Racial Disparities in Context: Student-,School-,and County-Level Effects on the Likelihood of Obesity among Elementary School Students

Abstract

Childhood obesity rates in the United States have risen since the 1980s and are especially high among racial minorities. Researchers document differentials in obesity rates by race, socioeconomic status, school characteristics, and place. In this study, the authors examine the impact of race on the likelihood of obesity at the student, school, and county levels and the interactions between student race and school racial composition. The data are from 74,661 third to fifth grade students in 317 schools in 38 North Carolina counties. Multilevel logistic regression models showed that racial differences in the likelihood of obesity persisted even when racial composition and socioeconomic disadvantage at the school level were controlled. The differences between white and nonwhite students slightly decreased once school-level measures were added. The magnitude of the effects of student-level race on the relative odds of obesity varied according to the racial composition of the school. These student- and school-level results held even when county-level race and socioeconomic variables were controlled. The results show that contextual factors at the school and county levels are important social determinants of racial disparities in the likelihood of childhood obesity.

Keywords

racial segregation public health health disparities children

Obesity rates of children have risen since the 1980s and are especially high among low-income and racial/ethnic minority children (Ogden et al. 2014). Data from the National Health and Nutrition Examination Survey show that 17 percent of U.S. youth (aged 2–19 years) were obese in 2011–2014. In addition, non-Hispanic black youth and Hispanic youth had higher obesity rates than non-Hispanic white youth in the 2011–2014 data. This pattern of higher rates for non-Hispanic black and Hispanic youth compared with non-Hispanic white youth is also present when the data are broken down by sex (Ogden et al. 2015). Racial disparities in childhood obesity rates are important for future health outcomes given that obese children are at higher risk for a number of physical and mental health problems, including high blood pressure, depression, and diabetes (Cornette 2008; Freedman et al. 2007; Friedemann et al. 2012).

Childhood obesity researchers examine how behaviors (i.e., eating and physical activity choices), individual and household demographics (e.g., race/ethnicity, socioeconomic status [SES]), exposure to stress and violence (Boynton-Jarrett et al. 2012), and, increasingly, the characteristics of schools and of the local geographic environment affect obesity rates (Brewer and Kimbro 2014; Carroll-Scott et al. 2013; Cetateanu and Jones 2014; Evans et al. 2012; Grow et al. 2010; Kershaw and Albrecht 2014; Kimbro and Denney 2013; Lee 2012; Lovasi et al. 2009; Rossen 2014). Scholars point to the necessity of including measures of the structural inequalities of the environments in which individuals are embedded in empirical analyses of racial differences in obesity rates (Chang 2006; Chang, Hillier, and Mehta 2009; Lee, Harris, and Lee 2013; Rossen 2014; Rossen and Talih 2014; Ruel et al. 2010). In our study, we use multilevel logistic regression to investigate how student-level, school-level, and county- or community-level variables affect the odds of being obese and the interaction among these levels. Specifically, we examine the effects of (1) student race, (2) the racial composition and socioeconomic disadvantage of schools, (3) the interaction between student race and school racial composition, and (4) the racial and socioeconomic inequalities and rurality of county of residence on the likelihood of obesity. Data come from 74,661 North Carolina elementary school students in 317 schools located in 38 urban and rural counties.

Background

The risk for obesity is highest among children who are racial/ethnic minorities (Flegal et al. 2010; Freedman et al. 2006), and there is evidence that “severe obesity” is increasing, particularly among Hispanic boys and non-Hispanic black girls (Wang, Gortmaker, and Taveras 2011). In addition to early life experiences (Taveras et al. 2013), racial and socioeconomic inequalities in schools and in the environments in which children live are important for understanding obesity disparities (Singh, Kogan, and van Dyck 2008; Williams and Sternthal 2010).

Public schools have, after a period of modest integration, become resegregated over the past 20 years (Kozol 2005; Porter and Bratter 2015). Much in the same way that residential segregation by race restricts the amount of resources available in communities, students in high-minority schools are likely to have fewer available resources (Condron 2009; Kozol 2005). Studies point to the widening of academic achievement gaps between white and minority students as segregation increases (Condron 2009; Kainz and Pan 2014; Lleras 2008; Mayer 2002) and to the health impacts of variations in the school environment (Walsemann, Gee, and Ro 2013). Schools with high-minority populations lack the resources of schools with high proportions of white students (Roscigno 1999; Roscigno and Ainsworth-Darnell 1999). One study of food options in American schools found that schools with high concentrations of minority or low-SES students had less healthy food options than higher SES schools and schools with fewer minority students (Delva, O’Malley, and Johnston 2007). Another study found that low school SES and rural location were school-level factors that influenced students’ obesity trajectories over time (Miyazaki and Stack 2015). A study using nationally representative data found that the concentration of poverty within a school was associated with students being overweight and obese and that increases in school-level poverty reduced the protective effect of parents’ education on student weight (Martin et al. 2012).

Scholars of race and health disparities have also pointed to the role of micro-aggressions in the school environment and their consequences for people of color (Pachter and García Coll 2009; Walsemann, Bell, and Maitra 2011). Discrimination, violence, or even the threat of these, can lead to health consequences for minority children (Paradies et al. 2015; Priest et al. 2013, 2014). For example, a study using the National Longitudinal Study of Adolescent Health found that depressive symptoms among black students varied with the racial composition of their schools, but this relationship disappeared after students’ perceptions of the discrimination and school attachment were controlled (Walsemann et al. 2011). A 2009 study found that minority girls in predominately white schools had lower values of body mass index (BMI) than minority girls who attended predominately nonwhite schools (Bernell, Mijanovich, and Weitzman 2009). The authors attributed these differences to social norms and the influence of white children on minority children, but there was no accounting for school socioeconomic disadvantage. An examination of differences in children’s BMI must therefore consider both school-level socioeconomic and racial composition, in addition to the student’s own race and sex, to fully understand the impacts of race on childhood obesity.

The impact of the neighborhood or community context on obesity rates is another area of interest for scholars of racial disparities. Ecological analyses using counties as the community or geographic context and multilevel analyses using census tracts as the neighborhood context have found mixed results in terms of the effects of the socioeconomic and racial neighborhood or locality environments on obesity rates (Carroll-Scott et al. 2013; Fan, Wen, and Kowaleski-Jones 2016). Differences in variables and in the types of data used may account for divergent results, but the need to consider measures of the local context in the study of obesity rates appears clear.

Residential segregation by race is persistent across space and time in the United States. For African Americans, it is associated with increased poverty, housing deterioration, unemployment, inferior public services, and lower quality schools (Massey and Denton 1993). Minority-segregated neighborhoods have less access to safe recreational facilities (Centers for Disease Control and Prevention 1999; Chung and Myers 1999) and have higher costs of living compared with integrated or white-segregated neighborhoods (Williams and Collins 2001). These structural conditions affect the health and well-being of individuals, and they cannot be accounted for by studying individual-level racial identity without including characteristics of place (Williams and Collins 2001). For African Americans, living in segregated spaces has detrimental health effects that are specific to segregation and not racial status or identity (Robert 1999; Williams and Collins 1995, 2001). These differences in social context may in part explain racial disparities in obesity among children (Rossen and Talih 2014). Similar to the contextual effects of school environments, a number of researchers have argued that the racism minorities encounter in integrated places is detrimental to their physical and mental health (Kelly, Becares, and Nazroo 2013; Priest et al. 2014; Ruel et al. 2010).

