Skip to main content
NCBI home page
HHS Author Manuscripts logoLink to HHS Author Manuscripts
. Author manuscript; available in PMC: 2021 May 3.
Published in final edited form as: Soc Sci Res. 2019 Apr 20;83:102301. doi: 10.1016/j.ssresearch.2019.04.014

Do income inequalities in higher weight status depend on social integration?

Anthony David Campbell a,*, Elizabeth H Baker b
PMCID: PMC8088976  NIHMSID: NIHMS1691029  PMID: 31422839

Abstract

Those with higher incomes tend to have better health outcomes, including healthy weight status. We use data from the 2003–2008 National Health and Nutrition Examination Survey (NHANES) to examine whether the association between higher weight status and social integration varies by income. We examine gender differences in weight status, measured by BMI and obesity, by social integration and income, and find evidence that high social integration is a risk factor for higher weight status among low-income men. The association between income and higher weight status operates differently for women and men and is dependent, in part, on their level of social integration. Income is negatively associated with weight status for men who are highly integrated, but is positively associated with weight status among men who have low integration. We conclude that higher numbers of close friends and family places low-income men at greater risk of higher weight status.

Keywords: Obesity, Body mass index (BMI), Income, Social relationships, Social integration, Social network size

1. Introduction

Higher weight status (overweight and obesity) is determined by a complex set of determinants, including social factors (McLaren, 2007). Two such social factors that influence higher weight status are income and social relationships. Prior research consistently demonstrates an inverse relationship between higher weight status and income; though this has only consistently been found for women (McLaren, 2007; Schafer and Ferraro, 2011; Wang and Beydoun, 2007). A proposed set of mechanisms linking income to health includes proximal factors such as diet and exercise, but also distal factors such as social relationships, which are composed of social networks, social support, and social integration (Thoits, 2011; Umberson and Montez, 2010). While numerous studies investigate direct relationships between income and higher weight status, no known studies focus on how this relationship may vary based on social networks. In this study, we investigate the relationship between of social network size, measured as number of close friends, and higher weight status, as well as whether this relationship varies by income for middle age U.S. adults.

The relevance of social relationships is a perennial focus with extensive research aimed at understanding how being socially connected works to influence health. Berkman and colleagues identify four broad mechanisms through which this health connection may operate: social support, social influence, social engagement and attachment, and access to resources (Berkman et al., 2000). These social mechanisms may impact weight status specifically, through the enactment of resources such as conveying health information, modeling healthy behaviors, social comparison, and communicating and modifying perceptions of healthy weight.

While few studies examine the specific interplay between social relationships and higher weight status, the research addressing this association finds mixed results depending on what aspect of social relationships is being measured. Deficits in social integration lead to higher weight status (Kouvonen et al., 2011), while positive social support protects against weight gain (Ball and Crawford, 2006). Although research specifically focused on social networks and weight status is sparse, there is a large body of research examining the relationship between income and high weight status. Social networks may play an important role in understanding this well-established relationship.

1.1. Income, social integration, and higher weight status

An inverse relationship between income and higher weight status is well documented, though this has only consistently been the case when examining women (McLaren, 2007; Wang and Beydoun, 2007; Sobal and Stunkard, 1989; Zhang and Wang, 2004). Using data from NHANES III (1988–1994), Zhang and Wang report considerable inequalities in obesity by family income, with variations by age, gender and race with significant inverse associations found among middle-aged women (Zhang and Wang, 2004). Chang and Lauderdale (2005), using data from NHANES between 1971 and 2002, find a consistent inverse association between poverty ratio and BMI for women and a weaker inverse association for men (age 18–64) (Chang and Lauderdale, 2005). In contrast, using the American’s Changing Lives Study, Ailshire and House (2011) find significant inverse relationships between income and higher weight status for both women and men age 25 to 64 (Ailshire and House, 2011).

Income can influence weight status through multiple mechanisms. Past research has examined lifestyle factors such as diet and exercise and has concluded that while such factors may mediate some of the relationship between higher weight status and income, a substantial proportion of the relationship remains even after controlling for these mechanisms. However, research examining multiple health outcomes finds that aspects of social relationships may represent a key factor for understanding the relationship between income and health. Social integration and network size are positively associated with income (Ichiro and Berkman, 2001; Thoits, 1995). Therefore, individuals from lower income groups may experience increased vulnerability due to lower levels of social integration (Thoits, 1995, 2011; Umberson and Montez, 2010; Berkman et al., 2000).

