Exploring the Link Between Students’ Sense of School Belonging and Shared Leadership in U.S. High Schools

Abstract

This study investigated whether and to what extent shared leadership in school was associated with a positive and equitable sense of school belonging among students from diverse backgrounds in the United States. We denoted shared leadership as collective practices that take place in schools in which principals enact inclusive practices in the decision-making process and teachers collaborate with their colleagues to support all students’ learning. We used linear mixed-effects modeling to analyze a nationally representative sample from the base-year data of the High School Longitudinal Study of 2009 (HSLS:09) provided by the National Center for Education Statistics (NCES). Our exploratory analyses revealed that shared leadership practices in school were positively associated with all students’ school belonging. We also found that shared leadership practices in school contributed to American Indian/Alaska Native students’ positive sense of school belonging through a cross-level interaction effect.

Introduction

A critical problem facing schools in the United States is the low level of sense of school belonging among students (Organization for Economic Co-operation and Development [OECD], 2019). According to the internationally comparable data from the 2018 Program for International Student Assessment (PISA), the mean index score of the sense of school belonging among 15-year-old students in the United States was −0.24, with an average of 0 for OECD countries, placing the United States 34th among the 38 OECD countries (OECD, 2019). This report defined the sense of school belonging as “how accepted, respected and supported students feel in their social context at school” (p. 130). The same study, as well as numerous others (e.g., Faircloth & Hamm, 2005; Gopalan & Brady, 2020; Walton & Cohen, 2007), report substantial disparities in the sense of school belonging experienced by various groups of students. These observations are particularly problematic for students in that a low sense of school belonging is detrimental to students’ overall physical, psychological, social, and academic outcomes (Catalano et al., 2004; Juvonen, 2006). Given that students’ sense of school belonging is essential for a variety of short- and long-term individual developmental and academic outcomes (Appleton et al., 2008; Bernat & Resnick, 2009), tackling the issue of inequities in students’ sense of school belonging may be instrumental in bridging the various opportunity and experience gaps in schooling (Darling-Hammond, 2010; Levinson et al., 2022; Riley, 2017, 2022).

Notably, there has been growing interest in students’ sense of school belonging in leadership and organizational studies (e.g., M. Lee et al., 2021; Louis et al., 2016; Osterman, 2000; Riley, 2017). Schools are central to fostering students’ sense of school belonging (Riley, 2017). They influence how school leaders shape culture; implement practices, processes, and structural arrangements; and transmit values of relatedness and connectedness to school community members, including teachers and students (Louis et al., 2016; Osterman, 2000; Riley et al., 2020). Similarly, inequities in the sense of school belonging among minoritized students are an important leadership concern because school leaders can play a key role in addressing the inequities in school life experienced by diverse students by creating more equitable learning opportunities and schooling experiences (Cambron-McCabe & McCarthy, 2005; Capper & Young, 2014; DeMatthews & Mawhinney, 2014; Furman & Gruenewald, 2004; Radd et al., 2021).

However, there is a lack of literature on whether and to what degree school principals and teachers collectively contribute to shaping a positive and equitable sense of school belonging across diverse groups of students. Despite the value of empirical studies of the links between school leadership and students’ sense of school belonging, most have been qualitative inquiries based on small case studies (e.g., Riley et al., 2020). Taking this into consideration, this study, based on a large quantitative dataset with a nationally representative sample, attempts to provide more broadly applicable knowledge about the capacity of school principals and teachers to make a difference in promoting a positive and equitable sense of school belonging for all students.

This study reflects on the nature of shared leadership (Drath, 2001; Pearce & Conger, 2003) with special attention paid to shared leadership practices in schools (Printy & Marks, 2006; Torres et al., 2020; Wahlstrom & Louis, 2008). Applied to schools, shared leadership denotes the collective activities and influences of individuals in a school (e.g., principals and teachers) for enhancing the quality of students’ schooling experiences and learning (Torres et al., 2020). Compared with a traditional understanding of leadership as an individual leader’s attributes or characteristics (Pearce & Conger, 2003), shared leadership emphasizes dynamic and multidirectional processes and activities among individuals (Printy & Marks, 2006). We use the term “shared leadership” to encompass a range of complex social and psychological attributes and practices of principals and teachers throughout schools, such as principals’ promotion of effective schoolwide communication (Torres et al., 2020; Wahlstrom & Louis, 2008) and teachers’ active interactions with their colleagues (Printy & Marks, 2006) and participation in decision-making (Carson et al., 2007; Day et al., 2004).

To identify the link between shared leadership in school and students’ equitable sense of school belonging, we first seek to identify the patterns of inequity in the sense of school belonging among students with diverse demographic characteristics. We specifically look at students’ race or ethnicity, gender, socioeconomic status (SES), and immigration status, because these are the most prevalent characteristics that are associated in the literature (DeNicolo et al., 2017; Gopalan & Brady, 2020) with inequities in students’ sense of school belonging.

This study focuses on the sense of school belonging, specifically among high school students, because of its importance for strengthening their identity, security, and recognition, which are critical in building their adult lives (Halse, 2018). It contributes to the literature by identifying on a large scale the unique needs of diverse high school students in relation to their sense of school belonging. Such an understanding is essential for school leaders and teachers to better support their students with diverse backgrounds. Furthermore, the findings regarding the link between shared leadership in school and students’ sense of school belonging, particularly that of minoritized students, have implications for research and practice in identifying the specific ways in which school leaders and teachers can collectively shape an equitable sense of school belonging among their students.

The next section presents a literature review focusing on shared leadership and students’ sense of school belonging and points to a potential link between them, which has been a blind spot in the literature. It also specifies a critical theoretical framework used in this study to understand the experiences of students of color, which includes critical race theory (Ladson-Billings & Tate, 1995), critical multiracial theory (MultiCrit) (Harris, 2016), and tribal critical race theory (TribalCrit) (Brayboy, 2005). The subsequent section describes the research design, data sources, and method of statistical analysis used in the study. The final two sections present the findings and conclude with an outline of the implications for researchers, policymakers, and educational leaders; the limitations of the study; and guidelines for future research.

Literature Review

Shared Leadership in School

A plethora of theoretical examinations of educational leadership have considered behaviors, styles, and traits, with different emphases in form and content (e.g., distributed leadership, instructional leadership, culturally responsive leadership). As the many varieties of leadership indicate, leadership is a complex process with multiple dimensions (Northouse, 2010). While complex and diverse, one commonality across the different types of leadership is that they all highlight the leadership role and responsibility of all members of the organization, albeit in varying degrees and with different purposes. In this sense, all leadership can be thought of as shared leadership (Drath, 2001) in some ways, although the distinctive features of each type of leadership should be appreciated. In other words, shared leadership practices can be fused into different types of leadership when applied in organizational settings.

Shared leadership emphasizes the ways in which school leaders work together with diverse members of a school organization to promote school improvement (Pearce & Conger, 2003; Printy & Marks, 2006; Torres et al., 2020). Specifically, teachers in schools with strong shared leadership practices tend to (1) actively collaborate with their colleagues; (2) seek to improve their professional practices by learning from each other; (3) have similar expectations of themselves; and (4) participate in collaborative decision-making (Day et al., 2004; Printy & Marks, 2006). Meanwhile, principals play important roles in distributing leadership activities among teachers, providing broad support for teachers’ participation in leadership responsibilities and decision-making tasks, setting clear expectations and organizational goals for staff and the whole school through effective communication, and encouraging innovation (Ingersoll et al., 2018; Leithwood & Jantzi, 1997; Printy & Marks, 2006; Robinson et al., 2008; Torres et al., 2020).

