Pushed Out and Drawn In: Exclusionary Discipline, Mental Health, and Protective Factors among Youth in Public Schools

Abstract

Background.

Exclusionary discipline (ED) has long been an educational equity concern, but its relationship with student health and protective factors is less understood.

Methods.

Using population-based public school student data (N=82,216), we examined associations between past-month ED and positive depression and anxiety screening instrument results,. We also assessed whether each of nine potential protective factors moderated the ED-mental health relationship by testing interaction effects.

Results.

Over 1 in 10 youth experienced past-month ED, with variation by sex, gender identity, special education status, poverty, region, race/ethnicity, and adverse childhood experiences. Net of sociodemographic factors, youth who experienced ED had higher likelihood for current depression (Adjusted Odds Ratio [AOR]: 1.64, 95% CI: 1.55, 1.73) and anxiety (AOR: 1.49, 95% CI: 1.41, 1.58) symptoms. Significant associations were robust across five racial/ethnic groups, except for anxiety among American Indian/Alaska Native youth. Individual, interpersonal, and school-level protective factors appeared to mitigate depression and anxiety regardless of disciplinary experience.

Implications for School Health Policy, Practice, and Equity.

Our findings document ED disproportionality and possible ramifications for emotional well-being.

Conclusions.

In concert with structural efforts to reduce reliance on ED, strategies that bolster protective factors may support youth already impacted by ED and/or mental health problems.

Schools in the United States (U.S.) are tasked with educating students within safe and nurturing learning environments. Following the advent of “zero tolerance” policies, educational systems have used exclusionary discipline (ED) practices such as detentions, suspensions, and expulsions in attempt to address undesired student behavior.^1,2 From 2017–2018, approximately 2.7 million children from pre-Kindergarten to 12^th grade were suspended or expelled, with significant differences by race/ethnicity and disability status. Specifically, Black and American Indian/Alaska Native (AI/AN) youth were estimated to be excluded at three times the rate of non-Hispanic White youth; students with disabilities were excluded at twice the rate of non-disabled peers.^3,4 Importantly, these disproportionate rates emerge as early as preschool,⁵ and persist even after accounting for the degree of students’ behavioral infraction.^6,7

Despite the scale and inequity of these practices, there remains limited evidence to suggest that these actions yield any improvements in student academic performance or behavior, school safety, or teacher-student relationships.⁸ Instead, youth who get suspended or expelled lose learning time and are longitudinally more likely to disengage from or drop out of school, be involved with the criminal legal system, and become unemployed.^9–13 These concerning associations have prompted some to describe this phenomenon as the “school to prison pipeline”¹⁴ and a persistent arm of structural racism.^1,15,16

Although ED has been highlighted as an educational system issue for decades, its place as a public health and healthcare matter has been raised more recently. The American Academy of Pediatrics’ 2013 policy statement highlighted suspensions and expulsions as an issue for child healthcare professionals to understand and address.¹¹ As the statement attests, greater collaboration across the education and health sectors stands to enhance the lives of young people affected by ED by assuring student medical, psychosocial, and educational needs are identified and intervened upon in an inclusive, supportive manner. Despite this call, intersectoral action to address and curtail the use of ED has largely remained relegated to education stakeholders, and research studies on the health of affected youth remain nascent.¹ A 2021 study of California schools¹⁶ documented longitudinal associations between school-level rates of student substance use, violence, and perceptions of safety/support with subsequent school discipline and police contact rates. Similarly, data on urban U.S. families showed that children’s rates of early adverse childhood experiences (ACEs) predicted later school suspension in adolescence¹⁷ and that Black children evidenced greater depressive symptoms when exposed to more punitive school environments.¹⁵

Collectively, these studies raise the possibility that ED constitutes a mismatched response to elevated student needs, largely attributable to individual and contextual risks. Additional population-based investigation leveraging large, representative samples of individual youth can add to our understanding of ED’s impacts on families and communities. Given the racial/ethnic and disability-based disparities outlined above, it would be valuable to carry out analyses with sufficient subsample sizes to permit inspection across these characteristics. In the current study, we sought to characterize the relationship between ED and emotional health among public school youth. Further, to elucidate other, less punitive points of intervention, we aimed to explore the potential role of psychosocial protective factors in attenuating the ED-mental health relationship.