Residential segregation is associated with increased obesity rates. One study found that African Americans living in segregated metropolitan neighborhoods were more likely to be obese (Chang 2006). A study of adults in Philadelphia found that BMI was higher among African American women (but not men) in racially isolated neighborhoods (Chang et al. 2009). Another study found that racial disparities in child and adolescent obesity rates were explained by economic disadvantage and demographic characteristics at the neighborhood level (Rossen 2014). A study of fifth and sixth graders in New Haven, Connecticut, found that students in affluent neighborhoods ate healthier and reported less screen time than their counterparts in concentrated disadvantage neighborhoods (Carroll-Scott et al. 2013). These studies move beyond individual-level race variables and highlight the effects of racial and socioeconomic segregation in urban neighborhoods. Racial and socioeconomic inequality and isolation are not unique to urban places. These inequalities are persistent in both suburban (Frey 2011) and rural areas (Wimberley and Morris 2003). The higher rates of obesity in rural as opposed to urban places are a major subject for scholars and health activists (McCormack and Meendering 2016; Scott and Wilson 2011).

To expand on previous studies of racial disparities in childhood obesity rates and to better understand the relationship among student-level race, school-level racial composition, and county-level race and socioeconomic contexts, we examine the following questions: (1) Do student-level racial differences persist when school-level racial composition and school-level socioeconomic disadvantage are controlled? (2) Do student-level racial differences in the likelihood of obesity vary on the basis of the racial composition of the school? Specifically, does the percentage of white students at a school affect the likelihood of obesity between nonwhite and white students? (3) Do county-level measures of socioeconomic inequality, racial segregation, and rurality explain school-level and/or student-level racial differences in obesity rates?

To answer these questions, we use data from third to fifth grade North Carolina students that include student- or individual-level, school-level, and community- or county-level variables. We use multilevel models to investigate the effects of the racial composition of the student body on the likelihood of obesity, the interaction between student-level race and racial composition at the school level on the likelihood of obesity, and if these effects are explained by county-level variables.

Methods

Study Population and Sample Data

The data used in this analysis were collected in third to fifth grade physical education (PE) classes as part of the In-school Prevention of Obesity and Disease Program, a statewide initiative in North Carolina to increase training and physical activity in schools.¹ As part of their participation, PE teachers were given, and trained to use, FitnessGram software to record demographic characteristics and the height and weight of their students. This software was also used in a study of students in New York City schools (Rundle et al. 2012). When available, PE teachers measured height and weight using a scale and attached height bar. However, not all schools were equipped with scales, and for these schools, teachers asked the students to report on their height and weight. The method of data collection for these variables is not indicated in the recorded information.

The student height and weight data come from the records collected during the first year (2009–2010) of the program. If a student’s FitnessGram record did not have valid height and weight data, the student was considered “missing” and eliminated from the data used in this analysis. Schools that submitted usable height and weight for fewer than 10 percent of their third through fifth grade students were omitted from this study. A majority of the omitted schools came from North Carolina counties in which no other schools had submitted data. In all, a little over 5 percent of the schools were dropped. The mean response rate for schools included in the data used in this study is 91 percent (SD = .13). This was calculated using the number of valid obesity records from PE teachers divided by the number of third to fifth grade students enrolled at participating schools. Students in lower and higher grades, or outside the age range of 7 to 13 years, were eliminated from this study’s sample data because they were not the target group for the program, and therefore data about them were limited.

The way in which individual-level student height and weight data were recorded introduces potential bias into the measurement of obesity. Asking a student for his or her height and weight is a common practice, but it is undoubtedly a source of error. Prior research shows that respondents tend to overestimate height and underestimate weight in self-reports, thereby underestimating obesity and making it a more conservative measure (Elgar and Stewart 2008; Field, Aneja, and Rosner 2007; Himes et al. 2005; Morrissey et al. 2006). Students could also reply that they “do not know.” Given that cases without measures for height and weight were dropped and that schools with substantial omitted data were also dropped, some of this type of underestimation error is eliminated, but not all of it.²

The final data set for this analysis included information on 74,661 students from 317 different schools. Thirty-eight North Carolina counties, representing all the geographic areas in the state, had one or more schools that participated in the project that produced the student-level data.

Measures

In addition to height and weight, teachers used FitnessGram software to record their students’ date of birth, race/ethnicity status, and grade. These student-level demographic variables were verified using the school’s administrative data. When there was a conflict between the teacher-entered data and the administrative data, the administrative data were used, because they come from parent or guardian reports and are more likely to represent the racial/ethnic category with which the student identifies. Neither citizenship status nor place of birth was available. The dependent variable, whether a student is obese, comes from the data recorded by the PE teachers. Each student’s recorded height and weight were used to calculate BMI (weight in kilograms divided by the square of height in meters). The student’s BMI was then plotted on a Centers for Disease Control and Prevention growth chart to determine his or her percentile for sex and age at the time of the test. In the statistical models, students with BMI values in the 95th percentile or higher were considered obese (Ogden et al. 2010) and were coded 1. Students were coded 0 if they were not obese.

The individual-level variables in the data set included the student’s grade (continuous),³ sex (female = 1, male = 0), and race. Students were grouped into the following categories for analysis: white, Hispanic,⁴ black, and other. Other was a multiracial category including, but not limited to, multiracial, American Indian, and Asian. These categories were grouped together because each of these groups were too small to examine separately in the models. “White” was used as the omitted reference category in the models. Because student records were deidentified in accordance with human subjects protocol and did not include any information on the student’s address or free and reduced-price lunch status, we were unable to measure the students’ household poverty status. Neither parents nor students were surveyed about family and household characteristics, so no information on parents or students’ households (e.g., parents’ education) was available.

The individual-level student information included the local education area and the school number. Using this information, we linked each student record with school- and county-level characteristics from publicly available administrative data (North Carolina Department of Public Instruction 2010a, 2010b; North Carolina Rural Economic Development Center 2012). North Carolina Department of Public Instruction (2010b) data were used to operationalize the racial composition of the school. Consistent with prior work, we used the percentage of white students in a school to create a measure of school racial composition (Walsemann et al. 2013). Percentage white is a continuous variable at the school level. School-level socioeconomic disadvantage was measured using the percentage of students receiving a free or reduced-price lunch at the school (North Carolina Department of Public Instruction 2010a). To minimize issues of multicollinearity between racial composition and socioeconomic disadvantage, two dummy variables were created to control for high- and low-disadvantage schools, respectively. Schools with 75 percent or more of their students receiving free or reduced-price lunch were considered to be at high economic disadvantage, schools with 25 percent or less were considered at low economic disadvantage, and all other schools made up the omitted category. This is consistent with other studies on school disadvantage (Condron 2009).

County-level measures of socioeconomic inequality and of racial segregation were adopted from the Rural Data Bank, which includes data on North Carolina counties from the U.S. Census and the American Communities Survey (North Carolina Rural Economic Development Center 2012). Our study differs from others (Rossen 2014) in that we did not have access to students’ residential addresses, and therefore we could not use census tract as the geographic unit for the measurement of place-based or neighborhood context. However, county is often used as a measure of place-based context in studies of the spatial distribution of inequality (Lobao 2004; Sparks, Sparks, and Campbell 2013; Tomaskovic-Devey and Roscigno 1997). An advantage of using county as the unit to measure place-based context is that it allows us to identify and compare urban and rural counties.

