Suspended Attitudes: Exclusion and Emotional Disengagement from School

Abstract

We know far less about the unintended social-psychological consequences of out-of-school suspensions on students than we do of the academic, behavioral, and civic consequences. Drawing on theories of socialization and deviance, I explore how suspension events influence students’ emotional engagement in school through changes in their attitudes. Using longitudinal middle school survey data connected to individual student administrative records, I find that students who receive out-of-school suspensions are psychologically vulnerable prior to their removal from school. Accounting for demographic characteristics of students, prior year disciplinary involvement, and students’ beginning-of-year attitudes, I find suspensions might further harm students by negatively changing their academic identities and perceptions of adults in school. A series of robustness checks add nuance and strengthen the claims I infer from the main analyses. I close by discussing how the engagement-related consequences of suspension inform social theory and educational policy.

Keywords

school discipline social psychology student attachment to school at-risk students adolescence

If schools serve a socialization function in society, disciplinary practices in schools should sanction students’ behaviors to deter them and others from acting in unacceptable ways (Dreeben 1968; Durkheim 1977; Parsons 1964). Because of education’s impact on many life outcomes, the implications of disciplinary practices reach beyond the school and into later adult political and civic life (Bruch and Soss 2018). Out-of-school suspension, or the temporary removal of a student from school for real or perceived prohibited offenses, is a relatively serious form of school discipline. Suspended students are likely to experience lower subsequent academic achievement and are more likely to drop or get pushed out of school, commit crime, and become incarcerated (Mizel et al. 2016; Noltemeyer, Ward, and Mcloughlin 2015; Wolf and Kupchik 2017). In practice, suspensions appear to be generally ineffective as a deterrent and might instead produce a litany of unintended consequences.

Yet the practice is persistent—and pervasive. The 2016 Digest of Education Statistics reports that 20 percent of all public school students and 48 percent of black male students in Grades 6 to 12 were suspended at least once by 2012 (National Center for Education Statistics 2016). Given the disproportionality of its use in schools toward historically underserved groups, research tends to study suspensions through the lenses of features like race and gender, socioeconomic background, and disability status. If schools effect psychological changes in young people, as Dreeben (1968) argues, psychological factors likely relate to suspensions as well. Students’ academic attitudes, or their perceptions about themselves in school and their interactions with others in school, could be signals that they are at risk of removal from school and the consequences of suspension.

In this article, I ask: How and for which students is suspension from school influential to attitudes about school? I begin by reviewing theoretical and empirical literature considering how students’ attitudes relate to becoming suspended and how suspensions can lead to changes in students’ later attitudes about school. Next, I describe the four academic attitudes investigated in this study. I use longitudinal data from middle school students in a diverse Midwestern urban school district. These data link students’ self-reported attitudes about school collected at the beginning and end of the school year to their individual-level administrative records. My analyses start with an exploration of how students’ attitudes about school can identify students at risk of experiencing suspension. I then shift to uncover the ways suspension might change students’ attitudes given their attitudes at the beginning of the year.

First, I corroborate prior research indicating suspensions are more likely to accrue to boys, students with disabilities, economically disadvantaged students, and African American students. Second, I find trust in school, belonging among peers, and feelings of personal agency are negatively associated with suspensions during the school year. These associations persist for feelings of school trust and social belonging independent of other observed background and school factors. Third, becoming suspended over the course of the school year is associated with lowered subsequent trust in school and academic identity, conditional on each student’s beginning-of-year attitudes, prior year disciplinary involvement, and demographic characteristics. Several robustness checks strengthen the main findings by revealing the degree to which the functional form of the suspension variable, prior suspensions, concurrent disciplinary involvement, and hypothetical unobserved confounders influence interpretation of the main results. In closing, I discuss implications of the results for social theory and schools’ use of exclusionary practices. I suggest schools focus on investments in socioemotional development and trust building, starting inchildhood, to ameliorate the need for exclusionary practices. When working with adolescents, schools should use compensatory school-based interventions to address declining attitudes about school that result from suspensions.

School Suspensions: Corrective or Exclusionary?

A large body of research has explored whether suspension is a useful sanction, an ineffective deterrent, or a harmful exclusionary policy justified by claims of upholding school safety and order. According to deterrence theory, individuals respond rationally and positively to rules and expectations because they are fearful of the consequences that might result (Gibbs 1975). In contrast, normative perspectives of compliance suggest the effectiveness of sanctions rests on whether individuals believe the rules are legitimate and fair (Tyler 2006). These normative perspectives emphasize the importance of context and relationships when considering the policies that enable disciplinary consequences. If students believe school discipline is unfair or weak, exclusionary punishments in schools can lead to greater defiance and lower average student achievement (Arum 2003; Perry and Morris 2014; Way 2011).

If harmful, suspension from school might be a form of systemic violence in which powerful school actors enforce rules that stigmatize and otherwise injure suspended students. Yet rather than being malicious, Epp (1996) argues that schools commit systemic violence inadvertently. Unintended consequences of continuing to impose harmful suspension policies could stem from one of two sources identified by Merton (1936). The first is an error in analysis of the situation: the faulty assumption that severe rule breakers who are suspended will learn from their mistakes. If this is the case, it means schools use suspensions habitually as a deterrent despite being ineffective and that suspensions are accurate and proportional reflections of proscribed student behaviors. The other potential source is an error resulting from favoring immediate interests over long-term interests: “Troublemakers” disrupt learning, and their removal maintains school and classroom order. This suggests the use of suspension can be exclusionary rather than corrective. Some researchers interpret suspension policies as the purposeful criminalization of school behavior leveraged against students whom teachers and administrators believe are on track for further punishment in the criminal justice system or are ill suited for further education (Casella 2001; Hirschfield 2008; Rios 2011). Regardless of their function, suspensions enacted for exclusionary reasons are not contingent on school officials believing the punishment is corrective or even proportional to the offense.

Which depiction of suspension from school is correct? Arum (2003) argues that the effectiveness of school discipline depends on school, local, legal, and political contexts. In some contexts, schools maintain a normative authority over students, reflected in students’ corrective responses to disciplinary sanctions. In other contexts, schools’ disciplinary measures are largely ineffective as correctives to behavior and instead serve to exclude students for the sake of school safety and order. To determine whether suspension practices are useful, harmful, or benign, we must understand whether suspensions change students’ attitudes and behaviors positively, negatively, or not at all.

Emotional Engagement in School

Many social scientists frame social and emotional development as part of what is learned in school (e.g., Dreeben 1968). Others argue that schooling shapes, rewards, and punishes students’ personality traits differently along class and economic lines, contributing to the social and economic reproduction of society (e.g., Bourdieu and Passeron 1990; Bowles and Gintis 1976; Foucault 1977). Yet despite the many theoretical connections between psychological changes through schooling and socialization, sociologists have focused mainly on functional outcomes like achievement and educational/occupational attainment or the social processes related to these outcomes. We know less about how schools shape social-psychological processes like emotional engagement, particularly in relation to disciplinary policies and practices.

Engagement, or students’ participation in and commitment toward school, enables students to benefit from the opportunities to learn that schools offer. Engagement in school has behavioral and emotional components (Johnson, Crosnoe, and Elder 2001). The behavioral components are composed of normative actions in school; the emotional components (also known as school attachment) are students’ affective responses to the learning environment. Students’ attitudes toward school are part of emotional engagement (Fredricks, Blumenfeld, and Paris 2004) and are precursors to the mindsets and skills that promote learning (Farrington et al. 2012). Academic attitudes are thoughts, feelings, and perceptions of the academic environment, which might differ from individuals’ attitudes about themselves or their relationships with others in different contexts.

As components of emotional engagement, academic attitudes are key mechanisms of informal social control in the desistence of adolescent delinquency (Sampson and Laub 2005). Students emotionally withdraw when they do not identify with school or they believe teachers and peers do not accept them (Hallinan 2008). Times of uncertainty and transition can amplify feelings of “fitting in” (Walton and Cohen 2007). Among many transitions across the life course that increase uncertainty, the transition into middle school is an important event involving environmental change and identity formation. Following this transition, a large proportion of students begin to withdraw from school (Eccles 2004). Declining attitudes in middle school and later can lead to suspension, push-out, or dropout (Finn 1989; Mizel et al. 2016; Voelkl 1997).

Attitudes and Suspension from School

Deviance and subsequent negative outcomes result not only from worsening behaviors but also from worsening relationships that lead to differences in the ways individuals perceive themselves and are perceived and treated by others (Becker 1963). Okonofua, Walton, and Eberhardt (2016) propose that negative behaviors and attitudes about school worsen through a “vicious cycle” of stereotyping and mistrust that plays out as school relationships break down through continued disciplinary actions. Their school-based perspective relates to three concepts in the sociological and criminological literatures on deviance: turning points, stigma, and labeling. These concepts can contribute to a theoretical explanation of how suspension from school might change students’ academic attitudes.