Several lines of evidence suggest that the elements of shared leadership in schools affect organizational outcomes (e.g., Walker et al., 2014). Specifically, principals’ effective communication of schools’ goals, their related expectations of the roles and responsibilities of staff members, and their adoption of a supportive approach to innovation and new ideas are keys to school improvement (J. C. K. Lee & Lo, 2007; Leithwood & Mascall, 2008; Printy & Marks, 2006; Walker et al., 2014). Specifically, research showed that transparent communication between principals and staff members was critical for developing a shared goal and vision (Wahlstrom & Louis, 2008). Previous studies also showed that principals’ promotion of effective school-wide communication was essential to implementing initiatives designed to change schools (J. C. K. Lee & Lo, 2007). The encouragement of innovation among principals and teachers facilitates improvement and changes in educational practice that students benefit from in their schooling (Printy & Marks, 2006). Furthermore, regular communication such as “keeping staff informed” (Earley et al., 2002, p. 11) and shared beliefs and responsibilities for all students’ learning among teachers are essential to enhance their professional learning and development, which is critical for promoting the quality of students’ experiences and learning (M. Lee & Louis, 2019; Printy & Marks, 2006).

Taken together, shared leadership practices among principals and teachers in school appear to be positively related to teachers’ instruction practices (Ingersoll et al., 2018; Wahlstrom & Louis, 2008), staff empowerment (Penlington et al., 2008), and students’ academic achievement (Walker et al., 2014). For example, Walker et al. (2014) found that transparent and effective communication structures as a mechanism of shared leadership put in place by principals explained 12% of the variation in academic achievement between schools.

For the purposes of this study and in line with the literature, we characterize shared leadership as school-wide practices and processes encompassing both principals and teachers. Shared leadership entails principals (1) communicating and consulting with staff members on a regular basis in decision-making procedures, (2) having clear expectations of staff members’ roles and responsibilities, (3) presenting and communicating the shared vision and values that the school pursues, and (4) encouraging innovation and new ideas, and teachers, (5) sharing responsibility for all students’ learning, (6) collaborating with each other for teaching, and (7) having high standards for their students and themselves. This conception of shared leadership reflects the findings of, and an emerging consensus among, the recent studies of shared leadership discussed previously.

Students’ Sense of School Belonging

Students’ sense of school belonging has been extensively explored in educational research (e.g., Allen & Kern, 2019; Goodenow & Grady, 1993; Korpershoek et al., 2020). The concept of school belonging has been used interchangeably with various terms, including school connectedness (Townsend & McWhirter, 2005), school attachment (Moody & Bearman, 2004), and school bonding (Oelsner et al., 2011). Despite such varied terminologies, the concept consistently represents an individual’s emotional, psychological, and physical attachment to people and places in school in terms of feeling accepted and included as well as fitting in (Gray et al., 2018; Riley, 2017).

Studies on the sense of school belonging among students have suggested its positive association with numerous aspects of their school experiences and learning outcomes (Appleton et al., 2008; Halse, 2018). In particular, students’ strong sense of school belonging is positively associated with their physical and emotional health and well-being (Sharma & Malhotra, 2010), self-esteem, self-identity, intrinsic motivation (Freeman et al., 2007; Nutbrown & Clough, 2009), school engagement and learning outcomes (Anderman, 2002; Appleton et al., 2008), and academic success (Osterman, 2000). In addition, scholars have documented the association between a strong sense of school belonging and fewer negative behaviors in schools, including truancy (Hallinan, 2008) and absenteeism (Croninger & Lee, 2001). Conversely, a low sense of school belonging can lead to negative health and developmental outcomes (Catalano et al., 2004), poor social and emotional well-being, and psychological depression and anxiety (Paradies, 2006), which have also been predictive of negative learning outcomes (Bücker et al., 2018).

Studies have identified key psychological and social determinants of students’ sense of school belonging, particularly at the school level. Students have a greater sense of belonging at school when their schools or classrooms meet their basic psychological needs, such as those for relatedness and autonomy (Korpershoek et al., 2020; Niemiec & Ryan, 2009). In addition, having good interpersonal relations with and support from teachers and peers are pivotal in forming a positive sense of school belonging among students (Allen et al., 2018; Roorda et al., 2017). A positive school climate (Ma, 2003) and a more egalitarian societal and organizational culture that values cooperation and mutual support, rather than a hierarchical culture emphasizing competition (Chiu et al., 2016), are both positively linked to students’ sense of school belonging.

A considerable number of studies have also identified significant disparities in students’ sense of school belonging by demographic characteristics, including race or ethnicity, gender, SES, and immigration status (Anderman, 2002; Gillen-O’Neel & Fuligni, 2013; Kim & Sax, 2009; Smerdon, 2002). Specifically, racial or ethnic minority students, including Black (Fisher et al., 2020; Gray et al., 2018), Latinx (Flores-González et al., 2014), American Indian/Alaska Native (Napoli et al., 2003), and mixed-race students (Rozek & Gaither, 2021), feel less connected to their schools than their White peers. These inequities in students’ sense of school belonging by their race or ethnicity arise from unique cultural and power relations in racially segregated contexts, such as schools (Yuval-Davis, 2011). For example, the weakened sense of school belonging experienced by American Indian/Alaska Native students (Napoli et al., 2003) is based on multiple factors, including a lack of representation and role models (Covarrubias & Fryberg, 2015), uncertainty about belonging (Walton & Cohen, 2007), experiences of structural exclusion and racism at their school (Tachine et al., 2017), and cultural divergence between their indigenous home culture and colonized education (Khalifa et al., 2018).

Shared Leadership and School Belonging

Although extensive leadership studies have focused on the effects of leadership on students’ academic achievement, a line of educational leadership research in recent years has paid substantial attention to whether and how principals influence students’ experience of school life in general and school belonging in particular (e.g., Kudlats & Brown, 2021; M. Lee et al., 2021; Louis et al., 2016; Riley, 2019, 2022; Ryu et al., 2020). Previously published studies have underscored the importance of principals’ leadership as a primary driver of the differential senses of school belonging found among diverse students (Osterman, 2000; Riley, 2017, 2019, 2022). While such researchers have documented that students’ relationships with and connectedness to their peers and teachers in school are essential for forming a positive sense of school belonging (Goodenow & Grady, 1993; Gowing, 2019; Osterman, 2000), studies focusing on school leadership have also suggested that school leaders can play an important role in developing their students’ sense of school belonging by creating a positive school climate and a safe learning environment (Klein et al., 2012; M. Lee et al., 2021; Wilson, 2004). Specifically, principals’ leadership contributes to the development of a school culture that emphasizes the relatedness of members of the school community, which includes caring and collective responsibility (Louis et al., 2016; Riley et al., 2020), and embraces the diverse voices and perspectives of members of the school community in their leadership practices (Flutter & Ruddock, 2004; Riley, 2019). For example, Riley et al. (2020) highlighted the interconnectedness of leadership, school culture, and school belonging. School leaders play a key role in shaping school culture and forming the agency of teachers, students, and communities by including their voices in decision-making (Wahlstrom & Louis, 2008). In turn, this creates a strong sense of school belonging among students (Riley et al., 2020).

Informed by the literature, we assume that given its nature, shared leadership practices in schools may improve the key domains of school life, such as school belonging and student agency experienced and developed by students. Specifically, from a heuristic model perspective, we view shared leadership as dynamic organizational practices that facilitate interactions among principals, teachers, and students, aimed at cultivating positive and inclusive school environments. Coupling fragmented findings from existing empirical studies (e.g., Kudlats & Brown, 2021; M. Lee & Louis, 2019; Printy & Marks, 2006; Riley et al., 2020; Wahlstrom & Louis, 2008; Walker et al., 2014), we assume that shared leadership influences students’ sense of belonging by promoting inclusive decision-making, transparent communication, and a strong school culture. In this heuristic model, principals enact inclusive practices by involving a wide range of school members in decision-making processes and communications, thereby fostering a sense of agency and belonging among students. Within this organizational culture, teachers are expected to actively engage in professional collaborations with their colleagues to create and sustain supportive learning environments for “all” students. We further assume that shared leadership can moderate disparities in school belonging among students from diverse backgrounds. For example, Quijada Cerecer (2013) emphasized that institutionalized forms of inclusive communication between principals and students can foster teacher–student relationships and school engagement, particularly among American Indian/Alaska Native students. Guided by recent educational leadership research on social justice (e.g., DeMatthews & Mawhinney, 2014; Furman, 2012; Theoharis, 2009) and on caring, positive leadership (e.g., Louis et al., 2016; J. F. Murphy & Louis, 2018), we examine whether shared leadership in school can contribute to the equitable sense of belonging, particularly among students of color. In their qualitative inquiry, DeMatthews et al. (2021) highlighted the role of principals in creating inclusive and equitable schools through such practices as building effective and transparent communication channels through discussion, dialogue, and debates with teachers and fostering a sense of shared responsibility within a whole school community. The emerging literature has suggested that the traditional conception of the leader as the sole decision-maker is not suitable for the creation of equitable schools in which all students feel connected regardless of their demographic characteristics (DeMatthews et al., 2021; Radd et al., 2021). These features of school leadership resonate with the concept of shared leadership discussed previously.

