Fixed Effect Estimates of Student-Teacher Racial or Ethnic Matching in U.S. Elementary Schools

Abstract

We used student fixed effects and statistical controls to investigate whether U.S. elementary students (N = 18,170) displayed greater academic achievement, social-emotional behavior, or executive functioning and were more likely to receive gifted or special education services when taught by teachers of the same race or ethnicity. We observed mostly null effects for student-teacher racial or ethnic matching across the study’s 12 dependent measures in analyses adjusting for Type 1 error. For the full sample, matching resulted in lower science achievement (effect size [ES] = −.03 SD). For Black students, matching resulted in fewer internalizing problem behaviors (ES = 0.18 SD). We observed null effects for Hispanic students. Robustness checks including those stratified by race or ethnicity and biological sex or by prior levels of low or high level of achievement, behavior, or executive functioning largely supported the study’s null findings. Exceptions were that matching resulted in fewer externalizing problem behaviors (ES = 0.22 SD) for Black girls and lower academic achievement (ES range = −0.04 to −0.14 SD) and fewer externalizing and internalizing problem behaviors (ES range = 0.24 to 0.33 SD) for students who had previously displayed low levels of academic, behavioral, or executive functioning. Collectively, the analyses provide limited support for student-teacher racial or ethnic matching as a school-based policy to address educational disparities in U.S. elementary schools.

Students who are Black or Hispanic are more likely to display lower academic achievement while attending U.S. elementary schools and so experience fewer educational and societal opportunities over time (Irizarry, 2015; McFarland et al., 2017; Morgan et al., 2016; National Assessment of Educational Progress [NAEP], 2020). These achievement gaps are large by kindergarten (Garcia, 2015; Morgan et al., 2016; von Hippel et al., 2018). Black and Hispanic students are disproportionality exposed to less family wealth, greater residential segregation, lower health and well-being, and lower access to well-resourced elementary schools because of historical and ongoing racialized policies and practices (Heilig & Holme, 2013; Krieger et al., 2020; Reardon, 2021; Schulz et al., 2020). Exposure to these factors substantially explains early achievement gaps (Fryer & Levitt, 2004, 2013; Morgan et al., 2016). Economic policies that reduce exposure to less-resourced environments (e.g., expanded benefits and additional child tax credits), which should disproportionately benefit Black or Hispanic students (Parolin et al., 2020), may potentially help to address early achievement gaps (Morgan 2016; Reardon, 2021).

Additional factors that also help to explain achievement gaps during elementary school include cultural misunderstandings, bias, and stereotypes that are more likely to occur when Black or Hispanic students are taught by White teachers (Dee, 2004, 2005; Gershenson et al., 2021). The resulting cultural asynchrony (Blair et al., 2016) may lead to lower quality teacher-student relationships, lower perceptions of student ability, and more biased and negative assessments of student social-emotional behavior (Nguyen & Le, 2022; Rasheed et al., 2020). Teachers are more likely to have lower expectations for Black and Hispanic students than White or Asian students (Tenenbaum & Ruck, 2007) including in analyses adjusting for independently assessed achievement (Cherng & Halpin, 2016). Black or Hispanic students are more likely to hold lower expectations for their own educational attainment when taught by such teachers (Cherng, 2017). Black and Hispanic elementary school students currently have few opportunities to be taught by teachers of the same race or ethnicity (Gershenson et al., 2017; Yarnell & Bohrnstedt, 2018) because of the limited racial and ethnic diversity in the U.S. teacher workforce (National Center for Educational Statistics, 2021; U.S Department of Education, 2016).

A potential focus of school-based policies to address achievement gaps and other educationally relevant disparities in U.S. elementary schools may be to increase opportunities for Black and Hispanic students to be taught by teachers of the same race or ethnicity (Gershenson et al., 2021; Joshi et al., 2018; Harbatkin, 2021). Below, we detail theoretical support for student-teacher racial or ethnic matching as well as rigorous and replicated evidence of matching’s positive effects on academic, behavioral, and social-emotional functioning (e.g., Dee, 2004, 2005; Nguyen & Le, 2022). We then report on additional evidence of null or negative effects (e.g., Joshi et al., 2018; Redding, 2021) and recently identified substantive and methodological limitations in the existing work (Driessen, 2015; Redding, 2019). We justify the current study based on the existing work’s conflicting findings and limitations.

Theoretical and Empirical Support for Student-Teacher Racial or Ethnic Matching’s Potential Benefits

Theoretically, exposure to teachers of the same race or ethnicity should provide Black and Hispanic students with teachers who have higher educational expectations (Fox, 2016; Gershenson et al., 2016) and who are better able to serve as role models, mentors, and cultural translators (Egalite et al., 2015; Goldhaber et al., 2015). Student-teacher racial or ethnic matching should result in greater cultural understanding and referents that increase instructional effectiveness by teachers, more positive student-teacher relationships, greater student engagement and responsiveness, lower stereotype threat, and greater teacher advocacy (Gershenson et al., 2021; Redding, 2019).

The mechanisms underlying student-teacher racial or ethnic matching’s potential benefits have been characterized as either “passive” and “active” teacher effects (Dee, 2004, 2005). Passive teacher effects refer to mechanisms by which teachers of the same race or ethnicity, regardless of their specific behaviors, result in students adopting more positive educational perceptions, increased receptivity, and greater engagement in classroom activities. Passive teacher effects can also occur from increasing feelings of academic self-worth by students, which should result in greater academic achievement (Joshi et al., 2018). Passive teacher effects may result from role-modeling and reduced stereotype threats (Dee, 2004, 2005; Joshi et al., 2018; Nguyen & Le, 2022).

Active teacher effects refer to specific behaviors that teachers engage in because they are of the same race or ethnicity as their students (Dee, 2004). Active teacher effects occur when teachers consciously or unconsciously interact more positively with students of the same race or ethnicity (Harbatkin, 2021). Examples include providing students of the same race or ethnicity with more frequent attention and feedback, having higher academic and behavioral expectations, being more likely to adopt asset-instead of deficit-based views, and to use more praise and fewer reprimands (Dee, 2004, Gershenson et al., 2021; Harbatkin, 2021).

A substantial body of rigorous empirical work repeatedly finds that Black and Hispanic students benefit from being taught by teachers of the same race or ethnicity. For example, analyses from a large experiment of Tennessee students from kindergarten to 3^rd grade indicated that Black students displayed greater reading and mathematics achievement when taught by Black teachers (Dee, 2004). The study’s findings suggested a dose-response relation in which students taught by same-race teachers for additional years displayed increasingly greater achievement. Additional analyses of these same data indicated that racial matching helped reduce achievement gaps during kindergarten (Penney, 2017). Students taught by teachers of the same race are also less likely to be rated as inattentive or disruptive in classrooms (Bates & Glick, 2013; Dee, 2005; Wright et al., 2017) and to be disciplined (Lindsay & Hart, 2017; Linday et al., 2021) as well as more likely to display higher course grades (Harbatkin, 2021). Analyses of a nationally representative sample of U.S. elementary school students followed from kindergarten to 2^nd grade self-reported closer relationships with their teachers when taught by teachers of the same race or ethnicity (Nguyen & Le, 2022). Positive effects of student-teacher racial or ethnic matching are evident across a wide range of educationally relevant indicators (e.g., academic achievement, classroom behavior, attendance, exclusionary discipline) including in analyses of state-wide or nationally representative datasets (Downey & Pribesh, 2004; Easton-Brooks., Lewis, & Zhang, 2009; Eddy & Easton-Brooks, 2011; Holt & Gershenson, 2019; Lindsay & Hart, 2017; Lindsay et al., 2021; Rasheed et al., 2020; Yarnell & Bohrnstedt, 2018).

Academically struggling as well as academically talented students are thought to especially benefit from being taught by teachers of the same race or ethnicity (Dee, 2004; Egalite et al., 2015; Gershenson et al., 2021). Being taught by teachers of the same race or ethnicity may result in disadvantaged Black or Hispanic students holding more positive beliefs about their educational or societal possibilities (Dee, 2004). Black and Hispanic teachers may be more likely to recognize the talents of students of the same race or ethnicity (Gershenson et al., 2021). Analyses using student fixed effects and statistical controls for potential confounds indicated that elementary students in Florida who had been academically struggling displayed greater reading and mathematics achievement (ES = 0.02 SD) if taught by teachers of the same race or ethnicity, suggesting that “race-matching may be particularly beneficial for lower performing students” (Egalite et al., 2015, p. 50). Students attending Wisconsin schools were more likely to receive special education services when attending schools with greater proportions of teachers of color (Fish, 2019). Regression analyses of nationally representative data indicated that Black students attending U.S. elementary schools were more likely to be assigned to gifted education for reading when taught by Black teachers (Grissom & Redding, 2016).

Existing Work’s Substantive and Methodological Limitations

Yet the extent to which student-teacher racial or ethnic matching should be the focus of school-based policies designed to address achievement gaps or other educational disparities in U.S. elementary schools is currently unclear (Driessen, 2015; Redding, 2019). Positive effects, when observed, are often quite small (Egalite et al., 2015; Harbatkin, 2021; Joshi et al., 2018; Redding, 2019). For example, analyses of data from 3^rd-5^th grade students attending schools in Florida indicated that racial and ethnic matching resulted in statistically significant effects of 0.004 and 0.01 SD on measures of reading and mathematics achievement in analyses adjusting for student and course fixed effects and additional statistical controls (Egalite et al., 2015). Among matching’s small positive effects, the largest effects typically observed are for Black students (Harbatkin, 2021; Joshi et al., 2018; Redding, 2019). Estimates of the impact of being taught by Black teachers on the academic achievement of Black elementary school students in analyses using student and course fixed effects as well as statistical controls for measured confounds were 0.01 and 0.03 of a SD in reading and mathematics, respectively (Egalite et al., 2015). Other estimates using student fixed effects and controls including for teacher quality indicted that elementary Black students in Tennessee taught by Black teachers averaged 0.04 and 0.08 of a SD higher on measures of reading and mathematics achievement, respectively (Joshi et al., 2018). Analyses of students in Tennessee found that being assigned Black teachers results in about a 3-6 percentile point increase in reading and mathematics achievement (Dee, 2004).

