Can Social Support Protect the Mental Health of College Students Who Experienced Bullying in High School?

Abstract

Transitioning from high school to college can be challenging and put young adults at risk for problematic mental health, particularly for those with bullying victimization histories. Bullying detrimentally impacts psychological well-being and mental health, both concurrently and in the future. Social support positively influences college students’ emotional, social, and academic performance. However, few studies have examined the extent to which different types of bullying, as well as different sources of social support, may interact to predict mental health symptoms. Further, few studies examine associations between bullying victimization and mental health symptoms across the transition from high school to college. We examined three sources of social support (and their composite) measured during college as moderators between high school bullying victimization (traditional, cyberbullying) and college mental health (anxiety, depressive symptoms) in a sample of young adults from the Southeast United States (n = 329). Thus, we hypothesized that social support may protect college students from the harmful effects of bullying on mental health. Multivariate linear regressions revealed that higher levels of both traditional and cyberbullying during high school predicted worse mental health during college, and higher levels of social support in college predicted better concurrent mental health. However, social support did not moderate links between either type of bullying victimization in high school and mental health in college. Our findings provide evidence for institutions and educators regarding the importance of fostering social support (e.g., developing new friendships, maintaining existing support systems with close family or friends) for young adults as they transition to college.

1. Introduction

College students’ struggles with stressors and mental health are well documented (Harrer et al., 2018; Meeks et al., 2023; Othman et al., 2019). The transition from high school to college involves unique challenges, such as greater independence, new living situations, and separation from friends and families, which can drastically undermine psychological well-being (Conley et al., 2014). Students have identified academic performance, post-graduation plans, sleep, finances, health, and relationships with friends and family as primary contributors to stress, anxiety, and depression (Beiter et al., 2015). There also appears to be a dose–response relationship between stressors and mental health for college students, such that more life challenges are associated with poorer mental and overall health for college students (Porru et al., 2022).

One possible stressor that may place college students at greater risk for poor mental health is bullying victimization. Extensive literature documents that bullying victimization can significantly impact well-being, both concurrently (Galindo-Domínguez & Losada Iglesias, 2023; Kalogerakis et al., 2021) and in the future (Copeland et al., 2013; Gorman et al., 2021; Lin et al., 2020). Adolescent bullying specifically is associated with problems with mental health, peer relationships, and internalizing symptoms (Kalogerakis et al., 2021) such as anxiety and depression (Vacca et al., 2023) and self-concept (Galán-Arroyo et al., 2023). Even more concerning, adolescents who are victimized exhibit higher rates of suicidal ideation (Galindo-Domínguez & Losada Iglesias, 2023). Studies of the long-term impact of bullying on children and adolescents indicate that they tend to experience mental illness as young adults, suggesting ongoing negative repercussions (Copeland et al., 2013; Gorman et al., 2021; Källmén & Hallgren, 2021; Lin et al., 2020; Manrique et al., 2020). Given the numerous stressors and heightened levels of anxiety and depression experienced by many college students, we might expect those who have been previously bullied to be at even greater risk for severe mental health problems compared to their non-bullied peers (Reid et al., 2016).

However, there may be unique factors that also mitigate the negative associations between bullying victimization and mental health symptoms. Social support comprises the degree of connectedness an individual experiences with various sources of support (Drageset, 2021) and plays a critical role in mental health (Bjørlykhaug et al., 2022). Social support from family and friends positively influences college students’ emotional, social, and academic performance (Awang et al., 2014; Conley et al., 2014). Conversely, students who lack social support tend to experience more mental health problems and a diminished quality of life (Alsubaie et al., 2019; Hefner & Eisenberg, 2009). College students often experience drastically changing social contexts in their first year or two of college, which are in turn marked by declines in psychological and social well-being (Awang et al., 2014; Conley et al., 2014).

The prevalence of bullying during high school is concerning, as is the high incidence of mental health problems in college students. Although bullying prevention may be a direct focus of high school interventions, college personnel may instead be focused on providing services and support for college students who have experienced bullying in high school to mitigate its adverse mental health outcomes and reduce symptomatology. Because of the positive relationship between social support and mental health, we used the stress-buffering hypothesis model (Cohen & Wills, 1985) to examine whether various sources of social support during college moderated the association between high school bullying victimization and college mental health symptoms in the current study. More specifically, we examined multiple types of bullying in high school, three sources of social support during college, and mental health symptoms during college.

1.1. Bullying and Its Related Mental Health Outcomes

Bullying is unwanted, repeated negative actions intended to harm an individual or their reputation, often occurring within a power imbalance (Manrique et al., 2020; Olweus, 1993; Volk et al., 2014). Although bullying has been examined across multiple ages and settings, a primary focus of bullying research is on adolescents who experience bullying in school settings; 19% of students ages 12–18 experience bullying at school (National Center for Education Statistics, 2024). Approximately 1 in 5 high school students report being bullied at school, with approximately 1 in 6 being cyberbullied (Centers for Disease Control, 2024). Other reports of cyberbullying are higher, with almost 34% of a large sample of U.S. middle and high school students reporting being cyberbullied during their lifetime and nearly 17% reporting having been bullied within the past month (Hinduja & Patchin, 2018).

Examinations of traditional bullying include both direct and indirect bullying, where direct bullying includes physical (e.g., kicking, punching, hitting) or verbal (e.g., name-calling, teasing, ignoring) behaviors, and indirect bullying is generally more social in nature (e.g., spreading rumors, telling lies, excluding someone from an activity or group; Manrique et al., 2020; Reid et al., 2016). Other research distinguishes types of bullying based on the context in which it occurs, differentiating traditional bullying observed since the 1800s and researched for more than 50 years (Koo, 2007) from cyberbullying, a relatively recently emerging construct (Peebles, 2014). Cyberbullying, sometimes referred to as online bullying (Fahy et al., 2016), is considered a distinct form that involves the use of electronic tools (e.g., computers, cell phones, social media) to execute spiteful and aggressive acts to intentionally harm others (e.g., spreading rumors, causing humiliation, posing a threat; Xu et al., 2024).