Theoretical Framework

CRT has been applied as a theoretical foundation to understand better the structural inequities in schooling experiences (e.g., sense of school belonging) among students of color (Ladson-Billings & Tate, 1995). CRT scholars have emphasized the institutionalized racism and oppression in the make-up of American society and schooling, which negatively affect students’ schooling experiences, including their sense of school belonging (Duncan, 2002; Ladson-Billings, 1999). They have argued that students’ schooling experiences cannot be separated from White privilege and have criticized claims of meritocracy, colorblindness, and race neutrality (Bell, 1995; Ladson-Billings & Tate, 1995; Solórzano & Solórzano, 1995).

Emerging from CRT, which was originally developed to challenge the structural inequity issues of African Americans, scholars have developed and applied successive theoretical frameworks focusing on other racial or ethnic groups, including MultiCrit and TribalCrit. For example, MultiCrit theorists have highlighted and challenged issues that influence mixed-race students’ daily lives, including monoracism that oppresses individuals who do not fit monoracial categories, identity, stereotyping, and structural exclusion from the interpersonal relationships with teachers in school organizations (Harris, 2016; Joseph-Salisbury & Andrews, 2017; Museus et al., 2016; Williams, 2011). Similarly, the fundamental tenet of TribalCrit is that colonization is endemic to society and schooling in the United States (Brayboy, 2005). This theoretical perspective is particularly powerful in explaining the experiences and challenges of American Indian/Alaska Native students in their daily lives. For example, Quijada Cerecer’s (2013) qualitative inquiry revealed that American Indian/Alaska Native students were forced to conform to dominant ways of being and learning in colonized schooling while school leaders and teachers reinforced the deficit-oriented perspective toward these students and negated their role as change agents and knowledge producers. TribalCrit scholars have underscored that such a decolonizing process of collective dialogue and communication in school can positively affect these students’ equitable schooling experiences, including their sense of belonging and the quality of their schooling experiences (Garcia & Shirley, 2012; Jacob et al., 2015; McConnell, 2013).

Guided by the literature and theoretical framework discussed previously, we explore the link between shared leadership practices in school and students’ sense of school belonging with the aim of addressing the following two research questions:

Research Question 1: To what extent are shared leadership practices in school associated with a positive sense of school belonging among students?

Research Question 2: To what extent do shared leadership practices in school moderate the relationship between students’ racial or ethnic characteristics and their sense of school belonging?

Data and Methods

Data

This study used base-year data collected in the fall term of the 2009–2010 school year from the High School Longitudinal Study of 2009 (HSLS:09) provided by the National Center for Education Statistics (NCES). This is a nationally representative sample of high school students, collected from 21,444 ninth graders (weighted N = 4,197,724) in 944 schools (weighted N = 22,897). In addition, HSLS:09 includes a math and science teacher sample (N = 17,882, weighted N = 4,042,879) and a principal sample (N = 944, weighted N = 22,897), which represent groups of teachers and principals who teach and lead a nationally representative sample of ninth graders. A stratified two-stage random sample design was used to obtain a representative sample of students (Lumley, 2011).

The HSLS:09 dataset includes specific student-, teacher-, and school-level information on diverse students’ sense of school belonging and how this relates to each school’s specific qualities (e.g., shared leadership practices). The student-level data include demographics and items to measure the perceived sense of school belonging among ninth graders. The teacher-level data contain survey data on math and science teachers’ perceptions of the degree to which their schools engage in shared leadership practices in terms of the communication and engagement of principals and teacher collaboration. Finally, the school-level data include information on the schools (community type, student demographic composition) and principals (years of service as a principal in this school). In this regard, HSLS:09 has considerable advantages over other more recent large-scale datasets that use similar measures of students’ sense of school belonging (e.g., Trends in International Mathematics and Science Study [TIMSS] 2019, Program for International Student Assessment [PISA] 2018). In particular, neither TIMSS nor PISA includes specific measures for principal- and school-related information (i.e., shared leadership practices in school) that is matched with students, particularly marginalized students (e.g., American Indian/Alaska Native American students). Thus, the use of HSLS:09 dataset in this study was essential for providing insights into how schools can make a positive and equitable difference in diverse students’ sense of school belonging. Although the HSLS:09 dataset consists of five sub-datasets tracking the same high school student samples over different waves of data collection, we specifically focused on the results of the 2009 base-year survey conducted among ninth graders. We chose this specific dataset because the subsequent surveys did not contain the necessary items to consistently measure the two core constructs of our study (i.e., the students’ sense of school belonging and shared leadership in schools).

Measures

Student characteristics

This study used students’ different demographic characteristics that have been found to be associated with their sense of school belonging in the literature, such as race or ethnicity, gender, socioeconomic status (SES), and immigration status (DeNicolo et al., 2017; Gopalan & Brady, 2020; Kim & Sax, 2009; Walton & Cohen, 2007). We created five categorical variables for students’ race or ethnicity (i.e., American Indian/Alaska Native, Black, Hispanic, Asian American/Pacific Islander, and mixed-race) based on each student’s race or ethnicity composite variable (X1RACE). White students, coded as 1, were used as a reference group to explore the systematic inequities in the sense of school belonging among students of color. In addition, the variable female was used to designate girls, with boys serving as the reference group (X1SEX). We used the following three categories from the P1USBORN9 dataset to determine students’ immigration status: (1) born in the United States, (2) born in Puerto Rico or another U.S. territory, and (3) born in another country. The dummy variable immigrant status was also used to distinguish students who were not born in the United States (category 3) from U.S.-born students, who served as the reference group (categories 1 and 2). More details about the characteristics of the samples, including the coding procedures for the categorical variables, are presented in Table 1.

Table 1

Characteristics of the Sample Students and Schools (Categorical Variables)

Variable	Original Database Variable Name	Original DatabaseData Label and Value	Newly Created Variable NameAfter Recoding(value in brackets)	Weighted n	Unweighted n
Students (weighted n = 4,197,724; unweighted n = 25,206)
Gender	X1SEX	Male (1)	Male (0)	2,119,192 (50.5%)	12,858 (51.0%)
		Female (2)	Female (1)	2,078,532 (49.5%)	12,296 (48.8%)
Race or ethnicity	X1RACE	White (8)	White (0)	2,165,839 (51.6%)	12,082 (53.7%)
		American Indian/Alaska Native (1)	American Indian/Alaska Native (1)	30,500 (0.7%)	168 (0.7%)
		Asian (2)	Asian American/Pacific Islander (1)	169,119 (4.0%)	2,206 (9.5%)
		Native Hawaiian/Pacific Islander (7)
		Black (3)	Black (1)	569,961 (13.6%)	2,648 (10.5%)
		Hispanic (4: no race specified, 5: race specified)	Hispanic (1)	942,019 (22.4%)	4,000 (17.2%)
		More than one race, non-Hispanic (6)	Mixed-race (1)	320,287 (7.6%)	1,952 (8.4%)
Immigration status	P1USBORN9	Born in the U.S. (1)	U.S.-born (0)	3,050,300 (72.7%)	16,602 (65.9%)
		Born in Puerto Rico or another U.S. territory (2)
		Born in another country (3)	Foreign-born (1)	230,908 (5.5%)	1,412 (5.6%)
Schools (weighted n = 23,022; unweighted n = 944)
Community type	X1LOCALE	City (1)	Urban (0)	4,869 (21.2%)	272 (28.8%)
		Suburb (2)	Suburban (1)	5,220 (22.7%)	335 (35.5%)
		Town (3)	Town (1)	3,825 (16.6%)	117 (12.4%)
		Rural (4)	Rural (1)	9,108 (39.6%)	220 (23.3%)

Note. Unweighted figures in the table are based on the original data with missing values. The reference groups (coded as 0) for race or ethnicity, female, immigration status, and school community type are White, male, U.S.-born, and urban, respectively. The subsample sizes for race or ethnicity and immigration status may not sum to the total student sample size (weighted n = 4,197,724; unweighted n = 25,206) because of missing values for these variables.