Other findings indicate that student-teacher racial or ethnic matching can have null or negative effects (Driessen, 2015; Ehrenberg et al., 1995; Nguyen & Le, 2022; Redding, 2019, 2022). Driessen’s (2015) synthesis indicated that, of 15 studies analyzing for student-teacher racial or ethnic matching’s effects on objective measures (e.g., academic achievement), 20% reported negative effects, 47% reported null effects, and 33% reported positive effects. Analyses of a nationally representative sample of kindergarten and first grade students failed to indicate that Black students who had Black teachers displayed relatively greater reading or mathematics achievement over time in analyses adjusting for teacher fixed effects (Fryer & Levitt, 2004). Although positive effects of student-teacher ethnic matching have been observed for both Black and Hispanic students on measures of externalizing problem behaviors, no statistically significant effects have been observed for either Black or Hispanic students on measures of internalizing problem behaviors, interpersonal skills, self-control, disruptive or inattention behaviors (Redding, 2019). Although student-teacher racial and ethnic matching results in more positive self-reports of closeness with teachers, null findings were concurrently observed on teacher ratings of behavioral or academic functioning as well as direct measures of academic achievement in analyses of a nationally representative sample followed from kindergarten to second grade (Nguyen & Le, 2022). Analyses of data from Florida using student and course fixed effects and statistical controls including for teacher quality indicated that Hispanic elementary students taught by Hispanic teachers displayed lower (i.e., −0.009 SD) reading achievement (Egalite et al., 2015). Null effects were observed for mathematics achievement. Positive effects of student-teacher racial or ethnic matching have failed to replicate in re-analyses using additional statistical controls (Howsen & Trawick, 2007). Results from studies analyzing nationally representative datasets provide less consistent evidence of matching’s potential benefits (Redding, 2019). Driessen’s (2015) synthesis of 24 quantitative studies indicated that “there is as yet little unambiguous empirical evidence” (p. 179) of positive effects of student-teacher racial or ethnic matching. When observed, positive effects are more likely to occur on subjective teacher evaluations rather than on objective achievement measures (Driessen, 2015). Little evidence was observed to indicate that Black or Hispanic students taught by teachers of color were more likely than White students to receive special education services (Fish, 2019). Although Black students attending U.S. elementary and middle schools were more likely to be assigned to gifted programs in reading when taught by Black teachers, this effect was not observed in mathematics. Hispanic students were not more likely to be assigned to gifted programs in either reading or mathematics when taught by Hispanic teachers (Grissom & Redding, 2016).

As reported in two recent syntheses, the existing work on student-teacher racial or ethnic matching also has substantive and methodological limitations (Driessen, 2015; Redding, 2019). Substantively, the few studies analyzing data from elementary school students have mostly been limited to examining for potential effects on measures of reading or mathematics achievement as well as of attention or disruptive behavior (Redding, 2019). Yet student-teacher racial or ethnic matching might be expected to positively impact other educationally relevant indictors during elementary school including interpersonal skills and self-control as well as whether teachers of the same race or ethnicity might be more likely to advocate for Black or Hispanic students to receive additional supports through gifted or special education (Egalite & Kisida, 2018; Fox, 2016; Gershenson et al., 2016; Redding, 2019). Although executive functioning independently predicts both academic achievement and classroom behavior during elementary school (Morgan, et al., 2019a), whether racial or ethnic matching positively impacts executive functioning is currently unclear. To date, no studies have examined whether Black or Hispanic students attending U.S. elementary schools who are taught by teachers of the same race or ethnicity are more or less likely to receive special education services (Redding, 2019). Assessing for the effects of student-teacher racial or ethnic matching on a relatively limited set of educationally relevant indicators may be providing the field with an incomplete assessment of matching’s potential benefits.

Methodologically, relatively few studies allow for causal inference by using fixed effects to account for unobserved confounds (Dee, 2004, Egalite et al., 2015; Joshi et al., 2018; Penney, 2017). No study is currently available that has examined matching’s effects on special or gifted education service receipt using fixed effects (Fish, 2019; Grissom & Redding, 2016). The few studies using fixed effects to provide causal estimates have mostly analyzed state-level datasets of middle and high school students, particularly those attending schools in the U.S. South (Dee, 2004; Redding, 2019). Positive effects may be especially likely to be observed in the U.S. South because the region’s history of de jure and de facto segregation may have resulted in stronger group identities between Black students and Black teachers (Redding, 2019). Null effects may be more likely to occur between Hispanic students and Hispanic teachers due to greater variation in national or immigration backgrounds that constrain shared cultural understanding (Redding, 2019). Thus, the benefits of student-teacher racial or ethnic matching may not generalize to the population of students attending U.S. elementary schools. To what extent the potential benefits of student-teacher racial or ethnic matching are moderated by other factors including biological sex or prior academic achievement, behavior, or executive functioning during elementary school is also unclear (Hart, 2020; Redding, 2019). Collectively, the existing work’s conflicting findings and limitations suggest that student-teacher racial and ethnic matching’s potential benefits for address achievement gaps and other educationally relevant disparities in U.S. elementary schools have yet to be well-established. Analyses of a nationally representative sample that address the existing work’s substantive and methodological limitations would help clarify the potential of student-teacher racial or ethnic matching as a focus of educational policy.

Study’s Purpose

We investigated student-teacher racial and ethnic matching’s effects by analyzing data from U.S. elementary students (N = 18,170) including those who are Black or Hispanic participating in a multiyear cohort and for whom an extensive battery of 12 measures of academic achievement, social-emotional behavior, and executive functioning as well as both gifted and special education service receipt were repeatedly administered. We used student fixed effects to control for time-invariant unmeasured confounds as well as statistical control for time-varying measured confounds including student and teacher characteristics (i.e., a student’s age or change of schools, a household’s poverty status, a teacher’s educational level) to better estimate the potential effects of student-teacher racial or ethnic matching. In addition to examining matching’s effects for the full sample as well as subsamples of students who are Black (n = 2,400) or Hispanic (n = 4,590), we conducted a series of robustness checks including those stratified on race or ethnicity and biological sex as well as prior levels of achievement, behavior, or executive functioning. Although we expected to observe some positive effects of student-teacher racial or ethnic matching based on findings reported in prior work including two recent reviews of existing work (Driessen, 2015; Redding, 2019), we also expected any observed effects to be small (Egalite et al., 2015; Joshi et al., 2018; Redding, 2019) and to be more frequently observed on subjective teacher evaluations rather than on objective achievement measures (Driessen, 2015).

Method

Dataset

We analyzed the Early Childhood Longitudinal Study, Kindergarten Class of 2010-11 (ECLS-K: 2011). The ECLS-K: 2011 is a nationally representative cohort of students who entered U.S. kindergarten classrooms in 2010 or 2011. The U.S. Department of Education’s National Center for Education Statistics (NCES) maintains the ECLS-K: 2011 (Tourangeau et al., 2019). The ECLS-K: 2011 collected data on individual student’s achievement, social-emotional behavior, and executive functioning as well as data from parents, teachers, and school administrators across kindergarten to fifth grade. The ECLS-K: 2011‘s (N = 18,170, rounded to the nearest 10 to follow NCES procedures for protecting participant confidentiality) sample is large and racially, ethnically, and economically diverse. Data for the current study were drawn from the spring of kindergarten and the springs of first, second, third, fourth, and fifth grade.

We created a dummy variable indicating whether students and their teachers were of the same race or ethnicity within each survey wave. We first analyzed the full sample (N = 18,170) using this “matched versus not matched” variable to evaluate the total effects of student-teacher racial and ethnic matching. We then restricted our analyses to student-level observations of subsamples of Black/African American students (n= 2,400) and Hispanic students (n= 4,590). We did so to evaluate for matching’s effects within these two racial and ethnic groups. We restricted these analyses to Black/African American and Hispanic students because there were very few same-race student-teacher matches for other racial or ethnic groups (e.g., Asian American, Native American, or Pacific Islanders). Table 1 displays the full sample’s descriptive statistics. Supplemental Table A1 reports additional descriptive statistics for the subsamples of Black or Hispanic students.

Table 1.