Methods

Participants and Procedure

The current study draws on the 2019 Minnesota Student Survey (MSS), a triennial school-based survey of students in 5^th, 8^th, 9^th, and 11^th grades. The anonymous survey addresses various topics concerning adolescents’ health, lives, and contexts. School districts throughout the state were invited to participate, with the option to opt-out given at the district, school, parent, and student levels. In 2019, 81% of eligible schools agreed to participate.¹⁸ Although letters sent home to families were written in English, Spanish, Hmong, and Somali, students completed the survey in English using a web-based format. Students with disabilities were afforded accommodations to support participation, such as screen readers for those with visual impairment. 4% of surveys indicating inconsistent, improbable, or exaggerated responses were removed to support validity of responses.¹⁸

Instrumentation

Exclusionary discipline.

We defined students’ experience with ED based on three items, and used a dichotomous approach based on literature showing both in-school (e.g., detention) and out-of-school (e.g., suspension) exclusionary practices predict adverse outcomes.¹⁹ Students were asked to report the last-month frequency of being (1) “sent out of the classroom for discipline” and (2) missing a full or partial day of school (not including school-sponsored activities). Those who missed ≥1 partial or full day of school were also asked to indicate the reason(s). Students who were sent out of the classroom for discipline or missed a part or full day of school due to being “suspended from school” at least once in the last 30 days were classified as having experienced past month ED; those who did not endorse either were classified as not having had ED.

Depression and anxiety symptoms.

For depression symptoms, we used the Patient Health Questionnaire-2 (PHQ-2), which asked youth how often they had been bothered by “little interest or pleasure in doing things” and “feeling down, depressed, or hopeless” in the last two weeks. Youth indicated responses on a 0 (not at all) to 3 (nearly every day) scale for each item, and these scores were combined to obtain an overall sum. Consistent with the PHQ-2’s clinical cutoff and prior validation work,^20,21 we set a cutoff of ≥3 to balance sensitivity and specificity. For anxiety, we leveraged the Generalized Anxiety Disorder scale-2 (GAD-2) that queried youth how often they had been bothered by “feeling nervous, anxious or on edge” or “not being able to stop or control worrying” in the last two weeks. Similar to the depression scale and other MSS studies,^22,23 a cutoff of ≥3 was used to classify adolescents screening positive for anxiety symptoms.

Protective factors.

Several constructs included in the MSS have been previously shown to hold potential for mitigating biologic and socio-emotional derangements implicated in mental health problems.^23–26 Defined here as “protective factors”, we examined nine binary psychosocial factors across individual, interpersonal, and school levels for use in moderation analyses as below. Individual-level factors included extracurricular youth development opportunities, positive identity, and social competency. Interpersonal-level factors included parent connectedness, friends caring, and other adults in the community caring. School-level factors included having an adult at school to talk to, teacher-student relationship quality, and feeling safe at school. These factors reflected both single survey items in some cases and sum scales in others (detailed in Appendix S1). Such variables were dichotomized in accordance with prior MSS studies wherever possible.^24,27,28

Sociodemographic factors.

We included survey items for various sociodemographic characteristics classified accordingly with prior work,^24,26,28 including biological sex and transgender/gender non-binary identity. Poverty status was determined based on students who responded affirmatively to either receiving free/reduced-price lunch or to having skipped meals due to family not having enough money. Students’ region was classified into either living in the seven counties constituting the Twin Cities metropolitan area or any other county in the state (i.e., Greater Minnesota); the latter category is largely comprised of non-urban areas. For race/ethnicity,¹ we leveraged four survey items to create five mutually exclusive groups: AI/AN, Black, Hispanic/Latinx, White, and Other (inclusive of Asian, Multiracial, and Pacific Islander students). We developed these racial/ethnic categories based on prior literature on which groups are disparately impacted by ED.^3,4,15,19 Special education status was based on endorsement of having “[received] special education services as part of an individual education plan” (IEP). Finally, youth were asked about their history of encountering nine distinct ACEs; each affirmative response was scored with one point to create an overall ACEs score (see Table 1).