It is worth noting that the sample size of the subgroup of American Indian/Alaska Native students is relatively small, compared to other racial or ethnic subgroups (unweighted n = 168, 0.7%; weighted n = 30,500, 0.7%). In fact, the most recent statistics indicate that American Indian/Alaska Native students comprised only 1% (0.5 million) of the 49.4 million students enrolled in public primary and secondary schools in the United States in 2020 (National Center for Education Statistics [NCES], 2022). Thus, the sample size of American Indian/Native American students (approximately 1%) in this study reflects the distributions found in other nationally representative samples (e.g., PISA, TIMSS). While some relatively small student groups, such as Asian and private school students, were over-sampled in HSLS:09, American Indian/Alaska Native students were not (Ingels et al., 2011). As a result, the relatively small sample size may increase the risk of false-positive (Type 1 error) and false-negative (Type 2 error) findings for this group (Forstmeier et al., 2017). Given this limitation in the sampling approach of HSLS:09 for American Indian/Alaska Native students, particular attention needs to be paid to our analytical procedures when interpreting our findings. As our study was exploratory in nature, we offer our findings and interpretations with caution where necessary, as well as further discussion of their limitations.

The SES index score was also used as an independent variable, which was created by NCES (X1SES) to represent the relative socioeconomic position of a student’s family in a vertical social hierarchy. It was measured by variables obtained from the parent/guardian survey, such as their education, occupation, and income.

School characteristics

This study incorporated two school backgrounds (community type, demographic composition) and principals’ backgrounds (number of years of service as a principal in the current school [A1YRSHSLSSCH]) as key covariates. The type of school community (X1LOCALE) indicates whether the school is located in an urban area, a suburban area, a town, or a rural area. Three categorical variables for community type were created in this study (Suburban, Town, Rural), and urban schools were established as the reference group for analysis (see Table 1). The demographic composition of a school refers to the percentage of disadvantaged students (i.e., free or reduced lunch eligible [FRL] students [A1FREELUNCH], special education students [A1SPECIALED], and English language learners [A1ELL]) who are enrolled.

Shared leadership practices in school

The construct of shared leadership practices in school was measured by the degree to which teachers perceived that their schools engaged in shared leadership practices (see Appendix 1). Specifically, the teachers reported the extent to which their principals consulted them before making decisions that could affect them, communicated the kind of school they wanted to promote, let staff members know what was expected of them, and showed interest in innovation and new ideas. The teachers also reported the extent to which teachers in their schools set high standards for themselves, had a strong responsibility for helping each other, and felt responsible for the learning of all students. Although these items may not be sufficient to encompass shared leadership practices enacted across various school contexts, they capture conceptually important and congruent features of shared leadership practices identified by the key literature (e.g., Printy & Marks, 2006; Torres et al., 2020). All of the items were rated on a 4-point Likert scale (1 = “strongly agree” to 4 = “strongly disagree”) and recoded so that a higher (lower) score indicated a more (less) positive perception. For the math and science teachers’ responses, we conducted a factor analysis with maximum likelihood estimation (Hirose et al., 2016) adjusted by the analytical weights of the math and science course enrollees, respectively (Cronbach’s α = 0.81). The combined items (factor scores) were then aggregated to the school level, with a higher value for this construct, thus indicating that the teacher perceived the school community to be engaging in more shared leadership practices in school. The HSLS:09 dataset includes specific student- and teacher-level information on science, technology, engineering, and mathematics (STEM) education.

Dependent variable

We used students’ sense of school belonging as the dependent variable, which was measured using six items from the student-level data (see Appendix 1). These six items assessed the extent to which the students felt safe at their school (S1SAFE), were proud to be a part of the school (S1PROUD), and agreed that there were always teachers or other adults they could talk to if there was a problem (S1TALKPROB). The items also evaluated whether the students felt that school was often a waste of time (i.e., reverse coded) (S1SCHWASTE) and whether their math (S1MTCHFAIR) and science teachers (S1STCHFAIR) treated them fairly. The third item (i.e., S1TALKPROB) inquired about the students’ perceptions of their relationships with all teachers and adults at the school. In contrast, the fifth and sixth items (S1MTCHFAIR and S1STCHFAIR) focused specifically on the students’ relationships with their math and science teachers, as the HSLS:09 dataset includes unique student-level information on STEM education. All items were rated on a 4-point Likert scale (1 = “strongly agree” to 4 = “strongly disagree”) and recoded so that a higher (lower) score indicated a more (less) positive perception, except for the fourth item, which was reverse-coded. A composite of students’ sense of school belonging was created based on a principal components factor analysis adjusted using analytical weights for the six items (Cronbach’s α = 0.69). This dependent variable was standardized with a mean of 0 and a standard deviation (SD) of 1 (see Appendix 2), allowing the coefficients to be interpreted as SD differences in the students’ sense of belonging, where a higher (lower) value of this composite indicates a greater (weaker) sense of school belonging. These six items used to gauge students’ sense of school belonging align with those used previously in the literature (e.g., Furlong et al., 2011; Goodenow & Grady, 1993; Hazel et al., 2013; Walton & Cohen, 2007) and other nationally representative secondary datasets (e.g., PISA, TIMSS, National Longitudinal Study of Adolescent Health). However, these items may not fully assess the nuanced and multidimensional nature of students’ sense of school belonging (e.g., extracurricular involvement, perceived discrimination; see Faircloth & Hamm, 2005) because they were obtained from these secondary data (HSLS:09) collected by NCES. We further discuss the limitations of this measure and suggest directions for future studies in our concluding section. The descriptive statistics of the continuous variables and the correlation matrix for the school-level variables are provided in Appendices 2 and 3, respectively.

Analytical Models

Linear mixed-effects modeling (LMM) was used to explore the relationships between students’ sense of school belonging and shared leadership practices in school when controlling for school-level characteristics. LMMs were used to address the nested data structure (students nested within schools). The calculated intraclass correlation coefficient (ICC) of the students’ sense of school belonging was 0.11, indicating that 11% of the variance in the students’ sense of school belonging was due to school-level differences. Organizational research has typically suggested using an ICC that is greater than 0.10 (10%) to account for organizational (e.g., school-level) differences (Vogt, 2011). This result justified the use of LMMs to examine the degree to which the school-level attributes used in this study accounted for the variance in the students’ sense of belonging across schools. For this purpose, this study followed Anderson’s (2012) model-building process both theoretically and empirically.