Much scholarship has focused on the potential intervening nature of social relationships in the association between income and health. This notion views social integration as a potential mechanism through which income impacts health (Ross and Mirowsky, 2006). However, a review of past literature reveals mixed support for this hypothesis (Uphoff et al., 2013). Rather research has found greater support for the notion that the association between social capital and health is stronger for the more disadvantaged groups, specifically low income (Gorman and Sivaganesan, 2007). Thus, social relationships are beneficial among low income individuals, but this association diminishes at higher income levels. These findings have been largely restricted to the health outcomes of self-rated health, mortality, and cardiovascular functioning. For these outcomes, social relationships may act as a substitutive resource for these health outcomes, buffering the negative impact of being low income. However, we argue that our focus on social integration and the health outcome measured weight status may result in dependency rather than substitution. The idea of dependency stems from Bourdieu’s notions on the interactions between social and economic capital. According to this theory social and economic capital are dependent on one another. Social groups are often homogeneous on many factors including their ability to access economic capital, thus the access to social capital and the quality of that capital is often determined by one’s access to economic capital. Additionally, one’s ability to effectively translate social capital into other forms of capital including, better health likely relies on access to economic capital (Bourdieu and Richardson, 1986). For example, one’s social capital may encourage health promoting norms, such as healthy diet and physical activity, but one would also need access to good grocery stores and recreation facilities to translate the social capital into better health. Thus, the idea of dependency predicts that greater social capital will have the greatest impact on health among those with the most economic resources.

Greater social integration may benefit those with higher incomes by providing resources, offering support, and modeling healthy behavior. However, social integration may also negatively impact health through stressful relationships, excessive demands and over-integration, and modeling negative health behaviors. The importance of social integration for modeling health behaviors is clearly seen in the example of adolescent smoking, where the best predictor of adolescent smoking is the number of her/his friends that smokes (Ali and Dwyer, 2009). The negative health behavior of smoking is socially transmitted through friendship networks by modeling negative behaviors and reinforcing group norms concerning health behaviors (i.e. whether smoking is seen as ‘cool’). Higher social integration would actually lead to worse health outcomes in this case. Research examining high weight status has found a similar association, showing that an individual who had a friend become obese had a 57% increase in the chance that they would also become obese (Christakis and Fowler, 2007). Additionally, greater integration may actually be associated with an increase in higher weight status, especially if those relationships are stressful, such as not being able to talk to, receive support from, or confide in close social relationships (Kouvonen et al., 2011). Low income individuals are more likely to have worse social integration and are more likely to be socially integrated into groups of other low income individuals. Health lifestyles theory builds off of Bourdieu’s theory of capital and suggests that certain social classes develop a collection of health behaviors that become indicative of that class (Bourdieu and Richardson, 1986; Cockerham, 2005). Increases in social capital or greater integration into these groups are beneficial when the health behaviors associated with that group are positive and harmful when they are negative. Thus, social integration is likely to amplify the associations found between class (e.g. income groups) and health. For example, researchers examining African-American mothers in Baltimore, found that greater social integration among mothers was associated with better child health when they were from high income neighborhoods and worse health when they resided in low income neighborhoods (Caughy et al., 2003). While greater integration may represent a benefit to low income individuals providing support when economic means fall short, they may also represent greater socialization into negative health habitus (Bourdieu, 1977).

2. Methods

2.1. Data source

Data were obtained from the National Health and Nutrition Examination Survey (NHANES), a large nationally representative sample of children and adults in the United States. NHANES is administered by the National Center for Health Statistics (NCHS) and includes a combination of in-home, computer-assisted personal interviews and standardized physical examinations that are conducted in standardized mobile laboratories. A major advantage of the dataset is its use of National Heart, Lung, and Blood Institute (NHLBI) guidelines for the objective measurement of height and weight. We combine three 2-year cycles of the continuous NHANES to create a 2003–2008 dataset per NCHS recommendations for higher statistical reliability (National Center for Health Statistics (NCHS), 2010). Although NHANES includes social support variables prior to 2003, they were only administered to respondents under the age of 60. Therefore, the analytic sample includes years 2003–2008 using data from respondents aged 40–65 years who were not pregnant and provided valid responses on all study variables, totaling 5073 respondents with roughly equal subsamples of women (n=2602) and men (n=2471).