Socioeconomic inequality at the county level is operationalized by using the Gini index, a summary measure of income inequality. The Gini index for each county was taken from the American Community Survey’s five-year estimates (U.S. Census Bureau 2010). “The Gini index varies between zero and one. A value of one indicates perfect inequality where only one household has any income. A value of zero indicates perfect equality, where all households have equal income” (Bee 2012:1). To calculate county-level residential racial segregation, we used dissimilarity indices that measure the evenness of the distribution of two groups among small geographic units (i.e., census tracts) within a larger geographic unit (i.e., county). For example, the formula for calculating the dissimilarity index between black and white households for a county is

\frac{1}{2} \sum_{i = 1}^{N} | \frac{b_{i}}{B} - \frac{w_{i}}{W} | .

This is a commonly used measure of residential segregation (Massey and Denton 1988). Similar to the Gini index, zero indicates complete residential integration, meaning that the two groups in question are equally distributed among census tracts throughout the county. A score of 1 indicates complete residential segregation from one another within the county. Two variables were calculated: a black/white dissimilarity index and a Hispanic/white dissimilarity index. Next, we used a dummy variable to identify rural counties (Bennett, Probst, and Pumkam 2011; North Carolina Rural Economic Development Center 2012). Rural counties were coded 1, with urban counties as the omitted category.

Analytic Strategy

Descriptive statistics were calculated for all variables, including the student-level, school-level, and county-level measures. To effectively model the multiple characteristics of students and place (both school and county), we use multilevel logistic regression (Guo and Zhao 2000). Estimation of these models was done using the PROC GLIMMIX procedure in base SAS 9.4 software (Li et al. 2011). To determine the amount of variation across students, across schools, and across counties, we first ran a three-level model without any predictor variables to confirm that there was statistically significant variation at each level. Consistent with other work, the majority of the variance was at the individual level, followed by the school (5 percent) and county (1 percent) (Sacker, Wiggins, and Bartley 2006; Snijders and Bosker 1999).⁵ Next we ran a model including student-level measures at level 1, followed by the addition of school-level measures at level 2. To investigate how school-level racial composition might be different once school-level economic disadvantage is controlled, the level 2 variables were added one at a time, with school-level racial composition variable added to the model first and then a second model that also included a control for school socioeconomic disadvantage.

Next we ran a model that included a cross-level interaction between student race (level 1) and school-level percentage of white students (level 2). Cross-level (level 1 and level 2) interaction terms were created for each student race/ethnicity variable and the school-level racial composition variable. A product term was created for each category of student-level racial category and the racial composition variable (i.e., percentage white students at the school). For example, an interaction term was created between black and percentage white (Black × % White), and a second term was created for Hispanic and percentage white (Hispanic × % White). The omitted reference category was white students. The final model included county-level measures at level 3.

Results

Table 1 shows the descriptive statistics for the North Carolina sample used in this study. Twenty percent of the students in this sample are obese. The sex balance in the sample is similar to the state population for the relevant student grade levels, with slightly fewer girls than boys. The sample data set also has slightly fewer black students than the state as a whole (30 percent statewide vs. 25 percent in this study). The average percentage of white students within the schools is 51 percent, although it is important to note that white students are not evenly distributed across schools. In about a quarter of the schools, white students make up more than 75 percent of the student population, and for a little over a quarter of schools, less than 25 percent of the population is white (24 percent and 27 percent, respectively). Twenty-nine percent of schools in the sample have high proportions of economically disadvantaged students, and about 14 percent have very low economic disadvantage. Almost 80 percent of the 38 North Carolina counties represented in the sample are rural. The mean Gini index for the counties in the sample was .45, which is slightly lower than the Gini index for the United States as a whole, .467 (Bee 2012). At the county level, the average dissimilarity index for white and black households was .37, meaning that 37 percent of residents would have to move to create an even racial distribution across the county. The mean Hispanic and white dissimilarity index was slightly lower (.27).

Table 1.

Descriptive Statistics for All Key Variables.

Variable	Mean/Frequency	SD	Minimum	Maximum
Student-level variables (level 1), n = 74,661
Obesity
Obese	20%		0	1.00
Not obese (omitted)	80%		0	1.00
Sex
Female	49%		0	1.00
Male (omitted)	51%
Race
Black	25%		0	1.00
Hispanic	13%		0	1.00
Other	8%		0	1.00
White (omitted)	53%
Grade	3.99	0.81	3.00	5.00
School-level variables (level 2), n = 317
Racial composition
Percentage white	51%		0	97%
School socioeconomic disadvantage
≥75% receive free or reduced lunch	29%		0	1.00
≤25% receive free or reduced-price lunch	14%		0	1.00
26%–74% receive free or reduced-price lunch (omitted)	57%		0	1.00
County-level variables (level 3), n = 38
Urban/rural
Rural county	79%		0	1.00
Urban county (omitted)	21%		0	1.00
Residential segregation
Black/white dissimilarity index	0.37	0.11	0.05	0.57
Hispanic/white dissimilarity index	0.27	0.11	0.08	0.48
Economic inequality
Gini index	0.45	0.03	0.40	0.53

Model 1 in Table 2 displays the results of the student-level measures from the multilevel analyses. The odds of obesity for black children in model 1 are about 1.7 times the odds for whites, and Hispanic youth’s odds are 1.99 times those of white children. Children of other races had odds for obesity that were 1.3 times the odds of white children in the model. Female students were less likely to be obese than male students. All of these findings were statistically significant at the .05 level or lower.

Table 2.

Multilevel Logistic Regression Results Predicting Odds of Obesity.

	Model 1: Student Characteristics	Model 2: School Racial Composition	Model 3: School Racial Composition and Socioeconomic Disadvantage	Model 4: Student Race and Racial Composition Interactions	Model 5: County Inequality and Rurality
Variable	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Constant	−1.67*** (.07)	−1.34*** (.09)	−1.77*** (.11)	−1.71*** (.11)	−2.09** (.68)
−2 res log pseudo-likelihood	353,575.9	353,608.7	353,739.6	353,823.9	353,845.1
Student level
Grade	.06*** (.01)	.06*** (.01)	.06*** (.01)	.06*** (.01)	.06*** (.01)
Sex
Female (=1)	−.16*** (.02)	−.16*** (.02)	−.16*** (.02)	−.16*** (.02)	−.16*** (.02)
Race
Black (=1)	.54*** (.03)	.51*** (.03)	.51*** (.03)	.36*** (.06)	.35*** (.06)
Hispanic (=1)	.69*** (.03)	.66*** (.03)	.65*** (.03)	.67*** (.07)	.67*** (.07)
Other (=1)	.28*** (.04)	.26*** (.04)	.26*** (.04)	.10 (.09)	.09 (.09)
School level
Racial composition
Percentage white		−.49*** (.08)	.16 (.12)	.07 (.14)	−.07 (.14)
Socioeconomic
≥75% ED			.20** (.06)	.21** (.06)	.17** (.06)
≤25% ED			−.50*** (.06)	−.51*** (.07)	−.46*** (.07)
County level
Rural county (=1)					.25* (.10)
Black/white DI					.25 (.40)
Hispanic/white DI					−.61 (.46)
Gini index					.85 (1.38)
Interactions
Black × % White				.35** (.11)	.37** (.11)
Hispanic × % White				−0.17 (.12)	−.17 (.12)
Other × % White				0.30* (.15)	.31* (.15)
School-level covariance	.11*** (.01)	.09*** (.01)	.07*** (.01)	.07*** (.01)	.07*** (.01)
County-level covariance	.05** (.02)	.08** (.03)	.04* (.02)	.04* (.02)	.02* (.01)

Note: N = 74,661 students, n = 317 schools, n = 38 counties. DI = dissimilarity index; ED = economically disadvantaged.