Turning Points

Critical events, or turning points, can positively or negatively alter delinquency and crime trajectories despite earlier experiences (Laub and Sampson 1993). Sampson and Laub (2005) summarize five important characteristics of turning points: They involve new situations that (1) clearly delineate past and current circumstances (i.e., “knifing-off”), (2) provide opportunity for investment in new social networks, (3) provide oversight in monitoring of behavior, (4) change and shape routine activities, and (5) allow for identity transformation.

Exclusionary school events might serve as negative turning points in adolescence, knifing off an individual’s past as a “student” from the current circumstance as a “delinquent.” Suspensions can also introduce students to networks outside of school that might encourage future negative behavior (i.e., “supportive deviant others”; Lemert 1951). Perhaps most importantly, exclusion from school alters routine activities by removing students from a key informal social control during adolescence: school attachment. Finally, if suspensions change students’ academic attitudes, this suggests such events can provide opportunities for identity transformation. Thus, suspension from school might redirect adolescents’ paths toward changes in identity and delinquency.

Stigma and Labeling

The sociological framing of stigma gives further insight into how turning points like exclusionary school events might alter individuals’ thoughts and feelings about themselves and others. Link and Phelan (2001) conceptualize stigma as the co-occurrence of labeling, stereotyping, exclusion, status loss, and discrimination due to a power structure that allows them to coexist. In their framing, all groups and individuals place social importance on and label human differences. However, when those in power link negative labels to stereotypes, it serves as a rationale for excluding individuals for being fundamentally different from those without the negative label (“us” vs. “them”). Following exclusion, stigmatized individuals experience status loss and discrimination. Status loss itself can serve as the basis for continued discrimination by characterizing stigmatized individuals as undesirable companions, peers, or fellow community members. Consequences of stigmatization range from the psychological to the structural. Individuals simply need to believe others have labeled them negatively for social-psychological consequences, like changes in attitudes, to occur.

Labeling theories further clarify the social-psychological consequences of stigmatization. The symbolic interactionist tradition of the labeling perspective argues that the simple act of labeling can lead to more deviant thoughts and actions (Becker 1963). Paternoster and Iovanni (1989) explain the process in probabilistic terms: A negative event increases the likelihood of publicly labeling an adolescent a delinquent and subsequently excluding the person from normal routines. Being excluded can lead to an alteration of identity, or deviance avowal, in which individuals accept the negative label ascribed to them. This change in identity can then increase the chances of further delinquency, or secondary deviance.

Theories of stigma and labeling inform the discussion of how suspensions lead to declining attitudes. An event or series of events (e.g., a fight, repeated classroom disruptions) can increase the likelihood of a student being publicly labeled and devalued as a troublemaker in the classroom or school. School and district administrators have the power to enact stigmatization and are motivated to identify and single out troublemakers to appease frustrated teachers and concerned parents (Staples 2014). Negative labeling and stereotyping by those in power thus turns whole and usual students into troublemakers or delinquents. Stereotyping can lead to exclusion from the normal routine of daily school attendance, distancing the outsider suspended student from insider peers and teachers. School authorities often rationalize exclusion by portraying suspended students as a threat to school safety and order (Hirschfield 2008). Link and Phelan’s (2001) framework suggests such a portrayal stigmatizes suspended students, who then expect others in school will devalue them once they return to school. Internalizing others’ devaluation leads to a degradation of academic identity, which could result in changes in attitudes among suspended students. Thus, exclusion serves as a turning point that stigmatizes students in part by changing their attitudes about school.

Empirical Evidence

Several ethnographic studies illuminate the processes connecting school punishments and subsequent psychological outcomes. Through observations of policed black and Latino boys, Rios (2011) documents how labeling, stereotyping, and stigma lead to adversarial attitudes, loss of trust, and resistance among adolescents. Observing responses to violence in two urban schools over several school years, Casella (2001:153) finds that suspended students often return to school with less respect for school staff and “a more combative attitude” than before the suspension. Casella explains these changes by noting, “[s]tudents know when schools do not want them.” In an ethnographic study of students at risk of high school dropout, Fine (1991:243) made similar observations: Students returning from a suspension often expressed a sense of embarrassment upon their return (e.g., in response to a sarcastic “You’re back!”), which was often quickly followed by truancy and eventually, dropout.

Quantitative studies attempting to identify the psychological effects of suspension are scant. Recently, Mittleman (2018) analyzed but found no conclusive association between childhood suspension and later feelings of connection to school. Other research has found suspended students report fewer positive school relationships (Morrison et al. 2001) and are more likely to say adults in school are not concerned about their academic well-being (Brown 2007). One limitation of these studies is they do not account for students’ perceptions about school prior to suspension, which might explain some or all of the association between suspensions and students’ subsequent attitudes.

Current Study

I pose two related questions in this study. First, to what degree do students’ previous attitudes matter for becoming suspended during middle school? The demographic characteristics of students at risk of suspension are well established in the literature; psychological characteristics less so. I pose this first question with psychological characteristics in mind to uncover whether academic attitudes explain becoming suspended independent of common demographic characteristics related to suspension. Because suspended students are already likely to be socially and psychologically vulnerable, suspensions over the school year might add to disadvantages these young people face.

Second, how do school suspensions matter to changing emotional engagement in middle school? Improved or unchanging student attitudes following suspension could mean labels and exclusion from school do not negatively influence students over the year. Link and Phelan (2001) and Paternoster and Iovanni (1989) certainly recognize this as a viable outcome in the labeling process given certain conditions. However, if students’ attitudes worsen following suspension, this suggests exclusion from school might have unintended psychological consequences. If so, theories of deviance help explain how changes in students’ attitudes following suspension could lead to changes in behaviors.

I answer these questions using attitudinal measures that are reliable, malleable, theoretically connected to disciplinary practices as inputs and outcomes, and responsive to evidence-based interventions (see Pyne, Rozek, and Borman 2018). The four attitudes I study are school trust, social belonging, external locus of control, and identification with school. Attitudes about school have interpersonal and intrapersonal qualities (Finn 1989), and I differentiate between the two types among the four attitudes. School trust and social belonging are interpersonal and represent how an individual relates to others in a particular context. External locus of control and identification with school are intrapersonal and represent how individuals think of themselves in that context.

Interpersonal and intrapersonal categories pose a useful distinction when considering the various effects suspensions might have on students and when choosing interventions that might mitigate the negative effects of suspensions. In schools, interpersonal academic attitudes are associated with a student’s relationship to teachers, administrators, support staff, and other students—which are important forces in the dispensation of and response to school discipline (Okonofua et al. 2016). Aggregated at the group or school level, interpersonal attitudes are important aspects of school climate, which can itself influence disciplinary policies (Arum 2003; Thapa et al. 2013). Intrapersonal academic attitudes are associated with how students view themselves in school, or their academic identities, with school relationships still in mind. Identity is central to the theoretical perspectives of labeling, turning points, and stigma; as self-perceptions, these attitudes are well suited to represent the identities students form in academic settings.

Method

Data

I use data collected from two cohorts of students in 10 middle schools in a diverse urban school district. Originally collected within a set of randomized controlled trials, these data include responses to a survey asking students about their academic attitudes, which was administered once at the beginning and once at the end of the implementation school year. The first cohort consists of seventh graders; the second cohort contains sixth graders. The consent rates for the surveys and interventions were 77 percent and 79 percent, respectively. For both cohorts, project assistants administered the survey of academic attitudes in September of the intervention academic year and then again in May of the same academic year. This ensures nearly all school events, such as suspensions, occurred between the two implementations. Project staff then linked survey data to individual student intervention and district administrative records. Because evidence suggests one of the interventions shaped students’ attitudes about school, I only examine control group students from both cohorts. Randomization of control and treatment groups shows no substantial differences on observable pretreatment characteristics (see Appendix A, Table A1).

Of the observations in this sample, 171 (16 percent) are missing data on at least one dependent or independent variable. Most of the reduction in sample size is due to students being absent from school on the days the study team administered the academic attitudes surveys and to a lesser extent, the district having no record of students’ prior disciplinary involvement. The minor differences between the full and reduced samples (less than .03 SD for all variables of interest) reveal students removed from the control sample for missing values were very similar to students for whom there was full control sample information (see Appendix A, Table A1).

The final data set consists of 885 control group students with complete information on all variables of interest. Sixth graders represent 54 percent of participants, and seventh graders represent 46 percent of participants. In the final sample, 52 percent of students are white, 19 percent African American, 18 percent Latino, and 12 percent Asian. Additionally, 48 percent are male, 36 percent are eligible for free or reduced-priced lunch, 11 percent are students with disabilities, and 16 percent are English language learners.