Table 1.

Sociodemographic characteristics by exclusionary discipline in a statewide sample of public school youth in grades 8, 9, 11 – Minnesota, 2019

Characteristic	Overall Sample	No Past Month ED	Past Month ED
N (%)	82216	72641 (88.4)	9575 (11.6)
Biological Sex, N (%)
Female	43463 (52.9)	40422 (55.6)	3041 (31.8)
Male	38631 (47.0)	32113 (44.2)	6518 (68.1)
Missing	122	106	16
Transgender/Gender Non-binary, N (%)
No	76376 (93.1)	67763 (93.5)	8613 (90.3)
Yes	1193 (1.5)	997 (1.4)	196 (2.1)
Not sure about gender identity	1365 (1.7)	1154 (1.6)	211 (2.2)
Not sure what question means	3079 (3.8)	2559 (3.5)	520 (5.5)
Missing	203	168	35
Age
Mean (SD) [NA]	14.8 (1.32) [66]	14.9 (1.3) [49]	14.3 (1.1) [17]
Median [Range]	14.0 [12.0, 19.0]	15.0 [12.0, 19.0]	14.0 [12.0, 19.0]
Poverty a, N (%)
No	51497 (70.5)	46898 (72.5)	4599 (55.5)
Yes	21518 (29.5)	17833 (27.5)	3685 (44.5)
Missing	9201	7910	1291
Region, N (%)
Twin Cities Metropolitan Area	43008 (52.3)	37898 (52.2)	5110 (53.4)
Greater Minnesota	39208 (47.7)	34743 (47.8)	4465 (46.6)
Missing	0	0	0
Race/Ethnicity, N (%)
American Indian/Alaska Native, NH	1109 (1.4)	843 (1.2)	266 (2.8)
Black, NH	5490 (6.7)	4333 (6.0)	1157 (12.2)
Hispanic/Latinx	4986 (6.1)	4235 (5.9)	751 (7.9)
Other, NHb	12448 (15.2)	10843 (15.0)	1605 (16.9)
White, NH	57690 (70.6)	51985 (72.0)	5705 (60.2)
Missing	493	402	91
Special Education and/or IEP, N (%)
No	58565 (71.5)	52707 (72.9)	5858 (61.6)
Yes	7894 (9.6)	6267 (8.7)	1627 (17.1)
Not sure	15397 (18.8)	13369 (18.5)	2028 (21.3)
Missing	360	298	62
ACEs, N (%)
0–1	56098 (72.6)	51057 (74.4)	5041 (58.3)
2 or more	21137 (27.4)	17533 (25.6)	3604 (41.7)
Missing	4981	4051	930

Notes. ED: Exclusionary discipline, ACEs: Adverse childhood experiences. NH: Non-Hispanic. IEP: Individualized Education Plan.

^aYouth were classified as living in poverty if they reported receiving of free/reduced price lunch or that they skipped meals due to family not having enough money.

^bIncludes youth who reported they were Asian, Multiracial, Native Hawaiian/Pacific Islander or Other.

^cAdverse childhood experiences included youth reports of: living with anyone who drinks too much alcohol; living with anyone who uses illegal drugs or abuses prescription drugs; living with anyone who is depressed or had mental health issues; living with a parent or other adult who swears at, insults, or puts down the youth; living with a parent or other adult who has ever hit, beat, kicked, or physically hurt the youth; living with a parent or other adult who has ever slapped, hit, kicked, or beat each other up; anyone who was not a family member/relative that pressured, tricked, or forced the youth to do something sexual or done something sexual to the youth against the youth’s wishes; anyone who was a family member/relative that pressured, tricked, or forced the youth to do something sexual or done something sexual to the youth against the youth’s wishes; and ever trading sex or sexual activity for money, food, alcohol, drugs, a place to stay, or anything else.