2.2. Measures

Weight status is operationalized as body mass index (BMI) and obesity. BMI is determined using measured height and weight as assessed by trained interviewers and calculated as weight in kilograms divided by height in meters squared and then rounded to the nearest tenth (BMI=kg/m2). Obesity is defined as having a BMI greater than or equal to 30. Income is operationalized by family income to poverty ratio (PIR), which accounts for the number of and structure of the respondent’s family. Social Integration is measured by the reported number of close friends (relatives or non-relatives) a respondent feels at ease with, can talk to about private matters, and can call on for help. Potential responses range from zero to fifty with a mean of seven close friends. This variable is dichotomized as having seven or more friends (=1) or fewer than seven friends (=0). Controls: We control for several factors associated with obesity including age, age-squared, marital status, race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, and non-Hispanic other), education (less than high school, high school, some college, college graduate), smoking status (never, former, and current), and number of household members. Additionally, to control for the quality of one’s social network supports we account for whether the respondent indicated that they had adequate emotional support and adequate financial support.

2.3. Hypotheses

We present three sets of hypotheses for testing the expected associations of income (family income to poverty ratio - PIR), social integration (number of close friends), and higher weight status (BMI and obesity). First, we expect that income is inversely associated with higher weight status. Second, we anticipate that social integration will be inversely associated with higher weight status. Third, we expect that the association between social integration and weight status will vary by income, such that social integration is the most beneficial for those with higher incomes and may represent a risk factor for those of lower incomes. Considering established gender differences in higher weight status and gender differences in the association between income and higher weight status, we examine each of these three hypotheses separately for women and men.

2.4. Statistical analysis

Linear regression and logistic regression models were performed to assess the associations between income, social integration, and weight status, while adjusting for age, age squared, race/ethnicity, education, marital status, emotional support availability, financial support availability, smoking status and household size. Model 1 examines whether there is support for the first hypothesis (PIR is negatively associated with weight status) and second hypothesis (social integration is negatively associated with weight status). Model 2 examines a moderation hypothesis by introducing an interaction term of PIR category and number of close friends and tests its association with weight status (holding constant all variables from Model 1). The specification of income and social integration was examined several different ways, including examining continuous and categorical specifications with multiple categories and by examining whether quadratics resulted in better model fit. This produced substantively similar results and did not result in significantly better model fit. These analyses are available upon request. Results are presented separately for women and men, following literature suggesting that higher weight status varies by gender. However, in additional analyses we examined whether the association between the predictor variables and weight status varied by gender by including gender interaction in a gender pooled model. Statistically significant differences in the coefficient between women and men (p < .05) are represented by a dagger. Statistical analyses were conducted using Stata 14.0 (StataCorp, 2009). All analyses are weighted and standard errors are corrected to account for the complex sampling design of NHANES using the svyset command in Stata.

3. Results

Table 1 presents descriptive characteristics, including weighted means or percentages for all study variables, displayed by gender subsamples and the full sample. The mean BMI for the full sample is 29.3 with no significant differences by gender (women 29.4; men 29.3). Additionally, about 39% of women and 37% of men are obese.

Table 1.

Weighted means and percentages (%) for study variables by gender, national health and nutrition examination survey 2003–2008.

