Exploring Racial Equity in Education: The Impact of Teacher Racial Bias on Student–Teacher Relationship Quality and the Opportunity for School Social Workers

Abstract

Despite the important role public schools serve in students’ lives, schools are an epicenter of racialized traumatic events for youth of color. School social workers are well positioned to lead racial equity in their schools and intervene to support more equitable outcomes for all students. Ample research has demonstrated educational inequities for these students, but no study to date has examined the impact of teacher racial bias on the quality of student–teacher relationships in U.S. pre-kindergarten through 12th grade schools. The purpose of this study was to understand the possible impact of interpersonal racism in the classroom, by examining whether teachers’ perceptions of their own implicit and explicit racial bias related to the quality of their relationships with students, and if teacher–student racial mismatch moderated this relationship. Current teachers (N = 115) completed an online computer survey that included measures of implicit and explicit racial bias, teacher–student racial mismatch, and the quality of student–teacher relationships. Data were analyzed using multiple linear regression in SPSS. Results show teachers’ explicit racial bias was a statistically significant predictor of the quality of student–teacher relationships. Implications are shared for school social work practice and research to advance educational justice.

Educational settings in the United States are increasingly scrutinized for their central role in sustaining educational inequities for students of color. Students’ educational and entangled lifetime success are determined heavily by their racial and ethnic identities due to the inherent racism embedded in U.S. public schools (Kohli, 2014; Kohli et al., 2017). Today, students of color who are marginalized based on their racial or ethnic identity may experience traumatic events in the school setting as a result of how peers, school staff, teachers, and others interact with students (Kohli et al., 2017). Nationally, state legislatures have targeted critical race theory (Crenshaw et al., 1996) across PK–12 and higher education contexts (Filimon & Ivănescu, 2024). As of 2024, 20 states, including Florida, Kentucky, Mississippi, and Texas, have placed formal restrictions on whether and how teachers can discuss politicized topics (Diliberti & Schwartz, 2024; Woo et al., 2023). Therefore, school social workers (SSWs) are increasingly called on to intervene in support of racial equity, through times of school integration (Colca et al., 1981), the school-to-prison pipeline (Phillippo & Crutchfield, 2021; Stone, 2017), and toward the pursuit of educational justice (Ball, 2021; Ball & Skrzypek, 2020).

Literature Review

Racism in Schools

Scholars argue that racism in schools, such as discrimination, race-based hate crimes, and differential treatment, are growing contributors to educational inequities in the preschool to 12th grade setting (Civitillo et al., 2024). For example, students of color who report experiencing discrimination, up to five times a day in some studies (English et al., 2020), have poorer academic outcomes (Civitillo et al., 2024). Differential treatment in schools has also led to parallel differences in high school graduation and dropout rates among students of color (McFarland et al., 2020). Racism in schools can impact student safety, academic outcomes, and their mental health (Golden et al., 2018; McWhirter et al., 2018; Weeks & Sullivan, 2019). To remedy student experiences of racism in schools, teachers, including teachers of color, often work to eliminate racism in the classroom but still experience internalized racism themselves (Kohli, 2014).

Teacher Racial Bias and Racial Mismatch in the Classroom

While there are several forms of racism, at the individual and system levels, teacher racial bias is one form considered as interpersonal racism occurring between two or more individuals (Race Forward, 2014). Teacher racial bias is defined for the purpose of this study as the attitudes and beliefs, both implicit and explicit, held by teachers based solely on a student’s race or ethnic identity (Chin et al., 2020; Warikoo et al., 2016). Despite the apparent nature of explicit biases (Warikoo et al., 2016), implicit biases cannot be readily recognized yet are critical to teachers’ evaluations and treatment of students (Benson & Fiarman, 2020; Chin et al., 2020). Broadly, teacher biases toward race, physical appearance, gender, and other identities have impacted student outcomes in schools repeatedly (Childs & Wooten, 2023), such as disciplinary and academic grades, test scores, and evaluations (Gershenson et al., 2016; Pena-Shaff et al., 2019; Quinn, 2017). Another growing concern building from discussions of racism in school is the discrepancy between teacher and student racial or ethnic identities, commonly referred to as teacher–student racial mismatch in the classroom, and the influence this may have on student outcomes. For instance, poor evaluations of student behavior, academic outcomes, and relationships among teachers and students have been found based on teachers’ race (Fox, 2015). However, no study to date has explored how teacher racial bias may relate to a crucial outcome essential for student learning and success in school (student–teacher relationship quality) and whether racial mismatch in the classroom changes this relationship.