Full Sample’s Descriptive Statistics (N = 18,170)

Variables	Kindergarten	1st Grade	2nd Grade	3rd Grade	4th Grade	5th Grade
Student-teacher racial or ethnic matching (reference = not matched)
Matched	0.63	0.63	0.63	0.62	0.62	0.63
Time-varying student characteristics
Age at assessment (in month)	73.56 (4.48)	85.49 (4.47)	97.57 (4.46)	109.09 (4.46)	121.07 (4.49)	133.10 (4.47)
School change status from prior grade (reference = did not change schools)
Changed schools	0.01	0.12	0.09	0.13	0.12	0.14
Household poverty status (reference = at or above 200% FPT)
Below 100% FPT	0.24	0.26	0.24	0.22	0.22	0.21
At or above 100% FPT, below 200% FPT	0.24	0.23	0.24	0.23	0.23	0.23
Time-varying teacher characteristics
Teacher’s biological sex (reference = female)
Male	0.02	0.03	0.06	0.08	0.10	0.12
Teacher’s highest education level (reference = bachelor’s degree or lower)
Master or higher degree	0.46	0.50	0.50	0.52	0.53	0.53
Years taught in school	14.20 (9.62)	14.84 (9.98)	15.19 (9.95)	14.38 (9.41)	14.22 (9.37)	14.18 (9.02)
Dependent variables
Reading achievement	69.40 (14.46)	95.23 (17.62)	112.66 (16.64)	121.18 (15.13)	129.59 (14.43)	136.63 (15.10)
Mathematics achievement	50.45 (13.26)	73.11 (15.48)	90.40 (17.73)	104.20 (17.56)	112.71 (17.47)	119.82 (17.24)
Science achievement	34.08 (7.36)	43.12 (10.27)	52.66 (11.56)	60.18 (11.87)	66.88 (11.84)	73.54 (12.64)
Externalizing problem behaviors (reverse scaled)	3.38 (0.62)	3.29 (0.61)	3.29 (0.62)	3.32 (0.61)	3.35 (0.60)	3.36 (0.59)
Internalizing problem behaviors (reverse scaled)	3.50 (0.49)	3.46 (0.50)	3.42 (0.52)	3.40 (0.53)	3.41 (0.55)	3.43 (0.52)
Self-control	3.20 (0.62)	3.23 (0.61)	3.23 (0.62)	3.26 (0.62)	3.28 (0.60)	3.29 (0.61)
Interpersonal skills	3.16 (0.64)	3.16 (0.64)	3.13 (0.66)	3.13 (0.66)	3.12 (0.65)	3.13 (0.65)
Approaches to learning	3.13 (0.68)	3.10 (0.69)	3.08 (0.70)	3.07 (0.71)	3.08 (0.70)	3.11 (0.70)
Cognitive Flexibility	15.25 (2.69)	16.12 (2.28)	6.75 (1.32)	7.14 (1.37)	7.65 (0.96)	8.00 (0.92)
Working memory	450.93 (30.12)	470.47 (24.77)	481.51 (21.81)	490.19 (21.30)	497.42 (21.05)	503.39 (21.80)
Gifted program	0.02	0.04	0.06	0.07	0.07	0.09
IEP	0.09	0.09	0.11	0.12	0.12	0.13

Note. Weight (w9c29p_2t290) applied. IEP = Individualized Education Program. FPT = Federal poverty threshold. SD reported in parenthesis.

Proportions are rounded to second decimal place based on NCES disclosure requirement.

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011(ECLS-K), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File

Student-teacher racial or ethnic matching may be a non-random event. That is, some students may be more likely to be taught by teachers of the same race or ethnicity during elementary school. Supplemental Table A3 displays socio-demographic characteristics of students and schools in the groups of students who did or did not experience matching with teachers of the same race or ethnicity during elementary school. Matched and un-matched students differed within each survey wave by race/ethnicity, primary language background, disability status, family socioeconomic status (SES), and school contexts. Contrasts between matched and non-matched students might be biased by these and additional but unmeasured confounds. This suggested the need for student fixed effects to better estimate the causal effects of student-teacher racial or ethnic matching.

Measures

Race and Ethnicity of Students and Teachers

A student’s race and ethnicity were surveyed during parent interviews in kindergarten and following waves if prior responses were missing. The ECLS-K: 2011 provided a series of dummy variables indicating whether student’s race was White, Black/African American, Asian, Native Hawaiian or Other Pacific Islander, American Indian or Alaska Native, and ethnicity was Hispanic or Latino. We further created six dummy variables indicating whether that student was (a) White, non-Hispanic, (b) Black/African American, non-Hispanic, (c) Asian, non-Hispanic, (d) Native Hawaiian or Other Pacific Islander, non-Hispanic, (e) American Indian or Alaska Native, non-Hispanic, and (f) Hispanic, no race specified.

Similarly, teacher race and ethnicity were self-reported during each survey wave. Each teacher was asked to indicate their ethnicity (“Are you Hispanic or Latino?”) and race (response options included American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, and White) separately. We further created the same six dummy variables indicating a teacher’s race and ethnicity. Based on these dummy variables, we were able to create final dummy variables (“Matched”) to indicating whether student and teacher’s race and ethnicity were matched or not across each wave. We separately restricted data to students whose race or ethnicity was Black/African American, non-Hispanic and Hispanic, in addition to the full sample analysis.

Academic Achievement

The ECLS-K: 2011 includes individually administered, untimed, and psychometrically well-validated measures of reading, mathematics, and science achievement. These measures were equated using item response theory. The reading achievement measure assessed basic reading skills (e.g., print familiarity), vocabulary, and reading comprehension. The mathematics achievement measure assessed a student’s conceptual knowledge, procedural knowledge, and problem-solving. The mathematics achievement measure included items on number sense, properties, and operations; measurement; geometry and spatial sense; data analysis, statistics, and probability; and patterns, algebra, and functions. The science achievement measure was designed to assess a student’s understanding of the physical, life, and Earth and space sciences as well as scientific inquiry. The reliability coefficients for these academic achievement measures were consistently high across each wave. These coefficients ranged from .86 to .95 for reading, .91 to .94 for mathematics, and .73 to .86 for science.

Social-emotional Behaviors

Teachers rated a student’s social-emotional behaviors using a modified version of the Social Skills Rating System (SSRS; Gresham & Elliott, 1990). The modified SSRS in the ECLS-K: 2011 consisted of five scales that represented a student’s classroom behaviors and social-emotional functioning. Externalizing Problem Behaviors assessed the extent of student engaged in arguing, fighting, acting impulsively, getting angry, and disturbing ongoing activities. Internalizing Problem Behaviors assessed the extent that a student exhibited anxiety, loneliness, low self-esteem, and sadness. Interpersonal Skills measured the extent that a student was able to get along with others, formed and maintained friendships, helped other children, showed sensitivity to the feelings of others, and expressed feelings, ideas, or opinions in positive ways. Self-control measured the extent that a student controlled his or her temper, respected the property of others, accepted the ideas of peers, and handled peer pressure. Approaches to Learning assessed the extent that a student displayed behavioral self-regulation by being able to keep his or her belongings organized, displayed eagerness to learn new things, adapted to change, persisted in completing tasks, paid attention, and followed classroom rules.

Teachers used a four-option frequency scale ranging from “never” to “very often” to rate a student’s behaviors. Higher scores indicated that the behavior occurred more frequently. We reserve coded the externalizing and internalizing problem behavior scales for consistency of interpretation with the other three behavioral measures so that higher scores indicated that positive behaviors occurred more frequently. The internal consistency reliability coefficients for these five measures were consistently high across waves. These coefficients ranged from .86 to .89 for externalizing behaviors, .73 to .79 for internalizing behaviors, .85 to .88 for interpersonal skills, .79 to .82 for self-control, and .91 to .92 for approaches to learning (Tourangeau et al., 2019).

Executive Functioning

Student executive functioning was individually assessed by trained NCES field staff. Executive functions are cognitive processes that help students to regulate attention, emotion, and behavior while learning in the classroom. Cognitive flexibility was individually assessed by the Dimensional Change Card Sort (DCCS) (Zelazo, 2006; Zelazo et al., 2013). Because the DCCS was changed from physical administration to computerized administration in second grade, we standardized the scores so that the distributions at each wave had a M of 0 and a SD of 1. The DCCS displays strong test-retest reliability (e.g., 0.90-0.94; Beck et al., 2011). Working memory was individually assessed using the Numbers Reversed subtest of the Woodcock-Johnson III Tests of Cognitive Abilities (Woodcock et al., 2001). The Numbers Reversed task has strong test-retest reliability (e.g., .69-.87; Vought, 2011). Students repeated sets of orally presented single-digit numbers in reverse order. Students were first given five two-digit sequences. We used the W scores as recommended by the ECLS-K: 2011 codebook (Tourangeau et al., 2019). The W scale is useful for measuring growth. The W scale is a standardized scale with an M of 500 and an SD of 100.

Gifted or Special Education Service Receipt

Gifted education services. Teachers reported whether the student received instruction and/or related services in a gifted program for reading or language arts across each survey wave. Special education services. Teachers answered, “does this child have an IEP [Individualized Education Program] on record with the school” in each wave’s questionnaire. We used this response to indicate whether students were receiving special education services.

Statistical Controls

Use of time-invariant covariates (e.g., biological sex) was unnecessary because the student fixed effects controlled for time-invariant individual characteristics (Allison, 2009). We controlled for a student’s age at assessment and its squared value to allow for non-linear effects of age growth. We did so because age is time-variant and was related to the study’s dependent variables on a longitudinal scale. Consistent with prior studies examining student-teacher racial or ethnic matching (Egalite et al., 2015; Joshi et al., 2018), we also controlled for a set of time-varying teacher characteristics including a teacher’s sex, highest degree, and years of experience. Teachers self-reported their biological sex during each grade’s questionnaire. We used female as the reference group. Teachers also reported their completed highest level of education. Response options ranged from not completing high school to any advanced professional degree beyond a master’s degree. We created a dummy variable to indicate whether the teacher had a graduate degree (i.e., a master’s degree or higher). We counted the teacher’s years of experience based on the answer to the question asking how many years the teacher had been teaching. We also controlled whether a student changed schools compared to the prior grade and the family’s poverty status. We created a dummy variable using school ID in each grade. We then used 1 to indicate whether the student’s school ID changed compared to the prior grade, and 0 to indicate whether the school ID remained unchanged. The ECLS-K: 2011 provided a composite score indicating a student’s household poverty status with three categories in each wave. These three categories were: (a) below 100% of the federal poverty threshold (FPT); (b) at or above 100% of the FPT while below 200% FPL; and (c) at or above 200% of the FPT.

