Sage Journals: Discover world-class research

Abstract

We examine whether students’ victimization (nonvictims, non-bias-based, and bias-based victims) relates to reports on sadness and hopelessness (SH) and suicidal ideation (SI) and how school size, ethnic and socioeconomic status composite, and school climate relate to these outcomes through multilevel analyses of the California Healthy Kids Survey (2017–2019; N = 2,569 schools; N = 1,025,876 students). Bias-based or non-bias-based victims experienced greater SH and SI compared to nonvictimized students. School climate negatively associated with both outcomes, yet positive school climate exacerbated SH and SI of victimized students, particularly those subjected to bias-based violence. The findings confirm the healthy context paradox, indicating that adverse effects may occur for violence victims in schools with more positive climates. Implications for interventions and future research are discussed.

Keywords

adolescence bias-based victimization cross-level interactions multilevel modeling sadness and hopelessness school climate school victimization suicidal ideation the healthy context paradox

Poor mental health (MH) among adolescents is a widespread public health issue (World Health Organization, 2021). National surveys of American middle and secondary school students have revealed significant increases in various MH symptoms in the past 15 years (Centers for Disease Control and Prevention, 2023; U.S. Surgeon General, 2021). School-based surveys often track two key indicators of youth MH: (a) persistent feelings of sadness and hopelessness (SH) and (b) suicide-related thoughts and behaviors. These symptoms, commonly linked to depression, are the most frequent signs of mental illness in this group. In 2021, more than 40% of high school students felt SH almost every day for at least 2 weeks in a row such that they stopped doing their usual activities and more than 20% had suicidal ideation (SI), thoughts, or behaviors (Centers for Disease Control and Prevention, 2023). Significant differences in prevalence of SH and SI were documented, with female, older, and nonheterosexual youths reporting greater SH and SI compared with male, younger, and heterosexual youths. Hispanic and multiracial students were more vulnerable to persistent feelings of SH, whereas Asian students were the least likely to seriously consider suicide compared to other ethnocultural groups (Centers for Disease Control and Prevention, 2023). Given the high proportion of youths experiencing SH and SI, it is important to explore such problems further and identify risk and protective factors.

School Victimization

School victimization significantly contributes to increased SH and suicidality among students (Newman et al., 2023). Students with greater vulnerability and multiple stigmatized identities often experience bias-based violence, which targets identity characteristics such as race, ethnicity, sexual and gender identity, religion, disability, physical appearance, and immigration status. The prevalence of bias-based violence in schools has varied widely across studies, with national reports indicating that from approximately 7% (Irwin et al., 2023) to 42% (L. M. Jones et al., 2019) of middle and secondary school students experienced some form of bias victimization in the past year.

Some researchers indicated that bias-based violence results in more severe MH outcomes for victims than non-bias-related victimization. Specifically, students who have experienced bias-based harassment are significantly more likely to exhibit SH and SI compared to those facing non-bias-based harassment (Russell et al., 2012). Additionally, victimization linked to a student’s identity is associated with broader cultural stigma and prejudice, which these students encounter outside of school (Joo et al., 2023), further entrenching social inequalities and oppression experienced by marginalized and vulnerable youths.

Previous research has highlighted the harmful health effects of both non-bias-based and bias-based victimization; however, studies have rarely compared SH and SI among nonvictims, bias-based victims, and non-bias-based victims of school violence. Furthermore, even fewer studies have used population samples to capture individual and school variations to investigate these widespread public health issues.

School Effects, SH, and SI

Research has suggested that the school context could potentially affect students’ victimization and associated negative MH outcomes (Astor & Benbenishty, 2019). The literature suggests three types of school effects: the structural effects of schools (e.g., school resources, school size, teacher–student ratio), the effects of school composition (e.g., school composite socioeconomic status [SES], proportion of ethnic minority students), and the effects of school practices and processes (e.g., school climate; Marks, 2010).

Research on school effects has presented an inconclusive picture of their role in youths’ MH outcomes, particularly SH and SI. Research on the structural effects of school size has yielded mixed results. One study found that smaller schools were associated with higher levels of depression and a greater likelihood of suicide attempts among male students (Watt, 2003). In contrast, larger schools were linked to a slightly increased risk of MH difficulties (Patalay et al., 2020). Other studies concluded that school size was not a significant predictor of depressive symptoms (Brière et al., 2013) or SI (Benbenishty et al., 2018). Student–staff ratio was not associated with students’ depression (Gutman & Feinstein, 2008). Adequate school resources (e.g., facilities, smaller class sizes) reduced the predictive power of low family SES on students’ depression (H. Wang et al., 2023).

Regarding the effects of school composite SES, several studies indicated that students in high-poverty schools are significantly more likely to experience persistent feelings of SH and SI (S. E. Jones et al., 2020) and poorer MH overall (Patalay et al., 2020). Benbenishty et al. (2018) found no association between school size and rates of SI. Mixed results have emerged concerning the proportion of ethnic minorities at the school level. Patalay et al. (2020) found no link between the proportion of ethnic minorities and children’s MH, whereas another study revealed fewer MH disorders in schools with higher percentages of White students (C. Wang et al., 2019). A systematic literature review indicated that greater racial or ethnic homogeneity among peers is associated with improved MH for all youths, although evidence of MH benefits in more diverse schools remains weak (DuPont-Reyes & Villatoro, 2019).

Last, school practices and processes related to positive school climate—broadly defined as the “quality and character of school life, which includes norms, values, and expectations that support people feeling socially, emotionally, and physically safe” (Cohen et al., 2009, p. 182)—were directly associated with desired student MH outcomes (Aldridge & McChesney, 2018). Similarly, subconstructs of positive school climate, including positive student–teacher relationships (Zhang et al., 2022), greater student–school connectedness (Raniti et al., 2022), and greater parent–school involvement (C. Wang et al., 2019), directly contributed to reduced SH and SI.