Conceptual Framework and Theoretical Underpinnings

School Social Work and Racial Equity in Schools

The updated National School Social Work Practice Model 2.0 (Tan & School Social Work Association of American [SSWAA], in press) makes explicit the historical narrative of SSWs of racial equities and barriers to equitable student outcomes and builds upon the original model’s purpose of uniting SSWs in a shared understanding of their roles and responsibilities (Frey et al., 2012). SSWs’ responsibility to antiracist practice is engrained in the model’s focus areas (e.g., school racial climate, disparate educational outcomes) and practice activities (e.g., facilitation of professional development). The model depicts the broad responsibilities that SSWs hold, yet the ways in which SSWs engage in antiracist practice and understanding racial equity in their local school(s) context are not well understood. Recently, researchers have called on SSWs to take a more prominent role in educational justice movements, leveraging the larger professions obligation to antiracist practice and suggesting racial equity frameworks specific to school social work (Ball, 2021; Phillippo & Crutchfield, 2021). Historically, school social work as a specialized area of practice has struggled to overcome barriers to moving beyond broad recommendations to action and changes in professional practice (Colca et al., 1981; Stone, 2017). Therefore, this study shares insight to one area of racial inequity in schools that SSWs should consider investigating, interpersonal racism and teacher racial bias, and the extent to which student experiences and outcomes are impacted.

Research to date and several theories that ground this research, including critical race theory (Crenshaw et al., 1996) and the theory of racialized organizations (Ray, 2019), provide the foundation for the research questions and included concepts of focus in this study. This framework considers the ways in which teacher racial bias might influence student–teacher relationships in the classroom, shown in Figure 1. This model is a portion of a larger conceptual model that is under development: A Model of Contributing Factors to Racial Inequities in School, which expands on school, teacher, and student factors contributing to educational inequities found at the school (e.g., grade-level retention) and the student (socioemotional) outcomes.

Current Study

Educational justice requires both advancement and critical examination of racism in schools (Ball, 2021; Kohli et al., 2017). Despite ample evidence of racial discrimination and differential educational outcomes for students of color, more research is needed to understand what school factors contribute to experiences of racism in schools (English et al., 2020). A cross-sectional quantitative research design was employed to examine racial equity practice in schools, specifically one aspect of teacher racial bias, its presence and possible relationship to the understudied outcome of student–teacher relationship quality, with the following two research questions: (1) Do U.S. teachers’ perceptions of their own implicit and explicit racial bias predict student–teacher relationship quality? (2) Does racial mismatch in the classroom between the student and teacher moderate the relationship between teacher racial bias and the student–teacher relationship quality?

Method

An online survey of teachers currently employed in a U.S. preschool to 12th grade public school setting was administered with measures of teacher racial bias (implicit and explicit), teacher–student racial mismatch in the classroom, and student–teacher relationship quality scales to understand the possible contribution of teacher racial bias to differences in the quality of student–teacher relationships (institutional review board approval #2098012). A cross-sectional survey was used as the research design, which collects data at one point in time, often used to test exploratory relationships between concepts (e.g., teacher racial bias and the quality of student–teacher relationships; Creswell & Creswell, 2018).

Data Collection and Participant Characteristics

A computer survey of teachers was conducted in the fall of 2022 using Qualtrics. Teachers were recruited using purposive sampling from any area in the United States via email, social media, and district communications to participate in the study. Due to recruitment through social media and broader efforts, a response rate cannot be calculated. Participants who completed the survey were entered into a raffle for six $50 gift cards. A majority of participants (N = 115) self-identified as female (63 percent), followed by male (36 percent) and nonbinary (1 percent). Participants were ages 30 to 39 years old (38 percent), 20 to 29 years old (27 percent), or over 40 years old (35 percent). Many participants self-identified as White (75 percent), African American (16 percent), or Latinx/Hispanic (17 percent). A few self-identified as biracial, Asian, Native American/Alaska Native, and Native Hawaiian/Pacific Islander. Teachers responded from 36 states with a majority of responses from those employed in South Carolina, California, Florida, and Texas. Most taught in elementary (55 percent) or middle schools (29 percent). Participants primarily held either a bachelor (n = 43, 38 percent) or a master’s degree (n = 54, 46 percent). Teachers taught for an average of 10.5 years (SD = 8.3).