There is some overlap between traditional and cyberbullying regarding definitions, harmful intentions, and some behaviors (Rodrigues et al., 2020). Moreover, children who perpetrate traditional bullying also tend to cyberbully, and victims of one type commonly experience victimization in the other (Peebles, 2014). However, some features distinguish cyberbullying from traditional bullying. Importantly, cyberbullying includes a range of behaviors beyond those perpetuated by traditional bullies, including impersonation, posting videos recorded in private places such as bathrooms or bedrooms (Hinduja & Patchin, 2018), posting humiliating comments on social media (Brailovskaia et al., 2018), or outing (revealing someone’s gender identity or sexual orientation publicly without their consent; Austen & Wellington, 1995). Moreover, the broader and more public nature of cyberbullying, its ease of sharing, the permanence of posts, and the potential anonymity and decreased empathy of the cyberbully increase its potential for greater harm (Fahy et al., 2016; Kim et al., 2018). These features contribute to the risk for cyberbullying leading to more stressful and serious consequences for victims than traditional bullying (Hellfeldt et al., 2020).

Indeed, a wealth of research documents the adverse outcomes associated with bullying victimization that persist into adulthood, such as future adjustment and mental health (e.g., Arseneault, 2017). Child victims of bullying tend to be at higher risk for mental health problems, such as anxiety and depression (Manrique et al., 2020) during young and middle adulthood (Wolke & Lereya, 2015), and adverse peer experiences such as bullying are associated with future relationship problems, financial stresses, and educational struggles (Wolke et al., 2013). Compared to their non-bullied peers, child and adolescent victims of bullying experience more psychological maladjustment (Guo et al., 2022) and difficulty with the adjustment to college (Jantzer & Cashel, 2017). Of particular concern is the finding that childhood bullying places individuals at greater risk for victimization during college (Felix et al., 2019), which may further exacerbate these problems.

Although cyberbullying has been less studied than traditional bullying, it has been linked to a similar range of psychological problems. In college students, cyberbullying victimization predicts a higher likelihood of psychological maladjustment, including anxiety and depression (Na et al., 2015; Schenk & Fremouw, 2012; Selkie et al., 2016; Wick et al., 2020). Moreover, compared to victims of traditional bullying, cyberbullying victims report higher levels of anxiety, depression, and social problems (Hamm et al., 2015). Additionally, college students who are cyberbullied are likely to engage in avoidant coping strategies, which in turn heightens their levels of psychological distress (Na et al., 2015; Schenk & Fremouw, 2012; Wick et al., 2020).

1.2. Social Support

Social support comprises the exchange of resources between individuals in a relationship (Zimet et al., 1988), reflecting the degree of connectedness that an individual experiences from their various sources of support (Drageset, 2021). Social support is essential for maintaining positive mental health, preventing adverse mental health outcomes, and promoting recovery from moderate and severe mental health problems across the lifespan (Bjørlykhaug et al., 2022). The positive impact of social support on quality of life (Alsubaie et al., 2019) may subsequently lower levels of mental health problems such as depression (Camara & Padilla, 2017; Dafaalla et al., 2016; Kugbey, 2015), anxiety (Scardera et al., 2020), and suicidal ideation (Galindo-Domínguez & Losada Iglesias, 2023). Conversely, individuals struggling with their mental health tend to have fewer social relationships, less robust social support, and a greater risk for social exclusion than their mentally healthy peers (Bjørlykhaug et al., 2022). Those who withdraw socially through avoidance are at heightened risk for internalizing disorders, relationship problems, and emotional dysregulation (Nelson, 2013), whereas well-adjusted individuals are more likely to engage in prosocial behavior, succeed in school, and be psychologically healthier (Arslan & Coşkun, 2020). College students who lack social support tend to experience more mental health problems and a diminished quality of life (Alsubaie et al., 2019) and have a high risk of experiencing mental health problems, with some estimates indicating they may be six times more likely to experience depressive symptoms compared to their peers with higher-quality social support (Hefner & Eisenberg, 2009).

Social support can come from various sources, such as peers, family, significant others, teachers, or colleagues (R. Evans et al., 2022; Song et al., 2023). Rueger et al. (2016) also found that the most substantial sources of support were family and peers, followed by teachers and close friends. Family, peer, and teacher social support each demonstrated reduced rates of suicidal ideation in adolescents; however, family social support demonstrated the strongest reduction, followed by peers (Galindo-Domínguez & Losada Iglesias, 2023). This suggests that different sources of social support may provide more or less protection against negative developmental outcomes, including psychological distress.

Beyond the well-documented relationship between social support and mental health outcomes, social support can play a role in bullying victimization more specifically. Lee et al. (2022) found that South Korean school-aged students (elementary through high school) who reported high familial support were less likely to be victimized or experience high rates of multiple types of bullying. Those with support from their peers also demonstrated less risk of verbal bullying and lower rates of multiple types of bullying victimization (Lee et al., 2022). Moreover, social support may serve as a protective factor for bullying victims, who are at higher risk for depression and anxiety. Studies focused on social support as a moderator of the relationship between bullying victimization and psychological distress indicate that although social support does not eliminate the consequences of bullying, it may mitigate the impact on the victim’s mental health (Rothon et al., 2011; Zhang et al., 2016). For example, peer support has also been found to moderate the relationship between victimization and depression, with higher support reducing the risk of depressive symptoms for those experiencing bullying victimization (Du et al., 2018).

1.3. Stress-Buffering Hypothesis Model

The stress-buffering hypothesis model purports that social support “buffers (protects) persons from the pathogenic influence of stressful events” (p. 310; Cohen & Wills, 1985), reducing the likelihood or severity of negative health outcomes in response to high-stress events. In contrast, individuals in similarly stressful conditions who lack social support may be more at risk for developing illness or psychopathology (Rueger et al., 2016). Although the model can be conceived as assuming that social support only helps people experiencing high stress, with limited or no benefits for individuals in less stressful situations, a modified model purports a more nuanced moderation effect, such that individuals in high- and low-stress situations both benefit from social support, but those experiencing higher stress will benefit the most (Stroebe & Stroebe, 1997).