Missing Data

We used the NCES-provided sampling weight w9c29p_2t290 to adjust for non-response associated with the student assessment and teacher questionnaire data from spring of kindergarten to the spring of fifth grade. Use of this weight allowed for the maximum number of data sources included in our analyses while maintaining a large sample size (Tourangeau et al., 2019). Supplemental Table A2 indicates substantial attrition on the study’s dependent variables. Missing data also increased from kindergarten to fifth grade. For example, 5.44% of the reading achievement data were missing in kindergarten. This increased to 37.12% in fifth grade.

Supplemental Table A17 provides a full correlation matrix between all our missing dummy variables (1 = missing, 0 = not missing). Students with missing data on reading achievement in kindergarten were likely to have missing data on mathematics (r = .98) and science (r = .89) achievement, cognitive flexibility (r = .97), and working memory (r = .97) in kindergarten. Students with missing data on any one of our directly assessed dependent variables (i.e., measures of academic and executive functioning) were likely to be missing data on other directly assessed dependent variables during the same survey wave. Students with missing data on any of our teacher-reported dependent variables (i.e., social-emotional behavior, receipt of gifted or special education services) were also likely to be missing data on any of the other teacher-reported dependent variables during the same survey wave.

We used multiple imputation (MI) to address missing data. Use of MI can lead to a reasonable assumption of data that is missing at random. Use of MI also adjusts for attrition bias more effectively than using attrition weights (Davis-Kean, 2015). We used chained equations in Stata v. 17.0 to adjust for non-response bias. We imputed m datasets until the fraction of missing information divided by m was less than 0.01. Doing so maximized the relative efficiency of imputations (Woods et al., 2021). This led to m = 50 imputations.

Analyses

We used a student fixed effects model to causally estimate relations between student-teacher racial or ethnic matching and student achievement, social-emotional behavior, and executive functioning as well as gifted or special education service receipt. Student fixed effects eliminates potentially biasing unobservable student characteristics by estimating the coefficients within rather than between students (Allison, 2009). The basic equation is as follows:

$$Y_{it}={\alpha }_i+{\beta }_1{MATCHED}_{it}+{\beta }_2{AGE}_{it}+{\beta }_3{SAGE}_{it}+{\beta }_4{CHGSCH}_{it}+{\beta }_5{FPT}_{it}+{\beta }_6{TSEX}_{it}+{\beta }_7{TDEGREE}_{it}+{\beta }_8{TEXP}_{it}+{\epsilon }_{it}$$

where Y_it is the outcome score of student i, during year t. The α_i are student-specific intercepts that capture heterogeneities across students, MATCHED_it is the main predictor indicating student-teacher matching status (reference group is not matched), AGE_it and SAGE_it represents a student’s age at assessment and its squared value, CHGSCH_it and FPT_it represents a student’s school change status and their household poverty status. TSEX_it, TDEGREE_it, and TEXP_it represents a teacher’s biological sex, highest degree, and experience in years. In addition, ε_it is the time-varying error term. Students therefore served as their own controls (Allison, 2009).

We used the command xtreg in Stata v. 17.0 because we were able to analyze multiple waves and a large amount of data. This command used ordinary least squares regression with mean deviation scores. The method computed scores deviated from the mean of a student’s score on a specific predictor over time, longitudinally comparing the student before and after the student was taught by a teacher of the same race or ethnicity (Allison, 2009).

We also conducted a Hausman test (Allison, 2009). Results indicated that the fixed-effects model was preferable to the random-effects model. We used fixed logistic regression with the clogit command for the binary dependent variables of gifted and special education service receipt. This command fits logistic regression models to longitudinal data with the conditional likelihood method (Allison, 2009). Because students were clustered within schools, we used the fifth grade school ID to adjust clustered standard errors allowing for heteroskedasticity and autocorrelated errors within an entity but not correlation across entities (Cameron & Miller, 2015). We report standardized regression coefficients. We standardized the continuous variables prior to the model with a M of 0 and SD of 1 in each survey wave. We used the Benjamini-Hochberg (1995) procedure to avoid potential Type I errors resulting from multiple hypotheses testing.

Robustness Checks

We also conducted multiple robustness checks. We re-estimated our models after reducing the number of multiple comparisons by averaging the continuous dependent variables. We did so to reduce the potential for over-correction using the Benjamini-Hochberg procedure. We used three composite categories. These three categories were: (a) academic achievement (i.e., measures of reading, mathematics, and science achievement); (b) social-emotional behaviors (i.e., measures of externalizing and internalizing problem behaviors, interpersonal skills, self-control, and behavioral self-regulation); and (c) executive functioning (cognitive flexibility and working memory). Table 6 displays the estimated results for the full sample as well as for subsamples of only Black or Hispanic students. To assess for potential variation based on biological sex, we also fit separate models for subsamples of Black boys, Black girls, Hispanic boys, and Hispanic girls (see Supplemental Table A4 to A7).

Table 6.

Effects of Student-Teacher Racial or Ethnic Matching, Combined Outcomes for Full Sample, Black Students, and Hispanic Students

	Full Sample (N = 18,170)			Black Students (n = 2,400)			Hispanic Students (n = 4,590)
	A	B	EF	A	B	EF	A	B	EF
Matched (vs. not matched)	−0.03** (0.01)	0.02 (0.02)	−0.01 (0.02)	−0.03 (0.02)	0.10† (0.05)	−0.09 (0.05)	−0.03 (0.02)	−0.02 (0.03)	0.01 (0.03)
Time-varying student characteristics
Age (z score)	0.31*** (0.02)	−0.04 (0.04)	0.15*** (0.03)	0.30*** (0.05)	−0.06 (0.11)	0.02 (0.11)	0.30*** (0.03)	−0.01 (0.06)	0.14* (0.06)
Age2 (z score)	0.02 (0.02)	−0.01 (0.03)	−0.02 (0.02)	0.04 (0.04)	−0.02 (0.08)	−0.12 (0.09)	0.02 (0.03)	0.03 (0.05)	−0.04 (0.05)
School change status (reference = did not change schools)
Changed schools	−0.02* (0.01)	0.02 (0.02)	−0.02 (0.02)	−0.06* (0.02)	0.04 (0.05)	0.05 (0.06)	−0.01 (0.01)	0.02 (0.03)	0.00 (0.03)
Household poverty status (reference = at or above 200% FPT)
Below 100% FPT	0.00 (0.01)	0.00 (0.03)	0.01 (0.03)	0.01 (0.04)	−0.12 (0.08)	−0.09 (0.08)	−0.02 (0.02)	−0.03 (0.04)	0.06 (0.04)
At or above 100% FPT, below 200% FPT	0.01 (0.01)	0.01 (0.02)	0.03 (0.02)	0.02 (0.04)	−0.07 (0.07)	0.01 (0.06)	0.00 (0.02)	−0.02 (0.04)	0.05 (0.03)
Time-varying teacher characteristics
Teacher’s biological sex (reference = female)
Male	−0.01 (0.01)	−0.03 (0.02)	−0.02 (0.01)	−0.04 (0.03)	−0.04 (0.06)	−0.09 (0.06)	−0.02 (0.02)	−0.01 (0.03)	0.01 (0.03)
Teacher’s highest education level (reference = bachelor’s degree or lower)
Master or higher degree	0.00 (0.01)	0.00 (0.01)	0.00 (0.01)	−0.01 (0.02)	0.01 (0.04)	0.01 (0.03)	0.00 (0.01)	0.01 (0.02)	−0.02 (0.02)
Years taught in school (z score)	0.00 (0.00)	0.03*** (0.01)	0.00 (0.00)	0.00 (0.01)	0.04† (0.02)	0.01 (0.02)	0.01 (0.00)	0.01 (0.01)	0.01 (0.01)
Constant	0.08*** (0.01)	0.02 (0.02)	0.05** (0.02)	−0.42*** (0.03)	−0.30*** (0.06)	−0.24*** (0.06)	−0.26*** (0.02)	0.08* (0.03)	−0.13*** (0.03)
rho (intraclass correlation)	0.88	0.58	0.56	0.88	0.56	0.56	0.87	0.55	0.54

Note. A = academic achievement (i.e., reading, mathematics, and science achievement). B = behavior (i.e., externalizing problem behaviors, internalizing problem behaviors, self-control, interpersonal skills, and approaches to learning). EF = executive functioning (i.e., cognitive flexibility, working memory). FPT = Federal poverty threshold. Age² = student’s age squared. Intraclass correlation (rho) = how much variance is due to differences across grades.

Standardized coefficients were reported (standard error in parenthesis). Weight (w9c29p_2t290) and standard error adjusted (fifth grade School ID). Multiple imputation included 50 imputed datasets. We used the Benjamini-Hochberg (B-H) correction procedure to reduce Type I errors.

The † indicates the p-value became non-significant (at 5% significance level) after B-H correction (false discovery rate set at 5% level).

^*p <0.05,

^**p < 0.01,

^***p < 0.001

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011 (ECLS-K: 2011), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File.