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

Models 2 and 3 in Table 2 present the results for the addition of the school-level measures of racial composition and socioeconomic disadvantage, respectively. All of the student-level variables from model 1 remained statistically significant and in the same direction. However, the size of the effects of the student-level variables on the odds of obesity decreased slightly. For example, comparing model 2 with model 1, the odds for black students being obese in the sample are 1.66 compared with the odds for whites in the sample, which are down slightly from 1.71 in the prior model. Comparing model 3 with model 1, the odds of obesity for Hispanic children slightly decreased (down from 1.99 to 1.93) relative to whites, net of the other variables in the model. The effects of grade and sex on the odds of obesity remained consistent across models. Students in higher grades showed slightly higher odds of being obese and being female as opposed to male decreased the odds of being obese across all the models in Table 2.

Model 2 shows that, net of the variables measuring student-level characteristics, the percentage of white students at the school level was associated with a lower likelihood of obesity. For each percentage increase in this school-level variable, the odds of being obese decreased (odds ratio [OR] = .61).

Once school-level socioeconomic disadvantage is controlled in model 3, racial composition at the school level (percentage white students) is no longer statistically significant. Students at socioeconomically disadvantaged schools were 1.22 times as likely to be obese compared with students in non–economically disadvantaged schools. Conversely, students in schools with very low socioeconomic disadvantage were less likely to be obese (OR = .61), net the other variables in the model. Student-level race, sex, and grade variables remain statistically significant predictors of obesity in model 3.

Model 4 adds cross-level interaction terms to test for the effects of interactions between a student’s race and school-level racial composition. There is a statistically significant interaction for black students and the percentage of white students in the school (see Table 2, model 4). Students reporting a race other than Hispanic, black, or white also have a statistically significant interaction with the percentage of white students at the school. Both interactions are positive, meaning that each percentage increase in white students at the school level is associated with increased odds of obesity among black and students of other race compared with white students (OR = 1.43 and 1.35, respectively). There is not a statistically significant interaction between school-level racial composition and student race for Hispanic students in the sample. Figure 1 provides a graphic summary of the ORs for the interaction variables of student race and school racial composition from model 4.

Figure 1.

Odds ratios for obesity by student race and percentage increase in white students at the school level.

Model 5 adds measures of county-level socioeconomic inequality, racial segregation, and rurality. Neither of the county-level dissimilarity indexes for black/white residential segregation or for white/Hispanic residential segregation was statistically significant. Likewise, the county-level Gini index was not statistically significant. However, students in rural counties were 1.28 times as likely to be obese compared with students in urban counties, and the effect was statistically significant. At the student level, sex, grade, and race/ethnicity remain statistically significant and show almost no change in their effect sizes from model 4. The ORs for school-level socioeconomic variables and for the interactions in Model 4 are also statistically significant in model 5. Concentration of socioeconomic disadvantage within a school increases the odds of obesity and concentration of socioeconomic advantage decreases the odds in both models 4 and 5, although the effect is slightly smaller once county-level variables are controlled.

Discussion and Conclusion

Summary of Main Findings

Our first research question was: Do student-level racial differences persist when school-level racial composition and school-level socioeconomic disadvantage are controlled? Racial differences in the likelihood of obesity persisted even when the racial composition and socioeconomic disadvantage at the school level were controlled. However, the differences between white and nonwhite students decreased slightly once school-level measures were added. This is similar to prior research that found social context explained some racial differences in obesity (Rossen and Talih 2014). Consistent with other work (Bernell et al. 2009), students attending schools with higher proportions of white students had a lower likelihood of obesity compared with those with lower levels of white students. However, high-minority schools in the North Carolina data were also disproportionately likely to be composed of large numbers of low-income students, and once school-level socioeconomic disadvantage was controlled, the effect of the school-level racial composition on the odds of obesity was no longer statistically significant. Therefore, although racial composition at the school level was associated with increases in the likelihood of obesity, net of a student’s race, the relationship appears to be due to socioeconomic disadvantage at the school-level rather than school-level racial composition in and of itself. Economic advantage at the school is a protective factor against obesity, net of the other variables in our multilevel models.

The findings for the school-level variables are consistent with studies showing that students in minority-segregated schools have poorer academic outcomes compared with similar students in white segregated or schools with no racial segregation (Lleras 2008; Roscigno and Ainsworth-Darnell 1999). Additional research studies show that minority students have less access to healthy foods compared with white students (Delva et al. 2007) and higher rates of obesity in minority segregated or economically disadvantaged school contexts (Bernell et al. 2009; O’Malley et al. 2007). Although economic disadvantage and racial segregation are highly correlated in the United States (Acevedo-Garcia and Osypuk 2008), other studies find that economic disadvantage is driving racial disparities (Rossen 2014) and that the concentration of poverty within a school is associated with adolescent overweight (Martin et al. 2012).

Our second research question was: Do student-level racial differences in the likelihood of obesity vary on the basis of the racial composition of the school? We find that student-level racial disparities in obesity did vary on the basis of the racial composition of the school. This is important because it indicates that there is not simply an additive effect of racial composition at the school level but that the impact of the school composition varies on the basis of the student’s race. We find that disparities between white and black students increased as the percentage of white students at the school increased. Students who reported their race as other also had an increased risk for obesity in schools with higher percentages of white students. There was no statistically significant difference for the interaction of Hispanic students and school-level racial composition. This builds on prior work showing that the racial composition of the school (Rossen 2014) and a student’s race are key predictors of obesity (Ogden et al. 2015) and extends it by examining the interaction between these levels for some students. Black students could be particularly vulnerable to white hostility in certain school contexts, but further investigation is required to better understand this finding.

With respect to our third research question (Do county-level measures of socioeconomic inequality, racial segregation and rurality explain school-level and/or student-level racial differences in obesity rates?), we did not find that county-level variables accounted for student or school characteristics. Student-level race, the interaction effects between student race and school racial composition, and school socioeconomic disadvantage all remained significant once county-level variables were controlled. Unlike a study using national data, we did not find that the Gini index for county-level inequality was associated with risk for obesity (Fan et al. 2016). The North Carolina data showed that students in rural counties did have an increased likelihood of obesity compared with their counterparts in urban counties, net all other variables in the multilevel models. This finding is consistent with other research on rural places that finds that living in a rural area is a risk factor for childhood obesity (Davis et al. 2011; Johnson and Johnson 2015; Liu et al. 2008, 2012; Lutfiyya et al. 2007)

Discussion

The multilevel models illustrate the importance of investigating school-level and community-level contexts in order to understand racial disparities in childhood obesity. The impact of a student’s race on the likelihood of childhood obesity varies on the basis of the racial composition of the school. Although the statistical significance of school-level racial composition disappears once school-level SES is controlled, the interactions of race (black and other categories) and school-level racial composition are statistically significant and remain significant even when county-level racial segregation and socioeconomic inequality variables are controlled. These results demonstrate the utility of multilevel approaches that incorporate individual, school-level, and community-level variables and the interactions between them.