Data Analysis

We restricted our sample to 8th, 9th, and 11th graders in Minnesota public schools that had valid responses for our dependent and independent variables; 5^th graders were not included as they were not asked about depression or anxiety symptoms. We calculated descriptive statistics and constructed a series of multiple logistic regression models to examine associations between ED and depression and anxiety symptoms. Models were carried out for overall average effects as well as for the five separate racial/ethnic groups noted above, each adjusted for child age, sex, region, and poverty. To assess whether protective factors might moderate the ED-mental health relationship, we included individual interaction terms for the nine protective factors in separate depression and anxiety models among the overall sample.

Missingness for our variables ranged from 0.1% (biological sex) to 11.2% (poverty; Table 1). We ran a total of 30 models, all of which were two-tailed and conducted using R (Version 4.0.2, Vienna, Austria) with a significance level set at 0.05. Adjustments for multiple comparisons were not made, as we were interested in each outcome separately and within pre-specified sub-groups. The University of Minnesota Institutional Review Board determined this study exempt from human subjects review.

Results

Descriptive Statistics

In the overall sample (N=82,216, mean age of 14.8 years (SD: 1.32)), 11.6% of youth had experienced past-month ED. The sample was racially/ethnically diverse, and roughly evenly divided between males and females and between those living in the Twin Cities metropolitan area vs. Greater Minnesota. Nearly 1/3 of students were living in poverty, and approximately 10% received special education services or had an IEP. Twenty-seven percent had experienced 2 or more ACEs (Table 1). There were differences in age, poverty, race/ethnicity, and special education status for youth with missing vs. non-missing ED responses (supplemental findings not shown).

We observed that among those with past month ED, students who were male; had non-binary or unsure gender; living in poverty; living in the Twin Cities metropolitan area; receiving or unsure about receiving special education services; and endorsing 2+ ACEs were more represented compared to their proportion of the overall sample. There were also more youth belonging to any non-White ethnic group (AI/AN, Black, Hispanic/Latinx, Other) among those who experienced ED relative to their population share. We observed a higher prevalence of positive depression symptom screens among youth who had experienced ED (30.7%) compared to those who had not (23.5%). Similarly, anxiety symptoms were more common among youth with past month ED (30.5%) compared to those without (27.6%).

Main Effects in Overall Sample

Overall multiple regression models indicated that ED was significantly associated with greater odds for positive depression (AOR: 1.64, 95% CI: 1.55, 1.73) and anxiety (AOR: 1.49, 95% CI: 1.41, 1.58) symptom screens, net of aforementioned sociodemographic factors (Table 2). These models also indicated that females and those in poverty or special education had a higher likelihood for both depression and anxiety symptoms. With each year of increasing age, odds for depression and anxiety symptoms increased by 7% (95% CI: 1.06, 1.09) and 9% (95% CI: 1.08, 1.11), respectively.

Table 2.

Adjusted odds ratios from logistic regressions predicting depression and anxiety symptoms from exclusionary discipline in a statewide sample of public school youth in grades 8, 9, 11 – Minnesota, 2019

	AORa	95% CI	p-value
Depression
Exclusionary Discipline: Yes	1.64	1.55, 1.73	0.00
Age (years)	1.07	1.06, 1.09	0.00
Biological Sex: Female	2.09	2.02, 2.17	0.00
Region: Greater Minnesota	0.99	0.96, 1.03	0.64
Poverty: Yes	1.71	1.64, 1.77	0.00
Special Education: Not Sure	1.16	1.11, 1.21	0.00
Special Education: Yes	1.38	1.30, 1.47	0.00
Anxiety
Exclusionary Discipline: Yes	1.49	1.41, 1.58	0.00
Age (years)	1.09	1.08, 1.11	0.00
Biological Sex: Female	3.19	3.01, 3.23	0.00
Region: Greater Minnesota	0.97	0.94, 1.00	0.05
Poverty: Yes	1.41	1.36, 1.46	0.00
Special Education: Not Sure	1.16	1.11, 1.22	0.00
Special Education: Yes	1.43	1.35, 1.52	0.00

Notes. AOR: Adjusted Odds Ratio, 95% CI: 95% Confidence Interval.