Measures

Academic attitudes come from a survey administered to students during instructional time. All survey items use 5-point Likert scales (1 = strongly disagree, 2 = disagree, 3 = neither agree nor disagree, 4 = agree, 5 = strongly agree). Although the survey is brief, each scale exhibits adequate internal consistency for research purposes (α = .72–.83; see Table 1). In an exploratory factor analysis conducted by Pyne and colleagues (2018) using a sample that includes the present participants, all item loadings were .40 or higher on each a priori factor, and each factor exhibited adequate measurement invariance (i.e., similar item response patterns) across contexts and demographic groups. An exploratory factor analysis using only students and items in the current study is consistent with published findings (Appendix A, Table A2).

Table 1.

Academic Attitude Scales, Items, and Alpha Reliability.

		Alpha Reliability
Scale	Items	Time 1	Time 2
School trust	The teachers at this school treat students fairly. The adults at this school care about the students. At this school, students are supported.	.77	.78
Social belonging	People in my school accept me. I feel comfortable in my school. I feel like I belong in my school. I feel like an outsider in my school (reverse coded).	.74	.72
External locus of control	Getting the grades you want is mostly a matter of luck. The main difference between students who get good grades and students who get bad grades is luck. It takes a lot of luck to be an outstanding student in most classes.	.83	.82
Identification with school	It is important for me to do well in school. I want to do well in school.	.76	.73

I measure two interpersonal academic attitudes: school trust and social belonging. School trust is the degree to which students believe they have positive relationships with adults in their school. In accordance with the concept of relational trust (Bryk and Schneider 2004), this measure focuses on students’ beliefs that adults in school care about them and treat them fairly. Because suspension means exclusion from the school community, a primary concern is whether students believe the adults meting out suspensions have positive regard for students, have personal integrity/honesty, and are benevolent rather than malicious. High levels of school trust are associated with higher student academic performance and motivation (Bryk and Schneider 2004; Goodenow 1993; Hallinan 2008). The school trust scale contains three items, adapted from the Add Health survey (Resnick et al. 1997) and the Psychological Sense of School Membership scale (Goodenow 1993).

Social belonging refers to students’ beliefs that they are connected socially to other students in school and that peers in school accept them. In an academic environment, socially stigmatized individuals and groups might be uncertain of social bonds with their peers, which can make belongingness salient to them (Walton and Cohen 2007). Suspensions are more likely related to trust in adults at school. Even so, I use this construct to determine the extent to which being suspended is a stigmatizing event among students, which might lower suspended students’ feelings of fitting in at school. Four items from Walton and Cohen’s (2007) Social and Academic Fit scale form the social belonging scale.

The remaining two constructs are intrapersonal attitudes: external locus of control and identification with school. External locus of control represents the belief that one does not have the ability to change life events. Although locus of control has a long history of use in education research (see Findley and Cooper 1983), the scale in the current study follows a narrow definition. Of interest here is the degree to which students believe that doing well in school is the result of forces out of an individual’s control. A person with high external locus of control in terms of academic pursuits might have low external locus of control in other contexts. The narrower conception used here clarifies if social exclusion through school suspension is associated with how individuals perceive their control over school work. The measure is a three-item scale adapted from the Work Locus of Control Scale (Spector 1988) and reworded for the school context.

Identification with school is a context-specific variation of competence valuation (Harackiewicz and Sansone 1991) representing the degree to which students care about and place importance on doing well in school. Prior research has framed identification with school as a function of both belonging in school and valuing school-related outcomes (e.g., Voelkl 1997). Here, I use only the latter definition, which aligns with the concept of competence valuation. It therefore differs from school trust or social belonging because it focuses primarily on affective states, not relationships within school (Finn 1993). Having been removed from the school environment, suspended students might adapt to the exclusion and respond to suspension by doubting the importance and value of school on their lives. Identification with school is a combination of two competence valuation items derived from the intrinsic motivation literature (Harackiewicz and Sansone 1991). Table 1 reports scale items and reliabilities.

I am primarily interested in how exclusion from the school environment serves as a crucial event affecting students’ attitudes, so I measure students’ out-of-school suspensions rather than other types of sanctions (e.g., detention, in-school suspension), and I code suspension as a dichotomous variable. This functional form represents the hypothesis that suspension is a turning point for middle school students, the effects of which should attenuate greatly after the first incident of the year. To add to the robustness of these analyses, I also include each student’s prior year number of suspensions and prior year number of office disciplinary referrals (ODRs). Prior year suspensions help account for spurious associations between current year suspensions and end-of-year attitudes not completely captured by beginning-of-year attitudes and other covariates. ODRs further help account for more minor disciplinary involvement, the accumulation of which might signal an escalation of student behaviors and responses to those behaviors prior to the suspension measurement window. In robustness checks, I use the same logic when applying current year ODRs to changes in attitudes.

Other covariates include free and reduced-price lunch participation for each student, disability status, English language proficiency, gender, and race—all of which prior literature suggests should be related to being suspended from school. Finally, I include school fixed effects to control for the influence of each of 10 middle school contexts in the study. Due to evidence of multicollinearity, I do not present models that control for prior achievement or grade level; however, their inclusion does not substantively alter any of the results of interest reported here.

Strategy and Design

I construct each academic attitude scale by taking the average score of all completed items in the scale. For descriptive results, I use the mean scale score; for analytic results, I standardize each scale score so the mean for each is zero and the standard deviation is one. I begin by determining the degree to which student demographic characteristics and academic attitudes differentiate students who do and do not have suspensions over the school year. Using logistic regression, I next enter each academic attitude individually and then enter all four academic attitudes simultaneously as predictors of suspension, including demographic and academic covariates and school fixed effects in each model. I then compare each individual attitude model to the fully saturated model. To adjudicate between each individual/full pair of models, I assess relative fit using Bayesian Information Criterion (BIC) and standards of their interpretation outlined by Raftery (1995). For interested readers, I also include the Akaike Information Criterion (AIC), area under the ROC curve, and McFadden’s pseudo R² fit statistics in Table B1.

Next, I investigate whether suspensions might change attitudes. Using end-of-year scales for each of the four academic attitudes of interest, I linearly regress attitudinal outcomes on receiving any suspensions at all, accounting for beginning-of-year attitudes, prior year disciplinary referrals and suspensions, demographic covariates, and school fixed effects. Suspension coefficients in these models are standardized mean differences in which each attitude outcome is standardized and compared between those who do and do not receive suspensions over the year.

Finally, I conduct four robustness checks to test the strength of the associations between suspension and changes in attitudes. First, I report tests of the functional form of the suspension variable, comparing the binary version I use in the main analyses to count versions in incidents and days of suspension. Second, I test the degree to which suspension’s association with changes in attitudes works through students with previous year suspensions versus students with suspensions in the current survey year but not the year prior. Third, I test the degree to which the prevalence of office disciplinary referrals received in school explains changes in attitudes. Fourth, to address remaining doubts, I quantify the influence a hypothetical unobserved confounder would need to have to invalidate the association between suspension and changes in attitudes independent of all other variables in the models.

Results

Characteristics of Suspended Students

Table 2 displays descriptive statistics for the analytic sample of students, differentiated between those who did and did not receive suspensions over the year. For Time 1 and Time 2 academic attitudes, asterisks denote a statistically significant (p < .05) unconditional mean difference between suspended and nonsuspended students’ reported levels. In the full analytic sample, about 6 percent of students (N = 53) were suspended between the two implementations of the academic attitudes survey. Students not suspended during the year had far less prior year disciplinary involvement compared to those who were. A third of suspended students experienced suspension the year prior, and 70 percent were given at least one ODR the year prior. Consistent with previous research, suspended students were much more likely than their peers to be eligible for free or reduced-price lunch and special education programs, were more likely to be male, and were disproportionately African American. Suspended students had lower levels of school trust and social belonging at the beginning of the school year and notably higher prior levels of external locus of control than nonsuspended students. Reported levels of identification with school were similar between suspended and nonsuspended students at the beginning of the school year. Unconditional mean differences in Time 2 (i.e., end of year) academic attitudes indicate suspended students had lower end-of-year school trust and identification with school and higher external locus of control than nonsuspended students. All of these differences are statistically significant at the p < .05 threshold.

Table 2.

Descriptive Statistics.