Men n = 2471 Women n = 2602 Total N = 5073
BMI 29.26 29.42 29.34
Obese 37.11 39.21 38.20
Family income to poverty ratio (PIR) 3.52 3.40 3.46
Number of close friends ≥ 7 32.30 32.60 32.50
Marital Status
 Not Married
 Married 73.46 65.06 69.11
Adequate emotional support 82.41 75.56 78.86
Available financial support 76.54 80.22 78.45
Highest educational attainment
 Less than high school 15.57 13.63 14.56
 High school 25.36 24.91 25.12
 Some college 29.97 33.77 31.94
 College or higher 29.10 27.70 28.38
Age 50.77 51.00 50.89
Race/Ethnicity
 Non-Hispanic White 76.84 73.86 75.30
 Non-Hispanic Black 10.16 11.61 10.91
 Hispanic 8.22 8.83 8.54
 Non-Hispanic Other 4.78 5.69 5.25
Number of household members 3.00 2.84 2.92
Smoking Status
 Never 42.64 54.67 48.87
 Former 30.70 24.38 27.43
 Current 26.66 20.95 23.70
Open in a new tab

Tables 2 and 3 present the ordinary least squares regression and logistic regression models predicting BMI and obesity for women and men, respectively. The control variables age, age-squared, race/ethnicity, education, marital status, emotional support, financial support, smoking status, and household size are included in all models. For women, the significant negative coefficient for PIR predicting BMI provides evidence for Hypothesis 1 that income is inversely associated with BMI. We find that a one unit increase in PIR is associated with a 0.29 reduction in BMI. However, among men, we do not find a significant relationship between PIR and weight status. The social integration measure, number of close friends, is not significantly associated with BMI or obesity for either gender, thus we do not find support for Hypothesis 2.

Table 2.

OLS (BMI) and logistic (obese) regression predicting weight status in women (age 40–65), national health and nutrition examination survey 2003–2008 (n = 2602).

BMI (b coefficients) Obesity (Odds Ratio)
Model 1 Model 2 Model 1 Model 2
Family income to poverty ratio (PIR) −0.29 * −0.23 *† 0.95 0.95 †
Number of close friends ≥ 7 0.34 1.039 0.99 1.12
 X PIR −0.2 0.97
Adequate emotional support −0.68 −0.69 0.93 0.93
Available financial support −0.79 −0.8 0.91 0.91
Married −0.88 * −0.88 *† 0.86 0.86
Highest educational attainment
 Less than high school 1.083 1.08 1.53 * 1.53 *
 High school 2.142 *** 2.13 *** 1.66 *** 1.66 **
 Some college 1.617 ** 1.605 ** 1.55 ** 1.54 **
 (College or higher) Age 0.003 0.00 1.00 1.00
Age-squared −0.01 * −0.01 *† 1 1.00 †
Race/Ethnicity
 (Non-Hispanic White)
 Non-Hispanic Black 2.819 *** 2.822 ***† 1.73 *** 1.73 ***†
 Hispanic 0.316 0.322 0.98 0.98
 Non-Hispanic Other −2.29 ** −2.32 ** 0.43 ** 0.43 **†
Number of household members −0.06 −0.06 1.05 1.05
Smoking Status
 (Never)
 Former 0.334 0.339 1.07 1.07
 Current −2.19 *** −2.19 *** 0.59 ** 0.59 **
Constant 31.4 *** 31.25 ***† 0.67 0.65 †
Open in a new tab
*

p < .05,

**

p < .01,

***

p < .001 (two-tailed tests).

†

significantly different from men, p < .05.

Table 3.

OLS (BMI) and logistic (obese) regression predicting weight status in men (age 40–65), national health and nutrition examination survey 2003–2008 (n = 2602).

BMI (b coefficients) Obesity (Odds Ratio)
Model 1 Model 2 Model 1 Model 2
Family income to poverty ratio (PIR) 0.152 0.277 *† 1.02 1.08 † †
Number of close friends ≥ 7 0.02 1.523 * 1.1 1.97 **
 X PIR −0.41 * 0.85 *
Adequate emotional support 0.204 0.176 1.05 1.02
Available financial support −0.24 −0.25 0.84 0.81
Married 0.555 0.533 † 1.16 1.13
Highest educational attainment
 Less than high school 1.359 * 1.303 * 1.56 1.53
 High school 1.665 *** 1.659 *** 1.75 ** 1.75 **
 Some college 1.801 *** 1.77 *** 1.97 *** 1.95 ***
 (College or higher)
Age 0.01 0.01 1.01 1.01
Age-squared 0.01 *† 0.01 * 1.00 *† 1.00 **
Race/Ethnicity
 (Non-Hispanic White)
 Non-Hispanic Black 0.043 † 0.071 1.03 † 1.04
 Hispanic −0.26 −0.22 0.82 0.83
 Non-Hispanic Other −0.57 −0.56 0.92 † 0.93
Number of household members −0.02 −0.01 1.03 1.04
Smoking Status
 (Never)
 Former −0.12 −0.11 0.97 0.98
 Current −1.8 *** −1.78 *** 0.63 * 0.63 *
Constant 27.48 ***† 27.08 *** 0.31 ** † 0.26 **
Open in a new tab
*

p < .05,

**

p < .01,

***

p < .001 (two-tailed tests).