We used the linear mixed-effects models, defined by the following generalized equation:

Student-level:

$$\begin{array}{c}{\left(\begin{array}{c}\mathrm{Students}’\mathrm{sense}\mathrm{of}\hfill \\ \mathrm{school}\mathrm{belonging}\hfill \end{array}\right)}_{ij}={\mathrm{\beta }}_{0j}+{\mathrm{\beta }}_{1j}{(\mathrm{Female})}_{1j}\hfill \\ +{\mathrm{\beta }}_{2j-6j}{(\mathrm{Race}\mathrm{or}\mathrm{ethnicity})}_{2j-6j}\hfill \\ +{\mathrm{\beta }}_{7j}{(\mathrm{Socioeconomic}\mathrm{status})}_{7j}\hfill \\ +{\mathrm{\beta }}_{8j}{(\mathrm{Immigrant}\mathrm{status})}_{8j}+e_{ij}\hfill \end{array}$$

School-level:

$$\begin{array}{c}{\mathrm{\beta }}_{0j}={\mathrm{\gamma }}_{00}+{\mathrm{\gamma }}_{01–03}{(\mathrm{Community}\mathrm{type})}_j\hfill \\ +{\mathrm{\gamma }}_{04–06}{(\mathrm{Demographic}\mathrm{composition})}_j\hfill \\ +{\mathrm{\gamma }}_{07}{\left(\begin{array}{c}\mathrm{Years}\mathrm{served}\mathrm{as}\mathrm{a}\mathrm{principal}\mathrm{in}\hfill \\ \mathrm{the}\mathrm{current}\mathrm{school}\hfill \end{array}\right)}_j\hfill \\ +{\mathrm{\gamma }}_{08}{\left(\begin{array}{c}\mathrm{Teachers}’\mathrm{perceptions}\mathrm{on}\mathrm{shared}\hfill \\ \mathrm{leadership}\mathrm{practices}\mathrm{in}\mathrm{school}\hfill \end{array}\right)}_j+u_{0j}\hfill \\ {\mathrm{\beta }}_{\mathrm{n}j}={\mathrm{\gamma }}_{\mathrm{n}0}+{\mathrm{\gamma }}_{\mathrm{n}1}{\left(\begin{array}{c}\mathrm{Teachers}’\mathrm{perceptions}\hfill \\ \mathrm{on}\mathrm{shared}\mathrm{leadership}\hfill \\ \mathrm{practices}\mathrm{in}\mathrm{school}\hfill \end{array}\right)}_j+u_{\mathrm{n}j}\hfill \end{array}$$

where the dependent variable is the sense of school belonging of student i in school j; β_nj is the difference in students’ average sense of school belonging by their different demographic characteristics compared to the respective reference groups; γ₀₀ is the average sense of school belonging of students from all schools in the sample; γ_01-08 are the effects of the school and principal characteristics on the mean of the school students’ sense of school belonging; γ_n0 is the average regression coefficients; β_{n j} for students with different demographic characteristics on their sense of school belonging; γ_n1 is the effects of teachers’ perceptions on the shared leadership in school on the student-level slopes β_nj; u_0j is the random effect associated with school j; u_nj is the random effect associated with the student-level coefficients β_nj; and e_ij is the random effect associated with student i in school j.

The base model includes the students’ demographic characteristics (race or ethnicity, gender, SES, and immigration status) as the key student-level predictors (Model 1) and school contexts (community type, demographic composition) as the school-level covariates (Model 2). Building on this model, the subsequent models consecutively included first the principal’s background (years of service as principal in the current school) and then the shared leadership practices in the school (Model 3) to answer RQ1. To answer RQ2, we first allowed all student-level slopes representing inequities in the sense of school belonging attached to students’ different demographic characteristics to vary as an exploratory building process (Anderson, 2012). Our aim was to understand whether there is such variance across schools for all of the students with different demographic characteristics and, if so, whether it can be explained by the shared leadership practice in the school. Subsequently, we allowed the student slopes with a statistically significant variance component to vary (Model 4), and we included the moderating effects of the shared leadership in school to explain such variance by using cross-level interaction terms (Model 5). Using these cross-level interaction terms for the nonsignificant student-level slopes is justifiable if there is a theory-based rationale (Aguinis et al., 2013). For example, although the fixed-slope models previously used may suggest that the pooled mean difference in the sense of school belonging between American Indian/Alaska Native American students and White students is not statistically significant, there can be significant variability in that difference between schools (e.g., Demmert et al., 2006). This may depend on the characteristics of the schools and their leaders. All of the continuous covariates and predictors were grand-mean-centered to enable straightforward interpretations. The combined single equation of the final model is presented in Appendix 4.

The performance of these multiple nested models was compared based on three criteria of the goodness-of-fit test: (1) a deviance test, (2) the Akaike information criterion (AIC), and (3) the Bayesian information criterion (BIC). Specifically, the deviance test statistically evaluates whether the additional interaction terms improve the fit of the model based on the number of parameters used in the two nested models, obtained from the maximum likelihood estimator. Furthermore, a model with lower AIC and BIC values indicates a model with a better fit.

We used Mplus software to address the complex sampling design of HSLS:09 by using an approach (TYPE = COMPLEX TWOLEVEL RANDOM) that considered both stratification and sampling weights (Lumley, 2011; Rabe-Hesketh & Skrondal, 2006). Specifically, we used the analytical weights and maximum likelihood estimation with robust standard errors in Mplus to address the potential issue related to the small subgroup sample size (see Asparouhov & Muthén, 2006; Muthén & Satorra, 1995; Rabe-Hesketh & Skrondal, 2006, for specific details). Regarding missing values, Ingels et al. (2011) emphasized that “HSLS:09 variables in general did not suffer from high levels of item nonresponse” (p. viii). The missing values in this study were treated using the model-based full information maximum likelihood procedure in Mplus. Schafer and Graham (2002) demonstrated that maximum likelihood estimation is usually unbiased. Therefore, the missing values should not significantly affect the results of our analysis.

Results

School Belonging by Student Demographics

Before presenting the results of our LMM analysis, we wish to capture how school belonging is related to student demographics without considering the school and principal characteristics. Model 1 in Table 2 reports the results of the student-level, variables-only model, using a composite variable of students’ sense of school belonging predicted by their different demographic characteristics. The results for Model 1 show that there were significant disparities in students’ sense of belonging by gender (b = 0.06, p < .01) and SES (b = 0.10, p < .001). We also found that, on average, the sense of school belonging among mixed-race students in the United States was significantly lower than that of their White counterparts (b = −0.11, p < .05). In terms of immigration backgrounds, the sense of belonging of foreign-born students was higher than that of U.S.-born students regardless of other demographic characteristics (b = 0.12, p < .01). The results indicate that, on average, the sense of school belonging among girls and foreign-born students was 0.06 SD and 0.12 SD higher than that of boys and U.S.-born students, respectively. The sense of school belonging among mixed-race students was 0.11 SD lower than that of White students without accounting for school and principal characteristics. In addition, the one-unit difference in students’ SES was positively associated with a 0.10 SD difference in their sense of school belonging.

Table 2

Results of Fitting Models for Students’ Sense of School Belonging Based on Their Different Demographic Characteristics (Model 1), School Contexts (Model 2), and Principal and School Characteristics (Model 3)

Fixed Effects	Model 1		Model 2		Model 3
	b	SE	b	SE	b	SE
Students’ different demographic characteristics
Female	0.06 **	0.02	0.06 **	0.02	0.07 **	0.02
Race or ethnicity
American Indian/Alaska Native	0.01	0.04	0.01	0.04	0.03	0.05
Black	−0.01	0.04	0.01	0.04	0.01	0.04
Hispanic	−0.01	0.03	−0.01	0.03	−0.02	0.03
Asian American/Pacific Islander (AAPI)	0.05	0.04	0.04	0.04	0.04	0.04
Mixed-race	−0.11 *	0.04	−0.12 **	0.04	−0.15 **	0.05
Socioeconomic status (SES)	0.10 ***	0.01	0.07 ***	0.01	0.07 ***	0.02
Immigrant status (foreign-born)	0.12 **	0.04	0.12 **	0.04	0.10 *	0.05
School contexts
Suburban schools			−0.06 *	0.03	−0.05	0.03
Town schools			−0.12 ***	0.04	−0.11 *	0.04
Rural schools			−0.14 ***	0.04	−0.11 **	0.04
% of free or reduced lunch–eligible students			−0.01 ***	0.00	−0.01 ***	0.00
% of special education students			−0.01 ***	0.00	−0.01 **	0.00
% of English language learners			−0.01	0.00	0.01	0.00
Principal and school characteristics
Years served as a principal in the current school					0.02 ***	0.00
Teachers’ perceptions on shared leadership practices in school					0.07 ***	0.02
(Intercept)	0.02	0.02	0.26 ***	0.03	0.16 ***	0.04
Deviance	35,163.30		35,068.20		29,385.05
AIC	35,185.30		35,102.20		29,423.05
BIC	35,232.38		35,174.97		29,501.15

Note. Weighted n = 4,197,724 (unweighted n = 25,206). (1) For all models, the reference groups for race or ethnicity, female, immigration status, school community type are White, male, U.S.-born, and urban schools, respectively. SE: Standard errors.