School Climate as a Moderator

Compositional and structural school effects often lie beyond the control of schools; however, school climate is more amenable to change. Consequently, extensive research has examined the potential for a positive school climate to diminish the negative impact of student victimization on various outcomes, although results have been inconclusive. Some studies indicated that a positive school climate buffers the adverse effects of violence exposure on student well-being (Shamrova et al., 2025) and behavioral (e.g., Haj-Yahia & Abo Basal, 2024) and MH outcomes, including depression and suicidality (Chan, 2022). These findings apply even to vulnerable student groups; a literature review on protective factors related to homophobic bullying identified a positive school climate buffered against depressive symptoms resulting from such victimization (Espelage et al., 2019). These findings are often interpreted through the socioecological theory of human development (Bronfenbrenner & Morris, 1998), which when applied to the school context, posits that the occurrence, characteristics, and severity of school violence—along with the negative outcomes for those affected—vary based on the interaction of factors in multiple subsystems both inside and outside the school. Such interactions can lead to both positive and negative developmental trajectories, depending on the internal school context (Astor & Benbenishty, 2019). Therefore, a positive school climate can directly reduce youth victimization and also mitigate its negative effects on MH.

In contrast, emerging research has found that the negative impact of student victimization on various outcomes is exacerbated in classrooms and schools with positive climates and lower overall violence levels. For example, the negative associations between victimization and engagement were intensified in more positive school environments (Yang et al., 2018). Similarly, victimized students reported greater MH difficulties when they perceived a better school climate (Zhu et al., 2024) and classrooms with reduced violence (Pan et al., 2021) and in countries with lower overall bullying prevalence (Odigie et al., 2025). These unexpected findings can be interpreted through social comparison theory (Festinger, 1954), which posits that individuals evaluate themselves by comparing their experiences with those of peers. Consequently, victims of school violence may experience negative self-assessments due to comparisons with peers who have more favorable experiences. Researchers theorized the “healthy context paradox” to describe why adverse outcomes for students worsen following violence exposure in more positive school or classroom settings (Salmivalli, 2018).

Whether school climate exacerbates or buffers the association between violence victimization and MH may depend on the type of violence being examined. Recently, Clark and colleagues (2022) found that although perceptions of more positive school climate exacerbated the link between non-bias-based cyber victimization and depressive symptoms, it buffered the link between bias-based homophobic victimization and depression. Additional research is needed to explore with greater nuance the moderating effect of school climate on the association between different types of school victimization and SH and SI.

This study used national data from California middle and secondary schools and students to explore whether students’ victimization (nonvictims, non-bias-based victims, and bias-based victims) relates to reports of SH and SI. It also examined how the three school effects—structural (school size), compositional (ethnic and SES composite), and school practices and processes (school climate)—relate to these MH outcomes. Another aim was to predict student SH and SI based on students’ victimization and school effects. Last, this study examined whether school climate exacerbates or buffers the relationship between victimization and SH and SI.

Method

Data Sources

Student-level data were drawn from the California Healthy Kids Survey, conducted by WestEd on behalf of the California Department of Education biannually since 2001. The survey is conducted as a census among all school districts, schools, and students in fifth, seventh, ninth, and 11th grades. A 2-year wave provides a representative data set that includes most school districts and schools in all counties across California. Student participation is voluntary, anonymous, and confidential (CalSCHLS, n.d.). Previous studies using this database have reported that approximately 85% of school districts in California participate in the data collection, with a student-level response rate of about 87% (e.g., Benbenishty et al., 2018). This study included students in middle (seventh grade) and secondary (ninth and 11th grades) school. The surveys were administered in the 2017–2018 and 2018–2019 academic years in 2,803 schools with 1,146,954 students. We removed student observations with missing values for the main study variables and schools with fewer than 30 students. Overall, 10.6% of the student sample was removed; thus, the total sample included 1,025,876 students from 2,569 schools. Because 513 schools did not include the SI item in their survey, analyses for this outcome were based on 854,506 students from 2,056 schools. School-level data were extracted from the California Department of Education California Basic Educational Data System (http://www.cde.ca.gov/ds/sd/cb/).

Measurements

Student level

SH was measured using a single dichotomous item: “During the past 12 months, did you ever feel so sad or hopeless almost every day for two weeks or more in a row that you stopped doing some usual activities?” (0 = no, 1 = yes). This item has been used in previous studies as an indication of persistent feelings of SH among students (e.g., S. E. Jones et al., 2020). SI was measured using a single dichotomous item: “During the past 12 months, did you ever seriously consider attempting suicide?” (0 = no, 1 = yes).

School level

We assessed the proportions of students in each school who reported feeling SH almost every day during the past 12 months and who reported that they seriously considered attempting suicide in the past 12 months.

Student-level independent variables

Victimization was measured using two sets of items; six referred to non-bias-based victimization, and an additional six referred to bias-based victimization. Answers for these 12 items were provided on a 4-point Likert scale (1 = 0 times, 4 = 4 times or more).

To examine whether each set of victimization items converged to a single construct, we conducted two exploratory factor analyses with varimax rotation (oblique rotation yielded the exact same results). For non-bias-based victimization items, one factor emerged for all six items (eigenvalue = 2.42, loadings = .76–.60). For bias-based victimization, one factor emerged for all six items (eigenvalue = 2.90, loadings = .68–.60). The cumulative explained variance by these two factors was 42.0%. Based on these results, we created 12 dichotomous variables indicating whether each student reported experiencing victimization at any frequency.

Preliminary analyses indicated that only 2.8% of students reported bias-based victimization without non-bias-based victimization. Therefore, we classified students into three groups: those who reported no form of victimization across the 12 items (nonvictims), students who reported at least one form of non-bias-based victimization but no bias-based victimization (non-bias-based victims), and students who reported at least one form of bias-based victimization (bias-based victims) regardless of their reports on non-bias-based victimization (i.e., this category included students who reported either bias-based victimization exclusively or both bias-based and non-bias-based victimization).