Research examining social support as a protective or buffering factor in the relationship between adverse life experiences and health outcomes is mixed; some studies have found social support to be a moderator while others have not, and some have found differences in the types of sources of support (although this literature is more limited). For example, Szkody and McKinney (2019) examined the relationship between adverse life events and physical and psychological health outcomes, which revealed a moderating effect of social support on the association between adversity life events and physical health. However, although a main effect was observed between social support and psychological health, social support failed to moderate the relationship between adverse life events and psychological health, namely depression (Szkody & McKinney, 2019). Rueger et al. (2016) conducted a meta-analysis of studies of social support and depression during childhood and adolescence that included a variety of stressors (e.g., family/medical/death, pregnancy, psychiatric illness). The stress-buffering hypothesis was supported only for moderating effects of social support from teachers, peers, and close friends on depression for youth with medical illnesses.

A similar model has been applied in studies of bullying and mental health directly as well. In an examination of bullying victimization and symptoms of anxiety, depression, and PTSD, Manrique et al. (2020) found that social support mediated some of the relationships. For example, both current and past social support mediated the relationship between physical and social (but not verbal) bullying and two of the mental health outcomes (depression and PTSD), but only current (not past) social support was a mediator between verbal bullying and those two mental health outcomes. There was no evidence of a mediation effect of social support on the relationship between bullying and anxiety.

Finally, Reid et al. (2016) examined the potential buffering effects of social support on the mental health of college students who were bullied during adolescence, including traditional bullying and online aggression. Like previous reports, they observed direct effects such that previous victimization predicted anxiety and depression during the first year of college and social support negatively predicted anxiety and depression (Reid et al., 2016). When testing the stress-buffering effect, overall social support (and family support, in particular) was observed to moderate the relationship between bullying and anxiety during the fall and spring semesters of the first year of college; however, moderating effects of social support were observed for depression only in the fall semester (Reid et al., 2016).

1.4. The Current Study

It is well documented that bullying occurs at high rates during childhood and adolescence and that victims tend to experience poor future mental health. It is also known that even in the absence of bullying, the incidence of anxiety and depression experienced by college students is high. When considered together, it seems that college students with a history of bullying victimization are at an even heightened risk for anxiety and depression. Our first aim was to replicate this finding in the current study. In doing so, we extended the current literature on the relationship between bullying and mental health by examining the effects of traditional bullying and cyberbullying separately. Despite the overlap between the two types of bullying, cyberbullying has been associated with more mental health problems than traditional bullying, especially for college students (Na et al., 2015; Schenk & Fremouw, 2012; Selkie et al., 2016; Wick et al., 2020). We expected to find a significant relationship between each form of bullying victimization during high school and levels of mental health symptoms in college, such that bullying victimization during high school would predict anxiety and depression in college.

Social support has demonstrated a positive impact on mental health, including depression (Camara & Padilla, 2017; Dafaalla et al., 2016; Kugbey, 2015) and anxiety (Scardera et al., 2020). College students who lack social support tend to experience more mental health problems (Alsubaie et al., 2019) and are more likely to experience depressive symptoms (Hefner & Eisenberg, 2009). Therefore, our second aim of the current study was to replicate this finding. Because social support can come from various sources (R. Evans et al., 2022; Song et al., 2023), we examined social support from family, friends, and special persons separately, as well as their composite, to better understand the relationships between different sources of social support and mental health. We expected to find significant relationships between each independent source of social support in college and mental health in college, such that stronger social support would predict lower levels of anxiety and depression symptoms in college.

With these primary relationships in mind, we next aimed to identify protective factors that could potentially mitigate the adverse effects of victimization on the mental health of college students. The existing literature suggests that individuals who believe they have sufficient social support may experience fewer mental health problems than their more isolated counterparts. Therefore, we used the stress-buffering model to guide our examination of social support as a moderator of the relationship between each type of victimization and mental health problems. Specifically, we expected each source of social support to moderate the relationships between traditional and cyberbullying in high school and mental health symptoms (anxiety and depression) in college.

In sum, the current study addresses two noted gaps in the literature. First, we distinguish between traditional bullying and cyberbullying due to the potentially more damaging effects of cyberbullying suggested in the literature because of its broader range of bullying behaviors. Second, we examine three sources of social support separately to strengthen our ability to detect more complex moderating relationships that may vary across support sources and types of bullying. This more nuanced approach may better inform targeted interventions for college students who have experienced traditional bullying versus those who have experienced cyberbullying in high school.

2. Materials and Methods

2.1. Participants

Participants included 329 college students attending a rural, Southeastern university who, out of a larger sample of 717, reported being victimized during high school. About 46% of the full sample indicated that they experienced bullying victimization in high school, which is slightly higher than national reported estimates for peer/bullying victimization in high school (~20%; National Center for Education Statistics, 2019). Participants in the full sample who reported no bullying victimization during high school were excluded from the analytic sample, as they were not able to answer any questions regarding bullying victimization in the study. The statistics and data analysis reported below are specific to the analytic sample. Most of the analytic sample identified as female, and most did not identify as sexual minorities. The majority age group in the analytic sample was 18–20 years, and most participants were freshmen or sophomores at the time of the study. Most participants had a grade point average in the B range. Detailed sample characteristics for the analytic sample are provided in

Table 1. Analytic sample demographic characteristics.

Variable		Frequency	Percent
Race/Ethnicity	White/Caucasian	289	87.8
	Spanish/Hispanic/Latinx	25	7.6
	Black/African American	10	3.0
	Asian/Asian American	10	3.0
	American Indian/Alaskan Native	3	0.9
	Native Hawaiian or Other Pacific Islander	1	0.3
	Not listed above	3	0.9
	Prefer not to say	6	1.8
Gender	Female	261	79.6
	Male	62	18.9
	Nonbinary/Third Gender	1	0.3
	Prefer to self-describe	1	0.3
	Prefer not to say	3	0.9
Sexual Minority Status	Indicate they identify as LGBT	71	21.6
	Indicate they do not identify as LGBT	251	76.3
	Prefer not to respond	7	2.1
Age	18–20	250	76.0
	21–23	76	23.1
	24–20	3	0.9
	30+	0	0.0
Grade Classification	Freshman	108	32.8
	Sophomore	96	29.2
	Junior	68	20.7
	Senior	57	17.3
	Non degree-seeking	0	0.0
GPA	3.7–4.0 (A− to A)	111	33.7
	2.7–3.69 (B− to B+)	179	54.4
	1.7–2.69 (C− to C+)	38	11.6
	0.7–1.69 (D− to D+)	1	0.3

Note. All frequencies and percentages are based on analytic sample (n = 329). All demographic characteristics were categorical variables.

