School influence
Variation in Obesity Among American Secondary School Students by School and School Characteristics

https://doi.org/10.1016/j.amepre.2007.07.001Get rights and content

Background

Body mass index (BMI) is known to vary by individual characteristics, but little is known about whether BMI varies by school and by school characteristics.

Methods

Nationally representative samples of United States schools and students are used to determine the extent to which BMI and percent of students at or above the 85th percentile of BMI vary by school and by school characteristics. Data from the 1991–2004 Monitoring the Future (MTF) study were analyzed in 2006 and 2007.

Results

A relatively small proportion of variance in BMI lies between schools; intraclass correlations are on the order of 3%. Still, this is sufficient variation to provide very different environments for students attending schools that are low versus high in average BMI. There is some modest variation by school type (public, Catholic private, non-Catholic private); school size (number of students in the sampled grade); region of the country; and population density. There is more variation as a function of school socioeconomic status (SES) and racial/ethnic composition of the school. School SES in particular was negatively associated with BMI levels, even after controlling individual-level SES and racial/ethnic status.

Conclusions

The residual differences in BMI by school suggest that some characteristic of the school and/or community environment—perhaps cultural factors or peer role modeling or differences in school food, beverage, or physical education policies—facilitate obesity in schools with a high concentration of lower socioeconomic students, beyond individual-level factors.

Introduction

The distribution of obesity among American adolescents is known to vary by important individual factors, including gender and race/ethnicity.1, 2, 3, 4 Little is known, however, about the extent to which obesity varies by school factors, and this represents an important gap for scientific and policy-related purposes. This article focuses on a description of: (1) the extent to which student obesity (measured by body mass index [BMI]) and the percentage of students who are at or above the 85th percentile (that is, overweight or at risk of overweight) vary among American secondary schools, and (2) how BMI and percentage of students at or above the 85th percentile vary by certain key characteristics of the schools. That is, this article describes the extent to which these problems cluster by school and by particular characteristics of the school, thereby providing indications of the potential importance of contextual factors in the school and community.

This study focuses on broad-based school characteristics, including school type (public, Catholic private, non-Catholic private); school size (measured by number of students in the sampled grade); school socioeconomic status (SES, as indicated by average parental education reported by students); and racial/ethnic composition (derived from student self-identification). Two other contextual characteristics that vary between schools (but not within schools) are also considered—the region of the country and the population density of the community in which they are located.

The extent to which obesity varies by school is an important issue because it sets outer limits to how much school-level factors could “explain” variations in individual-level obesity at the point in time at which measurement occurs. The degree of variation among schools could change over time to the extent that independent and/or dependent characteristics such as school policies about cafeteria offerings, vending machines, or required physical education become more or less homogeneous. The extent to which obesity varies by school characteristics is of interest primarily in a descriptive sense. Knowing whether obesity clusters by certain school characteristics can serve to focus future attention and resources on understanding the mechanisms by which these characteristics contribute to obesity in young people and to develop interventions that target these characteristics in order to prevent and reduce obesity.

Section snippets

Methods

Fourteen years of data (1991–2004) were examined from 8th-, 10th-, and 12th-grade students who participated in the University of Michigan’s Monitoring the Future (MTF) project, sponsored by the National Institute on Drug Abuse. Data analyses were conducted in 2006 and 2007.

Results

Table 1 shows mean BMI and the proportion at or above the 85th percentile for each grade for each year from 1991 to 2004. There is a clear general upward trend in both measures, as reported in more detail elsewhere.10 Table 2 provides the percentage of variance, also called the intraclass coefficient (ICC), that is between schools for BMI and for being at or above the 85th percentile, separately for grades 8, 10, and 12 from 1991 to 2004. Calculations were performed separately for each year,

Discussion

Although at present the great majority of the variation in BMI resides within schools, there remains enough variation between schools for school characteristics and school policies and programs to have had important effects on their students’ BMI. This is about equally true at all three grade levels included in this study. Although the ICC for BMI is only about 3%, it remains true that schools could have substantially more influence in the future. The figure of 3% reflects the maximum impact

Conclusion

Although a fair amount is known about how individual characteristics relate to BMI among adolescents, less is known about the extent to which BMI varies by school and by school characteristics. This study shows that although most variation in BMI lies within schools, there is sufficient between-school variation to be of interest to policymakers. School SES is shown to be of some importance, even after controlling for individual-level SES and race/ethnicity. In sum, the school one attends has

References (20)

There are more references available in the full text version of this article.

Cited by (70)

  • Cross-sector analysis of socioeconomic, racial/ethnic, and urban/rural disparities in food policy enactment in the United States

    2016, Health and Place
    Citation Excerpt :

    In several models, we found statistically significant interactions between median household income and other predictors (p<.05), and therefore all analyses of median household income were repeated after stratifying by predominant race/ethnicity and urbanicity. Predominant race/ethnicity was based on whether the community was at least 66.7% non-Hispanic White, 50% non-Hispanic Black, 50% Hispanic, or “other” (i.e., racially/ethnically mixed) if it met none of the first three criteria (O'Malley et al., 2007). In stratified analyses, communities were grouped into 6 categories – 1) urban and non-Hispanic White, 2) suburban and non-Hispanic White, 3) rural and non-Hispanic White, 4) non-Hispanic Black, 5) Hispanic, and 6) Other racial/ethnic composition.

View all citing articles on Scopus
View full text