Background characteristics (control variables)

Control variables were grade level (seventh, ninth, or 11th grades), gender (0 = female, 1 = male), sexual orientation (0 = heterosexual,¹ 1 = LGBTQ+), and race and ethnicity (1 = Hispanic, 2 = Asian, 3 = African American or Black, 4 = White, 5 = mixed, 6 = other).

School-level independent variables

Regarding structural attributes, we assessed school size (number of students). For compositional attributes, this study measured school SES using the proportion of students in each school eligible for free or reduced-price meals (FRPM). School ethnic composition was measured by computing the proportion of each ethnocultural group in each school. To distinguish schools with different levels of SES in the multivariate analyses, we followed S. E. Jones and colleagues (2020) by recoding school SES as low-poverty (≤25% of students eligible for FRPM), mid-poverty (26%–75% of students eligible for FRPM), and high-poverty (>75% of students eligible for FRPM) schools.

Regarding practices and processes, school climate was measured using five subscales referring to various aspects of the school socioemotional climate. First, teacher–student relationships was measured by six items (α = .90), and student participation was measured by five items (α = .85). These variables were measured on a 4-point Likert scale ranging from 1 (not at all true) to 4 (very much true). Next, student academic engagement was measured by four items (α = .88), student sense of belonging and attachment to school was assessed with four items (α = .83), and parental school involvement was measured with three items (α = .78). These variables were measured on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Five mean variables for each scale were computed such that higher scores indicated a more positive school climate. These scales were aggregated to reflect school-level climate. Because some school climate scales were highly correlated (rs = .73–.78), we conducted factor analysis with equamax rotation to examine the possibility of dimension reduction and avoid multicollinearity in model estimation. This analysis yielded a one-factor solution (eigenvalue = 2.72, loadings = .81–.66, cumulative explained variance = 55%). Given that the school climate constructs included in this analysis were measured on two different scales, factor scores were calculated for each respondent using least squares regression (M = 0, SD = 1; DiStefano et al., 2009) and then aggregated to reflect school-level scores.

Sample Characteristics: Student and School Levels

Descriptive statistics of student-level characteristics are shown in Table 1. Overall, most students were nonvictims (42.8%), about one third were non-bias-based victims (33.6%), and the smallest group was classified as bias-based victims (23.6%).

Table 1

Descriptive Statistics of the Study Background Variables, Overall and by Victimization Groups (N = 1,025,876)

	Overall	Nonvictims	Non-Bias-Based Victims	Bias-Based Victims
	(N = 1,025,876)	(n = 439,136)	(n = 344,610)	(n = 242,130)
	%	%	%	%	χ²	df	Cramér’s V
Group size	100.0	42.8	33.6	23.6
Grade level					14,263.46*	4	.084*
7	35.3	35.7	38.2	26.1
9	34.3	43.7	32.6	23.7
11	30.5	50.1	29.2	20.7
Gender					9,120.38*	2	.097*
Male	50.0	47.40	31.50	21.10
Female	50.0	37.90	35.90	26.20
Sexual orientation					29,413.18*	2	.172*
Heterosexual	82.4	45.0	34.7	20.3
LGBTQ+	17.6	32.1	28.5	39.4
Race and ethnicity					6,728.27*	10	.058*
Hispanic	51.6	44.8	33.	21.2
Asian	11.3	45.7	28.1	26.2
African American or Black	3.4	40.3	31.0	28.7
White	21.5	39.8	35.7	24.5
Mixed	10.0	36.9	34.6	28.5
Other	2.2	39.6	33.0	27.3

p < .001.

The schools in this sample had an average of 926 students (SD = 709.38), with an average of about 60% (SD = 25.80) of students eligible for FRPM. The largest proportion of students was Hispanic (M = 53.63, SD = 25.99), followed by White (M = 20.81, SD = 19.93). The smallest proportions were African American or Black (M = 3.86, SD = 5.84) and other race or ethnicity (M = 2.37, SD = 2.71). School-level mean scores were 2.91 (SD = 0.17) for teacher–student relationships, 2.01 (SD = 0.17) for student participation, 3.93 (SD = 0.18) for academic engagement, 3.57 (SD = 0.25) for student belonging and attachment, and 3.52 (SD = 0.26) for parental involvement. These results are detailed in the appendix available on the journal website.

Data Analyses

First, a bivariate analysis was used to examine differences in SH and SI by students’ demographic characteristics and victimization (chi-square tests) and school-level attributes. To examine school-level correlates with the outcome variables, we computed Pearson correlations. Second, multilevel binary logistic models (with students nested in schools) were estimated to predict SH and SI. For each outcome, we estimated a null model, main effect model, and cross-level interaction model between student victimization and school climate.

Results

Bivariate Results: Student Level

Overall, 31.8% of students (n = 339,002) reported feeling SH, and 16.1% (n = 137,680) reported having SI. Results of chi-square tests reveal that students who reported bias-based victimization and non-bias-based victimization were significantly more likely to report SH (53.8% and 36.4%, respectively) compared to nonvictims (15.7%), χ²(2) = 108,788.70, p < .001; Cramér’s V = .328, p < .001. Similar associations emerged between victimization and SI: Bias-based and non-bias-based victims were significantly more likely to report SI (32.0% and 17.6%, respectively) compared to nonvictim students (6.6%), χ²(2) = 60,026.16, p < .001; Cramér’s V = .271, p < .001.

Bivariate Results: School Level

Table 2 presents correlations between school attributes and the percentage of students in each school reporting SH and SI. A higher proportion of students eligible for FRPM was significantly correlated with higher rates of SH but not SI. The racial and ethnic composition of schools was weakly correlated with the proportion of students reporting SH or SI, excluding the proportion of students of other races and ethnicities, which was significantly correlated with higher school-level reports of SI. All school climate subscales exhibited significant negative associations with school-level SH, with the strongest correlations identified for academic engagement and belonging and attachment. Similar results were noted for SI, with the most pronounced association with school climate found with the academic engagement and belonging and attachment subscales.