*

p < .05, ^**p < .01, ^***p < .001.

To What Extent Are Principals’ Shared Leadership Practices in School Associated With a Positive Sense of School Belonging Among Students?

Based on the results of Model 1, we added school contexts (Model 2) and school and principal characteristics (Model 3), including principals’ shared leadership practices in school, in that order. We observed similar patterns of significant disparities in students’ sense of school belonging by gender (b = 0.06, p < .01), race or ethnicity of mixed-race students (b = −0.12, p < .01), SES background (b = 0.07, p < .001), and immigration status (b = 0.12, p < .01) after accounting for school contexts in Model 2. These results indicate that the sense of school belonging among girls and foreign-born students was 0.06 SD and 0.12 SD higher than those of boys and U.S.-born students, respectively, and the sense of school belonging among mixed-race students was 0.12 SD lower than that of their White counterparts, accounting for school contexts. The results of Model 2 also indicated that the one-unit difference in students’ SES was positively associated with a 0.07 SD difference in their sense of school belonging, holding school contexts constant. Furthermore, except for the percentage of English language learners in the school, the school contexts used in this study were significantly associated with the estimated school mean of students’ sense of school belonging. Specifically, the estimated school means of students’ sense of school belonging in suburban (b = −0.06, p < .05), town (b = −0.12, p < .001), and rural schools (b = −0.14, p < .001) were significantly lower than those of urban schools. In addition, the one-percentage-point differences for FRL-eligible students (b = −0.01, p < .001) and special education students (b = −0.01, p < .001) in school were negatively associated with the estimated school mean of students’ sense of school belonging. Adding these school contexts in Model 2 significantly improved the model fit compared to Model 1, which did not consider school contexts, with χ²(6) = 95.10, p < .001, and lower values of AIC and BIC.

Incorporating the two key school and principal characteristics (teachers’ perceptions on shared leadership practices, years served as a principal in the current school) into Model 3 further improved the model fit for the data based on the deviance test, χ²(2) = 5683.15, p < .001, and the lower values of AIC and BIC. The previously identified patterns related to students’ different demographic characteristics (mixed-race students, female, SES, and immigration status) and school contexts (community type, demographic composition) in Model 2 remained the same in Model 3, with the exception of the coefficient for suburban schools: in Model 3, the mean difference between suburban and urban schools was not statistically significant after controlling for the key principal and school characteristics. In Model 3, the magnitude of principals’ years of service (b = 0.02, p < .001) was small in association with students’ school belonging, whereas shared leadership practices in school (b = 0.07, p < .001) were significantly and positively associated with the estimated school mean of students’ sense of school belonging. This result indicates that one additional year of experience of service as a principal and a one-unit increase in teachers’ perceptions of shared leadership practices in school were associated with 0.02 SD and 0.07 SD increases in the estimated school means of students’ sense of school belonging, respectively.

To What Extent Do Principals’ Shared Leadership Practices in School Moderate the Relationship Between Students’ Racial/Ethnic Characteristics and Their Sense of School Belonging?

Based on the steps involved in building the analytical models discussed previously, Table 3 reports the results of the models that include only the student-level slopes with statistically significant variance across schools (i.e., the slopes for American Indian/Alaska Native and Black students) after allowing all student-level slopes to vary across schools (Models 4 and 5). In particular, the results of the random effects on the slopes of American Indian/Alaska Native and Black students in Model 4 (random intercept and random slope model) showed that the disparities in the sense of school belonging that were related to these students differed across schools. This result demonstrates the need to understand the aspects that may explain such variability. That is, the student-level slopes that have nonsignificant variance across schools were fixed.

Table 3

Results of Fitting Linear Mixed-Effects Models of Students’ School Belonging Predicted by Their Different Demographic Characteristics, School Contexts, and Principal and School Characteristics

	Model 4		Model 5
Fixed Effects	b	SE	b	SE
Students’ different demographic characteristics
Female	0.06 **	0.02	0.06 **	0.02
Race or ethnicity
American Indian/Alaska Native	0.02	0.05	0.02	0.04
Black	0.01	0.04	0.01	0.04
Hispanic	−0.01	0.03	−0.01	0.03
Asian American/Pacific Islander (AAPI)	0.04	0.04	0.04	0.04
Mixed-race	−0.14 **	0.05	−0.14 **	0.05
Socioeconomic status (SES)	0.07 ***	0.02	0.07 ***	0.02
Immigrant status (foreign-born)	0.10 *	0.05	0.10 *	0.05
School contexts
Suburban schools	−0.05	0.03	−0.05	0.03
Town schools	−0.11 *	0.05	−0.11 *	0.05
Rural schools	−0.11 **	0.04	−0.11 **	0.04
% of free or reduced lunch–eligible students	−0.01 ***	0.00	−0.01 ***	0.00
% of special education students	−0.01 **	0.00	−0.01 **	0.00
% of English language learners	0.00	0.00	0.00	0.00
Principal and school characteristics
Years served as a principal in the current school	0.02 ***	0.00	0.02 ***	0.00
Teachers’ perceptions on shared leadership practices in school	0.07 ***	0.02	0.07 ***	0.02
Cross-level interaction
American Indian/Alaska Native student × Teachers’ perceptions on shared leadership practices in school			0.10 *	0.05
(Intercept)	0.17 ***	0.04	0.17 ***	0.04
Random Effects (Variance Components)
Within-student	0.81***		0.81***
Student intercept	0.07***		0.07***
American Indian/Alaska Native student slope	0.07*		0.06***
Black student slope	0.21***		0.21***
Deviance	29,306.68		29,302.47
AIC	29,348.68		29,346.47
BIC	29,434.99		29,436.90

Note. Weighted n = 4,197,724 (unweighted n = 25,206). (1) For all models, the reference groups for race or ethnicity, female, immigration status, school community type are White, male, U.S.-born, and urban schools, respectively. (2) Models 4 and 5 include only statistically significant random effects and the significant cross-level interaction, respectively. SE: Standard errors.

*

p < .05, ^**p < .01, ^***p < .001.

Adding cross-level interaction terms to Model 4, the results of Model 5 (the cross-level interaction model) showed that the cross-level interaction effect between shared leadership practices in school as perceived by teachers and the slope of American Indian/Alaska Native students was statistically significant (b = 0.10, p < .05), whereas we did not find any other cross-level interaction effects. This finding indicates that while the pooled mean difference in the sense of school belonging between American Indian/Alaska Native and White students from all the schools was not statistically significant (b = 0.02, p = n.s.), the disparity depended on the teachers’ perception of the shared leadership practices in the school. In other words, in a school in which teachers’ perception of shared leadership practices is one unit higher, American Indian/Alaska Native students attending it tend to have a 0.10 SD stronger sense of school belonging than their White counterparts. The proportion of the variance in the slope of American Indian/Alaska Native students explained by the cross-level interaction in Model 5 was 0.14 (a small to medium effect, Cohen, 1992), accounting for approximately 14% of the total variance in the gap between the sense of school belonging of American Indian/Alaska Native and White students across schools. Our model comparison concluded that Model 5 (the final cross-level interaction model) fit the data better than the previous model (Model 4) based on a deviance test, χ²(1) = 4.21, p < .05, and the lower value of AIC. Nevertheless, as previously discussed, this finding regarding the American Indian/Alaska Native subgroup of students should be treated with caution given the relatively small subgroup sample size. We discuss this limitation further, especially in relation to the future direction of data collection and research focusing on this particular group of students.