Survey Measures and Study Variables

The online survey included seven sections with approximately 50 questions. The survey was pilot tested with a set of current teachers (see Childs, 2023, for more information).

Implicit Association Tests (IATs)

IATs (Greenwald et al., 1998; α = .82 in this study) are a common method for evaluating implicit bias and were used to measure teachers’ implicit bias toward students’ race, specifically Black–White implicit racial biases. IATs use reaction time to images and word associations to estimate unconscious or subconscious beliefs and attitudes toward people (Carpenter et al., 2019; Greenwald et al., 1998). The IATs were programmed into Qualtrics following Carpenter and colleagues’ (2019) detailed procedures for embedding IATs in survey collection software (e.g., Qualtrics, REDCap).

Feelings Thermometer

The feelings thermometer (warmth) has been utilized in nearly all prior research in tandem with IATs to assess explicit racial bias for or against particular racial and ethnic identities, although it is not a valid and reliable tool (Leitner et al., 2016). Participants rank from one to 10 their own feelings about White and Black people, then these ratings are subtracted to provide a warmth score.

Bayesian Racism Scale (Racism Scale)

The Racism Scale (Uhlmann et al., 2010) was used to understand teachers’ beliefs in traditional racial stereotypes or myths, where explicit racial bias emerges to justify differential treatment and standards for people who identify as a different racial or ethnic identity (Litam & Balkin, 2021). Greater scores on the Racism Scale indicate a higher likelihood the individual holds negative explicit racial biases toward racial and ethnic minority groups (Litam & Balkin, 2021). The scale contains six questions and has been validated in several studies showing good reliability (α = .80); however, lower reliability was seen in this study (α = .57 in this sample). Each statement is rated on a Likert scale, ranging from 1 = strongly disagree to 7 = strongly agree (Litam & Balkin, 2021).

Color-Blind Racial Attitudes Scale (Color-Blind Attitudes)

The Color-Blind Attitudes Scale includes 20 total items rated on a six-point Likert ranging from 1 = strongly agree to 6 = strongly disagree (α = .75 overall in this study; Neville et al., 2000). The scale is made up of three dimensions: racial privilege, institutional discrimination, and blatant racial issues; however, use of the subdimensions alone resulted in low reliability historically. Overall scores on the Color-Blind Attitudes Scale suggest high to no colorblind views, and levels of possible racial awareness.

Student–Teacher Relationship Scale (Relationship Scale): Dependent Variable

The Relationship Scale is a brief version of a longer measure often used to assess the quality of relationships between the teacher and child (Pianta, 2001) and was used to assess the outcome of interest in the study. The Relationship Scale includes two aspects of student–teacher relationships, conflict and closeness (α = .87 and α = .85, respectively, in this study). The Relationship Scale includes 15 questions on which teachers rate each question from 1 = definitely does not apply to 5 = definitely applies. The questions in this study were modified to replace “child” with “students” in order to approximate their relationship with their students collectively (rather than for each student).

Teacher–Student Racial Mismatch in the Classroom: Moderator

Racial mismatch in the classroom was approximated based on racial and ethnic categories commonly used by the National Center for Education Statistics (McFarland et al., 2020) with five options (e.g., African American, White) and then expanding the “two or more races” option into two options, “biracial” and “multiracial.” Responses were categorized into mismatch in the classroom or congruence (e.g., the same race and ethnicity between most of students and the teacher).

Demographic Control Variables and Data Security

Demographics included questions about teachers’ own identity (e.g., race, gender, age), teaching experience (e.g., grade, school setting, school state, and type of school), as well as some questions about their exposure to and awareness of the concept of teacher racial bias. Out of all the demographics, five were employed as controls in this study due to prior research that found differences in student outcomes (Hill et al., 2019; Phillippo et al., 2024). In the creation of the online Qualtrics survey, reCAPTCHA bot detection settings were added to the beginning of the survey and required completion prior to continuing on to the survey questions. This feature provided scores (0–1) with greater than 0.5 indicating high likelihood the participant was a human. Another question helped to ensure a human participant by asking participants to enter a 10-digit number.