Table 2

School-Level Correlations Between School Attributes and Percentage of Students Reporting SH and SI

	SH	SI
	(n = 2,570)	(n = 2,056)
Proportion, M (SD)	31.50 (7.89)	16.53 (5.95)
Structural
School size	–.021	–.139*
Compositional
FRPM	.220*	–.028
Race and ethnicity
Hispanic	.114*	–.193*
Asian	–.178*	–.045*
African American or Black	.034	–.001
White	–.028	.195*
Mixed	–.110*	.184*
Other	.118*	.248*
Practices and processes: school climate
Teacher–student relationships	–.274*	–.058*
Participation	–.367*	–.158*
Academic engagement	–.503*	–.335*
Belonging and attachment	–.434*	–.209*
Parental involvement	–.342*	–.163*

Note. Values are correlations unless otherwise noted. FRPM = free or reduced-price meal; SH = sadness and hopelessness; SI = suicidal ideation.

p < .001.

Multivariate Results

Table 3 shows the results of multilevel logistic models with random intercepts and fixed slopes predicting SH and SI. Coefficients of fixed effects (with robust standard errors) and odds ratios (ORs) are displayed for the predictors. For each outcome variable, we present null, main effect, and cross-level interaction models.

Table 3

Multilevel Logistic Regression Models for Prediction of SH and SI

	SH			SI
	b (SE)	OR	CI	β (SE)	OR	CI
Null model
Intercept (fixed effect)	–0.79* (0.00)	0.45	[0.44, 0.46]	–1.66* (0.00)	0.19	[0.18, 0.19]
Intercept variance component (random effect)	0.10* (0.00)			0.10* (0.00)
Main effect model
Level 1: student characteristics
Intercept	–2.03* (0.02)	0.13	[0.12, 0.13]	–2.71* (0.04)	0.06	[0.06, 0.07]
Grade level
Grade 9	0.39* (0.01)	1.48	[1.44, 1.52]	0.22* (0.01)	1.25	[1.20, 1.29]
Grade 11	0.68* (0.01)	1.98	[1.93, 2.03]	0.32* (0.01)	1.38	[1.33, 1.43]
Gender (male)	–0.66* (0.00)	0.51	[0.50, 0.51]	–0.44* (0.00)	0.63	[0.62, 0.64]
Sexual orientation (LGBTQ+)	0.73* (0.00)	2.09	[2.06, 2.11]	0.99* (0.00)	2.70	[2.66, 2.74]
Race and ethnicity
Hispanic	0.21* (0.00)	1.18	[1.16, 1.19]	0.09* (0.00)	1.10	[1.08, 1.12
Asian	0.01 (0.00)	0.96	[0.94, 0.98]	0.11* (0.01)	1.12	[1.09, 1.15]
African American or Black	–0.20* (0.01)	0.77	[0.75, 0.79]	–0.07* (0.02)	0.92	[0.88, 0.96]
Mixed	0.13* (0.00)	1.14	[1.12, 1.17]	0.20* (0.01)	1.22	[1.19, 1.25]
Other	0.18* (0.01)	1.20	[1.15, 1.24]	0.22* (0.02)	1.24	[1.19, 1.30]
Victimization
Non-bias-based	1.17* (0.00)	3.23	[3.19, 3.27]	1.09* (0.00)	2.97	[2.92, 3.03]
Bias-based	1.83* (0.00)	6.26	[6.18, 6.34]	1.74* (0.00)	5.72	[5.62, 5.82]
Level 2: School attributes
Intercept	0.04 (0.00)			0.06 (0.00)
School size	–0.00	0.99	[0.99, 0.99]	–0.00* (0.00)	0.99	[0.99, 0.99]
Proportion receiving FRPM
Mid-poverty schools	0.19* (0.01)	1.21	[1.17, 1.25]	0.15* (0.02)	1.16	[1.10, 1.23]
High-poverty schools	0.23* (0.02)	1.26	[1.20, 1.31]	0.13* (0.03)	1.14	[0.67, 0.77]
Proportion of non-White students	–0.00* (0.00)	0.99	[0.99, 1.00]	–0.00* (0.00)	0.99	[0.99, 0.99]
School climate	–0.30* (0.02)	0.73	[0.70, 0.77]	–0.31* (0.03)	0.72	[0.67, 0.77]
Interaction model
Level 1: student characteristics
Intercept	–2.03* (0.02)	0.13	[0.12, 0.13]	–2.72* (0.04)	0.06	[0.06, 0.07]
Victimization
Non-bias-based	1.18* 0(.00)	3.26	[3.22, 3.30]	1.09* (0.00)	2.99	[2.93, 3.04]
Bias-based	1.84* (0.00)	6.32	[6.24, 6.41]	1.75* (0.00)	5.78	[5.67, 5.89]
Level 2: School attributes
Intercept	0.04 (0.00)	0.04	[0.03, 0.04]	0.06 (0.00)	0.06	[0.05, 0.07]
School climate	–0.50* (0.02)	0.60	[0.56, 0.63]	–0.40* (0.04)	0.66	[0.61, 0.73]
Cross-level interactions
Non-Bias-Based × Climate	0.14* (0.02)	1.15	[1.09, 1.21]	0.05 (0.04)	1.05	[0.97, 1.14]
Bias-Based × Climate	0.46* (0.02)	1.59	[1.51, 1.68]	0.16* (0.04)	1.17	[1.08, 1.27]

Note. Reference categories were Grade 7 for grade level, White for race and ethnicity, nonvictim for victimization, and low-poverty schools for proportion receiving FRPM. OR = odds ratio; CI = confidence interval; FRPM = free or reduced-price meal; SH = sadness and hopelessness; SI = suicidal ideation.

p < .001.