Discussion

Disparities in School Belonging by Student Demographics

Our analysis, which was based on a nationally representative sample, revealed that students’ gender, SES, immigration status, and race or ethnicity (especially for mixed-race students) were the key demographic characteristics associated with the disparities in their sense of school belonging. This finding resonates with previous studies, indicating a positive association for girls (Anderman, 2002; Gillen-O’Neel & Fuligni, 2013; Smerdon, 2002) and for students from higher SES backgrounds (Kim & Sax, 2009; Ostrove & Long, 2007). Girls’ higher sense of school belonging may be the result of the gendered process of socialization for females, which gives more weight to connectedness and attachment to others (Gilligan, 1993; Goodenow, 1993). Further, the finding regarding students’ SES may suggest that students with a lower SES are perceived to be less connected to their teachers and schools because of their limited access to cultural and social capital (Jury et al., 2019; Lareau & Horvat, 1999; Ostrove & Long, 2007). We found a positive sense of school belonging among foreign-born students, which is consistent with previous evidence of a higher level of engagement and connectedness to schools among immigrant students (DeNicolo et al., 2017; Kao & Tienda, 2012). A plausible explanation for this finding may be related to foreign-born students’ more positive academic attitudes, motivation, and greater relatedness with their peers and teachers as compared to U.S.-born students (Greenman, 2013; M. Lee et al., 2014).

The relatively lower sense of school belonging among mixed-race students in comparison with their White counterparts in this study, regardless of schooling contexts, corroborates previous findings that have exposed systematic inequities in the schooling experiences of mixed-race students (e.g., Fisher et al., 2014), who have been characterized as “the fifth minority” (Tutwiler, 2016, p. x). This finding can be explained by the central tenet of MultiCrit (Harris, 2016), incorporating the literature that focuses on mixed-race students in the United States. Specifically, the theoretical framework of MultiCrit points out that the structural challenges and power relations experienced by mixed-race students in school are the major cause of inequities such as monoracism (Harris, 2016), prejudice and discrimination (Museus et al., 2016), teachers’ stereotyping (Williams, 2011), and exclusion from interpersonal relationships in school (Harris, 2016; Joseph-Salisbury & Andrews, 2017). These multiple factors hinder the satisfaction of mixed-race students’ basic psychological need for relatedness in a hostile organizational and social environment and are likely to lead to their reduced sense of school belonging in U.S. high schools.

Our finding with regard to Asian American/Pacific Islander (AAPI) students corroborates the quantitative findings of previous studies that have identified nonsignificant differences in these students’ sense of school belonging compared with their White counterparts (Gillen-O’Neel & Fuligni, 2013; Smerdon, 2002). This finding is likely to be because the AAPI subgroup included students from diverse backgrounds, which may have masked the nuances and complexities in their sense of school belonging attached to their specific ethnicities (Jang, 2023). In contrast, our findings using a nationally representative sample of ninth graders showed statistically nonsignificant differences in the sense of school belonging among Hispanic and Black students compared with their White peers, whereas previous studies using small-scale datasets and/or focusing on specific state contexts have identified significant racial or ethnic disparities (e.g., Fisher et al., 2020; Russell & Mantilla-Blanco, 2022; Walton & Cohen, 2007). It is important to note that our findings should not be interpreted simplistically as indicating that these racial or ethnic minority students do not experience structural challenges that negatively affect their sense of school belonging. Instead, this inconsistency could be attributed to varying contexts, such as studies focusing on specific states or regions (e.g., Walton & Cohen, 2007) versus the nationally representative states in our study. Specifically, whereas small-scale inquiries have identified the major challenges hindering Hispanic students’ sense of school belonging (e.g., Russell & Mantilla-Blanco, 2022), our findings for Hispanic students showing statistically nonsignificant differences between Hispanic students and their White peers are consistent with previous large-scale studies using nationally representative high school samples, such as the National Longitudinal Study of Adolescent to Adult Health (Add Health) (McNeely et al., 2002). Finally, although many quantitative studies (e.g., Anyon et al., 2016; McNeely et al., 2002) have reported a significantly lower sense of school belonging among Black students compared with their White peers, the statistically nonsignificant difference between Black and White students in our study may be due to differences in the students’ developmental stages and grades. For example, whereas previous studies have included students in all grades from 7 to 12 (e.g., McNeely et al., 2002), our study focused on ninth graders. These inconsistent findings may well represent the fluid and complex nature of students’ sense of school belonging ( K. A. Allen et al., 2021), warranting further qualitative investigations. In the following, we discuss specific directions for future data collection and research related to this observation.

Shared Leadership in School and Students’ Sense of School Belonging

Informed by emerging research on the role of principals and schools in students’ positive experiences of school learning environments broadly and school belonging in particular (e.g., M. Kudlats & Brown, 2021; Lee et al., 2021; Louis et al., 2016; Riley, 2017, 2019, 2022; Ryu et al., 2020), we investigated the extent to which shared leadership in school was associated with students’ sense of school belonging. We found that shared leadership practices in schools by both principals and teachers were significantly and positively associated with students’ sense of school belonging when controlling for other factors. This finding supports previous studies that have emphasized the crucial role that schools play in influencing the diverse social and psychological determinants of students’ sense of school belonging (e.g., Allen et al., 2018; Korpershoek et al., 2020; Ma, 2003). Specifically, our finding indicates that shared leadership practices in schools by both principals (i.e., enacting inclusive practices and communication, setting clear expectations and organizational goals, encouraging innovation) and teachers (i.e., actively collaborating with colleagues to offer high standards of learning to all students) are important in creating a strong sense of school belonging in all students. It resonates with previous studies that have documented the importance of these practices in improving the school environment and teachers’ relational skills, promoting agency among teachers, and leading to school reform and improvement (Flutter & Ruddock, 2004; Pantić, 2015; Pearce & Conger, 2003; Priestley et al., 2015; Radd et al., 2021; Wilkinson & Pearson, 2009). These factors, in turn, lead to the provision of stable learning experiences and agency among students (Printy & Marks, 2006). These are essential in creating a positive sense of school belonging in students (Riley, 2017).

Another important finding was that shared leadership practices in school were critical to a positive sense of school belonging, particularly among American Indian/Alaska Native students, a minoritized group in the U.S. context. As the central tenet of TribalCrit posits, colonized educational systems and policies in the United States may cause the systematic challenges, inequities, and conflicts that are experienced by this group of students (Bird et al., 2013; Brayboy et al., 2012; T. S. Lee & Quijada Cerecer, 2010; Martinez, 2014). Furthermore, TribalCrit scholars have argued that educators and school leaders can and should tackle such inequities in the diverse schooling experienced by American Indian/Alaska Native students, including in their sense of school belonging (Garcia & Shirley, 2012; Jacob et al., 2015; McConnell, 2013; Quijada Cerecer, 2013). In line with this theoretical framework and literature, our finding indicates that shared leadership practices in school can be a critical source for a positive sense of school belonging, particularly among American Indian/Alaska Native students. This finding is in line with the view that effective and authentic communication between members of the school community as an element of shared leadership practices is important for creating equitable institutions (DeMatthews et al., 2021; Theoharis, 2009). It also resonates with research that has highlighted school leaders’ equity-focused practices, including embracing diverse voices and perspectives through dialogue and activities involving the whole school (Radd et al., 2021).