Data Analysis

Survey responses were imported and de-identified in SPSS (Version 29.0.2.0). There were 1,097 raw responses prior to data management. After screening for current U.S. teachers, the eligibility criteria, and consent to participate in the study, 263 responses were removed (n = 834). Then, responses were reviewed for malicious actors such as bots or nonhuman completers by embedding the built in reCAPTCHA tool in Qualtrics and searching for indicators commonly found in research to suggest malicious actors (Teitcher et al., 2015; Xu et al., 2022). Fifteen malicious actor indicators were used to create a total score, then responses with 2+A were removed from the study sample (n = 349). No sole indicators were used to remove a survey response; however, most malicious actors were found to have a suspicious name or email, illogical response to the open-ended questions, a duplicate IP address, or low reCAPTCHA score indicating nonhuman (<.05).

Multivariate Statistical Analysis Approach

Multiple regression was used to understand the association between teacher racial bias and student–teacher relationship quality with one moderator: racial mismatch in the classroom. Analysis assumptions were reviewed and no violations were found. Survey responses were eliminated using case-wise deletion on study variables (Afghari et al., 2019) until only complete responses remained (N = 115). Post hoc power analysis using G*Power indicated the model held appropriate power to continue with the following analysis (0.98 to 0.99). Two study null hypotheses were examined: (1) The amount of variance contributed to factors included in the study model does not differ significantly from 0 (H₀: R² = 0). (2) Each model regression predictor variable coefficient does not differ significantly from 0 (H₀: β_k = 0).

Moderation Analysis

To assess racial mismatch in the classroom between the teacher and the majority of their students acting as a moderator, Baron and Kenny’s (1986) method was used. Five controls were included in the study regression model (gender, age, years taught, years at current school, and awareness of explicit racial biases) and the primary predictor variables, including implicit and explicit racial bias measures, were added to the model. Then, interaction terms were created by mean centering the predictor variables reducing issues of multicollinearity and multiplying by the moderator variable, racial mismatch in the classroom (Memon et al., 2019) to be input into the regression model.

Results

Statistical analyses included a multiple regression (N = 115), followed by a moderation analysis, to examine the relationship between five predictors and the outcome of interest (student–teacher relationship quality) when controlling for teacher demographic factors. Predictors included measures of teacher racial bias, both implicit (IATs) and explicit (warmth, Racism Scale, color-blind attitudes) measures. The regression model included five teacher demographics as control variables, and interaction terms were created for each predictor-moderator pair (e.g., Racism Scale*racial mismatch in the classroom). Two primary null hypotheses were tested: (1) the variance explained by the model (H₀: R² = 0) and (2) each predictor coefficient (H₀: β_k = 0), is not significantly different from zero. After conducting the regressions, post hoc assumptions including distribution of residuals and multicollinearity were checked and did not show indication of violations.

Variable Means and Correlations

Prior to running and analyzing the study regression model, descriptives and correlations were calculated for each of the included study variables with close examination of the dependent variable. Teachers held a preference toward White individuals in the Black–White IATs (d-score: M = –0.36, SD = 0.34). Teachers also held moderate racist views (Racism Scale: M = 3.30; SD = 1.01; Likert 1–7) and moderate color-blind attitudes (M = 3.10, SD = 0.65; rated 1–6). Teachers reported closeness in their relationships with students (M = 4.18, SD = 0.60) and some conflict in the student–teacher relationship from their perspective (M = 2.58, SD = 0.80; see Table 1).

Multiple Regression without Moderation

The study regression model with one implicit racial bias variable (Black–White IATs) and three explicit racial bias variables (warmth, Racism Scale, and color-blind attitudes), and racial mismatch in the classroom predicted 30 percent of the variability in the dependent variable: Relationship Scale (a) [R² = .302; adj. R² = .270, p < .001; F(5, 109) = 9.413, p < .001]; (b)[R² = .364; adj. R² = .303, F(5, 104) = 2.032, p = .080] with five controls. The study model hypothesis was rejected indicating at least one predictor variable showed statistical significance. Upon examination of the model, only the Racism Scale was statistically significant. Higher scores on the Racism Scale were associated with lower Relationship Scale scores (β = –2.694, p = .049). The remaining study predictor variables did not show statistical significance in the model regressions.