†

significantly different from women, p < .05.

Model 2 introduces the interaction term for number of close friends and PIR. Hypothesis 3 is partially supported such that the association between PIR and weight status is significantly different by level of social integration for men. Fig. 1 graphs this relationship and presents predicted probability of being obese by number of close friends and PIR for men and demonstrates that higher levels of social integration, perhaps over-integration, is a risk factor for obesity for men with low PIR. However, at higher levels of PIR the difference by social integration decreases. Additionally, we found that among men with low social integration, PIR is positively associated with weight status, but among men with high integration, PIR is negatively associated with weight status.

Fig. 1.

Fig. 1.

Open in a new tab

Predicted probability of obesity among men by poverty to income ratio and social integration, NHANES 2003–2008.

4. Discussion

In this study, we examined associations between income, social integration and higher weight status. We examined hypotheses that income (H1) and social integration (H2) are each independently and inversely associated with higher weight status. We found partial support for the first hypothesis in that there was an inverse association between income and higher weight status among women only. We found no independent association between social integration and higher weight status for neither gender. Finally, we found that the association between social integration and BMI varies by income, for men only (H3). This finding helps elucidate the relationship between income and BMI and social integration and BMI among middle-aged men.

We hypothesized that there would be an inverse association with income and higher weight status as demonstrated by past research. We find support for this hypothesis among women, such that, those with higher income had a significantly lower average BMI than women in with lower income even after controls are included in the model. However, among men, we actually find that those in with higher income have a significantly higher BMI than those with lower income (Model 2). Rather than being a protective factor, income represents a risk factor for increasing BMI among men and this is true in models with and without controls (models without controls not shown). Past research has consistently demonstrated that the association between income and weight status is weaker for men compared to women (McLaren, 2007; Wang and Beydoun, 2007; Sobal and Stunkard, 1989; Zhang and Wang, 2004). Past research has suggested there are several possible explanations for these gender differences in weight status. First, cultural ideals may be important factors in whether higher body weight is acceptable in women and men (Peralta, 2003). Phelan, Link, and Tehranifar (2010:S36) suggest that these cultural norms are “likely to be embedded in strong social norms and support” (Phelan Jo et al., 2010). In this scenario, higher relative weight may be advantageous for men and lower relative weight is beneficial for women. Kawachi, Adler, and Dow (2010:59) refer to this cultural explanation as the “fat bias” in which women are judged more harshly for being overweight (Kawachi et al., 2010). As a result, men who are heavier tend to earn higher incomes than thinner men, and conversely, women who are thinner enjoy higher income compared to women with higher weight status. Second, some research suggests that men in lower SES groups are more likely to engage in employment where physical activity is more frequent. Higher SES men tend to be employed in positions which require less physical activity. Third, stress may play a role in the development of higher weight status (Lee, 2011). A variety of stress-related factors such as gendered stress responses, differential stress associated with disadvantage, and differences in physiological responses to stress may impact differences in weight status.

While we did find that higher income was associated with higher average BMI among men, examining the interaction between income and number of friends demonstrates that this depends on social integration. As social integration increases, differences by income for men diminish. Additionally, at high levels of social integration there is a cross-over, such that men with the highest income have a lower average BMI than men in the lowest income. Among women, our findings mirror past research such that women at lowest income have a higher average BMI than women at highest income. However, we do not show that these differences for women are influenced by social integration. Examining the gender differences listed above, this research suggests that cultural norms may play an important role in the relationship between income and weight status. Past research has demonstrated that women pay a greater penalty for obesity but limited attention is given to the social protections obese men may experience, especially obese men with higher incomes. Men do not pay as hefty as a cultural cost for high weight status as women and may be why there is only a benefit to higher income in terms of lower weight status among the most socially integrated men. However, this does not explain the much lower risk of high weight status experienced by low income men, especially at low levels of integration. This may be due to differences in physical activity requirements by different occupations as suggested earlier. Future research should further examine the relationship between income and high weight status for men.