	Nonsuspended Students (N = 832)			Suspended Students (N = 53)
	Percentage	Mean	SD	Percentage	Mean	SD
Any prior year suspensions	4			32
Any prior year disciplinary referrals	18			70
Free/reduced-price lunch	34			70
Special education	9			30
Female	53			40
Race/ethnicity
White	53			21
Asian	12			4
African American	17			60
Latino	18			15
Number of current year suspensions		.00	(.00)		1.50	(.91)
Number of prior year suspensions		.07	(.39)		.81	(1.73)
Number of prior year disciplinary referrals		.83	(3.76)		7.19	(12.65)
Beginning of school year attitudes
School trust		4.12	(.63)		3.79*	(.74)
Social belonging		4.06	(.64)		3.84*	(.79)
External locus of control		2.13	(1.01)		2.75*	(1.12)
Identification with school		4.70	(.50)		4.65	(.53)
End of school year attitudes
School trust		3.75	(.74)		3.15*	(1.03)
Social belonging		3.91	(.74)		3.72	(.92)
External locus of control		1.91	(.89)		2.42*	(1.11)
Identification with school		4.66	(.51)		4.31*	(.71)

Note: For attitude variables only, an asterisk indicates a statistically significant difference between suspended and nonsuspended students.

p < .05.

I next construct logistic regression models with entry into suspension over the school year as the dependent variable. I enter each of the four academic attitudes scales one at a time and then simultaneously. BIC statistics indicate a superior model fit of each individual attitude model compared to the fully saturated attitude model (see Appendix B, Table B1). Figure 1 displays the association between attitudes and being suspended over the school year, in odds ratios. Each plot represents a model with only the attitude of interest and no others as a predictor, accounting for prior disciplinary involvement, students’ demographic characteristics, and school fixed effects. School trust and social belonging predict a suspension occurring at least once during the school year independent of covariates. In this sample, a standard deviation increase in school trust and social belonging corresponds to a change in the respective odds of suspension by factors of .74 (–26 percent) and .67 (–33 percent), all else equal.

Figure 1.

Associations between academic attitudes and being suspended during the school year.

Associations between Suspensions and Change in Academic Attitudes

These descriptive and logistic regression results suggest suspended students are socially and psychologically vulnerable prior to suspension during the observed school year. My primary interest is to understand whether current year suspensions change students’ attitudes about school. I thus shift to multiple linear regression models, regressing each z-scored academic attitude measure on a binary indicator of any out-of-school suspensions and other covariates (Figure 2). After accounting for prior attitudes, student demographics, and prior disciplinary involvement, only school trust and identification with school are associated with suspension earlier in the school year both substantively and within statistical significance. Being suspended during the school year is associated with an average decrease in school trust by 47 percent of a standard deviation and a decrease in identification with school by 54 percent of a standard deviation, all else equal. Both are equivalent to about a quarter of a scale point drop on these 5-point scales. Conversely, the conditional association between suspension and change in social belonging was functionally zero, and between suspension and change in external locus of control, it was a nonsignificant .07 standard deviations. Full regression results for academic attitude outcomes are in Appendix B, Table B2.

Figure 2.

Associations between suspension and change in academic attitudes.

Robustness Checks

Regarding the relationships between suspension and later academic attitudes, I am ultimately interested in how suspensions change attitudes about school. Observational data in the literature suffer from selection bias because suspensions are not random. In the following robustness checks, I test whether the associations between suspension and later academic attitudes hold up to the influence of (1) the functional form of the suspension variable, (2) prior year suspensions on changes in attitudes, (3) more minor incidents of disciplinary involvement, and (4) unobserved confounders in attitude outcome models.

The functional form of suspensions as predictors of attitudes

For the linear regression models presented in Figure 2, I use a binary suspension construct as an independent variable differentiating students with no suspensions versus those with any number of suspensions during the school year. In supplemental analyses not shown, I also construct linear splines for number of suspensions during the school year and number of days suspended during the year. Within these splines, the slope between 0 and 1 suspension or day suspended (depending on the model) is allowed to vary independent of the slope of subsequent suspensions received. I find little empirical support that the number of suspensions or days of suspension is useful for predicting attitudes beyond entry into suspension. Rather, in this sample, one incident or one day of suspension appears to carry most of the association with students’ changes in school trust and identification with school substantively and with statistical significance.

Previously suspended students

The models used in Figure 2 account for prior year suspensions but do not explicitly test the influence of suspension history on changes in attitudes. In this robustness check, I replace the binary suspension variables in these models with a categorical suspension variable. This categorical variable differentiates among (1) students with no suspensions in the school year under study, (2) students with suspensions in both the year under study and the prior school year, and (3) students who had a suspension in the school year under study but no suspensions in the prior school year. Recall from Table 2 that about a third of suspended students experienced a suspension the prior school year. Table 3 displays results regressing academic attitudes on the categorical suspension variable and other covariates. Results indicate school trust declined by over half a standard deviation more for study year only suspended students than students with no suspensions in the study year, all else equal. That is twice the magnitude of the relative decline of students suspended in both years compared to nonsuspended students. Results from the identification with school outcome model indicate study year only suspended students had steeper relative declines compared to their nonsuspended peers and their peers who were suspended both years. The difference in association between students with no suspensions in the current year and those with suspensions in both years was not statistically significant in either model.

Table 3.

Associations between Suspension and Changes in Select Academic Attitudes, Differentiating Current Year Suspended Students by Previous Year Suspensions (N = 885).

	School Trust	Identification with School
Suspension history (reference category = none in study year)
Study year only	−.53***(.16)	−.58***(.16)
Study and previous year	−.27(.26)	−.43(.26)
Lagged academic attitude	.40***(.03)	.36***(.03)
Free/reduced-price lunch eligible	−.13(.08)	−.11(.08)
Student with disability	.05(.10)	−.12(.10)
Limited English proficiency	.15(.11)	−.05(.11)
Male	.11(.06)	−.12*(.06)
Race/ethnicity
Asian	.13(.11)	.07(.11)
African American	−.15(.10)	.06(.10)
Latino	−.08(.11)	.03(.11)
Prior year disciplinary referrals	−.00(.01)	.00(.01)
Constant	−.30*(.12)	.02(.13)
R²	.25	.18

Note: Each column represents a linear model regressing each academic attitude on suspension history, beginning-of-year attitude, prior year disciplinary referrals, prior year suspensions, demographic characteristics, and school fixed effects. Academic attitudes are z-scored values. The suspension history variable indicates whether a student was suspended both years, suspended only during the survey year, or not suspended during the survey year. Standard errors are in parentheses.

p < .05. ***p < .001.

Office disciplinary referrals and changes in attitudes

Changes in attitudes associated with suspension during the school year might simply be due to an increasing progression of misbehavior over the course of the year rather than a suspension event creating a turning point for suspended students. This data set also contains ODRs measured during the year of the survey administrations. A limitation of using ODRs for this purpose is they (like suspension) do not directly represent misbehavior; rather, they measure teachers’ and other school staff’s responses to perceived student actions. Yet an advantage they have over suspension measures is the exclusionary mechanism—temporary removal from class—is more common than removal from school, and it likely captures less severe disciplinary involvement than that captured by suspensions. If the suspension-attitude associations simply reflect students’ involvement in school incidents that increase with severity over the year, ODRs should capture some or all of that association.

In the school trust and identification with school linear regression models, I replace the binary suspension variable with an indicator of number of ODRs received over the course of the school year. In these models, I also account for beginning-of-year attitudes, prior disciplinary involvement, and all other control variables. I then enter the suspension indicator back into both models to determine the degree to which suspensions explain the associations between ODRs and the two attitudes. The results of these checks are substantively similar whether using the whole sample of suspended and nonsuspended students, as in the analyses in Table 4, or when looking only at ODRs and attitudes of nonsuspended students (results not shown).

Table 4.

Associations between Office Disciplinary Referrals and Changes in Select Academic Attitudes (N = 885).

	School Trust		Identification with School
	1	2	3	4
Number of disciplinary referrals	−.03**(.01)	−.02*(.01)	.00(.01)	.01(.01)
Out-of-school suspension		−.40**(.14)		−.56***(.14)
Lagged academic attitude	.41***(.03)	.40***(.03)	.36***(.03)	.36***(.03)
Free/reduced-price lunch	−.11(.08)	−.11(.08)	−.12(.08)	−.12(.08)
Student with disability	.01(.10)	.04(.10)	−.15(.10)	−.11(.10)
Limited English proficiency	.16(.11)	.15(.11)	−.03(.11)	−.05(.11)
Male	.10(.06)	.11(.06)	−.14*(.06)	−.13*(.06)
Race/ethnicity
Asian	.11(.11)	.12(.11)	.07(.11)	.08(.11)
African American	−.14(.10)	−.11(.10)	.00(.10)	.05(.10)
Latino	−.08(.11)	−.08(.11)	.03(.11)	.03(.11)
Prior year disciplinary referrals	.01(.01)	.01(.01)	−.00(.01)	−.00(.01)
Prior year suspensions	.06(.08)	.07(.08)	−.03(.08)	−.01(.08)
Constant	−.24(.13)	−.25*(.13)	.02(.13)	.00(.13)
R²	.24	.25	.17	.18

Note: Each column represents a linear model regressing each academic attitude on current year number of office disciplinary referrals, current year suspension, beginning-of-year attitude, prior year disciplinary referrals, prior year suspensions, demographic characteristics, and school fixed effects. Academic attitudes are z-scored values. The suspension variable indicates whether a student was suspended at least once between the two administrations of the academic attitudes survey. Standard errors are in parentheses.

p < .05. **p < .01. ***p < .001.