A possible explanation for this might lie in the positive effects of shared leadership practices on improving teacher-student relationships, improving the school culture to offer diverse students positive schooling experiences, and boosting the agency of both teachers and students (Printy & Marks, 2006). These are factors that may uniquely benefit American Indian/Alaska Native students’ sense of school belonging. In particular, the literature has illustrated how various cultural and structural barriers and challenges in school obstruct the learning experiences of American Indian/Alaska Native students. Examples of these are the experience of structural racism in school (Palmer & Cooke, 1996), the cultural dissonance of colonized curricula (Rahman, 2013), and culturally insensitive education (Khalifa et al., 2018). Specifically, American Indian/Alaska Native students are typically forced to assimilate to mainstream schooling, policies, and ideologies in which “principles of competition, neoliberalism, [and] managerialism” prevail (Jacob et al., 2015, p. 6), which negatively affects these students’ sense of school belonging (Khalifa et al., 2018). Meanwhile, to meet the need for relatedness and foster agency and engagement in the schooling of American Indian/Alaska Native students, the institutionalized communication and reflective dialogue of school leaders and teachers is crucial (Quijada Cerecer, 2013). In other words, the reflective, inclusive, and interactive teaching and learning environments fostered by shared leadership practices may support indigenous ways of knowing and learning (Jacob et al., 2015; Martinez, 2014), thereby benefiting the sense of school belonging of American Indian/Alaska Native students. This speculation rests on the premise of a culturally responsive leadership that highlights inclusivity in decision-making and communication as being the fundamental characteristics of the decolonization of Indigenous leadership (Khalifa et al., 2018). In addition, strong shared leadership practices could create a school culture that may undermine negative stereotyping (e.g., deficit-oriented perspective) and microaggressions against American Indian/Alaska Native students (Hughes & Morrison, 2022; Printy & Marks, 2006; Riley et al., 2020), which in turn can protect their sense of school belonging (Martinez, 2014).

Limitations and Future Research

This study has several limitations that remain to be addressed. These limitations mainly stem from the use of secondary data collected by NCES, which meant that this study was exploratory in nature. First, the sample of the American Indian/Alaska Native students in these data was relatively small. As such, the result does not appear to be generalizable. Nevertheless, given that the majority of quantitative studies focusing on this group of students have been descriptive (Demmert et al., 2006), our exploratory study has specific value in providing a strong foundation for addressing the critical issue of the sense of school belonging, which has not been thoroughly investigated in this population previously, as indicated by Swedberg (2020). Our exploratory study further suggests the potential of shared leadership practices in schools to achieve a positive and equitable sense of school belonging among ninth graders, which has implications for future research in terms of determining sampling and data collection methods, guiding theory-building, and developing leadership practice (e.g., Louis et al., 2016), as we specify later. Second, similar to other nationally representative datasets that include items related to students’ sense of school belonging (e.g., Add Health and TIMSS), we were unable to use more comprehensive instruments to measure students’ sense of school belonging, such as an 18-item measure of the Psychological Sense of School Membership Scale (PSSM, Goodenow & Grady, 1993) and a 17 item-measure of the Sense of Social Fit Scale (SSF, Walton & Cohen, 2007). This limitation is related to the time and cost concerns associated with including a large number of survey items in large-scale, nationally representative datasets. Although our modified scale included the items most commonly used in the literature and other nationally representative datasets, the lack of a comprehensive set of items may limit fruitful comparisons and syntheses of the major findings across studies (Meyer et al., 2023). In addition, as noted, we focused on ninth graders from one year of the longitudinal dataset due to data limitations. Although the construct of students’ sense of school belonging may vary over time (Allen et al., 2021), the use of survey items from a single time point cannot reveal the longitudinal relationship between shared leadership practices and students’ sense of school belonging. Third, we were unable to include a comprehensive set of students’ demographic characteristics, identities, and contexts that may be important for their sense of school belonging (e.g., sexuality, disability), given the limited data in the secondary dataset.

To address these limitations, we propose directions for future research. First, increasing the sample size of American Indian/Alaska Native students would lead to higher-quality research in this group. This could be carried out using an over-sampling strategy similar to that used in HSLS:09 with other relatively small populations (e.g., Asian students). Second, future studies could collect longitudinal datasets using the same scales to measure the construct of students’ sense of school belonging and use such data to examine whether shared leadership practices moderate changes in high school students’ sense of belonging over time. Furthermore, future studies using data collected during specific periods, such as during and after the COVID-19 pandemic, could reveal how different contexts influence students’ sense of belonging. Furthermore, future small-scale studies could use more comprehensive and internationally validated instruments to measure students’ sense of school belonging, such as the PSSM (Goodenow & Grady, 1993), the SSF (Walton & Cohen, 2007), and the Hemingway Measure of Adolescent Connectedness (Karcher, 2001), to produce more comparable findings. This would enable the synthesis of insights from various studies to enhance a positive and equitable sense of school belonging among diverse students. Third, future research could collect data on other diverse identities to further explore patterns of inequities and their associations with schools and the attributes of their principals.

In addition, although the results of this study indicated that there are variations across schools in the inequities in school belonging experienced by Black students, this study did not find any evidence that shared leadership practice in school plays a moderating role. This suggests that other school-level factors (e.g., other types of school leadership, different aspects of school culture) not addressed in this study might play a moderating role in the inequities in the sampled students’ sense of school belonging. In this regard, future research can benefit from the literature on school leaders’ practices from an asset-based perspective and a strengths-based perspective for students of color, in particular for Black students (e.g., Wright, 2018), which may reduce the inequities in their sense of school belonging. This speculation warrants future research to identify the moderating effects of various leadership practices in cultivating an equitable sense of school belonging among diverse students. Furthermore, studies exploring the experiences and academic performance of mixed-race students have revealed significant variation among this group of students based on their different combinations of racial and ethnic backgrounds (e.g., Joseph-Salisbury & Andrews, 2017). Future research should unpack the complexity of mixed-race students’ sense of school belonging by disaggregating them based on these combinations of racial and ethnic identities (e.g., Black & White, American Indian/Alaska Native & Asian/Pacific Islander; see Lopez, 2003). A more nuanced understanding will be helpful in addressing the distinctive schooling challenges and disadvantages that mixed-race students experience.

To actualize the proposed line of future research, we echo the suggestion of Perrone et al. (2022) to collect nationally representative information about the practices and policies related to principals and schools. To the best of our knowledge, the HSLS:09 is the most recent nationally representative dataset based on a nested data structure that includes principals, schools, and teachers matched with detailed information about the demographic characteristics, schooling experiences (e.g., sense of school belonging), and academic achievement of their students. Thus, we recommend that state education agencies and the U.S. Department of Education collect nationally representative data on schools and principals matched with their students, including longitudinally consistent measures of students’ diverse schooling experiences, such as their sense of school belonging. The availability of such data would enable researchers to further scrutinize to what extent and in what way schools and principals can facilitate a more equitable sense of school belonging among diverse groups of students.

Conclusion

Ensuring a sense of school belonging, especially among racial or ethnic minority students, is a critical challenge for school leaders in the United States, given the wide gaps in students’ sense of school belonging. Therefore, we suggest that it is important to enhance principals’ awareness of the specific challenges and barriers faced by marginalized students that underlie the patterns of inequities in their sense of school belonging. Our study suggests that shared leadership can be a vehicle for principals to identify and tackle the context-specific challenges and barriers facing minoritized students by enacting inclusive practices and efficient communication in decision-making, as well as creating a school culture in which teachers interact with each other to promote the learning of all students. We also suggest that leadership preparation programs should seek to ensure that aspiring principals understand that the creation of schools to which students feel connected cannot be done alone but is a collective and inclusive effort. Furthermore, such programs should ensure that aspiring principals have effective, engaged communication skills that embrace diverse voices and stimulate interactions between school community members (Radd et al., 2021), who are critical contributors to innovation and change. Providing such preparation and professional development opportunities to principals serving in rural schools and the states where American Indian/Alaska Native students are concentrated, such as Alaska, Oklahoma, Montana, and New Mexico (NCES, 2022), will be crucial for creating a positive and equitable sense of school belonging among students. Finally, based on our study’s findings, we conclude that students’ lack of sense of school belonging is not based on any individual deficiency. Instead, shared school practices of principals and teachers are essential for students to have a positive and equitable sense of school belonging in U.S. high schools.