^aAdjusted for age, biological sex, region, poverty, and special education status.

Main Effects by Race/Ethnicity

In race/ethnicity-stratified multiple regression analyses, ED remained associated with higher odds of depression symptoms within the five groups, to slightly varying magnitudes (Figure 1). Significant ED-depression AORs ranged from 1.48 (95% CI: 1.25, 1.74; Black youth) to 1.72 (95% CI: 1.61, 1.85; White youth). For anxiety, the same patterns generally held for individual subgroups except among AI/AN youth, for whom a significant relationship was not observed (Figure 2). Significant ED-anxiety AORs ranged from 1.32 (95% CI: 1.11, 1.57; Black youth) to 1.59 (95% CI: 1.48, 1.70; White youth).

Figure 1.

Adjusted odds ratios from logistic regression predicting depression symptoms from exclusionary discipline in a statewide sample of public school youth in grades 8, 9, 11, overall and by race/ethnicity – Minnesota, 2019

Figure 2.

Adjusted odds ratios from logistic regression predicting anxiety symptoms from exclusionary discipline in a statewide sample of public school youth in grades 8, 9, 11, overall and by race/ethnicity – Minnesota, 2019

Racially/ethnically stratified models showed some deviations from findings in the overall sample (Figure 1). For example, whereas region was not associated with mental health in the overall sample, White youth had greater odds for depression or anxiety symptoms among those in the Twin Cities metropolitan area relative to those in Greater Minnesota. Similarly, the overall sample showed that youth in special education had greater likelihood for depression symptoms compared to those not in special education, but stratified models demonstrated this association did not generally hold for AI/AN, Black, and Hispanic/Latinx students. In other cases, effect sizes remained significant across race/ethnicity but the strength of association within groups varied. For instance, the sex-based discrepancy in odds for mental health symptoms was more pronounced among AI/AN youth – AI/AN females were 3 and 4 times as likely to experience depression and anxiety compared to males.

Interaction Effects in Overall Sample

We observed that numerous individual, interpersonal, and school-level factors were associated with lower likelihood for depression and anxiety symptoms – both for youth with and without ED. Further, interaction terms between ED and extracurricular youth development opportunities, positive identity, social competency, parent connectedness, friends caring, other adults in community caring, adults at school to talk to, teacher-student relationship quality, and feeling safe at school were all significantly associated with depressive symptoms. In general, the strength of relationship between ED and depression symptoms was slightly less pronounced for youth with a given protective factor, compared to youth without that characteristic (Table 3, Appendix S2). Most protective factors also evidenced significant interactions for anxiety, except for social competency and having an adult at school to talk to (Table 3, Appendix S3).

Table 3.

Adjusted ratios of odds ratios for interaction between exclusionary discipline with select protective factors on depression and anxiety symptoms in a statewide sample of public school youth in grades 8, 9, 11 – Minnesota, 2019