First, we employed null models to examine school-level variance in the dependent variables. The variance component for the random intercept (.10 for both outcomes) was statistically significant, indicating significant variance among schools in the probability of reporting SH and SI. In addition, the likelihood ratio test for the null hypothesis (that the mixed effects and fixed logistic regression models would provide the same goodness of fit) was significant (p < .001) for both estimations; thus, the mixed effects (multilevel) model was appropriate. Next, we included student-level and school-level predictors. Finally, controlling the entire set of predictors, we incorporated cross-level interaction terms between student victimization and school climate. To avoid cumbersome and unnecessary repetition, for the interaction models, we present only the coefficients, standard errors, and ORs for student victimization, school climate, and the two interaction terms. The ORs of other predictors remained identical to those of the main effect models. Because school-level ethnicity and race showed weak and mixed associations with the study outcome variables, the multivariate analyses incorporated a dichotomous school-level variable indicating the percentage of non-White students.

Prediction of SH

According to the main effect model, grade level was associated with SH. Compared to seventh grade, students in ninth (OR = 1.48) or 11th (OR = 1.99) grades were more likely to report SH. Male students were far less likely to report SH than female students (OR = 0.51). LGBTQ+ students were more likely than heterosexual students to report SH (OR = 2.08). Race and ethnicity showed mixed patterns. Compared to White students, those in the Hispanic (OR = 1.24), mixed (OR = 1.15), and other (OR = 1.20) categories were more likely to report SH, whereas African American and Black students were less likely to report SH (OR = 0.81).

Student victimization greatly increased the likelihood of experiencing SH. Compared to nonvictims, both non-bias-based victims (OR = 3.23) and bias-based victims (OR = 6.26) were far more likely to report SH. Bias-based victims were more likely to report SH than non-bias-based victims (OR = 2.02; not shown in Table 3).

Students in larger schools were less likely to report SH (OR = 0.99). Compared to those in low-poverty schools, students in mid-poverty schools (OR = 1.21) and high-poverty schools were more likely to report SH (OR = 1.26). School climate showed a strong association with SH: Students in schools with more positive climate were far less likely to report SH (OR = 0.73).

The cross-level interaction model revealed that whereas school climate decreased the probability of SH (OR = 0.60) for nonvictims, this reduction was of a lesser extent among both non-bias-based victims (OR = 1.15) and bias-based victims (OR = 1.60). To illustrate the interaction between victimization and school climate in predicting SH, we calculated the marginal predicted means of SH for the three victimization groups across five values of school climate (–2 SD to +2 SD; Figure 1). Although the risk of SH decreased among the three victimization groups as school climate increased, this reduction differed by group. Compared to the slope of the nonvictims’ group (i.e., reference category), the slopes of non-bias-based victims and bias-based victims were more moderate. That is, the reduction in the probability of SH due to an increase in school climate was larger among nonvictims (6.4 percentage points difference between schools with –2 SD and +2 SD climate scores, equivalent to a 32.6% reduction in the probability of SH) compared to both non-bias-based victims (7.7 percentage points difference, 18.8% reduction in probability of SH) and bias-based victims (0.9 percentage points difference, 1.7% reduction in probability of SH). Thus, bias-based victims derived the least benefit from a positive school climate.

FIGURE 1.

Marginal predicted means of sadness and hopelessness (SH) by victimization across levels of school climate (M, +2 SD, –2 SD).

Prediction of SI

According to the main effect model, grade level was associated with SI. Compared to seventh graders, students in ninth (OR = 1.25) or 11th (OR = 1.38) grades were more likely to report SI. Male students were far less likely to report SI than female students (OR = 0.63). LGBTQ+ students were far more likely than heterosexual students to report SI (OR = 2.70). Compared to White students, those in the Hispanic (OR = 1.10), Asian (OR = 1.12), mixed (OR = 1.22), and other (OR = 1.24) categories were more likely to report SI. African American and Black students were less likely than White students to report SI (OR = 0.92).

Student victimization greatly increased the probability of SI. Compared to nonvictims, non-bias-based victims (OR = 2.97) and bias-based victims (OR = 5.72) were far more likely to report SI. School size and ethnic composition, measured by the percentage of non-White students, were associated with a decreased risk of SI (OR = 0.99 for each). Compared to those in low-poverty schools, students in mid-poverty (OR = 1.16) and high-poverty (OR = 1.14) schools were more likely to report SI. School climate showed strong association with SI. Students in schools with higher socioemotional climate were far less likely to report SI (OR = 0.72).

The cross-level interaction model revealed that the effect of school climate on SI varied between nonvictims and bias-based victims. No such interaction effect was found with respect to non-bias-based victimization. Whereas school climate decreased the probability of SI (OR = 0.66 for nonvictims), this reduction was smaller among bias-based victims (OR = 1.18). Similarly, to illustrate the interaction effect, we calculated the marginal predicted means of SI for the three victimization groups across five values of school climates (Figure 2). Compared to the slope of nonvictim students, the slope of bias-based victims was less steep. That is, the reduction in the probability of SI due to an increase in school climate was larger among nonvictim students (2.3 percentage points difference between schools with –2 SD and +2 SD climate scores, equivalent to a 27.38% reduction in the probability of SI) compared to bias-based victims (4.1 percentage points difference, equivalent to 13.18% reduction in probability of SI). Thus, bias-based victims derived the least benefit from a positive school climate.

FIGURE 2.

Marginal predicted means of suicidal ideation (SI) by victimization across levels of school climate (M, +2 SD, –2 SD).

Discussion

Our data indicated that the largest proportion of the sampled students were nonvictims, whereas a smaller group—approximately one third—experienced non-bias-based victimization. Resembling earlier national reports (e.g., Turner et al., 2017), the smallest group, accounting for about one quarter of the sample, experienced bias-based victimization. Although bias-based victims represented the smallest group, the potential for bias-based victimization to be particularly devastating underscores the unique vulnerability of these students. Indeed, consistent with prior research, our findings confirm that experiencing victimization, particularly bias-based victimization, was associated with increased SH and SI (Newman et al., 2023; Russell et al., 2012).