To assess for heterogeneous treatment effects in which academically struggling or academically talented students including those who are Black or Hispanic might especially benefit from exposure to teachers of the same race or ethnicity (Egalite et al., 2015), we limited our data to low-performing (see Supplemental Table A8 to A10) and high-performing samples of students (see Supplemental Table A11 to A13) for each continuous dependent variable. We used a student’s kindergarten score of each continuous dependent variable and applied a 25% cut-off to identify students who had initially displayed either low or high levels of academic achievement, social-emotional behavior, or executive functioning. For example, we conducted the same fixed effect analyses for each continuous dependent variable for students whose kindergarten reading, mathematics, science achievement, externalizing problem behaviors (reverse coded), internalizing problem behaviors (reverse coded), self-control, interpersonal skills, approaches to learning, cognitive flexibility, or working memory scores were at the lowest 25% of the total score distribution on each measure of the dependent variable administered during kindergarten (see Supplemental Table A8). We conducted similar analyses for students who had displayed high (i.e., top 25%) levels of academic achievement, social-emotional behavior, or executive functioning in kindergarten (see Supplemental Table A11). We further restrained these specific analyses to only Black or Hispanic students who displayed low or high levels of academic achievement, social-emotional behavior, or executive functioning in kindergarten. We were unable to conduct fixed logistic regressions of receipt of gifted or special education services receipt for this subsamples of only Black or Hispanic students. This is because too few of these student groups experienced both receipt of gifted or special education service receipt and fluctuations in their exposure to Black or Hispanic teachers. To assess whether our results were sensitive to the use of student fixed effects, we alternatively fit conventional ordinary linear regression (OLS) and logistic regression models to the first-to-fifth grade data while controlling for a set of covariates (see Supplemental Table A14 to A16). Results from these alternative analyses were consistent with our main findings using student fixed effects.

Results

Descriptive Statistics

Table 1 indicates that about 63% of the full sample’s students experienced a teacher of the same race or ethnicity during any grade. Approximately half of the teachers had at least a master’s degree or higher across the surveyed grades. The average teaching experience was around 15 years.

Table 2 displays the prevalence of student-teacher racial or ethnic matching. Among White students, 92% experienced having a teacher of the same race during at least one elementary grade. The contrasting rate for Black or Hispanic students was 45%. White students were especially likely to experience matching throughout elementary school. Among White students, 16% and 39% had a teacher of the same race during five or six elementary grades, respectively. The contrasting rates for Black students were only 2% and 1%, respectively. Among Hispanic students, only 5% had a Hispanic teacher during five or six elementary grades.

Table 2.

Prevalence of Student-Teacher Racial or Ethnic Matching from Kindergarten to Fifth Grade for White, Black, and Hispanic Students

Student Race or Ethnicity	0 times	1 time	2 times	3 times	4 times	5 times	6 times
White, non-Hispanic (n = 8,490)	0.08	0.14	0.07	0.07	0.09	0.16	0.39
Black/African American, non-Hispanic (n = 2,400)	0.55	0.22	0.10	0.06	0.04	0.02	0.01
Hispanic (n = 4,590)	0.55	0.16	0.08	0.06	0.05	0.05	0.05

Note. Proportions are rounded to second decimal place based on NCES disclosure requirement.

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011(ECLS-K: 2011), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File

Estimates for the Full Sample

Table 3 presents the fixed effect models for the full sample. Student-teacher racial or ethnic matching displayed non-significant effects in reading and mathematics achievement, externalizing and internalizing problem behaviors, self-control, interpersonal skills, approaches to learning, cognitive flexibility, working memory, and gifted education service receipt. We observed a statistically significant negative effect of −0.03 SD (p < .05) on science achievement. Students were less likely to receive special education services when they were taught by a teacher of the same race or ethnicity. However, the effect for special education service receipt was statistically non-significant after Type I error adjustment. Repeatedly significant covariates included age, years of teaching experience, and teacher biological sex.

Table 3.

Effects of Student-Teacher Racial or Ethnic Matching, Full Sample (N = 18,170)

	Reading	Mathematics	Science	EPBa	IPBa	Self-control	Interpersonal skills	ATL	CF	WM	Gifted	IEP
Matched (vs. not matched)	−0.02 (0.01)	−0.02 (0.01)	−0.03* (0.01)	0.02 (0.02)	0.03 (0.02)	0.00 (0.02)	0.02 (0.02)	−0.01 (0.02)	−0.01 (0.02)	−0.01 (0.02)	1.09 (0.87-1.37)	0.78† (0.61-0.99)
Time-varying student characteristics
Age (z score)	0.25*** (0.02)	0.31*** (0.02)	0.26*** (0.02)	−0.05 (0.04)	−0.05 (0.04)	−0.04 (0.04)	0.00 (0.04)	0.00 (0.04)	0.18*** (0.04)	0.06† (0.03)	1.38 (0.71-2.69)	1.24 (0.80-1.92)
Age2 (z score)	0.02 (0.02)	0.00 (0.02)	0.04† (0.02)	−0.02 (0.03)	−0.02 (0.03)	0.01 (0.03)	0.00 (0.03)	0.01 (0.02)	−0.03 (0.03)	−0.01 (0.02)	1.35 (0.91-2.00)	1.02 (0.71-1.47)
School change status (reference = did not change schools)
Changed schools	−0.03† (0.01)	−0.03† (0.01)	−0.02† (0.01)	0.04† (0.02)	−0.01 (0.02)	0.04† (0.02)	−0.02 (0.02)	0.01 (0.02)	−0.03 (0.02)	−0.01 (0.02)	0.87 (0.67-1.13)	1.18 (0.96-1.45)
Household poverty status (reference = at or above 200% FPT)
Below 100% FPT	−0.02 (0.02)	0.01 (0.02)	0.01 (0.02)	−0.01 (0.03)	−0.05 (0.03)	0.02 (0.03)	0.01 (0.03)	0.04 (0.02)	−0.06 (0.04)	0.07† (0.03)	0.76 (0.50-1.16)	0.89 (0.61-1.31)
At or above 100% FPT, below 200% FPT	−0.01 (0.02)	0.01 (0.01)	0.02 (0.01)	0.00 (0.02)	0.00 (0.03)	0.02 (0.02)	0.02 (0.02)	0.03 (0.02)	0.01 (0.02)	0.03 (0.02)	0.82 (0.61-1.10)	0.98 (0.72-1.33)
Time-varying teacher characteristics
Teacher’s biological sex (reference = female)
Male	0.00 (0.01)	−0.02 (0.01)	−0.01 (0.01)	−0.01 (0.02)	0.02 (0.02)	−0.05* (0.02)	−0.07*** (0.02)	−0.07*** (0.02)	−0.02 (0.02)	−0.01 (0.01)	1.35† (1.05-1.75)	1.75*** (1.39-2.20)
Teacher’s highest education level (reference = bachelor’s degree or lower)
Master or higher degree	0.01 (0.01)	0.00 (0.01)	0.00 (0.01)	0.01 (0.01)	−0.02 (0.01)	0.00 (0.01)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	−0.01 (0.01)	1.18 (0.99-1.40)	1.12 (0.98-1.29)
Years taught in school (z score)	0.00 (0.00)	0.00 (0.00)	0.00 (0.00)	0.03*** (0.01)	0.00 (0.01)	0.03*** (0.01)	0.03*** (0.01)	0.02*** (0.01)	0.00 (0.01)	0.00 (0.00)	1.13** (1.04-1.23)	1.00 (0.93-1.07)
Constant	0.05*** (0.01)	0.03** (0.01)	0.06*** (0.01)	−0.03 (0.02)	0.01 (0.02)	−0.01 (0.02)	−0.01 (0.02)	0.00 (0.02)	0.04 (0.02)	0.01 (0.01)
rho (intraclass correlation)	0.77	0.83	0.79	0.58	0.36	0.49	0.47	0.59	0.38	0.56

Note. EPB = externalizing problem behaviors. IPB = internalizing problem behaviors. ATL = approaches to learning. CF = cognitive flexibility. WM = working memory. IEP = Individualized Education Program. FPT = Federal poverty threshold. Age² = student age squared. Intraclass correlation (rho) = how much variance is due to differences across grades.

For continuous outcomes, linear fixed effect regression models were fitted, and standardized coefficients were reported (standard error in parenthesis). For binary outcomes (gifted program and IEP), fixed logistic regression models were fitted, and odds ratio were reported (95% CI in parenthesis). Weight (w9c29p_2t290) and standard error adjusted (fifth grade School ID). Multiple imputation included 50 imputed datasets. We used the Benjamini-Hochberg (B-H) correction procedure to reduce Type I errors.

The † indicates the p-value became non-significant (at 5% significance level) after B-H correction (false discovery rate set at 5% level).

^aWe reverse coded externalizing and internalizing problem behaviors to have consistent direction with other outcomes. Higher scores indicate better behaviors.

^*p <0.05,

^**p < 0.01,

^***p < 0.001

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011 (ECLS-K: 2011), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File.

Estimates for Black Students

Table 4 displays the fixed effect models for Black students. Matching between Black students and Black teachers displayed non-significant effects across reading, mathematics, and science achievement. We observed statistically significant positive effects of 0.14 SD in externalizing problem behaviors (p < .01), and 0.18 SD in internalizing problem behaviors (p < .01). Thus, Black students were rated as displaying fewer externalizing and internalizing problem behaviors when taught by Black teachers versus non-Black teachers. Racial matching’s effect on externalizing problem behaviors but not on internalizing problem behaviors was statistically non-significant after Type I error adjustment. We observed no consistently significant student-teacher racial matching effects on self-control, interpersonal skills, approaches to learning, cognitive flexibility, or working memory. Directionally, Black students were more likely to receive gifted education services and less likely to receive special education services versus when taught by Black versus non-Black teachers. However, neither of these estimates were statistically significant.