	Depression			Anxiety
	RORa	95% CI	p-value	RORa	95% CI	p-value
Individual
Extracurricular activities	0.83	0.75, 0.93	0.00	0.89	0.80, 1.00	0.04
Positive identity	0.57	0.50, 0.65	0.00	0.69	0.60, 0.79	0.00
Social competency	0.78	0.69, 0.88	0.00	0.90	0.80, 1.01	0.08
Interpersonal
Parent connectedness	0.75	0.67, 0.83	0.00	0.87	0.78, 0.97	0.01
Friends care	0.75	0.67, 0.83	0.00	0.79	0.71, 0.88	0.00
Other adults care	0.68	0.60, 0.78	0.00	0.71	0.62, 0.80	0.00
School
Adult at school to talk to	0.79	0.70, 0.89	0.00	0.90	0.80, 1.02	0.10
Teacher-student relationship quality	0.78	0.69, 0.88	0.00	0.87	0.77, 0.98	0.02
Safe at school	0.78	0.69, 0.89	0.00	0.80	0.70, 0.91	0.00

Notes. ROR: Ratio of Odds Ratio, 95% CI: 95% Confidence Interval.

^aAdjusted for age, biological sex, region, poverty, and special education status

Discussion

Adding to growing evidence on the potential harms of ED for youth and families,^9–13 our investigation documented associations between past-month ED and current internalizing mental health symptoms. Although findings reflect youth’s recent experiences, other work suggests that emotional problems can often persist beyond initial ED exposures,^15,16 consistent with the notion that educational exclusion fosters risk for a cascade of adverse and mutually reinforcing consequences, such as increased anger and apathy towards education leading to school disengagement²⁹ or disruption of social cohesion with subsequent isolation.³⁰

We observed higher rates of ED for males, youth of color, and those in special education as highlighted elsewhere.^3,4 Similarly, there were higher ED rates among youth in metropolitan areas, those meeting criteria for living in poverty, nonbinary gender identity, and ACEs, contributing to the general picture that youth affected by ED are more often those with pre-existing disadvantages.^31,32 Significant relationships held even accounting for demographic factors that may have explained the putative relationship, and also generally persisted within individual racial/ethnic groups. Thus, although socially disadvantaged students are disciplined with exclusion at disparate rates, ED practices may affect emotional health regardless of one’s background.

In addition, we saw that females had greater odds for depression and anxiety even within separate racial/ethnic groups and when holding other factors constant. Thus, although males more often experienced ED, females are more susceptible to poor mental health irrespective of ED exposure. Our study also had sufficient subsamples to examine understudied groups, such as AI/AN youth, for whom we observed higher rates of ED and correlations with depression, but not anxiety. Although ED disproportionality for AI/AN youth has been previously reported,^3,4 the ways in which disciplinary action affects this group’s mental health may be distinct, especially in light of the legacy of the federal Indian boarding school program. Overall, these findings speak to nuanced experiences with ED and mental health across sociodemographic groups, emphasizing the need for special attention to future intersectional analyses³³ and tailored psychosocial supports.

Interestingly, we found that the ED-mental health relationship depended on the presence of a given protective factor, such that individual protective factors correlated with reduced depressive/anxiety symptoms but to a lesser degree in ED-affected youth. This finding may point to a difference that reached statistical significance, potentially due to our large sample, but may not necessarily translate to clinical significance. For example, the likelihood for depression symptoms among youth with ED who had an adult at school to talk to was 5% lower than those without an adult at school to talk to; among youth without ED, this difference was 8%. As shown in Appendices S1–S2, lines were typically non-parallel with different slopes, but did not intersect, possibly reflecting “removable” interactions.³⁴ Ultimately, results imply that protective factors support the health of youth irrespective of disciplinary experience.

Implications for School Health Policy, Practice, and Equity

The findings have numerous implications. Caregivers, educators, and health care providers can pay attention to the emotional health of youth subjected to ED, particularly those with repeated suspensions/expulsions or who possess additional risk factors. Health care providers might also advocate for supportive approaches directly to schools, such as evaluation for special education services, for youth encountering behavioral challenges.¹¹ This approach could assist a subset of youth in receiving coordinated educational (e.g., learning accommodations) and/or clinical (e.g., psychotherapy) interventions that may have otherwise been under-utilized. School health professionals – such as school nurses, counselors, or school-based health center providers – occupy a unique position to address ED, owing to their physical location on school campuses and potential involvement in administering students’ Section 504 accommodation plans and/or IEPs.