.

2.2. Measures

2.2.1. Bullying

College-High School Bullying Questionnaire (CHSBQ). The College-High School Bullying Questionnaire (CHSBQ) is an unpublished, modified version of the Swearer Bully Survey-Student Version (unpublished; Swearer, 2001) and the Self-Report Coping Scale (Kochenderfer-Ladd & Skinner, 2002) that was developed for this study. The CHSBQ includes 103 questions that address students’ experiences with bullying during college and retrospectively during high school. The questionnaire begins with the following definition: “Bullying is unwanted, aggressive behavior that is intentional (i.e., on purpose) and involves a real or perceived power imbalance (i.e., the person being bullied has a hard time defending himself or herself; we acknowledge that the preferred terminology should be “themself”). Bullying behavior is repeated, or has the potential to be repeated, over time”. The CHSBQ consists of four parallel sections: college victim, college bully, high school victim, and high school bully. Each section begins with a simple yes or no question asking whether the respondent was involved in bullying (e.g., “Were you bullied in high school?” and “Did you bully anyone in high school?”).

Only participants who responded “Yes” to questions about bullying were directed to subsequent questions about specific types of bullying—both traditional and cyber/online—that they experienced (victims). Thus, if participants selected “No”—indicating that they did not experience bullying based on the definition provided—then participants were not able to answer subsequent question about various types of bullying victimization. All individuals who answered “Yes” to the query about bullying victimization in high school were included in the analytic sample (n = 329). All 20 items regarding traditional and cyberbullying/online bullying were rated on a Likert scale (1 = Never to 4 = Daily). The first four items describe physical contact or harm or damage to property. The remaining 16 items include teasing, name-calling, inappropriate sexual comments, threatening harm, ignoring, spreading rumors, telling others not to be friends with the victim, and embarrassing the victim; participants indicate how often they experienced each of these actions, both in-person/traditionally or online/cyber.

The factor structure of a portion of the CHSBQ was previously examined using principal components factor analysis with a varimax rotation and revealing two factors (Swearer Napolitano, 2008). The first factor was characterized as verbal bullying (α = 0.85) and explained 34% of the variance in the full scale; a second factor was characterized as physical bullying (α = 0.79) and explained 23% of the variance in the full scale. Because one aim of the current study was to independently examine traditional and online bullying, we chose to exclude the physical bullying items for the current study, because they were limited to occurring in-person and could not be replicated in online contexts. Therefore, we instead duplicated the verbal bullying items in our survey and asked participants to report on their traditional and online bullying experiences separately, allowing us to distinguish bullying across these contextual settings.

For the current study, we created mean scores from the 16 items described above to form “traditional bullying” (8 items; α = 0.82) and “cyberbullying” scores (8 items; α = 0.87) during high school (retrospectively reported). The questions that follow, which were not used in this study, are dichotomous: victims indicate who bullied them, why they think they were bullied, and what they did to cope. Those who report having bullied others indicate why and whom they bullied.

2.2.2. Mental Health Symptoms

Patient Health Questionnaire (PHQ). The Patient Health Questionnaire (PHQ; Kroenke et al., 1999) is a 9-item self-report that screens for symptoms of depression assessed during college. The scale corresponds with the diagnostic criteria for depression specified in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV TR; American Psychiatric Association, 2000). Respondents report symptoms they have experienced over the previous two weeks, such as “feeling tired or having little energy” and “trouble concentrating on things, such as reading the newspaper or watching television” (Kroenke et al., 1999). Items are rated on a Likert scale from 0 (Not At All) to 3 (Nearly Every Day), and a total score is provided; higher scores indicate more depressive symptoms. For the current study, eight of the ten items were used and coded with a range of 1 (Not At All) to 4 (Nearly Every Day); item #9, which addresses thoughts of hurting oneself, was not included. These modifications resulted in a possible range of scores from 8 to 32 for the current sample. Internal consistency reliability for the measure is high (α = 0.89).

Generalized Anxiety Disorder-7 (GAD-7). The Generalized Anxiety Disorder-7 (GAD-7; Spitzer et al., 2006) is a 7-item scale that screens for anxiety-related symptoms assessed in college. Items are rated on a Likert scale from 0 (Not At All) to 3 (Nearly Every Day), and a total score is provided that ranges from 0 to 21. Using this score range, totals ranging from 0 to 4 reflect minimal anxiety, totals between 5 and 9 represent mild anxiety, those from 10 to 14 reflect moderate anxiety, and totals of 15–21 represent severe anxiety (Spitzer et al., 2006). For the current study, a scale of 1 (Not At All) to 4 (Nearly Every Day) was used, resulting in a possible range of scores from 7 to 28 for the current sample. The scale’s psychometric properties are strong (α = 0.90).

Given the high correlations between depressive and anxiety symptoms in the current study (r(325) = 0.72, p < 0.001), we created a higher-order mean composite of the PHQ and GAD-7 scores to represent broader mental health symptoms during college as an outcome in analyses. Higher scores represented greater anxiety and depressive symptoms or more mental health symptoms.

• Social Support

Multidimensional Survey of Perceived Social Support (MSPSS). The Multidimensional Survey of Perceived Social Support (MSPSS; Zimet et al., 1988) is a 12-item self-report of respondents’ perceptions of social support available from different people in their lives and was assessed when participants were in college. The MSPSS provides three subscales (4 items each), and a total score (12 items; α = 0.88). Higher scores on the MSPSS indicate greater levels of social support. Items are rated on a Likert scale from 1 (Very Strongly Disagree) to 7 (Very Strongly Agree). The Family subscale (α = 0.90) includes items such as “My family really tries to help me” and “I get the emotional help & support I need from my family”. The Friends subscale (α = 0.94) includes items such as “My friends really try to help me” and “I can count on my friends when things go wrong”. The Significant Other subscale (α = 0.94) includes items such as “There is a special person in my life who cares about my feelings” and “There is a special person who is around when I am in need”. We tested each subscale as a moderator in the analyses. Because the items on the Significant Other subscale refer to “a special person”, with no mention of a “significant other”, we use the term “special person” in our presentation of the results. Given high correlations between all three sources of social support in the current study (and typical scoring of the MSPSS; rs(325) = 0.50 − 0.81, ps < 0.001), we also created a higher-order mean composite of the family, friend, and special person social support scores to represent total/overall social support during college as a moderator in the analyses. Higher scores represented greater total social support across all sources.