In regards to social integration, the results highlight the positive and negative potential of this for weight status. Having more close friends appears to have an advantageous impact on weight status among higher income men, but represents a disadvantage among those with lower incomes. These positive and negative consequences of social integration for weight status may occur for a variety of reasons. First, homophily within social networks, or the degree of similarity among selected peers, has been shown to contribute to higher weight status in adolescents (de la Haye et al., 2011; Simpkins et al., 2013) as well as adults (Christakis and Fowler, 2007). In other words, people tend to associate with similar others (e.g. weight status) and as a result, having more friends of lower SES (who may also be heavier) may contribute to higher weight status (McPherson and Smith-Lovin, 2001; de la Haye et al., 2011; Simpkins et al., 2013; Centola, 2011). The current examination does not assess the mechanisms by which social networks influence weight status, but this would be important to consider in future studies of weight status of middle-aged populations. Second, socialization may also play an important role in the development of health behaviors and is patterned by different markers of socioeconomic status including income (Cockerham, 2005). Greater integration may result in more socialization such that the health behaviors of one’s social connections highly influence one’s own health behaviors (Centola, 2011). Finally, having more close friends (and family) may place additional strain on one’s already limited resources, perhaps prompting an elevated physiological response to stress and ultimately leading to higher weight status. Each of these explanations is rooted in social integration and may help to explain differences in weight status.

An important limitation of this study is the use of cross-sectional data which does not allow for causal inferences. In addition, NHANES has limited variables representing social integration. Despite these limitations, this study has numerous strengths including the use of measured height and weight, a large nationally representative sample, and comparisons by gender. Further study would ideally use longitudinal data with more sensitive and improved measures of social integration. Additionally, past research has indicated that income and weight status may operate differently for different race/ethnic groups. Additional analyses indicated that none of the interactions between income and race/ethnic group were significant for women or men. This suggests that among middle aged adults, the association between income and weight status is similar across race/ethnic groups.

This study contributes to the large body of literature examining the relationship between income and weight status and adds new information that social integration plays a role in that association. Perhaps the most important contribution to the literature is a confirmation of a gendered pattern of associations for income and social integration with higher weight status, particularly for men. Each of these social factors has varied associations with higher weight status by gender. Future research would examine data to determine appropriate interventions based on women’s and men’s tendencies to use social resources differently. Findings suggest that health professionals may maximize the effectiveness of healthy weight promotion efforts by paying special attention to their clients’ available supports through friends and family members. Finally, public health initiatives aimed at decreasing obesity rates might benefit from incorporating the role of social relations.

5. Conclusion

The association between income and higher weight status operates differently for women and men. This association between one’s income and body weight is dependent, in part, on the degree that they are socially integrated with friends and family. Interactions between income and social integration reveal that higher numbers of close friends and family may present as protective or risk factors for higher weight status, especially among men. We conclude that higher numbers of close friends and family is a protective factor against higher weight status for men with higher income but places low-income men at greater risk.

Acknowledgements

Funding: This work was supported by the Agency for Healthcare Research and Quality [grant number 5T32HS013852].

Footnotes

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ssresearch.2019.04.014.