Relatedly, given inverse associations between protective factors with depression and anxiety symptoms, efforts to bolster protective factors may promote emotional well-being irrespective of individual youth’s personal ED exposure. These actions can range from informal efforts to connect youth to resources to structured programs targeting these buffering protections. For instance, caregivers and educators can facilitate involvement in extracurricular activities or mentor relationships for youth with, or at-risk for, ED experience. They might also connect youth to evidence-based interventions addressing social competency (e.g., social skills group training³⁵) or parent-child connectedness (e.g., positive parenting programs³⁶), for example, where deficits are identified. Such individual-level efforts can help support emotional health in the near-term, particularly as structural changes to ED policies and practices may require significant investment of time and resources (see below). As well, because protective factors have also been shown to benefit youth with existing mental health problems,³⁷ our findings suggest they might also help such young people avoid ED and its sequelae.

Although the aforementioned strategies are important actions for educators and service providers, the greatest impact will likely be realized through systemic change. Our study corroborates other recommendations for institutions to reduce reliance on ED as an educational and health equity imperative.^1,38 Institutions can also explore and fund alternative disciplinary frameworks such as multi-tiered systems of supports and restorative practices, which may positively influence student and school-level outcomes inclusive of ED.^11,19,38 These models reimagine student misbehaviors as inappropriate individual choices towards understanding them as developmentally appropriate coping behaviors to adverse circumstances.³⁹ Structures governing school staff, such as school resource officer policies or teacher training, can also be modified to align with principles of child development and trauma-informed care. This is particularly crucial given staff’s role in determining the circumstances under which students get excluded,^30,40,41 as well as the current study’s findings that school-level factors such as teacher-student relationship quality correlate with mental health symptoms. Finally, incorporating both primary (e.g., universal student training on coping skills) and secondary prevention (e.g., school-based mental health services) approaches into schools can translate to improved developmental and behavioral health for students across levels of baseline need.^39,42

Limitations

Despite merits, the study has limitations. First, although we looked at ED using a broad, dichotomized approach to support adequate subsamples and inform public health actions, this approach may have ignored differences in subtypes of ED (e.g., in-school vs. out-of-school suspension). A binary determination also precluded us from examining dose-response effects of ED on mental health. Future studies can help clarify the extent to which different types and quantities of ED may matter for mental health, as implied by prior studies.^8,32 Second, the cross-sectional study design precluded our ability to tease apart temporality or causality. Prior literature points to a potential bidirectional relationship^16,43 in which ED both affects and reflects mental health. Future studies should probe these relationships further with attention to mechanistic targets for disruption of both ED and emotional problems. Relatedly, examined variables primarily reflected youth’s contemporary experiences (e.g., past month ED), potentially ignoring time-varying features of youth’s lives or introducing recency bias. Third, although differences found between youth with and without valid ED responses for certain demographic characteristics can suggest biased associations, this finding is likely not clinically meaningful. Given removal of invalid surveys noted above,¹⁸ other scholars have presumed 2019 MSS data are missing at random.⁴⁴ Ultimately, a non-response bias analysis, as done elsewhere,⁴⁵ may be warranted. Finally, we focused on youth in public schools to support broad generalizability of our findings, but additional study is needed to understand ED and mental health in other school contexts, including beyond the state of Minnesota.

Conclusions

We examined student perspectives across a Midwestern state to uncover modestly elevated greater likelihood for mental health symptoms among youth facing ED, within and across diverse groups of young people. Our population-based study furthers the conversation on punitive school discipline practices through illuminating links to clinically validated measures of mental health, leveraging a broad measure of ED, and carrying out analyses by racial/ethnic group. Although characterization of disparities among minoritized youth is a necessary step towards health and educational equity, it must be paired with institutional action. Our analyses demonstrate both the promise of modifying various protective factors in youth’s social ecology and the urgency of pursuing non-punitive disciplinary strategies.