2.2.3. Demographics/Covariates

Demographic variables that have been associated with individual differences in mental health symptoms in the previous literature were included in bivariate correlations to determine whether they should be included in main analyses (e.g., see review by Argyriou et al., 2021; BlackDeer et al., 2023; Cole et al., 2002; Daly, 2022). Demographic variables were included in main analyses if they showed a significant correlation with mental health symptoms (see

Table 2. Zero-order correlations and descriptive statistics.

	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Traditional Bullying	-
2. Cyber Bullying	0.66 ***	-
3. Family Social Support	−0.26 ***	−0.21 ***	-
4. Friend Social Support	−0.21 ***	−0.17 **	0.50 ***	-
5. Special Person Social Support	−0.15 **	−0.14 *	0.35 ***	0.47 ***	-
6. Social Support Total	−0.27 **	−0.22 ***	0.79 ***	0.81 ***	0.77 ***	-
7. Mental Health Symptoms	0.30 ***	0.28 ***	−0.41 **	−0.31 **	−0.14 *	−0.37 ***	-
8. Race/ethnicity	−0.07	−0.06	0.08	0.05	0.18 ***	0.14 *	−0.01	-
9. Gender	−0.07	0.09	0.05	0.09	0.22 ***	0.14 **	0.02	0.16 ***
10.Sexual Minority Status	−0.24 ***	−0.13 *	0.23 ***	0.15 **	0.09	0.20 ***	−0.17 **	0.02	0.01
11. Age	0.02	0.05	0.04	0.07	0.06	0.08	−0.03	0.07	0.07	0.09
12. Grade	0.06	0.07	−0.002	−0.02	0.07	0.01	0.01	0.02	0.14 **	0.04	0.38 ***
13. GPA	−0.004	−0.04	−0.06	−0.03	−0.15 **	−0.10	0.08	−0.08	−0.10	−0.02	−0.04	−0.22 ***
Mean	1.96	1.66	5.30	5.59	5.72	5.54	0.00
Standard Deviation	0.62	0.68	1.37	1.20	1.38	1.03	0.93
Minimum	1.00	1.00	1.00	1.75	1.00	2.00	−1.42
Maximum	4.00	3.86	7.00	7.00	7.00	7.00	2.52
Skewness	0.85	1.29	−0.72	−0.81	−1.31	−0.70	0.65
Kurtosis	0.17	1.00	0.09	0.09	1.33	0.33	−0.37

Note. Demographic variables were coded into dichotomous variables for analyses such that, Race/ethnicity: 0 = all other or multiple races, 1 = White/Caucasian; Gender: 0 = male, 1 = female; Sexual minority Status: 0 = yes/prefer not to respond, 1 = no; Age: 0 = all other age groups, 1 = 18–20 years, Grade: 0 = all other grade classifications, 1 = freshman; GPA: 0 = all other grade point averages, 1 = 2.7–3.69 average. Percentages for dichotomous demographic variables are provided in Table 1. Mental Health Symptoms are a z-score composite for the PHQ and GAD-7. * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001.

). Demographic variables that were tested included race/ethnicity (non-Hispanic White/Caucasian = 1), gender (female = 1), sexual minority status (does not identify as sexual minority = 1), age (18–20 years = 1), and grade point average (GPA; 2.7–3.69 average = 1).

2.3. Procedure

This study was conducted as part of a larger study of correlates of college and high school bullying. Participants were recruited through the Psychology Department’s Research Participant Pool, which consists of students enrolled in introductory and intermediate Psychology classes who may participate in research studies to earn Experiential Learning Credits (ELCs) for course requirements. To avoid coercion to participate in research studies to fulfill their ELC requirements, students could perform an alternative activity, such as writing a research paper. Students who wished to participate registered on the Student Research Participation software system (SONA) and signed up for the study. Once enrolled, participants viewed an Informed Consent Form and were instructed that by clicking the “Agree” button, they acknowledged that they had read the information and voluntarily agreed to participate. Students ages 18 and younger were excluded from participation. Participants completed eight questionnaires combined into one online survey on Qualtrics; four of those were examined in the current study. The expected time to complete all eight questionnaires was approximately 30–60 min.

After completing the survey, participants could enter a drawing for a USD 50 Amazon gift certificate. To help ensure the anonymity of the data, students who elected to be entered in the drawing provided their email addresses through a process separate from the survey data collection. All participants were awarded two Experiential Learning Credits (whether they completed all surveys or not). This research was approved by the University’s Institutional Review Board and was conducted in a manner consistent with its guidelines for research.

3. Results

3.1. Preliminary Results

Analyses were conducted in Jamovi 2.3.22 (The Jamovi Project, 2024). Descriptive statistics and zero-order correlations are reported in Table 2. As previously noted, covariates used in analyses were identified based on significant correlations between demographic variables and the mental health composite (Table 2). Only sexual minority status was significantly associated with mental health symptoms; as such, all other covariates not associated with the mental health composite score were excluded from the main analyses to maintain degrees of freedom in multiple regression models. Both traditional and cyberbullying were correlated with all sources of social support as well as with mental health symptoms.

3.2. Main Analyses

We conducted eight independent multivariate linear regression models to test associations between the following:

(a)

Traditional bullying, family social support, and mental health symptoms;

(b)

Traditional bullying, friend social support, and mental health symptoms;

(c)

Traditional bullying, special person social support, and mental health symptoms;

(d)

Traditional bullying, overall social support, and mental health symptoms;

(e)

Cyberbullying, family social support, and mental health symptoms;

(f)

Cyberbullying, friend social support, and mental health symptoms;

(g)

Cyberbullying, special person social support, and mental health symptoms;

(h)

Cyberbullying, overall social support, and mental health symptoms.