Table 4.

Effects of Student-Teacher Racial Matching, Black Students (n = 2,400)

	Reading	Mathematics	Science	EPBa	IPBa	Self-control	Interpersonal skills	ATL	CF	WM	Gifted	IEP
Matched (vs. not matched)	−0.06 (0.03)	0.00 (0.03)	−0.03 (0.03)	0.14† (0.06)	0.18* (0.06)	0.01 (0.05)	0.09 (0.06)	−0.01 (0.04)	−0.13 (0.08)	−0.02 (0.05)	1.83 (0.88-3.81)	0.76 (0.41-1.44)
Time-varying student characteristics
Age (z score)	0.31*** (0.07)	0.28*** (0.07)	0.24** (0.07)	−0.02 (0.14)	0.00 (0.11)	−0.08 (0.13)	−0.04 (0.13)	−0.09 (0.11)	0.07*** (0.14)	−0.04† (0.10)	2.96 (0.57-15.39)	1.80 (0.52-6.24)
Age2 (z score)	0.02 (0.05)	0.00 (0.06)	0.10† (0.05)	−0.05 (0.09)	−0.01 (0.10)	−0.02 (0.09)	0.04 (0.09)	−0.05 (0.09)	−0.22 (0.13)	0.02 (0.08)	2.74 (0.66-11.42)	1.28 (0.61-2.69)
School change status (reference = did not change schools)
Changed schools	−0.06 (0.03)	−0.06† (0.03)	−0.08* (0.03)	0.06 (0.05)	0.05 (0.07)	0.09 (0.05)	−0.04 (0.05)	0.00 (0.05)	0.04 (0.07)	0.03 (0.05)	0.56 (0.25-1.26)	1.24 (0.79-1.95)
Household poverty status (reference = at or above 200% FPT)
Below 100% FPT	−0.02 (0.06)	0.01 (0.05)	0.04 (0.05)	−0.02 (0.09)	−0.21† (0.10)	−0.11 (0.09)	−0.11 (0.09)	−0.01 (0.08)	−0.22 (0.12)	0.06 (0.08)	0.47 (0.16-1.39)	0.31 (0.10-1.01)
At or above 100% FPT, below 200% FPT	0.00 (0.05)	−0.01 (0.04)	0.06 (0.04)	−0.02 (0.08)	−0.11 (0.09)	−0.09 (0.07)	−0.04 (0.07)	0.01 (0.07)	0.01 (0.08)	0.01 (0.07)	0.98 (0.46-2.08)	0.66 (0.23-1.85)
Time-varying teacher characteristics
Teacher’s biological sex (reference = female)
Male	−0.06 (0.02)	−0.05 (0.03)	−0.01 (0.04)	0.03 (0.07)	0.04 (0.08)	−0.08 (0.07)	−0.09 (0.06)	−0.09 (0.06)	−0.11 (0.08)	−0.04 (0.05)	1.29 (0.53-3.11)	1.35 (0.58-3.11)
Teacher’s highest education level (reference = bachelor’s degree or lower)
Master or higher degree	0.02 (0.02)	−0.01 (0.02)	−0.02 (0.02)	0.04 (0.04)	−0.01 (0.04)	0.00 (0.05)	0.01 (0.04)	0.00 (0.04)	0.06 (0.04)	−0.05 (0.03)	0.96 (0.57-1.61)	0.99 (0.65-1.51)
Years taught in school (z score)	−0.01 (0.01)	0.02 (0.01)	0.00 (0.01)	0.05* (0.02)	−0.01 (0.02)	0.05* (0.02)	0.04† (0.02)	0.02 (0.02)	−0.01 (0.02)	0.03 (0.02)	0.98 (0.73-1.32)	0.97 (0.79-1.20)
Constant	−0.25*** (0.04)	−0.51*** (0.03)	−0.48*** (0.04)	−0.45*** (0.07)	0.01 (0.08)	−0.33*** (0.07)	−0.28*** (0.06)	−0.31*** (0.06)	−0.16 (0.08)	−0.28*** (0.06)
rho (intraclass correlation)	0.79	0.82	0.79	0.56	0.35	0.49	0.46	0.55	0.42	0.55

Note. EPB = externalizing problem behaviors. IPB = internalizing problem behaviors. ATL = approaches to learning. CF = cognitive flexibility. WM = working memory. IEP = Individualized education program. FPT = Federal poverty threshold. Age² = student’s age squared. Intraclass correlation (rho) = how much variance is due to differences across grades.

For continuous outcomes, linear fixed effect regression models were fitted, and standardized coefficients were reported (standard error in parenthesis). For binary outcomes (gifted program and IEP), fixed logistic regression models were fitted, and odds ratio were reported (95% CI in parenthesis). Weight (w9c29p_2t290) and standard error adjusted (fifth grade School ID). Multiple imputation included 50 imputed datasets. We used the Benjamini-Hochberg (B-H) correction procedure to reduce Type I errors.

The † indicates the p-value became non-significant (at 5% significance level) after B-H correction (false discovery rate set at 5% level).

^aWe reverse coded externalizing and internalizing problem behaviors to have consistent direction with other outcomes. Higher scores indicate better behaviors.

^*p <0.05,

^**p < 0.01,

^***p < 0.001

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011 (ECLS-K: 2011), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File.

Estimates for Hispanic Students

Table 5 presents the fixed effect models among Hispanic students. Matching between Hispanic students and Hispanic teachers yielded null effects across most of the dependent variables including reading, mathematics, externalizing and internalizing problem behaviors, self-control, interpersonal skills, approaches to learning, cognitive flexibility, and working memory. We observed a significant negative effect of −0.05 SD on science achievement (p < .05). This effect was not statistically significant after adjusting for Type I errors. Directionally, Hispanic students were less likely to receive gifted or special education services when they were taught by Hispanic teachers. However, neither of these estimates was statistically significant.

Table 5.

Effects of Student-Teacher Ethnic Matching, Hispanic Students (n = 4,590)

	Reading	Mathematics	Science	EPBa	IPBa	Self-control	Interpersonal skills	ATL	CF	WM	Gifted	IEP
Matched (vs. not matched)	0.00 (0.02)	−0.02 (0.02)	−0.05† (0.02)	0.00 (0.03)	0.03 (0.03)	−0.05 (0.03)	−0.01 (0.03)	−0.03 (0.03)	0.00 (0.03)	0.02 (0.03)	0.81 (0.57-1.14)	0.90 (0.64-1.26)
Time-varying student characteristics
Age (z score)	0.26*** (0.04)	0.33*** (0.04)	0.22*** (0.05)	−0.06 (0.06)	−0.05 (0.08)	−0.06 (0.07)	0.05 (0.07)	0.10 (0.06)	0.22** (0.07)	0.02 (0.06)	2.49 (0.81-7.70)	1.04 (0.49-2.22)
Age2 (z score)	0.03 (0.03)	−0.02 (0.03)	0.04 (0.03)	0.03 (0.06)	0.06 (0.07)	0.05 (0.06)	−0.03 (0.05)	0.02 (0.05)	0.02 (0.06)	−0.09 (0.05)	0.86 (0.33-2.20)	1.09 (0.64-1.85)
School change status (reference = did not change schools)
Changed schools	−0.07*** (0.02)	0.00 (0.02)	0.01 (0.02)	0.05 (0.03)	−0.05 (0.04)	0.06 (0.03)	−0.01 (0.03)	0.02 (0.03)	−0.01 (0.04)	−0.01 (0.03)	0.77 (0.49-1.21)	1.30 (0.88-1.93)
Household poverty status (reference = at or above 200% FPT)
Below 100% FPT	−0.04 (0.03)	0.00 (0.03)	−0.03 (0.03)	−0.01 (0.04)	−0.07 (0.05)	−0.02 (0.05)	−0.05 (0.05)	0.05 (0.05)	0.03 (0.05)	0.07 (0.05)	0.68 (0.32-1.42)	1.20 (0.64-2.28)
At or above 100% FPT, below 200% FPT	−0.02 (0.03)	0.01 (0.02)	0.00 (0.03)	−0.02 (0.04)	−0.04 (0.05)	−0.02 (0.04)	−0.03 (0.05)	0.05 (0.04)	0.01 (0.04)	0.06 (0.04)	0.59 (0.31-1.10)	1.15 (0.67-1.96)
Time-varying teacher characteristics
Teacher’s biological sex (reference = female)
Male	−0.06† (0.02)	−0.03 (0.02)	0.00 (0.02)	−0.03 (0.03)	0.06 (0.04)	−0.03 (0.04)	−0.05 (0.04)	−0.05 (0.03)	−0.03 (0.03)	0.03 (0.03)	1.34 (0.87-2.06)	1.97* (1.34-2.90)
Teacher’s highest education level (reference = bachelor’s degree or lower)
Master or higher degree	0.02 (0.01)	−0.01 (0.01)	−0.01 (0.01)	0.02 (0.02)	−0.03 (0.03)	0.01 (0.02)	0.01 (0.02)	0.01 (0.02)	−0.01 (0.02)	−0.02 (0.02)	1.15 (0.81-1.63)	1.10 (0.86-1.40)
Years taught in school (z score)	0.00 (0.01)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	−0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	0.00 (0.01)	0.02 (0.01)	1.12 (0.95-1.31)	1.00 (0.88-1.14)
Constant	−0.21*** (0.02)	−0.25*** (0.02)	−0.32*** (0.02)	0.07 (0.03)	0.08 (0.04)	0.04 (0.04)	0.04 (0.04)	−0.03 (0.03)	−0.06 (0.04)	−0.20*** (0.03)
rho (intraclass correlation)	0.78	0.82	0.75	0.57	0.34	0.44	0.44	0.57	0.39	0.52

Note. EPB = externalizing problem behaviors. IPB = internalizing problem behaviors. ATL = approaches to learning. CF = cognitive flexibility. WM = working memory. IEP = Individualized Education Program. FPT = Federal poverty threshold. Age² = student’s age squared. Intraclass correlation (rho) = how much variance is due to differences across grades.