About 19 percent of students in the sample received one or more ODRs during the school year under study—roughly three times the number who received suspensions during the year. The median number of ODRs among students with any was two, and among those with out-of-school suspensions, it was six. Table 4 presents analytic results of linear models regressing school trust and identification with school on ODRs and other covariates. Each additional ODR a student received is associated with a .03 standard deviation reduction in school trust. Assuming linearity, this means it would take about 16 ODRs to equal the association between being suspended and change in school trust (see Figure 2)—a far higher number of ODRs than the median amounts for suspended or nonsuspended students. Suspensions also explain about a third of the association between prevalence of ODRs and end-of-year school trust, all else equal. End-of-year identification with school is not at all associated with prevalence of ODRs. This is particularly striking given that beginning-of-year measures of identification with school are nearly identical between students who are and are not suspended over the year. I found no statistically significant or meaningful associations between number of ODRs and social belonging or external locus of control (results not shown).

Hypothetical unobserved confounder analyses

Although I control for numerous priors, including beginning-of-year attitudes, prior year suspensions, and prior year office disciplinary referrals, other unobserved factors might still confound the interpretation of observed associations between suspension and later attitudes. To quantify the magnitude of the hypothetical omitted variable bias, I draw on Frank (2000) and Frank and colleagues (2013) to calculate the independent effect an unobserved confounding variable would need to have to render the observed association between suspension and changes in attitudes invalid. These checks essentially turn critiques of causal inference into quantifiable thresholds.

To invalidate the inference of the school trust outcome, 42 percent of the estimated effect would have to be due to unobserved variable bias, and 365 of the 885 cases would have to be replaced with cases that have an effect of zero. The independent association of the unobserved covariate would have to be over twice as strong as the association between prior school trust on suspension and the school trust outcome. To invalidate the inference of the identification with school variable, 49 percent of the estimated effect would have to be due to bias, and 435 observations would have to be replaced with zero-effect cases. The independent association between an unobserved confounding variable and the suspension and outcome variables would need to be about seven times stronger than the association between lagged attitude on suspension and the identification with school outcome.

Discussion

Schools can affect students psychologically, shaping adolescents’ academic engagement and later trajectories into adulthood (Bowles and Gintis 1976; Dreeben 1968). Attitudes are important early indicators of school engagement in adolescence; they serve as precursors to more stable mindsets students will eventually develop about schooling that influence behavior and learning (Farrington et al. 2012). I created several analytic models to learn how and for whom out-of-school suspensions affected emotional engagement in school through changes in academic attitudes. To accomplish this, I used data from a diverse sample of middle school students that contained their individual responses to two academic attitudes surveys. I linked students’ responses from the survey to their individual-level school administrative records. This resulted in a novel data set containing students’ self-reported attitudes at the beginning and end of the school year combined with official district reports of student suspensions between those two timepoints. I divide the results from this study into four main topics for discussion: attitudes as predictors and outcomes of suspension, implications of this work for social theory, implications for social policy, and limitations of the study.

Attitudes as Predictors and Outcomes of Suspension

Suspended students are a socially vulnerable population prior to their exclusion from school (Losen and Martinez 2013). Consistent with prior literature, I confirm that suspended students in this sample are more likely to be male, African American and economically disadvantaged and have a reported disability. Adding to prior research, I find suspended students tend to have lower school trust and social belonging and higher external locus of control than nonsuspended students prior to suspension. Interpersonal attitudes (school trust and social belonging) remain statistically significant predictors of suspension later in the school year after accounting for prior year disciplinary referrals, prior year suspensions, and demographic covariates (Figure 1). Although the effect sizes appear small in each case, suspended students represent only 6 percent of students in the sample. Therefore, changes in the odds of suspension by 20 percent to 30 percent are substantial for predicting who ends up in this small group of students.

Results of models predicting changes in attitudes suggest suspensions are associated with later emotional disengagement from school, as represented by negative changes in school trust and identification with school (Figure 2). Suspensions have no independent association with changes in either social belonging or external locus of control after accounting for other factors. Social belonging results are consistent with Mittleman’s (2018) finding that suspension does not predict later feelings of school connection, yet students’ reports of school connection predict subsequent delinquency. Although it is unclear why this might be the case, the corroboration of these recent findings in the current study is worth noting.

Unlike previous studies examining the psychological consequences of suspensions, I account for beginning-of-year levels of student attitudes and prior disciplinary involvement. Two sets of results demonstrate the strength of these lagged dependent variable models. First, prior year disciplinary referrals and prior year suspensions are strongly unconditionally associated with current year suspensions and end-of-year school trust and identification with school. Yet the lagged dependent variable models account for much of these associations (see Table B2). Second, suspended students report substantially higher levels of external locus of control at the end of the year than nonsuspended students (Table 2), but beginning-of-year levels of external locus of control and other covariates account for 87 percent of that unconditional association (see Figure 2). On its own, the lagged external locus of control variable accounts for 71 percent of that unconditional association. The same is not true of end-of-year school trust and identification with school. The end-of-year differences between suspended and nonsuspended students on these two variables remain statistically significant and moderate in magnitude even when accounting for beginning-of-year measures of attitudes and prior disciplinary involvement.

Robustness checks further strengthened the results of this study. First, the suspension-attitude associations for school trust and identification with school outcome models are greater in magnitude for suspended students who had no suspensions the prior year relative to those who did (Table 3). This suggests the moderate associations between suspension and later attitudes observed in Figure 2 are not due to ongoing issues associated with prior year disciplinary involvement. Second, more minor disciplinary involvement over the school year does not fully account for the associations between suspension and changes in school trust and identification with school (Table 4). In fact, the number of ODRs received during the surveyed school year is only weakly associated with end-of-year school trust and is virtually unassociated with later identification with school, all else equal. This suggests the suspension event itself might change attitudes beyond changes due simply to escalating patterns of disciplinary involvement over the school year. Third, to invalidate the claims made in this study, roughly two-fifths of the conditional association between suspension and school trust and roughly half the association between suspension and identification with school would need to be due to unobserved variable bias. These supplementary analyses add to the robustness of the main findings in Figure 2, providing strong evidence in favor of a meaningful relationship between suspension and changes in attitudes.

Social Theory Implications

In describing what students learn in schools, Dreeben (1968:38) notes, “If the child at home wonders whether he is loved, the pupil wonders whether he is a worthwhile person. In both settings he can find some kind of answer by observing how others treat him.” Dreeben argues that school relationships can affect students psychologically, but those psychological effects are not always positive or productive. Functionalists assert that school sanctions should correct unwanted behaviors among students. Yet much research documents the negative academic, behavioral, civic, and economic outcomes associated with school suspensions. The current study adds more evidence by identifying potential psychological consequences related to suspension from school. This study contributes to social theory in six distinct ways, all of which highlight how and why changing attitudes about school might be important psychological consequences of school suspensions.

First, the school trust results identify how students’ relationships with adults in school might deteriorate due to repeated disciplinary involvement. Interpersonal academic attitudes like school trust refer to thoughts and feelings about relationships with adults in school. Sampson and Laub (2005) argue that adults’ social bonds are important explanatory mechanisms linking childhood inequalities to later adult crime. The evidence in this study implicates suspension as a likely mechanism in degrading adolescent social bonds, as proxied by suspended students’ changes in school trust. Consistent with Okonofua and colleagues (2016), results suggest relationships with adults in school slowly degrade through a “vicious cycle” of mistrust as students continue to experience discipline. This is clear in the robustness checks of the school trust outcome, which reveal that office disciplinary referrals have a modest and statistically significant negative association with students’ later trust in adults at school conditional on prior attitudes, prior disciplinary involvement, and student demographics (Table 4). Suspension appears to add to the decline in students’ perceptions of adults at school beyond the effect of ODRs alone. Suspension thus might serve as a shock that accelerates the vicious cycle of mistrust and stereotyping in some school relationships.

Second, and relatedly, end-of-year identification with school and school trust outcome results point to out-of-school suspension as a negative turning point in young adolescents’ lives. Consistent with the turning point perspective, the first suspension a student received during the school year was a much stronger predictor of changes in attitudes than subsequent suspensions. This suggests knifing off students from normal routines changes their school trajectories. The strongest evidence for this is the greater associations between suspension and these later attitudes for students who did not have suspensions the year prior compared to those who did. If the opposite were true and students suspended in both years carried much or all of the association between suspension and later attitudes, this would weaken the argument that suspensions serve as crucial turning points in adolescents’ lives that can change their academic identities. Instead, students who also experienced suspension the previous year had weaker substantive negative declines in attitudes compared to those who had no suspension experience the year before, all else equal. Previously suspended students might still experience psychological harm from continued suspensions; however, with the promise of a fresh start at the beginning of each school year, a suspension could stigmatize students in terms of their new relationships, further degrading their identity through repeated declines in academic attitudes year after year.