Multiple Regression Testing for Moderation Results

Following the addition of four interaction terms, one for each predictor and the moderator (racial mismatch in the classroom), to the overall study model, no significant change in the amount of variance accounted. Neither the model predictors nor the interaction terms were statistically significant in the model. These results show racial mismatch in the classroom did not act as a moderator between teacher racial bias and the quality of student–teacher relationships (see Table 2).

Discussion

The purpose of this survey of pre-kindergarten through 12th grade teachers was to contribute to the growing involvement of SSWs in schoolwide racial equity practices to support the educational justice movement by better understanding interpersonal racism in school classrooms. Despite including several indicators of teacher racial bias, in this study only one survey measure of teachers’ explicit racial bias, as measured by the Racism Scale, was a significant contributor to the quality of student–teacher relationships as reported by teachers. Teachers’ higher reports of racist attitudes predicted poorer student–teacher relationships from the teachers’ perspective. This study finding was similar to prior research in the area of racial biases, which has indicated that teachers with negative views toward students of color act on those beliefs and attitudes in their classroom decision making and treatment of students (Quinn, 2017; Rueda, 2015). For example, in another study, teachers’ explicit racial attitudes were found to influence how teachers viewed student motivation and academic ability (Quinn, 2017; Rueda, 2015). Further, results of this current study suggest that teachers’ explicit racial biases, where teachers hold and possibly perpetuate racial stereotypes (see Valencia, 2010, for information on the theory of deficit thinking), could manifest in continued oppression of students of color in school settings. Yet, in contrast to racist attitudes, teachers who held color-blind attitudes or race-neutral views did not have the same effect as holding racist attitudes on the quality of student–teacher relationships, a finding that requires further examination to understand color-blind attitudes.

More surprising were the contrasting findings related to teachers’ implicit racial biases, which did not relate to differences in the quality of student–teacher relationships in this study sample. Teachers’ IATs captured a preference toward White people (87 percent), over Black or African American people, which suggests implicit racial bias is present in the classroom. This finding aligns with ample research to date showing everyone, including teachers, holds these racial biases (Chin et al., 2020; Denessen et al., 2022). However, the evidence on the possible impacts of teacher implicit racial bias have been mixed (Denessen et al., 2022), and similarly in this study, did not demonstrate a significant impact on one outcome, the quality of student–teacher relationships. In past research, these biases often resulted in poor academic test scores but had differential effects depending on the subject area (e.g., math versus reading; Chin et al., 2020; Pena-Shaff et al., 2019). Despite these findings, several explanations can be offered to help explain why implicit racial biases were not influential in this specific study. For instance, the nature of racial biases, and particularly those that are unconscious or subconscious, is difficult to capture, and teachers may not fully understand their biases’ presence in or influence on their classroom environment. Given recent student reports of school experiences and student–teacher relationships show differences in quality based on student race or ethnicity (Burrell-Craft et al., 2022), it is possible that teachers in this study simply reported higher quality relationships because they were not aware of the racial biases they hold. Therefore, there is a dire need to further explore whether teachers currently employ strategies to become aware of and reduce their racial bias in the classroom, through professional development, trainings, or personal efforts and education. Further, SSWs might be leveraged as a resource for working to better understand how racial bias is showing up in schools.

Racial mismatch in the classroom between teacher and the majority of their students was also investigated in this study to understand if this factor acted as a moderator between teacher implicit and explicit racial biases and their perceptions of the quality of student–teacher relationships. Unlike past studies, which found racial mismatch at a school or classroom level impacted teacher perceptions of grades, standardized test scores, and student motivation, racial mismatch did not act as a moderator to the quality of student–teacher relationships (Fox, 2015; Gershenson et al., 2016; Pena-Shaff et al., 2019). Teacher racial mismatch in the classroom did not have a positive or negative impact on the student outcome of interest, meaning teachers whose race or ethnicity matched their students did not equate to higher quality relationships. It is important this study finding be examined further, as this teacher sample was limited in terms of teacher racial and ethnic diversity. And therefore, variations in teacher–student racial mismatch could not be explored further in this study sample (i.e., Black teachers with majority Black student classrooms). However, prior research has shown teachers who self-identify as Black did not hold any implicit racial biases toward students from any racial or ethnic group (Chin et al., 2020). Yet, it is not clear why racial congruence, or racial or ethnic representation as the teacher, might influence these student outcomes or if this finding holds as additional school-level factors and multiple perspectives (e.g., students) are examined.