In line with prior research (Benbenishty et al., 2018; Brière et al., 2013), the current results further indicate that the structural effect of school size was not a significant predictor of SH and SI. Yet compositional school effects of students’ poverty level, as measured based on the percentage of students eligible for FRPM, were significantly correlated with higher rates of reported SH (S. E. Jones et al., 2020) but not school-level SI. Because increased SH among youths has been correlated with elevated SI (Romanelli et al., 2022), the current results raise significant concerns regarding the protection of youth MH in impoverished conditions. There is a bidirectional causal relationship between poverty and mental distress (Ridley et al., 2020). Key factors contributing to youth poverty that could be targeted include financial hardship, neighborhood violence, exposure to crime, overcrowded housing, and housing instability (S. E. Jones et al., 2020). In accordance with prior findings (Patalay et al., 2020), the racial and ethnic composition of schools did not demonstrate a significant association with levels of SH and SI. Last, the current results align with a substantial body of evidence highlighting the direct contribution of positive school climate to youths’ desired MH (e.g., Aldridge & McChesney, 2018). Specifically, students’ academic engagement and sense of belonging and attachment to their school were most strongly associated with reduced SH and SI.

Given increasing emphasis on school as a crucial context for MH prevention and intervention efforts and considering that schools are ideally positioned to implement interventions that reach many youths, these results indicate schools should alter their climate to be more supportive as a strategy to decrease SH and SI among students. However, an examination of the moderating role of school climate in the relationship between victimization and SH and SI reveals a more complex picture. Whereas school climate decreased the probability of SH and SI for nonvictims, this effect was less pronounced among both non-bias-based and bias-based victims, with the latter deriving the least benefit from a positive school climate. These results confirm previous evidence (e.g., Zhu et al., 2024) suggesting that paradoxically, schools with healthier, more positive climates might have adverse effects on victims’ adjustment. In contrast to some studies suggesting that a positive school climate mitigates the negative effects of violence victimization on student depression and suicidality (Chan, 2022), healthier school contexts reduced SH and SI among nonvictims but did not provide similar benefits for victims of school violence. Unlike Clark and colleagues (2022), we found that victims of all types of violence derived less benefit from a positive school climate, particularly those subjected to more severe forms of bias-based violence, who gained the least advantage. Based on social comparison theory (Festinger, 1954), we postulate that a positive school climate can exacerbate feelings of inadequacy and isolation in victimized students by highlighting the discrepancy between their experiences and those of their nonvictimized peers, leading to negative self-evaluations and loneliness. In concordance with Salmivalli’s (2018) explanation of the healthy context paradox, it is likely that in schools with more positive climates, peers are more prone to blame victims for their circumstances and victims may engage in self-blaming attributions. This dynamic can lead to negative self-perceptions, ultimately resulting in detrimental effects on the victims’ MH. That is, in healthier contexts where fewer students typically experience victimization, it is easier to stand out and experience peer and self-blame, SH, and SI.

Study Implications

Given that a positive school climate can exacerbate the distress of victimized students, particularly those subjected to bias-based violence, it is essential to identify protective contextual factors that provide support. Schools should foster a peer culture where students stand up against school violence (Yun & Juvonen, 2020). Alongside implementing anti-violence interventions, it is also crucial to address prejudice and discrimination, which are fundamental contributors to bias-based violence. Furthermore, promoting a school climate that encourages tolerance and acceptance of all individuals regardless of their identities is vital.

The current scientific literature on interventions to combat bias-based violence emphasizes the necessity of a multitiered school approach that targets both individual and school-wide changes (e.g., Robinson et al., 2023). Students who exhibit higher levels of victimization, especially bias-based victimization, should be identified early for more comprehensive follow-up. Interventions should aim to change the perceived causes of behavior among victimized students and their peers by emphasizing that victims are not responsible for others’ negative behaviors (Salmivalli, 2018).

Encouraging students to engage in defending behaviors and providing increased support to those who have been victimized can significantly enhance their emotional well-being and improve their ability to cope with violence. When assessing the effectiveness of anti-violence interventions, merely reducing the overall prevalence of victimization is insufficient because this may obscure increases in suffering among those who continue to experience victimization despite the intervention (Salmivalli, 2018). Evaluations should monitor additional related effectiveness indicators beyond victimization levels, such as students’ experiences of SH and SI.

Study Limitations and Directions for Future Research

Although informative and based on a population sample of secondary students and schools in California, the current research has several limitations. It employed a cross-sectional design, limiting the ability to determine the directionality of associations among victimization, school climate, and student SH and SI. A longitudinal design could reveal whether students who continue to experience victimization or mental distress in positive school climates differ from those facing similar issues in negative contexts. Additionally, research has suggested that experiencing multiple forms of bias victimization is correlated with higher levels of negative outcomes (Joo et al., 2023); however, this study classified students as bias-based victims without considering the number of social identities targeted. Further research is needed to investigate how targeting based on one or multiple facets of social identity relates to students’ SH and SI and whether school climate affects these associations differently based on the number of identities involved. Last, the current results are consistent with some prior research (e.g., Zhu et al., 2024) yet inconsistent with others (e.g., Chan, 2022). These similarities and differences may be attributed to factors such as sample size and type, country, culture, time frame, or the comprehensiveness of ecological levels of schools. This study used national data from California, incorporating both individual- and school-level information. Further large-scale, multilevel studies that encompass most students in a country or state are necessary to investigate alignments and discrepancies.

Supplemental Material

sj-pdf-1-edr-10.3102_0013189X251393659 – Supplemental material for The Contribution of School Victimization to Sadness, Hopelessness, and Suicidal Ideation Among Bias-Based and Non-Bias-Based Victims and the Moderating Role of School Climate

Supplemental material, sj-pdf-1-edr-10.3102_0013189X251393659 for The Contribution of School Victimization to Sadness, Hopelessness, and Suicidal Ideation Among Bias-Based and Non-Bias-Based Victims and the Moderating Role of School Climate by Ruth Berkowitz, Netta Achdut, Rami Benbenishty and Ron Avi Astor in Educational Researcher

Footnotes

Acknowledgements

The authors wish to thank WestEd Science and Innovation for sharing the data used in this manuscript.