Third, end-of-year identification with school outcome results are consistent with sociological framings of stigma and labeling, identifying suspension as a stigmatizing exclusionary event that alters individuals’ identities. Behavioral events can result in others publicly labeling students as troublemakers, negatively stereotyping them as such, and subsequently excluding them from normal routines. Exclusion increases the odds these students will internalize and accept such an attribution. Suspended students’ beginning-of-year identification with school is nearly identical to that of their peers who had no suspensions that school year (Table 2). However, among students who became suspended, their academic identities (e.g., “I want to do well in school”) were lower than those not suspended, on average, conditional on students’ attitude reports at the beginning of the year, prior disciplinary involvement, and demographic covariates. Unlike with school trust, ODRs in general do not appear to affect changes in identification with school over the year (Table 4). This suggests the suspension event itself, not increasingly severe disciplinary involvement over the year, changes academic identity. Exclusionary suspension policies thus might induce disengagement from school by isolating students from the work ethic and culture they would otherwise be expected to embrace.

Fourth, this work implicates suspension as a mechanism of cumulative disadvantage: The sequential and additive effects of earlier experiences and (dis)advantages contribute to long-term outcomes. Consistent with prior research, descriptive results reveal suspensions disproportionately accrue to African American students, students from economically disadvantaged families, and students with documented disabilities. The associations between suspensions and later attitudes in the current study do not differ substantively based on race, gender, disability, or economic disadvantage (see Limitations section for more on this). However, even if the psychological effects of suspension are equally negative regardless of group membership, disproportionate exposure to suspension adds to prior disadvantages, leading to greater social inequality based on origins and disability. Thus, the dispensation of suspensions could serve as a turning point that leads vulnerable adolescents to become even more socially and academically isolated.

Fifth, this study highlights suspension’s influence on attitudes as a potential mechanism for gender differences in social and behavioral development, as proposed by DiPrete and Buchmann (2013). Because boys are more likely than girls to be suspended, resulting declines in boys’ attitudes and emotional engagement might help explain gender gaps in later academic achievement and attainment.

Finally, this study adds evidence in support of normative perspectives of discipline. Some scholars frame school suspension policy as an extension of the criminal justice system, in which those in power dispense punitive measures as if offenders and would-be offenders are rational actors influenced by punishment and deterrence (Garland 2002). Yet rather than deferring to school authorities’ decisions, which according to deterrence theory should lead to corrective behaviors, suspended students’ trust in school authorities lessened after exclusion, and they were more likely to doubt the general intentions of adults in school. This study, coupled with theories of deviance described earlier, casts doubt on the theory of deterrence and reinforces the normative perspective that context determines whether suspended students think of schools as authoritatively inconsequential institutions with unfair disciplinary practices.

Social Policy Implications

Debates about suspension policies persist between advocates promoting school safety and order and critics arguing against discrimination and exclusion. These are not abstract debates but play out concretely in communities and schools. For example, in Framing Dropouts, Fine (1991:50) describes a conversation she prompted among school administrators and staff that represents these kinds of debates. In an unlikely analogy, a dean of students likened his job to that of a pilot throwing hijackers off a plane. Two administrators derided the “liberal tendencies” of a guidance counselor who condemned disproportionality in school suspensions. These administrators stressed “how really dangerous these kids are.” To some degree, students with problem behaviors do distract teachers and reduce instructional time (Davis and Jordan 1994). Violence in school affects test scores and grades by creating threatening and unstable learning environments (Burdick-Will 2013). Yet school staff’s perceptions of student disobedience or disrespect, not school violence, account for the vast majority of office referrals resulting in suspension (Lewis et al. 2010; Skiba et al. 2011).

For these and many other reasons, proponents of social justice reforms (e.g., Ryan 1976) are hesitant to direct attention toward individual students’ beliefs and behaviors because those most likely to receive such attention are usually the victims of social forces that constrain their actions. It would be ill advised to implicate student beliefs and behaviors as the only components of the suspension question. Yet those beliefs and behaviors reveal a great deal about school contexts and the relationships students enjoy or endure. As England (2016) reminds us, “sometimes the social becomes personal.” This means focusing on psychological processes that affect individuals’ life outcomes need not be victim blaming. Students’ attitudes about school are by definition personal, but we can also view them as a signal of other issues present in schools (e.g., discrimination, disproportionate responses to disrespectful students). Bryk and Schneider (2004) identify relational trust as a key factor in advancing high educational expectations and outcomes for students in US schools. My findings suggest suspensions harm students’ trust in adults at school. With this in mind, an examination of students’ beliefs about relationships in school might say as much or more about the school as they do about its students.

Regardless of the causes, state- and school-level administrators have the power to change or reduce exclusionary practices that are counterproductive to growth in students’ emotional engagement. Critics of exclusionary school policies have called for replacing suspensions with more inclusionary solutions to student disciplinary problems, such as restorative justice and positive behavior support programs (e.g., Gonzalez 2012; Safran and Oswald 2003). Other institutional strategies seek to reduce barriers for marginalized students and their families so they feel welcomed and accepted in schools before perceived problem behaviors arise. For example, investments in early social and emotional support for children can mitigate students’ psychological vulnerabilities, foster positive relationship building in school, and head off subsequent disciplinary consequences (Carrell and Carrell 2006; Reback 2010). Further solutions include implementation of compensatory school-based interventions that help build trust between students and teachers (Okonofua et al. 2016; Yeager et al. 2014), school-wide initiatives to build positive and trusting relationships among students and school staff (Bryk and Schneider 2004), and interventions that help improve the negative attitudes of students who are disengaging from school (see Pyne et al. 2018). The effects of such investments will be clearer through further study of students’ attitudes in relation to school discipline.

Limitations

The primary limitation of this work is the threat of unobserved variables confounding the relationship between suspension and changes in attitudes. Because this study uses correlational data, it suffers from the same critiques about omitted variable bias as much of the previous literature on the effects of school suspensions. Concerning the associations between beginning-of-year attitudes and later suspension, I reiterate that my interest is more descriptive than diagnostic. Even if earlier or concurrent experiences explain part or all of the associations between attitudes and later suspension, they would only obscure or supplant the signal that proximal measurements of attitudes send about students at risk of exclusion from school.

Concerning end-of-year attitudes, the influence of unobserved priors is much more detrimental to the interpretation of my findings. Unlike previous suspension studies, I address this limitation in three ways. First, I mitigate the potential effects of prior experiences on future attitudes through the inclusion of lagged student attitudes captured prior to the suspension measurement window. Experiences prior to this window that might affect end-of-year attitudes are likely reflected in beginning-of-year attitudes, which I account for in these models. Second, I account for prior year disciplinary referrals and suspensions, which in addition to measuring disciplinary involvement can serve as proxies for behavior patterns that occurred prior to the observed suspensions and attitudes in the year under study. However, a remaining weakness is that I do not observe disciplinary involvement in even earlier prior grades, so I cannot conclusively identify study year suspended students who have had no school suspensions in the past. Third, to address remaining weaknesses, I calculate the bias necessary to invalidate the statistically significant results for the effect of suspensions on academic attitudes. These checks suggest the association between suspensions and changes in academic attitudes is a moderate and legitimate effect that is unlikely to be invalidated by the most relevant unobserved covariates, conditional on observed priors.

Additional limitations are due to sample selection, sample size, and assignment to treatment. These results come from a single urban school district. School climate and suspension policies and practices vary across districts, so results might not represent the experiences of students in other locations. Additionally, some readers might wonder whether suspensions shape attitudes differently by student race, gender, disability status, or family income. The number of suspended students during the measured period was quite small (n = 53), and moderation analyses (not shown) were inconclusive. Whether suspensions psychologically affect all students in similar ways or differences are simply undetectable due to very small cell sizes is an open question. Finally, although the current study uses only control group students to examine the associations between suspensions and later attitudes, bias might still exist due to spillover effects of treatment within schools. Given treatment group students in the sixth-grade cohort generally reported increases in school trust and identification with school due to the intervention, any spillover effects would likely have resulted in an underestimation of the associations between suspension and changes in attitudes observed in the current control only sample.