Study Limitations

This study is not without limitations, primarily, the limited sample of teachers predominantly from one region of the United States and elementary school settings. Greater generalizability could be attained by sampling a larger and more geographically representative sample of preschool through 12th grade teachers. In addition, best practices to ensuring data integrity were used through employing the reCAPTCHA and additional indicators of possible malicious actors, yet some may still be present in the final sample despite these safeguards because of social media recruitment methods. Future research must carefully consider the drawbacks of certain recruitment methods. Last, measures used in this study were valid and reliable, but study concepts such as teacher implicit bias are known to be difficult to measure and subject to immense social desirability bias. Discussion of racial bias continues to be a divisive and vulnerable topic for teachers. Due to this sensitivity, teachers could have attempted to choose “correct” answers when responding to IATs, in particular (Carpenter et al., 2019; Greenwald et al., 1998). Future research might include additional data sources on teacher beliefs and attitudes toward diversity, equity, and inclusion, and more reliable measures of warmth toward racial and ethnic groups in the United States.

Implications for Research and Practice

Racial equity in education can only be advanced with specific school social work training and support to reduce or eliminate teacher and school staff racial bias in schools. Findings of this study provide important insight to the critical role SSWs may hold in educational spaces. The National School Social Work Practice Model 2.0 is particularly calling on SSWs to adopt antiracist practices, work to eliminate racism, and employ clinical skills to help dismantle inequities and advance educational justice for students of color (Ball, 2021; Crutchfield et al., 2020; Tan & SSWAA, in press; Teasley et al., 2021). Results of this study demonstrate how student outcomes and teacher reports can be used to examine the impact of racial bias in the classroom. Moving forward, SSWs and school staff together can begin to explore teacher racial bias, school racial climate, and other indicators of the presence of racism in schools to identify an intervention point. A starting point for schools may also include the accountability measures and nonacademic indicators in the Every Student Succeeds Act (2015). For instance, SSWs may help school leadership disaggregate school data, such as student safety, relationships, school connectedness, and others, not typically reported by student racial and ethnic subgroup in the same ways as academic indicators. These nonacademic indicators may be a launching point for SSWs to leverage in discussions, and then action, to improve student experiences and school outcomes, particularly for marginalized youth.

Schools of social work may be the first exposure or formal opportunity to train and support SSWs to engage in and lead racial equity practices in schools. Yet, division has infiltrated school contexts, which increasingly demonstrate how critical a comprehensive social work response is to confrontation of inequities in education settings (Ball & Skrzypek, 2020; Crutchfield et al., 2020). Further, recent accreditation standards are in support of propelling antiracist practices among current and future social workers, and these explicit competencies aim to ensure social workers are adequately prepared to enter the workforce (Council on Social Work Education, 2022). One way that SSWs in training may be better supported is through exposure to coursework and practicum activities that will develop self-awareness of their own racial bias, identity, and the influence those will have on their future social work practice to ultimately foster these conversations among school staff as referred to across professional activities in the National School Social Work Practice Model 2.0 (Murphy et al., 2024; Tan & SSWAA, in press). More research and practical approaches are needed to both evaluate existing preparation programs’ efforts to embed antiracist practice and leverage SSWs as racial equity agents to explicitly address one of the key social work challenges of working to eliminate racism (Teasley et al., 2021). More collaboration is also needed to understand how current practitioners are engaged in antiracist practice and implement racial equity approaches in schools.

Conclusion

Educational spaces are critical areas for the exploration of teacher bias and opportunities for SSWs to implement with support of school staff evidence-based interventions to eliminate racial bias in the classroom and help alleviate racism in schools. Findings of this study demonstrate teacher explicit racial bias predicted significantly lower student–teacher relationship quality, while implicit racial bias and teacher–student racial mismatch in the classroom did not impact teachers’ own perceptions of relationship quality. Findings also provide additional evidence of racism in schools and recommendations for how to train and prepare SSWs to engage in racial equity practices in the context of the National School Social Work Practice Model 2.0.

Acknowledgment

This work was supported by the University of South Carolina College of Social Work, which provided funding for research incentives.

References