The results from the North Carolina data fit within a growing body of research that examines the role of place and social context in racial health disparities, particularly relating to the issue of childhood obesity. Racial segregation cannot be examined or understood without accounting for socioeconomic disadvantage. Second, this shows that the effects of individual-level racial status must be understood within the racial context of students’ schools, neighborhoods, and communities. To our knowledge, this study is the first to examine the interaction between a student’s race and racial context and how this affects health disparities. The differences between white and black students in particular need further investigation to better understand what mechanisms lead to greater disparities in majority-white schools, but this is an important first step. The different results for the interaction of student-level race and school racial composition for black students and for Hispanic students highlights the way race/ethnicity may work differently for these groups in these contexts with higher proportions of white students. This contributes to the growing body of research and analysis on the impacts of school racial and socioeconomic composition on student behavior, mental health, and status attainment (Bernell et al. 2009; Story et al. 1995; Walsemann and Bell 2010). Health disparities should also be understood not as an individual issue but as part of an examination of children within their schools and communities.

Previous research has shown that residential segregation by race is associated with poor health outcomes among adults, including an increased risk for obesity among women (Chang 2006; Williams and Collins 2001; Williams and Sternthal 2010). In this study, we did not find any statistically significant effects for county-level residential segregation once school-level characteristics were controlled. Because school-level segregation is caused in part by residential segregation patterns, it could be that county-level racial segregation was accounted for at the school level. However, this could be a result of the limitations of the data that required us to use county-level segregation and not a neighborhood- or tract-level segregation measure. But this research shows that disparities persist at the school and student levels even when community factors are controlled, suggesting that schools are an important site to exacerbate or minimize health disparities.

Although our study provides important new information on the effects of race and place on childhood obesity rates, we acknowledge several limitations. As we noted previously, the data used were not a random sample of students in North Carolina or the United States. Although the students closely matched the North Carolina student population on key demographic characteristics, a true stratified random sample is needed for these findings to be generalizable to students in the state or the United States at large. We also lack data on students’ place of birth and time in the United States and therefore cannot investigate how acculturation may affect BMI. Second, we were unable to control for pubertal timing, which can influence obesity and varies by race and sex (Kaplowitz et al. 2001; Lee et al. 2007). Our measure of student BMI is based on self-report in some cases and thus may underestimate actual BMI. A final limitation of our data is the lack of measures of income and education at the parental or household level.

To better understand the causal link between social context and childhood obesity, a longitudinal or ecological momentary assessment (EMA) analysis would be an appropriate next step. Other studies have found that racism and discrimination have detrimental effects on health (Saffron and Nazroo 2004), and those could explain why the racial disparities increase as the percentage of white students increases. Identifying the mechanisms by which a student’s race and school-level racial composition are associated with a higher likelihood of obesity is beyond our data and requires in depth examination of the daily interactions, micro-aggressions, and experiences of students either through ethnographic examination or through EMA studies. An EMA design could ask students to complete short surveys throughout the day gathering everyday experiences of discrimination, diet, exercise, and other health behaviors. This kind of examination of the daily lives of adolescents would offer a better insight into immediate impacts of micro-aggressions on mood and health and how they may be affected by the social context. Prior EMA studies have even been able to examine how genetics may interact with both experiences and behaviors in the daily lives of adolescents (Russell, Wang, and Odgers 2015).

Conclusion

Our study contributes to the continuing work being done on racial disparities and child obesity. It extends beyond individual racial/ethnic identity and incorporates measures of structural inequalities of the school and community and county contextual environments and how they interact. The history of race in America involves segregation, discrimination, and systematic exclusion of minority groups from accessing the same spaces and resources as whites. Racial status and its impact on health must be understood within this historical context and the continuing barriers faced by students of color. As efforts are made to dismantle the structures of racial exclusion, special attention must be paid to the impacts of racism on minority students. By understanding the effects of social context and place on health, we can better address racial disparities among children and provide them the safety and protection required for health and wellbeing, including a decrease in obesity.

Racial disparities in health are complex and involve structural inequalities at multiple levels of analysis. Public policies to deal with obesity must also be multilevel and address individual behavior (e.g., eating healthy or unhealthy foods, screen time, racist bullying), school-level factors (race and socioeconomic composition, resources and racism), and locality factors (food environment, access to recreation, and social cohesion). School-based programs directed at student diet, nutrition, and exercise can be effective interventions in preventing childhood obesity, but evaluations often do not report results by population subgroup (Wang et al. 2015) or by the racial composition of the school. Advocates for student health should focus their efforts on targeted interventions directed at the ways that race and racism affect students and actively seek to minimize these impacts through curriculum and training and changes in the school and neighborhood environments.

Footnotes

Acknowledgements

Dr. Schulman’s work is supported by the U.S. Department of Agriculture National Institute for Food and Agriculture Hatch Project (1007488), by the College of Agriculture and Life Sciences at North Carolina State University, and by the North Carolina Agricultural Research Service.

Notes

Author Biographies

Joy Rayanne Piontak is a sociologist and health researcher who studies mental and physical health disparities. Her past work has examined the role of socioeconomic segregation, exposure to violence, and rurality on health and health behaviors among children and adolescents as well as the role of multigenerational households in maternal depression. Currently, she is working on several projects aimed at better understanding obesity, food insecurity and access, and other characteristics of space in rural and urban environments.

Michael D. Schulman is a rural sociologist whose work details the social psychological and behavioral adaptations made by individuals and families to changes in economic conditions, the organization of work, and the labor process. He has completed research on stress and adaptation to economic hardship among rural families, on changes in the structure of agriculture, and on textile worker responses to technological change and firm consolidation. He has examined the impact of economic restructuring on occupational mortality rates and the hazards and injuries experienced by young workers. Currently, he is collaborating on a series of research projects on obesity, food access and availability, and the social and economic characteristics of local spatial environments. He is a past editor of the journal Rural Sociology and a past president of the Rural Sociological Society.

References

Acevedo-Garcia

Dolores

Osypuk

Theresa L.

2008. “Invited Commentary: Residential Segregation and Health—The Complexity of Modeling Separate Social Contexts.” American Journal of Epidemiology 168(11):1255–58.

Bee

Adam

. 2012. “Household Income Inequality within U.S. Counties: 2006–2010.” American Community Survey Briefs ACSBR/10-18. Washington, DC: U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau.

Bennett

Kevin J.

Probst

Janice C.

Pumkam

Chaiporn

. 2011. “Obesity among Working Age Adults: The Role of County-level Persistent Poverty in Rural Disparities.” Health & Place 17(5):1174–81.

Bernell

S. L.

Mijanovich

Weitzman

B. C.

2009. “Does the Racial Composition of the School Environment Influence Children’s Body Mass Index?” Journal of Adolescent Health 45(1):40–46.

Boynton-Jarrett

Renée

Rosenberg

Lynn

Palmer

Julie R.

Boggs

Deborah A.

Wise

Lauren A.