Conclusion

This study contributes to knowledge of how exclusion influences the lives of young people. Adolescents’ emotional engagement, measured by their academic attitudes, might decline due to suspension from school. Students’ behaviors can contribute to suspension rates, but institutional policies and the perceptions of school decision makers also contribute to which students experience suspension. Regardless of the causes, suspensions are likely to have unintended psychological consequences for adolescents. Rather than acting as corrective sanctions, suspensions might harm students’ trust in school authorities and their academic identities. This harm likely extends beyond the school itself, adding to the overall effects of disadvantage in their adult lives. If the results observed here hold in future studies, more research could begin to focus on addressing suspended students’ engagement in school in addition to confronting disciplinary practices and the social constraints affecting students at risk of exclusion.

Footnotes

Appendix A. Supplemental Descriptive Tables

Table A2.

Exploratory Factor Analysis of Academic Attitude Items.

Variable	School Trust	Social Belonging	External Locus of Control	Identification with School
Teachers are fair	.63	.20	−.05	.20
Adults at school care	.67	.21	−.04	.16
Students are supported	.54	.36	−.07	.16
People accept me	.14	.53	−.04	.06
Feel comfortable	.31	.62	−.09	.20
I belong in my school	.29	.63	−.02	.17
Outsider in my school	−.13	–.53	.25	−.18
Getting grades you want is luck	−.02	−.05	.77	−.07
Difference in good and bad grades is luck	−.04	−.07	.79	−.12
Takes luck to be outstanding	−.04	−.04	.75	.00
Important to do well	.17	.15	−.06	.68
Want to do well	.19	.16	−.16	.68

Note: Loadings over .40 are in bold. Factors rotated using promax method.

Appendix B. Full Regression Results (Online)

Table B2.

Associations between Suspension and Changes in Academic Attitudes (N = 885).

	School Trust	Social Belonging	External Locus of Control	Identification with School
Out-of-school suspension	−.47*** (.14)	.01 (.14)	.07 (.11)	−.54*** (.14)
Lagged academic attitude	.40*** (.03)	.51*** (.03)	.54*** (.03)	.36*** (.03)
Free/reduced-price lunch eligible	−.13 (.08)	.03 (.08)	.31*** (.07)	−.11 (.08)
Student with disability	.05 (.10)	.06 (.10)	.32*** (.08)	−.12 (.10)
Limited English proficiency	.15 (.11)	.06 (.11)	.35*** (.09)	−.05 (.11)
Male	.10 (.06)	.22*** (.06)	−.04 (.05)	−.13* (.06)
Race/ethnicity
Asian	.13 (.11)	.08 (.11)	−.06 (.09)	.07 (.11)
African American	−.14 (.10)	−.10 (.10)	.07 (.08)	.06 (.10)
Latino	−.08 (.11)	−.08 (.10)	−.13 (.08)	.03 (.11)
Prior year disciplinary referrals	−.00 (.01)	−.02 (.01)	−.00 (.01)	.00 (.01)
Prior year suspensions	.03 (.08)	.08 (.08)	−.07 (.06)	.00 (.08)
Constant	−.31* (.12)	−.05 (.12)	.02 (.10)	.02 (.13)
R²	.25	.29	.48	.18

Note: Each column represents a linear model regressing each academic attitude on current year suspension, beginning-of-year attitude, prior year disciplinary referrals, prior year suspensions, demographic characteristics, and school fixed effects. Academic attitudes are z-scored values. The suspension variable indicates whether a student was suspended at least once between the two administrations of the academic attitudes survey. Standard errors are in parentheses.

p < .05. ***p < .001.

Acknowledgements

This work was supported by: U.S. Department of Education (grant numbers R305A110136, R305C050055 to Geoffrey Borman) and the Spencer Foundation (grant number 201500044 to Geoffrey Borman). The content is solely the responsibility of the author and does not necessarily represent the official views of the supporting agencies. I first thank Geoffrey Borman for supporting this study and Christopher Rozek for his insights into the social-psychological concepts I present in this study. Special thanks to Geoffrey Borman, Sarah Bruch, Daanika Gordon, Eric Grodsky, Michael Halpin, and Christopher Rozek for their comments during the development of this manuscript. Additional thanks to Geoffrey Borman, Sarah Bruch, Dominique Bradley, Sara Dahill-Brown, Adam Gamoran, Jeffrey Grigg, Paul Hanselman, Christopher Rozek, and the rest of the research project team for conducting the work that produced these data. Finally, I thank the editor and three anonymous reviewers for their constructive and critical feedback, which resulted in a much stronger paper.

Research Ethics

An institutional review board approved the research conducted on human subjects used for this manuscript, and this research was carried out in a way that is consistent with the ethical standards articulated in the 1964 Declaration of Helsinki and Section 12 of the American Sociological Association Code of Ethics. All human subjects and their parent or guardian gave their informed consent prior to participation in the research. Participants were allowed to leave the study at any time. Adequate steps have been taken to protect participants’ confidentiality.

Author Biography

Jaymes Pyne is a Ph.D. candidate in the Department of Sociology and a former Interdisciplinary Training Program IES predoctoral fellow at the University of Wisconsin-Madison. His research uses a life-course approach to examine how social policies and inequalities shape personal beliefs and actions.

References

Arum

Richard

. 2003. Judging School Discipline: The Crisis of Moral Authority. Cambridge, MA: Harvard University Press.

Becker

Howard

1963. Outsiders: Studies in the Sociology of Deviance. New York: The Free Press.

Bourdieu

Pierre

Passeron

Jean-Claude

. 1990. Reproduction in Education, Society and Culture. Thousand Oaks, CA: Sage.

Bowles

Samuel

Gintis

Herbert

. 1976. Schooling in Capitalist America: Educational Reform and the Contradictions of Economic Life. New York: Basic Books.

Brown

Tara M.

2007. “Lost and Turned out: Academic, Social, and Emotional Experiences of Students Excluded from School.” Urban Education 42(5):432–55.

Bruch

Sarah K.

Soss

Joe

. 2018. “Schooling as a Formative Political Experience: Authority Relations and the Education of Citizens.” Perspectives on Politics 16(1):36–57.

Bryk

Anthony

Schneider

Barbara

. 2004. Trust in Schools: A Core Resource for Improvement. New York: Russell Sage Foundation.

Burdick-Will

Julia

. 2013. “School Violent Crime and Academic Achievement in Chicago.” Sociology of Education 86(4):343–61.

Carrell

Scott E.

Carrell

Susan A.

2006. “Do Lower Student to Counselor Ratios Reduce School Disciplinary Problems?” The B.E. Journal of Economic Analysis & Policy 5(1):1–24.

10.

Casella

Ronnie

. 2001. “Being down”: Challenging Violence in Urban Schools. New York: Teachers College Press.

11.

Davis

James Earl

Jordan

Will J.

1994. “The Effects of School Context, Structure, and Experiences on African American Males in Middle and High School.” Journal of Negro Education 63(4):570–87.

12.

DiPrete

Thomas

Buchmann

Claudia

. 2013. The Rise of Women: The Growing Gender Gap in Education and What It Means for American Schools. New York: Russell Sage Foundation.

13.

Dreeben

Robert

. 1968. On What Is Learned in School. Boston, MA: Addison-Wesley.

14.

Durkheim

Emile

. 1977. “On Education and Society.” Pp. 92–105 in Power and Ideology in Education, edited by Karabel

Halsey

A. H.

New York: Oxford University Press.

15.

Eccles

Jacquelynne S.

2004. “Schools, Academic Motivation, and Stage-environment Fit.” Pp. 125–53 in Handbook of Adolescent Psychology, edited by Lerner

R. M.

Steinberg

Hoboken, NJ: John Wiley & Sons.

16.

England

Paula

. 2016. “Sometimes the Social Becomes Personal: Gender, Class, and Sexualities.” American Sociological Review 81(1):4–28.

17.

Epp

Juanita R.

1996. “Schools, Complicity, and Sources of Violence.” Pp. 1–23 in Systemic Violence: How Schools Hurt Children, edited by Epp

J. R.

Watkinson

A. M.

Washington, DC: The Falmer Press.

18.

Farrington

Camille A.

Roderick

Melissa

Allensworth

Elaine

Nagaoka

Jenny

Keyes

Tasha Seneca

Johnson

David W.

Beechum

Nicole O.

2012. Teaching Adolescents to Become Learners: The Role of Noncognitive Factors in Shaping School Performance: A Critical Literature Review. Chicago, IL: University of Chicago Consortium on Chicago School Research.

19.

Findley

Maureen

Cooper

Harris M.

1983. “Locus of Control and Academic Achievement: A Literature Review.” Journal of Personality and Social Psychology 44(2):419–27.

20.

Fine

Michelle

. 1991. Framing Dropouts. Albany: State University of New York Press.

21.

Finn

Jeremy D.

1989. “Withdrawing from School.” Review of Educational Research 59(2):117–42.

22.

Finn

Jeremy D.

1993. School Engagement and Students at Risk. Washington, DC: National Center for Education Statistics.

23.