3.3. Associations Between Traditional Bullying, Social Support, and Mental Health

Four independent multivariate regression models were conducted to determine the extent to which traditional bullying during high school, various sources of social support during college, and their interaction predict mental health symptoms during college over and above the effect of significant covariates (see

Table 3. Regression analyses for the associations between traditional bullying, social support from family, friend, and special person, and mental health symptoms.

Model Predictors	Mental Health Symptoms Est. (SE)
Intercept	0.06 (0.10)
Sexual Minority Status	−0.08 (0.11)
Traditional Bullying	0.28 (0.07) ***
Family Social Support	−0.24 (0.04) ***
Traditional Bullying × Family Social Support	−0.02 (0.05)
Intercept	0.15 (0.10)
Sexual Minority Status	−0.19 (0.12) +
Traditional Bullying	0.33 (0.08) ***
Friend Social Support	−0.20 (0.04) ***
Traditional Bullying × Friend Social Support	−0.04 (0.05)
Intercept	0.16 (0.10)
Sexual Minority Status	−0.21 (0.12) +
Traditional Bullying	0.39 (0.08) ***
Special Person Social Support	−0.06 (.04) +
Traditional Bullying × Special Person Social Support	0.00 (0.06)
Intercept	0.13 (0.10)
Sexual Minority Status	−0.16 (0.11)
Traditional Bullying	0.30 (0.08) ***
Total Social Support	−0.30 (0.05) ***
Traditional Bullying × Total Social Support	−0.00 (0.07)

Note. All models run independently. Covariates and predictors were grand mean centered. Interaction terms were computed after centering the predictors. Est. = unstandardized partial regression coefficient estimate. SE = robust standard error. ⁺ p ≤ 0.10; *** p ≤ 0.001.

). Traditional bullying, social support variables, and covariates were grand-mean centered at zero in multivariate linear regression models. Unstandardized beta estimates and standard errors were reported for multivariate linear regression models. Across all models, higher levels of traditional bullying were associated with higher scores on mental health symptoms on average. Similarly, three of the four models showed that higher levels of social support (from family, friends, and overall support, but not from a special person) were associated with lower scores on mental health symptoms on average. However, identifying as a sexual minority was not associated with mental health symptom scores in any of these regression models, and there were no significant interactions between traditional bullying during high school and any source of social support during college when predicting mental health symptoms in college.

3.4. Associations Between Cyberbullying, Social Support, and Mental Health

Four additional independent multivariate regression models were conducted to determine the extent to which cyberbullying experienced during high school, various sources of social support reported during college, and their interaction predict mental health symptoms during college over and above the effect of significant covariates (see

Table 4. Regression analyses for the associations between cyberbullying, social support from family, friend, and special person, and mental health symptoms.

Model Predictors	Mental Health Symptoms Est. (SE)
Intercept	0.09 (0.10)
Sexual Minority Status	0.12 (0.11)
Traditional Bullying	0.25 (0.07) ***
Family Social Support	−0.24 (0.04) ***
Traditional Bullying × Family Social Support	0.05 (0.05)
Intercept	0.18 (0.10) +
Sexual Minority Status	−0.22 (0.11) *
Traditional Bullying	0.28 (0.07) ***
Friend Social Support	−0.20 (0.04) ***
Traditional Bullying × Friend Social Support	−0.07 (0.05)
Intercept	0.19 (0.10) +
Sexual Minority Status	−0.25 (0.12) *
Traditional Bullying	0.33 (0.08) ***
Special Person Social Support	−0.06 (0.04)
Traditional Bullying × Special Person Social Support	−0.04 (0.05)
Intercept	0.14 (0.10)
Sexual Minority Status	−0.18 (0.11)
Traditional Bullying	0.25 (0.07) ***
Total Social Support	−0.27 (0.05) ***
Traditional Bullying × Total Social Support	−0.09 (0.06)

Note. All models run independently. Covariates and predictors were grand mean centered. Interaction terms were computed after centering the predictors. Est. = unstandardized partial regression coefficient estimate. SE = robust standard error. ⁺ p ≤ 0.10; * p ≤ 0.05; *** p ≤ 0.001.

). Cyberbullying, social support variables, and covariates were grand-mean centered at zero in multivariate linear regression models. Across all models, higher levels of cyberbullying were associated with higher scores on mental health symptoms on average. Similarly, three of the four models showed that higher levels of social support (from family, friends, and overall support, but not from a special person) were associated with lower scores on mental health on average. Further, in the models with social support from friends and a special person, identifying as a sexual minority was associated with worse/higher mental health symptoms in college, and there were no significant interactions between cyberbullying during high school and any source of social support during college when predicting mental health scores reported during college.

4. Discussion

College students face many challenges as they transition from high school into a new and unfamiliar environment (Conley et al., 2014), with many experiencing poor mental health (Porru et al., 2022). Compared to their non-bullied peers, college students who have been previously bullied often experience an exacerbated risk for more severe mental health problems (Reid et al., 2016). It is possible that factors such as social support could attenuate the adverse outcomes associated with victimization. The purpose of this study was to examine the relationships between bullying victimization (traditional and online), perceived social support (from family, friends, a special person, and overall support), and mental health (anxiety and depression) and to determine whether social support moderates the associations between victimization experienced during high school and mental health symptoms in college. As expected, our findings provide support for the links between bullying victimization and poor mental health. However, while various sources of social support were associated with positive mental health symptoms, social support did not emerge as a protective factor overall.

Specifically, higher levels of both traditional and cyberbullying predicted worse mental health symptoms over and above other covariates and interactions we tested, with a slightly more robust main effect for traditional bullying experienced during high school on mental health symptoms during college. This also indicates that, regardless of the level or source of social support, as well as other demographic factors, greater bullying is associated with worse self-reported mental health symptoms on average for the entire sample. These findings are consistent with past research demonstrating that both traditional and cyberbullying victimization is associated with higher levels of psychological distress, including anxiety and depression (Na et al., 2015; Schenk & Fremouw, 2012; Selkie et al., 2016; Wick et al., 2020). As noted previously, we excluded physical items from the traditional bullying section of the CHSBQ to ensure the comparability of our two measures of bullying. In doing so, our measures had the same number of items and addressed the same types of bullying behaviors across each type of victimization.