2012. “Child and Adolescent Abuse in Relation to Obesity in Adulthood: The Black Women’s Health Study.” Pediatrics 130(2):245–53.

Brewer

Mackenzie

Kimbro

Rachel Tolbert

. 2014. “Neighborhood Context and Immigrant Children’s Physical Activity.” Social Science & Medicine 116:1–9.

Carroll-Scott

Amy

Gilstad-Hayden

Kathryn

Rosenthal

Lisa

Peters

Susan M.

McCaslin

Catherine

Joyce

Rebecca

Ickovics

Jeannette R.

2013. “Disentangling Neighborhood Contextual Associations with Child Body Mass Index, Diet, and Physical Activity: The Role of Built, Socioeconomic, and Social Environments.” Social Science & Medicine 95:106–14.

Centers for Disease Control and Prevention. 1999. “Neighborhood Safety and the Prevalence of Physical Inactivity—Selected States, 1996.” Morbidity and Mortality Weekly Reports 48(7):143–46.

Cetateanu

Andreea

Jones

Andy

. 2014. “Understanding the Relationship between Food Environments, Deprivation and Childhood Overweight and Obesity: Evidence from a Cross Sectional England-wide Study.” Health & Place 27(1):68–76.

10.

Chang

Virginia W.

2006. “Racial Residential Segregation and Weight Status among US Adults.” Social Science & Medicine 63(5):1289–1303.

11.

Chang

Virginia W.

Hillier

Amy E.

Mehta

Neil K.

2009. “Neighborhood Racial Isolation, Disorder and Obesity.” Social Forces 87(4):2063–92.

12.

Chung

Chanjin

Myers

Samuel L.

1999. “Do the Poor Pay More for Food? An Analysis of Grocery Store Availability and Food Price Disparities.” Journal of Consumer Affairs 33(2):276–96.

13.

Condron

Dennis J.

2009. “Social Class, School and Non-school Environments, and Black/White Inequalities in Children’s Learning.” American Sociological Review 74(5):683–708.

14.

Cornette

Robert

. 2008. “The Emotional Impact of Obesity on Children.” Worldviews on Evidence-based Nursing 5(3):136–41.

15.

Davis

Ann McGrath

Bennett

Kevin J.

Befort

Christie

Nollen

Nikki

. 2011. “Obesity and Related Health Behaviors among Urban and Rural Children in the United States: Data from the National Health and Nutrition Examination Survey 2003–2004 and 2005–2006.” Journal of Pediatric Psychology 36(6):669–76.

16.

Delva

Jorge

O’Malley

Patrick M.

Johnston

Lloyd D.

2007. “Availability of More-healthy and Less-healthy Food Choices in American Schools: A National Study of Grade, Racial/Ethnic, and Socioeconomic Differences.” American Journal of Preventive Medicine 33(4 Suppl. 1):S226–39.

17.

Elgar

Frank J.

Stewart

Jennifer M.

2008. “Validity of Self-report Screening for Overweight and Obesity: Evidence from the Canadian Community Health Survey.” Canadian Journal of Public Health 99(5):423–27.

18.

Evans

Gary W.

Jones-Rounds

McKenzie L.

Belojevic

Goran

Vermeylen

Francoise

. 2012. “Family Income and Childhood Obesity in Eight European Cities: The Mediating Roles of Neighborhood Characteristics and Physical Activity.” Social Science & Medicine 75(3):477–81.

19.

Fan

Jessie X.

Wen

Ming

Kowaleski-Jones

Lori

. 2016. “Tract- and County-level Income Inequality and Individual Risk of Obesity in the United States.” Social Science Research 55(1):75–82.

20.

Field

Alison E.

Aneja

Parul

Rosner

Bernard

. 2007. “The Validity of Self-reported Weight Change among Adolescents and Young Adults.” Obesity 15(9):2357–64.

21.

Flegal

Katherine M.

Ogden

Cynthia L.

Yanovski

Jack A.

Freedman

David S.

Shepherd

John A.

Graubard

Barry I.

Borrud

Lori G.

2010. “High Adiposity and High Body Mass Index-for-age in US Children and Adolescents Overall and by Race-ethnic Group.” American Journal of Clinical Nutrition 91(4):1020–26.

22.

Freedman

David S.

Khan

Laura Kettel

Serdula

Mary K.

Ogden

Cynthia L.

Dietz

William H.

2006. “Racial and Ethnic Differences in Secular Trends for Childhood BMI, Weight, and Height.” Obesity 14(2):301–308.

23.

Freedman

David S.

Mei

Zuguo

Srinivasan

Sathanur R.

Berenson

Gerald S.

Dietz

William H.

2007. “Cardiovascular Risk Factors and Excess Adiposity among Overweight Children and Adolescents: The Bogalusa Heart Study.” Journal of Pediatrics 150(1):12–17.

24.

Frey

William H.

2011. “New Racial Segregation Measures for Large Metropolitan Areas: Analysis of the 1990–2010 Decennial Censuses.” Ann Arbor: Population Studies Center, Institute for Social Research, University of Michigan; 2011. Retrieved July 22, 2017 (http://www.psc.isr.umich.edu/dis/census/segregation2010.html).

25.

Friedemann

Claire

Heneghan

Carl

Mahtani

Kamal

Thompson

Matthew

Perera

Rafael

Ward

Alison M

. 2012. Cardiovascular Disease Risk in Healthy Children and Its Association with Body Mass Index: Systematic Review and Meta-Analysis.” BMJ 345:e4759.

26.

Grow

H. Mollie Greves

Cook

Andrea J.

Arterburn

David E.

Saelens

Brian E.

Drewnowski

Adam

Lozano

Paula

. 2010. “Child Obesity Associated with Social Disadvantage of Children’s Neighborhoods.” Social Science & Medicine 71(3):584–91.

27.

Guo

Guang

Zhao

Hongxin

. 2000. “Multilevel Modeling for Binary Data.” Annual Review of Sociology 26:441–62.

28.

Himes

John H.

Hannan

Peter

Wall

Melanie

Neumark-Sztainer

Dianne

. 2005. “Factors Associated with Errors in Self-reports of Stature, Weight, and Body Mass Index in Minnesota Adolescents.” Annals of Epidemiology 15(4):272–78.

29.

Johnson

James Allen

Johnson

Asal Mohamadi

. 2015. “Urban-rural Differences in Childhood and Adolescent Obesity in the United States: A Systematic Review and Meta-analysis.” Childhood Obesity 11(3):233–41.

30.

Kainz

Kirsten

Pan

. 2014. “Segregated School Effects on First Grade Reading Gains: Using Propensity Score Matching to Disentangle Effects for African-American, Latino, and European-American Students.” Early Childhood Research Quarterly 29(4):531–37.

31.

Kaplowitz

Paul B.

Slora

Eric J.

Wasserman

Richard C.

Pedlow

Steven E.

Herman-Giddens

Marcia E.

2001. “Earlier Onset of Puberty in Girls: Relation to Increased Body Mass Index and Race.” Pediatrics 108(2):347–53.

32.

Kelly

Yvonne

Becares

Laia

Nazroo

James

. 2013. “Associations between Maternal Experiences of Racism and Early Child Health and Development: Findings from the UK Millennium Cohort Study.” Journal of Epidemiology and Community Health 67(1):35–41.