Foucault

Michel

. 1977. Discipline and Punish. New York: Pantheon.

24.

Frank

Kenneth A.

2000. “Impact of a Confounding Variable on a Regression Coefficient.” Sociological Methods and Research 29(2):147–94.

25.

Frank

Kenneth A.

Maroulis

Spiro

Duong

Minh Q.

Kelcey

Ben

. 2013. “What Would It Take to Change an Inference? Using Rubin’s Causal Model to Interpret the Robustness of Causal Inferences.” Educational Evaluation and Policy Analysis 35:437–60.

26.

Fredricks

Jennifer A.

Blumenfeld

Phyllis C.

Paris

Alison H.

2004. “School Engagement: Potential of the Concept, State of the Evidence.” Review of Educational Research 74(1):59–109.

27.

Garland

David

. 2002. The Culture of Control: Crime and Social Order in Contemporary Society. Chicago, IL: The University of Chicago Press.

28.

Gibbs

Jack P.

1975. Crime, Punishment, and Deterrence. New York: Elsevier.

29.

Gonzalez

Thalia

. 2012. “Keeping Kids in Schools: Restorative Justice, Punitive Discipline, and the School to Prison Pipeline.” Journal of Law & Education 41(2):281–336.

30.

Goodenow

Carol

. 1993. “The Psychological Sense of School Membership among Adolescents: Scale Development and Educational Correlates.” Psychology in the Schools 30:79–90.

31.

Hallinan

Maureen T.

2008. “Teacher Influences on Students’ Attachment to School.” Sociology of Education 81(3):271–83.

32.

Harackiewicz

Judith M.

Sansone

Carol

. 1991. “Goals and Intrinsic Motivation: You Can Get There from Here.” Advances in Motivation and Achievement 7:21–49.

33.

Hirschfield

Paul

. 2008. “Preparing for Prison? The Criminalization of School Discipline in the USA.” Theoretical Criminology 12(1):79–101.

34.

Johnson

Monica Kirkpatrick

Crosnoe

Robert

Elder, Jr.

Glen H.

2001. “Students’ Attachment and Academic Engagement: The Role of Race and Ethnicity.” Sociology of Education 74(4):318–40.

35.

Laub

John H.

Sampson

Robert J.

1993. “Turning Points in the Life Course: Why Change Matters to the Study of Crime.” Criminology 31(3):301–25.

36.

Lemert

Edwin M.

1951. Social Pathology: A Systematic Approach to the Theory of Sociopathic Behavior. New York: McGraw Hill.

37.

Lewis

Chance W.

Bonner

Fred A.

Butler

Bettie Ray

Joubert

Marcus

. 2010. “African American Male Discipline Patterns and School District Responses Resulting Impact on Academic Achievement: Implications for Urban Educators and Policy Makers.” Journal of African American Males in Education 1(1):7–25.

38.

Link

Bruce G.

Phelan

Jo C.

2001. “Conceptualizing Stigma.” Annual Review of Sociology 27(1):363–85.

39.

Losen

Daniel J.

Martinez

Tia E.

2013. “Out of School and off Track: The Overuse of Suspensions in American Middle and High Schools.”Los Angeles, CA: The Civil Rights Project.

40.

Merton

Robert K.

1936. “The Unanticipated Consequences of Purposive Social Action.” American Sociological Review 1(6):894–904.

41.

Mittleman

Joel

. 2018. “A Downward Spiral? Childhood Suspension and the Path to Juvenile Arrest.” Sociology of Education 91(3):183–204.

42.

Mizel

Matthew L.

Miles

Jeremy N. V.

Pedersen

Eric R.

Tucker

Joan S.

Ewing

Brett A.

D’Amico

Elizabeth J.

2016. “To Educate or to Incarcerate: Factors in Disproportionality in School Discipline.” Children and Youth Services Review 70:102–11.

43.

Morrison

Gale M.

Anthony

Suzanne

Storino

Meri

Dillon

Cynthia

. 2001. “An Examination of the Disciplinary Histories and the Individual and Educational Characteristics of Students Who Participate in an In-school Suspension Program.” Education & Treatment of Children 24(3):276–93.

44.

National Center for Education Statistics. 2016. Digest of Education Statistics. Washington, DC: U.S. Department of Education.

45.

Noltemeyer

Amity

Ward

Rose Marie

Mcloughlin

Caven

. 2015. “Relationship between School Suspension and Student Outcomes: A Meta-analysis.” School Psychology Review 44(2):224–40.

46.

Okonofua

Jason A.

Walton

Gregory M.

Eberhardt

Jennifer L.

2016. “A Vicious Cycle: A Social-psychological Account of Extreme Racial Disparities in School Discipline.” Perspectives on Psychological Science 11(3):381–98.

47.

Parsons

Talcott

. 1964. Social Structure and Personality. New York: The Free Press.

48.

Paternoster

Raymond

Iovanni

Leeann

. 1989. “The Labeling Perspective and Delinquency: An Elaboration of the Theory and an Assessment of the Evidence.” Justice Quarterly 6(3):359–94.

49.

Perry

Brea L.

Morris

Edward W.

2014. “Suspending Progress: Collateral Consequences of Exclusionary Punishment in Public Schools.” American Sociological Review 79(6):1067–87.

50.

Pyne

Jaymes

Rozek

Christopher S.

Borman

Geoffrey D.

2018. “Assessing Malleable Social-psychological Academic Attitudes in Early Adolescence.” Journal of School Psychology 71:57–71.

51.

Raftery

Adrian E.

1995. “Bayesian Model Selection in Social Research.” Sociological Methodology 25:111–63.

52.

Reback

Randall

. 2010. “Noninstructional Spending Improves Noncognitive Outcomes: Discontinuity Evidence from a Unique Elementary School Counselor Financing System.” Education Finance and Policy 5(2):105–37.

53.

Resnick

Michael D.

Bearman

Peter S.

Blum

Robert W.

Bauman

Karl E.

Harris

Kathleen M.

Jones

Tabor

Joyce

Beuhring

Trish

Sieving

Renee E.

Shew

Marcia

Ireland

Marjorie

Bearinger

Linda H.

Udry

J. Richard

. 1997. “Protecting Adolescents from Harm.” JAMA 278(10):823–32.

54.

Rios

Victor M.

2011. Punished: Policing the Lives of Black and Latino Boys. New York: NYU Press.

55.

Ryan

William

. 1976. Blaming the Victim. New York: Vintage.

56.

Safran

Stephen P.

Oswald

Karen

. 2003. “Positive Behavior Supports: Can Schools Reshape Disciplinary Practices?” Exceptional Children 69(3):361–73.

57.

Sampson

Robert J.

Laub

John H.

2005. “A Life- course View of the Development of Crime.” The ANNALS of the American Academy of Political and Social Science 602(1):12–45.

58.

Skiba

Russell J.

Horner

Robert H.

Chung

Choong-Geun

Rausch

M. Karega

May

Seth L.

Tobin

Tary

. 2011. “Race Is Not Neutral: A National Investigation of African American and Latino Disproportionality in School Discipline.” School Psychology Review 40(1):85–107.

59.

Spector

Paul E.

1988. “Development of the Work Locus of Control Scale.” Journal of Occupational Psychology 61:335–40.

60.

Staples

William G.

2014. Everyday Surveillance: Vigilance and Visibility in Postmodern Life. Lanham, MD: Rowman & Littlefield.

61.

Thapa

Amrit

Cohen

Jonathan

Guffey

Shawn

Higgins-D’Alessandro

Ann

. 2013. “A Review of School Climate Research.” Review of Educational Research 83(3):357–85.

62.

Tyler

Tom R.

2006. Why People Obey the Law. Princeton, NJ: Princeton University Press.

63.

Voelkl

Kristin E.

1997. “Identification with School.” American Journal of Education 105(3):294–318.

64.

Walton

Gregory M.

Cohen

Geoffrey L.

2007. “A Question of Belonging: Race, Social Fit, and Achievement.” Journal of Personality and Social Psychology 92(1):82–96.

65.

Way

Sandra M.

2011. “School Discipline and Disruptive Classroom Behavior: The Moderating Effects of Student Perceptions.” The Sociological Quarterly 52(3):346–75.

66.

Wolf

Kerrin C.

Kupchik

Aaron

. 2017. “School Suspensions and Adverse Experiences in Adulthood.” Justice Quarterly 34(3):407–30.

67.

Yeager

David S.

Purdie-Vaughns

Valerie

Garcia

Julio

Apfel

Nancy

Brzustoski

Patti

Master

Allison

Hessert

William T.

Williams

Matthew E.

2014. “Breaking the Cycle of Mistrust: Wise Interventions to Provide Critical Feedback Across the Racial Divide.” Journal of Experimental Psychology 143(2):804–24.