identity bullying as well as cyber bullying, which supports the work by Berlan [35], Norris [36] Cantone et al. discuss other possible interventions for cyber bulling [37]. Goodenow et al. suggest that interventions for gender identity may also have a positive effect [38].

In most cases, social bullying was positively correlated with both feeling sad or hopeless as well as suicide attempts which suggestions interventions should include specific information on name calling, social exclusion, and rumormongering.

While our results did not show much correlation between racial and religious bullying and mental health, it should be pointed out that Iowa is largely homogenous with respect to both race (White) and religion (Christian), and by no means are we suggesting that anti-bullying programs should not combat racial and religious bullying, especially in states or districts with a more heterogeneous student population than Iowa. We would also note that the data in this study were collected before the murder of George Floyd and the rise of the Black Lives Matter movement. Subsequent studies may find that racial bullying has stronger correlations than we did with the need for specific programming to be developed.

The lowest priority for program development appears to be in the area of physical bullying. Few positive correlations between physical bullying and mental health were seen. That said, we do not mean to imply that physical bullying should be ignored in program design.

Finally, the aORs for medication use and bully-victim status tended to be lower than unadjusted ORs. That said, it does not appear that anti-bullying programs narrowly targeted to students in these groups need to be developed. Unlike Brunstein and colleagues [16], we did not find that bully-victims were more prone to suicide than other student groups. As for unadjusted ORs, the aORS for bullying on sexual or gender identity had the strongest correlations with sadness or suicidality.

Limitations

Given the number of independent variables we analyzed, we did not disaggregate our results by grade, sex, gender identity, race etc. It is unclear if our results apply equally across all cohorts of students in the IYS. Methodologically, this is a cross sectional study. Such methods can evaluate the correlation between variables but cannot be used to imply causation. We did not study the associations of poly-victimization in students who may have suffered (for example) sexual violence in addition to the types of bullying evaluated in the IYS. Our data does not allow us to determine if psychotropic medicines were started before or after a student experienced bullying. Nor do we know the type of medications respondents were prescribed. It should be noted, too, that the 2018 IYS did not specifically ask about gender identity.

As with any survey, social desirability bias may have limited respondents’ willingness to provide truthful responses. Similarly, recall bias may have affected respondents’ ability to accurately state their feelings or number of episodes of bullying in the past. These limitations notwithstanding, it is quite possible that with over 70,000 respondents, regression to the mean could reduce these potential biases.

Conclusions

Perhaps the greatest take-home message from this study is that not all forms of bullying are equal in the harm they do. As educators or policy makers propose interventions to reduce bullying, it appears that interventions designed to address sexual identity or gender-based bullying are the most urgent.

Although there are differences between ORs and aORs, the data make it fairly clear that anti-bullying campaigns should target cyberbullying and identity-based bullying first, and more specifically sexual identity or gender-based joking. Despite the trend that aORs tend to return lower values, it is essential to observe that bullying problems in both the medicated and victim-bullies often resemble the unadjusted ORs. This suggests that educators need not create anti-bullying interventions directed exclusively towards students on medication or bully-victims. However, while degrees of suffering may vary somewhat between groups, there are also important similarities, so any victims of bullying should of course be included in interventions.

The role of medication in lessening the impact of bullying on victims is not clear. While our research suggests that medications may mitigate the effects of some types of bullying, we cannot draw any such conclusion from it. Finally, our results confirm that many bully-victims are also among those whose suffering is part of the dynamics of bullying.