33.

Kershaw

Kiarri N.

Albrecht

Sandra S.

2014. “Metropolitan-level Ethnic Residential Segregation, Racial Identity, and Body Mass Index among US Hispanic Adults: A Multilevel Cross-sectional Study.” BMC Public Health 14:283.

34.

Kimbro

Rachel Tolbert

Denney

Justin T.

2013. “Neighborhood Context and Racial/Ethnic Differences in Young Children’s Obesity: Structural Barriers to Interventions.” Social Science & Medicine 95:97–105.

35.

Kozol

Jonathan

. 2005. The Shame of the Nation: The Restoration of Apartheid Schooling in America. New York: Crown.

36.

Lee

Hedwig

Harris

Kathleen M.

Lee

Joyce

. 2013. “Multiple Levels of Social Disadvantage and Links to Obesity in Adolescence and Young Adulthood.” Journal of School Health 83(3):139–49.

37.

Lee

Helen

. 2012. “The Role of Local Food Availability in Explaining Obesity Risk among Young School-aged Children.” Social Science & Medicine 74(8): 1193–1203.

38.

Lee

Joyce M.

Appugliese

Danielle

Kaciroti

Niko

Corwyn

Robert F.

Bradley

Robert H.

Lumeng

Julie C.

2007. “Weight Status in Young Girls and the Onset of Puberty.” Pediatrics 119(3):e624–30.

39.

Bayoue

Lingsma

Hester

Steyerberg

Ewout

Lesaffre

Emmanuel

. 2011. “Logistic Random Effects Regression Models: A Comparison of Statistical Packages for Binary and Ordinal Outcomes.” BMC Medical Research Methodology 11:77.

40.

Liu

Jihong

Bennett

Kevin J.

Harun

Nusrat

Probst

Janice C.

2008. “Urban-rural Differences in Overweight Status and Physical Inactivity among US Children Aged 10–17 Years.” Journal of Rural Health 24(4):407–15.

41.

Liu

Ji-Hong

Jones

Sonya J.

Sun

Han

Probst

Janice C.

Merchant

Anwar T.

Cavicchia

Philip

. 2012. “Diet, Physical Activity, and Sedentary Behaviors as Risk Factors for Childhood Obesity: An Urban and Rural Comparison.” Childhood Obesity 8(5):440–48.

42.

Lleras

Christy

. 2008. “Race, Racial Concentration, and the Dynamics of Educational Inequality across Urban and Suburban Schools.” American Educational Research Journal 45(4):886–912.

43.

Lobao

Linda

. 2004. “Continuity and Change in Place Stratification: Spatial Inequality and Middle-range Territorial Units.” Rural Sociology 69(1):1–30.

44.

Lovasi

Gina S.

Hutson

Malo A.

Guerra

Monica

Neckerman

Kathryn M.

2009. “Built Environments and Obesity in Disadvantaged Populations.” Epidemiologic Reviews 31(1):7–20.

45.

Lutfiyya

May Nawal

Lipsky

Martin S.

Wisdom-Behounek

Jennifer

Inpanbutr-Martinkus

Melissa

. 2007. “Is Rural Residency a Risk Factor for Overweight and Obesity for U.S. Children?” Obesity 15(9):2348–56.

46.

Martin

Molly A.

Frisco

Michelle L.

Nau

Claudia

Burnett

Kristin

. 2012. “Social Stratification and Adolescent Overweight in the United States: How Income and Educational Resources Matter across Families and Schools.” Social Science & Medicine 74(4):597–606.

47.

Massey

Douglas S.

Denton

Nancy A.

1988. “The Dimensions of Residential Segregation.” Social Forces 67(2):281–315.

48.

Massey

Douglas S.

Denton

Nancy A.

1993. American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press.

49.

Mayer

Susan E.

2002. “How Economic Segregation Affects Children’s Educational Attainment.” Social Forces 81(1):153–76.

50.

McCormack

Lacey Arneson

Meendering

Jessica

. 2016. “Diet and Physical Activity in Rural vs Urban Children and Adolescents in the United States: A Narrative Review.” Journal of the Academy of Nutrition and Dietetics 116(3):467–80.

51.

Miyazaki

Yasuo

Stack

Maria

. 2015. “Examining Individual and School Characteristics Associated with Child Obesity Using a Multilevel Growth Model.” Social Science & Medicine 128(1):57–66.

52.

Morrissey

Susan L.

Whetstone

Lauren M.

Cummings

Doyle M.

Owen

Lynda J.

2006. “Comparison of Self-reported and Measured Height and Weight in Eighth-grade Students.” Journal of School Health 76(10):512–15.

53.

North Carolina Department of Public Instruction. 2010a. “Free and Reduced Application Data by Site, All Year-to-date Data 2009–10.” Raleigh: Child Nutrition Services, North Carolina Department of Public Instruction. Retrieved February 22, 2011 (http://www.ncpublicschools.org/docs/fbs/resources/data/freereduced/2010–11freereduced.xls).

54.

North Carolina Department of Public Instruction. 2010b. “Grade, Race, and Sex Data (by LEA Grade/by School Grade)–Collected from Each School at the End of the First School Month.” Retrieved December 11, 2011 (http://www.ncpublicschools.org/fbs/accounting/data/grs10mod.xls).

55.

North Carolina Rural Economic Development Center. 2012. “Rural Data Bank.” Retrieved June 5, 2012 (http://www.ncruralcenter.org).

56.

Ogden

Cynthia L.

Carroll

Margaret D.

Curtin

Lester R.

Lamb

Molly M.

Flegal

Katherine M.

2010. “Prevalence of High Body Mass Index in US Children and Adolescents, 2007–2008.” JAMA 303(3):242–49.

57.

Ogden

Cynthia L.

Carroll

Margaret D.

Fryar

Cheryl D.

Flegal

Katherine M.

2015. “Prevalence of Obesity among Adults and Youth: United States, 2011–2014.” Hyattsville, MD: National Center for Health Statistics.

58.

Ogden

Cynthia L.

Carroll

Margaret D.

Kit

Brian K.

Flegal

Katherine M.

2014. “Prevalence of Childhood and Adult Obesity in the United States, 2011–2012.” JAMA 311(8):806–14.

59.

O’Malley

Patrick M.

Johnston

Lloyd D.

Delva

Jorge

Bachman

Jerald G.

Schulenberg

John E.

2007. “Variation in Obesity among American Secondary School Students by School and School Characteristics.” American Journal of Preventive Medicine 33(4 Suppl. 1):S187–94.

60.

Pachter

Lee M.

García Coll

Cynthia

. 2009. “Racism and Child Health: A Review of the Literature and Future Directions.” Journal of Developmental and Behavioral Pediatrics 30(3):255–63.

61.

Paradies

Yin

Ben

Jehonathan

Denson

Nida

Elias

Amanuel

Priest

Naomi

Pieterse

Alex

Gupta

Arpana

Kelaher

Margaret

Gee

Gilbert

. 2015. “Racism as a Determinant of Health: A Systematic Review and Meta-analysis.” PLoS One 10(9):e0138511.

62.

Porter

Jeremy R.

Bratter

Jennifer L.

2015. “Investigating Race, Class and Context through Historical Evidence: Segregation and the Ecology of Aspirations of Black College Students in the 1960s South.” Sociological Spectrum 35(3):286–308.