For continuous outcomes, linear fixed effect regression models were fitted, and standardized coefficients were reported (standard error in parenthesis). For binary outcomes (gifted program and IEP), fixed logistic regression models were fitted, and odds ratio were reported (95% CI in parenthesis). Weight (w9c29p_2t290) and standard error adjusted (fifth grade School ID). Multiple imputation included 50 imputed datasets. We used the Benjamini-Hochberg (B-H) correction procedure to reduce Type I errors.

The † indicates the p-value became non-significant (at 5% significance level) after B-H correction (false discovery rate set at 5% level).

^aWe reverse coded externalizing and internalizing problem behaviors to have consistent direction with other outcomes. Higher scores indicate better behaviors.

^*p<0.05,

^**p < 0.01,

^***p < 0.001

Source: U.S. Department of Education, National Center for Education Statistics (NCES), Early Childhood Longitudinal Study, Kindergarten Class of 2011 (ECLS-K: 2011), Kindergarten Through Fifth Grade Full Sample Restricted-Use Data File.

Robustness Checks

Table 6 presents estimates of matching’s effects on the combined continuous dependent variables for the full sample as well as for subsamples of only Black or Hispanic students. We observed a statistically significant negative effect of −0.03 SD (p < .01) on achievement for the full sample. For the subsample of only Black students, we observed statistically significant positive effect of 0.10 SD (p < .05) in behavior. However, this effect was non-significant after adjusting for Type I errors. For the subsample of only Hispanic students, we observed non-significant effects of student-teacher ethnic matching on achievement, social-emotional behavior, and executive functioning. These findings were consistent with those from our main analyses.

In supplemental analyses (Tables A4–A7), we further restrained our sample to Black boys, Black girls, Hispanic boys, and Hispanic girls. We then fitted the same student fixed effects regression models. Supplemental Table A4 displays the effect of student-teacher matching within a subsample of only Black boys. We initially observed a statistically significant positive effect of 0.17 SD (p < .05) on internalizing problem behaviors. However, this effect was not statistically significant after adjusting for Type I errors. Supplemental Table A5 displays the effect of student-teacher racial matching within a subsample of only Black girls. We observed statistically significant positive effects of 0.22 SD (p < .05) and 0.18 SD (p <.05) on externalizing and internalizing problem behaviors, respectively. The positive effect on internalizing problem behaviors was not statistically significant after adjusting for Type I errors. Supplemental Table A6 and A7 displays the effect of student-teacher matching within subsamples of only Hispanic boys or girls, respectively. We initially observed that Hispanic boys displayed lower mathematics achievement and were less likely to receive gifted program service when taught by Hispanic teachers. However, neither of these effects was statistically significant after adjusting for Type I errors. We observed no statistically significant effects of matching between Hispanic girls and Hispanic teachers.

We also stratified the analyses based on whether students including those who are Black or Hispanic had previously displayed low or high levels of achievement, social-emotional behavior, or executive functioning (Table A8–A13). Matching’s effects among students who had previously displayed low levels of academic achievement, social-emotional behavior, or executive functioning (Table A8) were largely null, with some evidence of negative effects. Among Black students (Table A9), being taught by Black teachers mostly resulted in null effects. However, there was some evidence of positive effects on measures of externalizing (ES range = 0.26 to 0.30, p < .05) and internalizing problem behaviors (ES = 0.24 to 0.33, ps < .05 or lower) as well as some negative effects on reading achievement (ES range = −0.13 to −0.14, p < .05). We observed only null effects among Hispanic students (Table A10) in analyses adjusted for Type 1 errors.

Student-teacher racial or ethnic matching yielded mostly null effects for students including those who are Black or Hispanic who had previously displayed high levels academic achievement, social-emotional behavior, or executive functioning (Tables A11–A13). We observed some evidence of lower likelihoods of receiving special education services when students were taught by teachers of the same race or ethnicity (OR range = 0.35 to 0.42, p < .001) for the full sample (Table A11). Estimates among Black or Hispanic students were mostly null (Tables A12–A13). The two exceptions were: (a) Black students previously rated as displaying high levels of self-control displayed higher mathematics (ES = 0.20, p < .05), but not reading or science achievement, when taught by Black teachers; and (b) Hispanic students previously rated as displaying high behavioral self-regulation displayed higher working memory (ES = 0.14, p < .05) but not cognitive flexibility when taught by Hispanic teachers. We interpret the statistically significant effects cautiously due to the lack of internal replication across other measures of achievement, social-emotional behavior, or executive functioning.

Discussion

We examined whether and to what extent U.S. elementary school students including those who are Black or Hispanic displayed greater academic achievement, social-emotional behavior, or executive functioning as well as were more likely to receive gifted or special education services when they were taught by teachers of the same race or ethnicity. We found that most Black and Hispanic students attending U.S. elementary schools do not experience having teachers of the same race or ethnicity at any time between kindergarten and fifth grade. This finding from our longitudinal analyses of nationally representative data is consistent with cross-sectional analyses of nationally representative data (e.g., Yarnell & Bohrnstedt, 2018) and longitudinal analyses of state-level data (Gershenson et al., 2017). Our descriptive results further support reports of limited racial and ethnic diversity in the U.S. elementary school teacher workforce (National Center for Educational Statistics, 2021; U.S Department of Education, 2016).

Consistent with other studies including those analyzing nationally representative data (Driessen, 2015; Nguyen & Le, 2022; Redding, 2019), our analyses repeatedly failed to indicate that student-teacher racial or ethnic matching resulted in U.S. elementary school students displaying greater achievement, social-emotional behavior, or executive functioning or were more likely to receive gifted or special education services. The full sample’s results yielded mostly null effects. The only exception was an estimated negative effect on science achievement. This effect did not internally replicate across reading or mathematics achievement. For Black students, being taught by a Black teacher positively resulted in being rated as displaying fewer internalizing behaviors. Null effects were observed for academic achievement, other indicators of behavior or executive functioning, and receipt of gifted or special education services. Consistent with prior work (Bates & Glick, 2013; Downey & Pribesh, 2004; Driessen, 2015; Redding, 2019; Wright, 2017), we initially observed positive effects of racial student-teacher matching on subjective teacher ratings of externalizing problem behaviors. However, this effect was not statistically significant in analyses adjusting for Type 1 errors. We observed no evidence to indicate that Hispanic students taught by Hispanic teachers displayed greater achievement, social-emotional behavior, or executive functioning or were more likely to receive gifted or special education services.

The study’s robustness checks mostly supported the study’s null findings. We did observe a positive effect on teacher ratings of externalizing problem behaviors for Black girls taught by Black teachers. We did not find this to be the case for Black boys. This finding extends prior work analyzing data from the ECL-K: 2011 of students in kindergarten classrooms (Wright et al., 2017), which found that Black and Hispanic students taught by teachers of the same race or ethnicity were rated as having fewer externalizing problem behaviors in analyses using classroom fixed effects and statistical controls, with no indication that the effect was moderated by biological sex. Our analyses using student fixed effects and statistical controls of data from the ECLS-K: 2011 of students attending kindergarten through fifth grade classrooms indicated that racial matching’s positive effects on externalizing problem behaviors during elementary school occur for Black girls, but not for Black boys, Hispanics boys, or Hispanic girls. We also found some evidence to indicate that, for Black students who had previously displayed low levels of academic achievement, social-emotional behavior, or executive functioning, being taught by a Black teacher resulted in positive effects on externalizing and internalizing problem behaviors, but negative effects on reading achievement. Effects were mostly null for Hispanic students who had previously displayed low levels of achievement, social-emotional behavior, or executive functioning as well as for Black and Hispanic students who had previously displayed high levels of achievement, social-emotional behavior, or executive functioning. Further research should examine whether these specific intersectional findings replicate.

Study’s Strengths and Limitations

Our study has several strengths. We used student fixed effects to account for unmeasured time-invariant confounds as well as statistical controls for measured time-varying confounds. Doing so provides plausibly causal estimates of the effects of student-teacher racial or ethnic matching. By analyzing a nationally representative cohort and assessing for matching’s effects across 12 dependent variables including multiple indicators of academic achievement, social-emotional behavior, and executive functioning as well as gifted or special education service receipt, our study allows for an unusually extensive evaluation of matching’s potential to positively impact U.S. elementary school students including those who are Black or Hispanic. We also evaluated for heterogenous treatment effects. To date, estimates of the extent to which other factors including biological sex moderate the effects of student-teacher racial or ethnic matching have been largely unavailable (Redding, 2019).

Our study also has several limitations. We were unable to examine for positive effects of student-teacher racial or ethnic matching on some educational relevant indicators. For example, and although prior work indicates that Black teachers have significantly higher educational attainment expectations for Black students than White teachers by high school (Fox, 2016; Gershenson et al., 2016), such expectation measures were unavailable in the ECLS-K: 2011’s survey of elementary school teachers. We also were unable to examine to what extent racial or ethnic matching may have helped reduce racial and ethnic disparities in exclusionary discipline (Morgan et al., 2019b; Redding, 2019). We are unable to report on the specific types of passive or active teacher effects resulting in our study’s findings. This is a consistent limitation of studies investigating student-teacher racial or ethnic matching (e.g., Dee, 2004; Harbatkin, 2021; Joshi et al., 2018, although see Nguyen & Lee, 2022, for evidence of role modeling as a potential mechanism). We also were unable to use other types of fixed effects (e.g., students-by-subject fixed effects) analyses (e.g., Harbatkin, 2021). Direct observations of classroom behavior, which would have allowed us to establish the accuracy of the teacher ratings, were unavailable in the ECLS-K: 2011.