A major strength of our study is the consistent pattern that emerged across traditional bullying and cyberbullying. Our findings add to the existing literature by illustrating that regardless of the format, bullying during high school predicts mental health problems during college; indeed, our study findings are consistent with recent empirical studies showing that both traditional and cyberbullying predict mental health outcomes, with slightly more robust effects demonstrated between traditional/in-person bullying and mental health outcomes (e.g., Prowten & Breitenstein, 2023). Many studies do not differentiate between these types of bullying (C. B. R. Evans et al., 2019; Kozasa et al., 2017), reporting on just one type or the other or combining across contexts. Yet there is evidence that traditional and cyberbullying are distinct (Jones et al., 2024; van den Eijnden et al., 2014), and specific types of bullying might be stronger predictors of various outcomes, including mental health. For example, given the ubiquitous and damaging nature of cyberbullying (e.g., Nick et al., 2018), we might have expected it to emerge as a stronger predictor than traditional bullying. However, similar to prior studies (Mehari et al., 2020; Prowten & Breitenstein, 2023), we did not find that one type of bullying was necessarily a stronger predictor of mental health outcomes, suggesting that perhaps any or all types of bullying are similarly negative or damaging to adolescent mental health.

We also found that participants who perceived more general social support and support from peers, parents, and overall social support (but not from a special person) during college experienced more positive mental health during college, suggesting that regardless of the level or type of bullying experienced and other demographic factors, greater perceptions of social support from various sources are associated with better self-reported mental health scores on average for the entire sample during college. While we anticipated this main effect, we also hypothesized a moderating effect of social support that did not emerge as a protective factor against either traditional bullying or cyberbullying. However, in their description of the buffering model, Cohen and Wills (1985) explain that for social support to play a buffering role, the individual must be experiencing high levels of stressful responses to the situation. In other words, social support may not help individuals who are not experiencing high stress levels. It is possible that our sample did not meet a sufficient threshold of stress (e.g., poor mental health), as average levels of anxiety and depressive symptoms in the current study were relatively mild and below clinical levels. However, there does not appear to be any established threshold or minimum in the current literature that qualifies “high stress levels”, so it is difficult to tell whether the participants in the current study met some sort of minimum stress level under which social support may act or function as a buffer. This being said, at least one new study shows that parental warmth and positive school environments did not buffer against the negative long-term effects of peer victimization on various mental health outcomes in adolescence (Martínez et al., 2024), which seems to be consistent with our lack of findings regarding the moderating effects of social support on links between bullying victimization and mental health symptoms.

A possible explanation for social support not emerging as a buffer in our study involves the timing of our data collection. Our participants provided retrospective reports of victimization during high school (past) yet reported on their mental health during college (current). Had we been able to survey participants’ bullying experiences while they were in high school and more proximal to the event, their stress responses likely would have been higher, and we might have observed a buffering effect. Although real-time ratings of victimization may be preferable, support for using retrospective data is provided in other published studies of prior victimization in college students (Chen & Huang, 2014; Lin et al., 2020; Waechter et al., 2017). Additionally, some evidence points to the persistence of memories of difficult childhood events. For example, when remembering instances of childhood teasing, adults with social anxiety recalled those instances accurately, even when their anxiety symptoms resolved (Waechter et al., 2017).

Another reason that social support during college may not have emerged as a significant moderator could be related to the social support instrument used in the current study. Specifically, social support was measured using a bipolar rating scale in which response options from one to three indicated varying degrees of no social support from specific sources, option four indicated neutral or no opinion regarding support from specific sources, and response options five to seven indicated mild to significant social support from specific sources. Therefore, responses from any of the first four options on the scale all indicate a lack of support and are indistinguishable from one another, leaving only three response options indicating degrees of social support. Unlike the scales for the other instruments used in this study, this bipolar response scale potentially introduces construct-irrelevant variance, artificially lowers the means, and obscures the meaning of responses regarding social support. For the current study, participants showed an average of just above five on all social support scales (and the total social support mean scale; see Table 2), suggesting some support (but not significant support) on average from either friends, family, or a special person. Thus, means for each social support source and overall social support suggest participants reported mild to moderate levels of social support at most, which might explain the null effects of social support as a moderator in the current study.

As previously described, the literature supporting social support as a protective factor is mixed, with some studies failing to find evidence for the buffering effect. For example, although they did not specifically focus on bullying, Szkody and McKinney (2019) found that neither perceived nor received social support moderated the association between negative life events and depression as they expected, concluding that the stress-buffering effect may not generalize to depression and, conceivably, psychological problems altogether. These findings from Szkody and McKinney (2019) seem to align with the current study’s findings, rather than other findings and theoretical models that suggest social support buffers against negative developmental outcomes. Further, in Rueger et al.’s (2016) review, the only support for the stress-buffering hypothesis was the benefit of social support from teachers, peers, and close friends on depression for youth with medical illnesses. When pondering the disparate findings, one issue to consider is the type of social support measured (e.g., tangible, seeking, perceived). Teasing apart the different types of support may provide different results. For example, as noted above, Szkody and McKinney (2019) examined both received and perceived support, which might lead to divergent findings. With these and other conflicting results in mind, it seems that social support needs further investigation to fully understand its potential contribution as a buffer between bullying and mental health.

Limitations and Future Directions

Sample and measurement issues may limit the generalizability of our results. Our sample is relatively small and homogeneous, consisting primarily of white participants with limited gender or sexual diversity between 18 and 20 years who were freshmen and sophomores attending a rural Southeastern university. It is noteworthy, however, that our sample reflects the larger university population. Additionally, we studied a nonclinical sample, which, as noted previously, may have precluded our ability to find a buffering effect. Future studies could examine these associations with more heterogeneous populations to determine whether social support matters to different groups (e.g., gender nonconforming, LGBTQ, race, SES, family structure). For example, existing findings have identified family support as a protective factor for LGBTQ adolescents who are at risk for depression and suicidal ideation from homophobic victimization; peer support and having a trusted adult at school played significant yet smaller roles (Rivas-Koehl et al., 2022). Additional findings identified support from peers, but not family, as a buffer between bullying victimization and depression in college students (Moran et al., 2018).