63.

Priest

Naomi

Paradies

Yin

Trenerry

Brigid

Truong

Mandy

Karlsen

Saffron

Kelly

Yvonne

. 2013. “A Systematic Review of Studies Examining the Relationship between Reported Racism and Health and Wellbeing for Children and Young People.” Social Science & Medicine 95:115–27.

64.

Priest

Naomi

Perry

Ryan

Ferdinand

Angeline

Paradies

Yin

Kelaher

Margaret

. 2014. “Experiences of Racism, Racial/Ethnic Attitudes, Motivated Fairness and Mental Health Outcomes among Primary and Secondary School Students.” Journal of Youth and Adolescence 43(10):1672–87.

65.

Robert

Stephanie A.

1999. “Socioeconomic Position and Health: The Independent Contribution of Community Socioeconomic Context.” Annual Review of Sociology 25:489–516.

66.

Roscigno

Vincent J.

1999. “The Black-white Achievement Gap, Family-school Links, and the Importance of Place.” Sociological Inquiry 69(2): 159–86.

67.

Roscigno

Vincent J.

Ainsworth-Darnell

James W.

1999. “Race, Cultural Capital, and Educational Resources: Persistent Inequalities and Achievement Returns.” Sociology of Education 72(3):158–78.

68.

Rossen

Lauren M.

2014. “Neighbourhood Economic Deprivation Explains Racial/Ethnic Disparities in Overweight and Obesity among Children and Adolescents in the USA.” Journal of Epidemiology and Community Health 68(2):123–29.

69.

Rossen

Lauren M.

Talih

Makram

. 2014. “Social Determinants of Disparities in Weight among US Children and Adolescents.” Annals of Epidemiology 24(10):705–13.

70.

Ruel

Erin

Reither

Eric N.

Robert

Stephanie A.

Lantz

Paula M.

2010. “Neighborhood Effects on BMI Trends: Examining BMI Trajectories for Black and White Women.” Health & Place 16(2):191–98.

71.

Rundle

Andrew

Richards

Catherine

Bader

Michael D M.

Schwartz-Soicher

Ofira

Lee

Karen K.

Quinn

James

Lovasi

Gina S.

Weiss

Christopher

Neckerman

Kathryn

. 2012. “Individual- and School-level Sociodemographic Predictors of Obesity among New York City Public School Children.” American Journal of Epidemiology 176(11):986–94.

72.

Russell

Michael A.

Wang

Lin

Odgers

Candice L.

2015. “Witnessing Substance Use Increases Same-day Antisocial Behavior among At-risk Adolescents: Gene-environment Interaction in a 30-day Ecological Momentary Assessment Study.” Development and Psychopathology 28(4 Pt. 2):1441–56.

73.

Sacker

Amanda

Wiggins

Richard D.

Bartley

Mel

. 2006. “Time and Place: Putting Individual Health into Context. A Multilevel Analysis of the British Household Panel Survey, 1991–2001.” Health & Place 12(3):279–90.

74.

Saffron

Karlsen

Nazroo

James Yzet

. 2004. “Fear of Racism and Health.” Journal of Epidemiology and Community Health 58(12):1017–18.

75.

Scott

Alison J.

Wilson

Rebecca F.

2011. “Upstream Ecological Risks for Overweight and Obesity among African American Youth in a Rural Town in the Deep South, 2007.” Preventing Chronic Disease 8(1):A17.

76.

Singh

Gopal K.

Kogan

Michael D.

van Dyck

Peter C.

2008. “A Multilevel Analysis of State and Regional Disparities in Childhood and Adolescent Obesity in the United States.” Journal of Community Health 33(2):90–102.

77.

Snijders

Tom A. B.

Bosker

Roel J.

1999. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. Thousand Oaks, CA: Sage.

78.

Sparks

P. Johnelle

Sparks

Corey S.

Campbell

Joseph J. A.

2013. “Poverty Segregation in Nonmetro Counties: A Spatial Exploration of Segregation Patterns in the US.” Spatial Demography 1(2):162–77.

79.

Story

Mary

French

Simone A.

Resnick

Michael D.

Blum

Robert W.

1995. “Ethnic/Racial and Socioeconomic Differences in Dieting Behaviors and Body Image Perceptions in Adolescents.” International Journal of Eating Disorders 18(2):173–79.

80.

Taveras

Elsie M.

Gillman

Matthew W.

Kleinman

Ken P.

Rich-Edwards

Janet W.

Rifas-Shiman

Sheryl L.

2013. “Reducing Racial/Ethnic Disparities in Childhood Obesity: The Role of Early Life Risk Factors.” JAMA Pediatrics 167(8):731–38.

81.

Tomaskovic-Devey

Donald

Roscigno

Vincent J.

1997. “Uneven Development and Local Inequality in the US South: The Role of Outside Investment, Landed Elites, and Racial Dynamics.” Sociological Forum 12(4):565–97.

82.

U.S. Census Bureau. 2010. “Factfinder 2010 ACS 5-year Estimates.” Retrieved July 22, 2017 (http://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml).

83.

Walsemann

Katrina M.

Bell

Bethany A.

2010. “Integrated Schools, Segregated Curriculum: Effects of Within-school Segregation on Adolescent Health Behaviors and Educational Aspirations.” American Journal of Public Health 100(9):1687–95.

84.

Walsemann

Katrina M.

Bell

Bethany A.

Maitra

Debeshi

. 2011. “The Intersection of School Racial Composition and Student Race/Ethnicity on Adolescent Depressive and Somatic Symptoms.” Social Science & Medicine 72(11):1873–83.

85.

Walsemann

Katrina M.

Gee

Gilbert C.

Annie

. 2013. “Educational Attainment in the Context of Social Inequality: New Directions for Research on Education and Health.” American Behavioral Scientist 57(8):1082–1104.

86.

Wang

Claire Y.

Gortmaker

Steven L.

Taveras

Elsie M.

2011. “Trends and Racial/Ethnic Disparities in Severe Obesity among US Children and Adolescents, 1976–2006.” International Journal of Pediatric Obesity 6(1):12–20.

87.

Wang

Cai

Wilson

R. F.

Weston

Fawole

Bleich

S. N.

Cheskin

L. J.

Showell

N. N.

Lau

B. D.

Chiu

D. T.

Zhang

Segal

2015. “What Childhood Obesity Prevention Programmes Work? A Systematic Review and Meta-analysis.” Obesity Reviews 16(7):547–65.

88.

Williams

David R.

Collins

Chiquita

. 1995. “US Socioeconomic and Racial Differences in Health: Patterns and Explanations.” Annual Review of Sociology 21:349–86.

89.

Williams

David R.

Collins

Chiquita

. 2001. “Racial Residential Segregation: A Fundamental Cause of Racial Disparities in Health.” Public Health Reports 116(5):404–16.

90.

Williams

David R.

Sternthal

Michelle

. 2010. “Understanding Racial-ethnic Disparities in Health: Sociological Contributions.” Journal of Health and Social Behavior 51(Suppl.):S15–27.

91.

Wimberley

Ronald C.

Morris

Libby V.

2003. “U.S. Poverty in Space and Time: Its Persistence in the South.” Sociation Today 1(2). Retrieved July 22, 2017 (http://www.ncsociology.org/sociationtoday/v2/wimmor.htm).