The ECLS-K: 2011‘s data collection ended as students completed fifth grade. We were unable to assess for student-teacher racial or ethnic matching effects as the students attended middle or high school or postsecondary schools. Recent work finds that Black students taught by Black teachers are more likely to later graduate from high school and to be enrolled in college, particularly two-year colleges (Gershenson et al., 2021). Still other work finds that Hispanic high school students taught by Hispanic teachers are less likely to be absent (Gottfried et al., 2021). Racial or ethnic matching’s beneficial effects may be more evident in analyses of life-course data, although such evidence is currently limited (Redding, 2019). We aggregated the study’s results to provide estimates throughout elementary school. We may have observed positive effects if we had analyzed specific grades (Penney, 2017, but see Fryer & Levitt, 2004). Our analytical subsamples are relatively small in comparison to other work analyzing state-level administrative datasets (Dee, 2004; Egalite et al., 2015; Harbatkin, 2021; Joshi et al., 2018). As a result, our analyses may have been relatively underpowered. Although the Benjamini-Hochberg (1995) procedure increases statistical power by using false discovery rates instead of family-wise error rates, and our results were robust to use of composite variables and so correction for fewer comparisons, our adjustments for Type 1 error resulting from multiple comparisons may have been overly conservative.

Contributions and Implications

Increasing exposure to teachers of the same race or ethnicity has been suggested as a potential focus of educational policy because it may help address achievement gaps and other educationally relevant disparities (e.g. Gershenson et al., 2021; Milner, 2006). In contrast to White students, Black and Hispanic students are unlikely to experience being taught by teachers of the same race or ethnicity (Yarnell & Bohrnstedt, 2018). Theoretically, exposure to teachers of the same race or ethnicity may positively impact academic achievement and other educational relevant indicators (e.g., attendance, advanced course taking) through active or passive teacher effects (Dee, 2004; Hart, 2020; Wright et al., 2017). Being taught by Black or Hispanic teachers may provide Black or Hispanic students with more attentive and culturally sensitive teachers who hold higher educational expectations (Fox, 2016; Gershenson et al., 2016) and who better serve as role models, mentors, and cultural translators (Egalite et al., 2015; Goldhaber et al., 2015; Nguyen & Lee, 2022). Student-teacher racial or ethnic matching may also increase the instructional effectiveness of teachers as well as increase engagement and responsiveness by students who are less susceptible to stereotype threat (Redding, 2019). Greater cultural synchrony (Blair et al., 2016) may result in better teacher-student relationships, more accurate perceptions of student ability, and more positive assessments of student behavior (Cherng & Halpin, 2016; Rasheed et al., 2020).

Student-teacher racial or ethnic matching has been repeatedly shown to positively impact Black and Hispanic students including during elementary school (Dee, 2004, 2005; Easton-Brooks et al., 2009; Eddy & Easton-Brooks, 2011; Egalite & Kisida, 2018; Harbatkin, 2021; Hart, 2020; Wright et al., 2017), although the positive effects are often small (Dee, 2004; Egalite et al., 2015) as well as more likely to be observed on subjective measures of behavior (Driessen, 2015; Redding, 2019). Yet there is also repeated evidence of matching’s negative or null findings (Driessen, 2015; Fryer & Levitt, 2004; Nguyen & Lee, 2022; Redding, 2019; Wright et al., 2017). For example, analyses of the ECLS-K: 2011 data from a sample of students attending kindergarten classrooms yielded positive effects on externalizing problem behaviors but no evidence of positive effects on measures of internalizing problem behaviors, interpersonal skills, approaches to learning, or self-control (Wright et al., 2017). Still other studies reporting positive effects of racial or ethnic matching have used methods unable to account for unmeasured confounds (Bates & Glick, 2013; Easton-Brooks et al., 2009; Eddy & Easton-Brooks, 2011). Although some studies find that matching may be particularly helpful for academically struggling students (Egalite et al., 2015), other work fails to find this to be the case (Joshi et al., 2018). Findings from studies analyzing nationally representative datasets have been less consistent than those analyzing state-level datasets, particularly those from the U.S. South (Redding, 2019). The potential benefits of matching as a policy designed to addressing racial and ethnic disparities in academic achievement and other educational relevant indicators among students attending U.S. elementary schools has been unclear (Driessen, 2015; Redding, 2019).

Collectively, and consistent with prior studies (Buddin & Zamarro, 2009; Downey & Pribesh, 2004; Howsen & Trawick, 2007; Jennings & DiPrete, 2010; McGrady & Reynolds, 2013; Wright et al., 2017), our findings suggest that educational policies designed to increase student-teacher racial or ethnic matching in U.S. elementary schools would have limited impact including for students who are Black or Hispanic, and so unlikely to meaningfully address large achievement gaps or other educationally relevant disparities during this time period. The positive effects we observed were often small in magnitude and not robust to corrections for Type 1 errors resulting from multiple comparisons. Consistent with prior work (Driessen, 2015; Egalite et al., 2015; Redding, 2019), the positive effects that we did observe were mostly specific to Black students, particularly girls or those with prior histories of low academic achievement, social-emotional behavior, or executive functioning.

Our results are consistent with prior studies analyzing nationally representative data, which have repeatedly provided less conclusive evidence to support student-teacher racial or ethnic matching (Fryer & Levitt, 2004; Jennings & DiPrete, 2010; Redding, 2019; but see Yarnell & Bohrnstedt, 2018). An explanation of the consistently null findings for student-teacher racial or ethnic matching based on nationally representative datasets may be the use of relatively smaller samples and so lower statistical power relative to analyses of state-level administrative datasets. Alternatively, it may be that racial and ethnic matching’s effects do not generalize to the U.S. elementary school population. Such effects may instead to specific to contexts and populations (Clotfelter et al., 2006; Dee, 2004; Egalite et al., 2015; Redding, 2019) in which shared group identities between students and teachers of color may be especially salient and so result in the hypothesized active or passive teacher effects (Dee, 2005; Redding, 2019). The U.S. South is an example region where such shared identities may be especially salient because of the region’s history of discrimination as well as de jure and de facto practices of racial segregation (Harbatkin, 2021). Student-teacher racial or ethnic matching’s positive effects are most consistently observed in analyses of students attending schools in the U.S. South (Redding, 2019). The student and teacher populations in this region also may be relatively more racially diverse and so allow for greater estimate precision of student-teacher racial or ethnic matching’s effects (Joshi et al., 2018; Schaeffer, 2021; U.S. Census Bureau, 2021). Consistent with these possibilities, Dee (2005) reported that the effects of student-teacher racial or ethnic matching on measures of student behaviors were positive in the U.S. South but not in the U.S. regions of the Northeast, North-Central, or West.

It may also be that positive findings observed in U.S. South may not only be explained by passive or active teacher effects (Dee, 2004, 2005; Harbatkin, 2021). The positive effects previously observed for racial or ethnic matching instead may be explained by the types of educational institutions that teachers attend. New analyses (Edmonds, 2022) using student and year fixed effects as well as additional statistical controls find that Black students attending elementary schools in North Carolina display greater mathematics achievement (ES = 0.01 SD) when taught by Black versus non-Black teachers—but only when these students were taught by Black teachers who had been trained at historically Black colleges or universities (HBCUs). Black students taught by non-HBCU-trained Black teachers displayed lower achievement (ES = −0.02 SD), suggesting that types of teacher educator practices used in HBCUs may be particularly relevant in explaining previously reported positive effects for student-teacher racial or ethnic matching observed in the U.S. South. Black students also displayed greater mathematics achievement when taught by HBCU-trained White teachers than by non-HBCU-trained White teachers. Further research clarifying why the positive effects for student-teacher racial or ethnic matching are especially likely to be observed in analyses of students attending schools in the U.S. South is currently needed (Edmonds, 2022; Redding, 2019).

Conclusion

Our analyses repeatedly yielded null effects of student-teacher racial or ethnic matching on academic achievement, social-emotional behavior, or executive functioning as well as gifted and special education service receipt for U.S. elementary school students including those who are Black or Hispanic. Of the few statistically significant effects observed, students taught by teachers of the same race or ethnicity displayed lower science achievement, but this finding was not observed in the subsamples of only Black or Hispanic students. Black students taught by Black teachers were rated as displaying fewer internalizing problem behaviors. Robustness checks suggested that the previously reported positive effect of being taught by Black teacher on the externalizing problem behaviors of Black students may be specific to Black girls. We did observe some positive behavioral effects for Black but not Hispanic students who had previously displayed low levels of academic achievement, social-emotional behavior, or executive functioning. Collectively, the study’s analyses provide limited empirical support for student-teacher racial or ethnic matching as a policy for addressing achievement gaps or other educationally relevant disparities among students attending U.S. elementary schools.

Highlights.

• Student-teacher racial or ethnic matching may help address educational disparities

• Matching’s impacts were examined using student fixed effects and controls

• Matching had mostly null effects across the study’s 12 dependent measures

• There was some inconsistent evidence of effect heterogeneity across subsamples

• The observed positive and negative effects were small in size