ORCID IDS

Ruth Berkowitz

Netta Achdut

Rami Benbenishty

Ron Avi Astor

Notes

Authors

RUTH BERKOWITZ, PhD, is a tenured senior lecturer at the School of Social Work, Faculty of Social Welfare & Health Sciences, University of Haifa, 199 Abba Khoushy Ave.Mount Carmel, Haifa, Israel; ruthberko@univ.haifa.ac.il . Her research focuses on school climate, school violence, bullying, parental school involvement, and social, emotional, and academic outcomes.

NETTA ACHDUT, PhD, is a tenured senior lecturer at the Spitzer Department of Social Work, Ben-Gurion University of the Negev, Israel, 1 David Ben-Gurion Boulevard, Be’er Sheva, Humanities and Social Sciences Building; nettaach@bgu.ac.il . Her research focuses on child welfare and on socioeconomic inequalities in health and well-being outcomes among children, adolescents, and young adults.

RAMI BENBENISHTY, PhD, is professor emeritus of social work, School of Social Work and Social Welfare, The Hebrew University of Jerusalem, Israel, and visiting professor at Universidad Andrés Bello, Santiago, Chile, Faculty Building of Social Work; rami.benbenishty@mail.huji.ac.il . His research focuses on the welfare and safety of children in the community and in- out-of-home care.

RON AVI ASTOR, PhD, is a professor at the Luskin School of Public Affairs-Social Welfare and School of Education and Information Studies, University of California, Los Angeles, 533 Luskin Building, School of Public Policy, Department of Social Welfare; rastoravi@ucla.edu . His research focuses on school safety across the globe.

References

Aldridge

J. M.

McChesney

(2018). The relationships between school climate and adolescent mental health and wellbeing: A systematic literature review. International Journal of Educational Research, 88, 121–145.

Astor

R. A.

Benbenishty

(2019). Bullying, school violence, and climate in evolving contexts: Culture, organization, and time. Oxford University Press.

Benbenishty

Astor

R. A.

Roziner

(2018). A school-based multilevel study of adolescent suicide ideation in California high schools. The Journal of Pediatrics, 196, 251–257.

Brière

F. N.

Pascal

Dupéré

Janosz

(2013). School environment and adolescent depressive symptoms: A multilevel longitudinal study. Pediatrics, 131(3), e702–e708.

Bronfenbrenner

Morris

P. A.

(1998). The ecology of developmental processes. In Damon

Lerner

R. M.

(Eds.), Handbook of child psychology Vol 1, theoretical models of human development (Vol. 1, pp. 993–1028). Wiley.

CalSCHLS. (n.d.). California school climate, health, and learning surveys. https://calschls.org/about/the-surveys/#chks

Centers for Disease Control and Prevention. (2023). Youth Risk Behavior Survey: Data summary and trends report 2011-2021. https://safesupportivelearning.ed.gov/sites/default/files/YRBS_Data-Summary-Trends_Report2023_508.pdf

Chan

R. C.

(2022). Gender nonconformity, peer victimization, and internalizing problems among youth: Differential moderating effects of school climate by sex assigned at birth. Psychology of Violence, 12(5), 347–360.

Clark

K. N.

Strissel

Demaray

M. K.

Ogg

Eldridge

M. A.

Malecki

C. K.

(2022). Victimization and depressive symptoms in early adolescence: The role of perceived school climate. School Psychology, 37(5), 355–366.

10.

Cohen

McCabe

E. M.

Michelli

N. M.

Pickeral

(2009). School climate: Research, policy, practice, and teacher education. Teachers College Record, 111(1), 180–213.

11.

DiStefano

Zhu

Mindrila

(2009). Understanding and using factor scores: Considerations for the applied researcher. Practical Assessment, Research & Evaluation, 14(1), Article 20. https://doi.org/10.7275/da8t-4g52

12.

DuPont-Reyes

M. J.

Villatoro

A. P.

(2019). The role of school race/ethnic composition in mental health outcomes: A systematic literature review. Journal of Adolescence, 74, 71–82.

13.

Espelage

D. L.

Valido

Hatchel

Ingram

K. M.

Huang

Torgal

(2019). A literature review of protective factors associated with homophobic bullying and its consequences among children & adolescents. Aggression and Violent Behavior, 45, 98–110.

14.

Festinger

(1954). The social comparison theory. Human Relations, 7, 117–140.

15.

Gutman

Feinstein

(2008). Children’s well-being in primary school: Pupil and school effects. Centre for Research on the Wider Benefits of Learning, Institute of Education, University of London.

16.

Haj-Yahia

M. M.

Abo Basal

(2024). Exposure to family violence and school violence among Palestinian high schoolers: The moderating effect of gender and school climate. Child Abuse & Neglect, 158, Article 107148. https://doi.org/10.1016/j.chiabu.2024.107148

17.

Irwin

Wang

Cui

Thompson

(2023). Report on indicators of school crime and safety: 2022. National Center for Education Statistics and Bureau of Justice Statistics. https://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2023092

18.

Jones

L. M.

Turner

H. A.

Mitchell

Cuevas

C. A.

Farrell

A. M.

(2019). A comprehensive measure of youth experiences with bias victimization: Findings from the Youth Bias Victimization Questionnaire (YBVQ). Crimes Against Children Research Center, University of New Hampshire.

19.

Jones

S. E.

Underwood

J. M.

Pampati

V. D.

DeGue

Demissie

Hocevar Adkins

Barrios

L. C.

(2020). School-level poverty and persistent feelings of sadness or hopelessness, suicidality, and experiences with violence victimization among public high school students. Journal of Health Care for the Poor and Underserved, 31(3), 1248–1263.

20.

Joo

Lee

Rodriguez

(2023). Trends in bias-based bullying from 2015 to 2019 and the associations among bias-based bullying, school avoidance, and supportive adults at school. Journal of School Violence, 22(4), 543–555.

21.