Studies of individuals involved in different bullying roles—bully, bully-victim, and bystander—may also shed light on the importance of social support. In their study of cumulative exposure to bullying and mental health in adolescents, C. B. R. Evans et al. (2019) found both similarities and differences in outcomes for students in different bullying roles. For example, victims, bullies, and bystanders tended to experience internalizing problems, aggression, and low optimism about the future, but victims also experienced low self-esteem, and bystanders’ academic achievement declined.

Our measurements of the constructs of interest present both strengths and limitations. Relying solely on self-reports can potentially introduce systematic biases such as method variance or self-serving response set biases that can, in turn, obscure the relationships among constructs. However, we employed measures (GAD-7, PHQ-9, MSPSS) that have been widely used in research across various disciplines, which we consider a strength. Additionally, although the MSPSS allowed us to examine different sources of social support, items on its “Significant Other” subscale refer to a “special person” but do not specify who that person is. Therefore, interpretations of that subscale should consider that participants may have reported on various people (partner, parent, teacher, etc.), which might further explain why this source of social support did not demonstrate either a significant main effect on mental health symptoms in most models or a moderation effect in any statistical models. Indeed, the predictive power of support from a special person was markedly decreased in both sets of models, as compared to family, friend, or overall support.

Our bullying survey presents both strengths and limitations. Compared to many bullying measures described in the literature (C. B. R. Evans et al., 2019; Kozasa et al., 2017), the CHSBQ captures many more bullying behaviors, thereby providing a broader sample of the construct. Despite that strength, the structure of the CHSBQ may have reduced our sample size. Participants who selected “no” to the initial question (Were you bullied in high school?) did not see the subsequent questions describing various bullying behaviors. Consequently, they might not have realized that certain incidents, such as telling others not to be friends with someone, are considered bullying. Had participants viewed the list of specific bullying behaviors first, a broader sample may have been identified to include in the sample. Volk et al. (2014) purport that assessing specific bullying behaviors provides a better measurement of bullying than general questions. Future studies using the CHSBQ should consider eliminating the initial question to capture as many participants with bullying histories as possible. Another concern about the CHSBQ is the lack of psychometric data supporting its use, as it was developed for the current study. Still, it is modified from Swearer’s (2001) Bully Survey-Student Version and the Self-Report Coping Scale (Kochenderfer-Ladd & Skinner, 2002) surveys, which have demonstrated reliability and validity across several studies (Flanagan et al., 2012; Swearer & Cary, 2003; Swearer et al., 2008). Future studies might examine a broader range of the CHSBQ’s psychometric properties beyond the strong internal consistency reliabilities we observed in this study.

Further, about 46% of the larger, original study population, reported experiencing peer victimization, which is higher than reported rates nationally of bullying victimization in adolescents (20% on average). This may be due to the working definition of bullying used in the CHSBQ, which may have led to more participants reporting experiences of bullying than we would typically see in national surveys or other data sets. Further, higher bullying victimization rates in the current study may have also been influenced by self-selection because the sign-up was advertised as “Correlates of Bullying in College and High School”. Any of these factors may limit the generalizability of our findings. However, it is also possible that by broadening definitions of bullying behaviors, researchers are able to capture participants who may not always report bullying victimization that is actually occurring. It is also possible that in our particular region and geographic location that rates of bullying victimization are slightly higher than national averages.

Temporal issues should also be noted as potential limitations. Almost half the data for our analytic sample was collected during the COVID-19 pandemic, with some surveys completed during the spring of 2020 and others during the following fall, the first full semester of online instruction for many college/university students. This context substantially changes the backdrop for the sample and requires consideration when interpreting the results. An additional temporal concern surrounds the use of retrospective reports of high school bullying. However, as mentioned previously, it is not unusual for researchers to rely on retrospective reports of bullying (Chen & Huang, 2014; Lin et al., 2020). Longitudinal studies could provide more accurate information about the impact of high school bullying on college mental health, but that type of research is challenging and costly to carry out.

5. Conclusions

Bullying is a serious public health concern, and its deleterious outcomes are widely documented in the literature. Our study findings are in line with the prior literature demonstrating that victims of bullying experience heightened symptoms of anxiety, depression, suicidal thoughts and ideation, and other mental health and psychological problems (Galán-Arroyo et al., 2023; Kalogerakis et al., 2021; Vacca et al., 2023). Even more concerning, our findings support prior work showing that adolescents who are victimized exhibit higher rates of mental health symptoms (Galindo-Domínguez & Losada Iglesias, 2023) that can persist into college (Copeland et al., 2013; Gorman et al., 2021; Källmén & Hallgren, 2021; Lin et al., 2020; Manrique et al., 2020). Given their heightened levels of anxiety and depression, college students who have been previously bullied may experience continuing detrimental outcomes later in college. Although prevention is a best practice for educators and administrators at all levels of schooling, once bullying occurs, the damage is done, and tertiary measures are required. Therefore, this study examined social support as a factor that could potentially serve to protect the mental health of college students who had been previously bullied.

Our findings replicated previous studies demonstrating the relationship between both traditional and cyberbullying and later reports of anxiety and depression. Moreover, two of the three independent sources of social support we examined (and the composite of overall social support) were predictive of more positive mental health outcomes. However, we did not find evidence that social support moderated the relationship between experiencing bullying in high school and mental health in college, as was predicted from a stress-buffering model (Cohen & Wills, 1985); rather, the protective role of social support was limited to its main effect. Moving forward, researchers and practitioners should conder designing interventions that emphasize relationship-building to facilitate improved social support. Emphasizing relationship-building in interventions may hold promise for improved mental health of college students, including those who have experienced bullying previously (as demonstrated in the current study). However, researchers and practitioners should remain cautious of the overall impact of these interventions with college students, as the anticipated efficacy of any such interventions might not be expected to vary significantly between those who have and those who have not experienced bullying previously (based on our study findings).