Marks

G. N.

(2010). What aspects of schooling are important? School effects on tertiary entrance performance. School Effectiveness and School Improvement, 21(3), 267–287.

22.

Newman

K. L.

Alexander

D. S.

Rovers

J. P.

(2023). Sadness, hopelessness and suicide attempts in bullying: Data from the 2018 Iowa youth survey. PLOS One, 18(2), Article e0281106. https://doi.org/10.1371/journal.pone.0281106

23.

Odigie

Elsden

Hosozawa

Patalay

Pingault

J. B.

(2025). The healthy context paradox: A cross-country analysis of the association between bullying victimisation and adolescent mental health. European Child & Adolescent Psychiatry, 34, 215–224. https://doi.org/10.1007/s00787-024-02483-x

24.

Pan

Malamut

Zhang

Salmivalli

(2021). Why does classroom-level victimization moderate the association between victimization and depressive symptoms? The “healthy context paradox” and two explanations. Child Development, 92(5), 1836–1854.

25.

Patalay

O’Neill

Deighton

Fink

(2020). School characteristics and children’s mental health: A linked survey-administrative data study. Preventive Medicine, 141, Article 106292. https://doi.org/10.1016/j.ypmed.2020.106292

26.

Raniti

Rakesh

Patton

G. C.

Sawyer

S. M.

(2022). The role of school connectedness in the prevention of youth depression and anxiety: A systematic review with youth consultation. BMC Public Health, 22, Article 2152. https://doi.org/10.1186/s12889-022-14364-6

27.

Ridley

Rao

Schilbach

Patel

(2020). Poverty, depression, and anxiety: Causal evidence and mechanisms. Science, 370(6522), Article eaay0214. https://doi.org/10.1126/science.aay0214

28.

Robinson

L. E.

Clements

Drescher

El Sheikh

Milarsky

T. K.

Hanebutt

Graves

Delgado

A. V.

Espelage

D. L.

Rose

C. A.

(2023). Developing a multi-tiered system of support-based plan for bullying prevention among students with disabilities: Perspectives from general and special education teachers during professional development. School Mental Health, 15, 826–838.

29.

Romanelli

Sheftall

A. H.

Irsheid

S. B.

Lindsey

M. A.

Grogan

T. M.

(2022). Factors associated with distinct patterns of suicidal thoughts, suicide plans, and suicide attempts among US adolescents. Prevention Science, 23(1), 73–84.

30.

Russell

S. T.

Sinclair

K. O.

Poteat

V. P.

Koenig

B. W.

(2012). Adolescent health and harassment based on discriminatory bias. American Journal of Public Health, 102(3), 493–495.

31.

Salmivalli

(2018). Peer victimization and adjustment in young adulthood: Commentary on the special section. Journal of Abnormal Child Psychology, 46, 67–72.

32.

Shamrova

Lee

J. M.

Choi

(2025). School bullying victimization and child subjective well-being in east Asian countries and territories: Role of children’s participation in decision-making in schools and community. Children & Society, 39(1), 1–18.

33.

Turner

H. A.

Mitchell

K. J.

Jones

Shattuck

(2017). Assessing the impact of harassment by peers: Incident characteristics and outcomes in a national sample of youth. Journal of School Violence, 16(1), 1–24.

34.

U.S. Surgeon General. (2021). Protecting youth mental health: The U.S. surgeon general’s advisory. https://www.hhs.gov/sites/default/files/surgeon-general-youth-mental-health-advisory.pdf

35.

Wang

La Salle

T. P.

K. A.

Sullivan

K. E.

(2019). Does parental involvement matter for students’ mental health in middle school? School Psychology, 34(2), 222–232.

36.

Wang

Xiong

Liu

(2023). How are family resources and school resources related to low-income adolescents’ psychological adjustment? The moderating role of belief in a just world. Child Indicators Research, 16(2), 655–670.

37.

Watt

T. T.

(2003). Are small schools and private schools better for adolescents’ emotional adjustment? Sociology of Education, 76(4), 344–367.

38.

World Health Organization. (2021). Mental health of adolescents. https://www.who.int/news-room/fact-sheets/detail/adolescent-mental-health

39.

Yang

Sharkey

J. D.

Reed

L. A.

Chen

Dowdy

(2018). Bullying victimization and student engagement in elementary, middle, and high schools: Moderating role of school climate. School Psychology Quarterly, 33(1), 54–64.

40.

Yun

H. Y.

Juvonen

(2020). Navigating the healthy context paradox: Identifying classroom characteristics that improve the psychological adjustment of bullying victims. Journal of Youth and Adolescence, 49(11), 2203–2213.

41.

Zhang

Jin

Cui

(2022). Do teacher autonomy support and teacher–student relationships influence students’ depression? A 3-year longitudinal study. School Mental Health, 14(1), 110–124.

42.

Zhu

Cheong

Wang

(2024). Relation between peer victimization and mental health among Chinese students: Perceived school climate and covitality as potential moderators. School Psychology Review, 53(5), 538–551.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.11 MB

The Contribution of School Victimization to Sadness,Hopelessness,and Suicidal Ideation Among Bias-Based and Non-Bias-Based Victims and the Moderating Role of School Climate

Abstract

Keywords

School Victimization

School Effects, SH, and SI

School Climate as a Moderator

Method

Data Sources

Measurements

Student level

School level

Student-level independent variables

Background characteristics (control variables)

School-level independent variables

Sample Characteristics: Student and School Levels

Data Analyses

Results

Bivariate Results: Student Level

Bivariate Results: School Level

Multivariate Results

Prediction of SH

Prediction of SI

Discussion

Study Implications

Study Limitations and Directions for Future Research

Supplemental Material

sj-pdf-1-edr-10.3102_0013189X251393659 – Supplemental material for The Contribution of School Victimization to Sadness, Hopelessness, and Suicidal Ideation Among Bias-Based and Non-Bias-Based Victims and the Moderating Role of School Climate

Footnotes

Acknowledgements

ORCID IDS

Notes

Authors

References

Supplementary Material