References

  1. Ailshire Jennifer A., House James S., 2011. The unequal burden of weight gain: an intersectional approach to understanding social disparities in BMI trajectories from 1986 to 2001/2002. Soc. Forces 90 (2), 397–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ali Mir M., Dwyer Debra S., 2009. Estimating peer effects in adolescent smoking behavior: a longitudinal analysis. J. Adolesc. Health 45, 402–408. [DOI] [PubMed] [Google Scholar]
  3. Ball Kylie, Crawford David, 2006. An investigation of psychological, social and environmental correlates of obesity and weight gain in young women. Int. J. Obes 30 (8), 1240–1249. [DOI] [PubMed] [Google Scholar]
  4. Berkman Lisa F., Glass Thomas, Brissette Ian, Seeman Teresa E., 2000. From social integration to health: durkheim in the new millennium. Soc. Sci. Med 51,843–857. [DOI] [PubMed] [Google Scholar]
  5. Bourdieu Pierre, 1977. Outline of a Theory of Practice. Cambridge University Press. [Google Scholar]
  6. Bourdieu Pierre, 1986. The forms of capital. In: Richardson J (Ed.), Handbook of Theory and Research for the Sociology of Education. Greenwood, New York, pp.241–258. [Google Scholar]
  7. Caughy Margaret O., O’Campo Patricia J., Muntaner Carles, 2003. When being alone might Be better: neighborhood poverty, social capital, and child mental health. Soc. Sci. Med 57 (2), 227–237. [DOI] [PubMed] [Google Scholar]
  8. Centola Damon, 2011. An experimental study of homophily in the adoption of health behavior. Science 334 (6060), 1269–1272. [DOI] [PubMed] [Google Scholar]
  9. Chang Virginia W., Lauderdale Diane S., 2005. Income disparities in body mass index and obesity in the United States, 1971–2002. Arch. Intern. Med 165 (18), 2122–2128. [DOI] [PubMed] [Google Scholar]
  10. Christakis Nicholas A., Fowler James H., 2007. The spread of obesity in a large social network over 32 years. N. Engl. J. Med 357, 370–379. [DOI] [PubMed] [Google Scholar]
  11. Cockerham William C., 2005. Health lifestyle theory and the convergence of agency and structure. J. Health Soc. Behav 46 (1), 51–67. [DOI] [PubMed] [Google Scholar]
  12. de la Haye Kayla, Robins Garry, Mohr Philip, Wilson Carlene, 2011. Homophily and contagion as explanations for weight similarities amongst adolescent friends. J. Adolesc. Health 49 (4), 421–427. [DOI] [PubMed] [Google Scholar]
  13. Gorman Bridget K., Sivaganesan Ahilan, 2007. The role of social support and integration for understanding socioeconomic disparities in self-rated health and hypertension. Soc. Sci. Med 65, 958–975. [DOI] [PubMed] [Google Scholar]
  14. Ichiro Kawachi, Berkman Lisa F., 2001. Social ties and mental health. J. Urban Health: Bull. N. Y. Acad. Med 78 (3), 458–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Kawachi Ichiro, Adler Nancy E., Dow William H., 2010. Money, schooling, and health: mechanisms and causal evidence. Ann. N. Y. Acad. Sci 1186, 56–68. [DOI] [PubMed] [Google Scholar]
  16. Kouvonen Anne, Stafford Mai, De Vogli Roberto, Shipley Martin J., Marmot Michael G., Cox Tom, Vahtera Jussi, Väänänen Ari, Heponiemi Tarja, Singh-Manoux Archana, Kivimäki Mika, 2011. Negative aspects of close relationships as a predictor of increased body mass index and waist circumference: the whitehall II study. Am. J. Public Health 101 (8), 1474–1480. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lee Hedwig, 2011. Inequality as an explanation for obesity in the United States. Sociol. Compass 5 (3), 215–232. [Google Scholar]
  18. McLaren Lindsay, 2007. Socioeconomic status and obesity. Epidemiol. Rev 29, 29–48. [DOI] [PubMed] [Google Scholar]
  19. McPherson Miller, Smith-Lovin Lynn, 2001. Birds of a feather: homophily in social networks. Annu. Rev. Sociol 27, 415–444. [Google Scholar]
  20. National Center for Health Statistics (NCHS), 2010. NHANES Web Tutorial Frequently Asked Questions. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, Hyattsville, MD. [Google Scholar]
  21. Peralta Robert L., 2003. Thinking sociologically about sources of obesity in the United States. Gend. Issues 21 (3), 5–16. [Google Scholar]
  22. Phelan Jo C, Link Bruce G., Tehranifar Parisa, 2010. Social conditions as fundamental causes of health inequalities: theory, evidence, and policy implications. J. Health Soc. Behav 51 (Suppl. l), S28–S40. [DOI] [PubMed] [Google Scholar]
  23. Ross Catherine E., Mirowsky John, 2006. Sex difference in the effect of education on depression: resource multiplication or resource substitution? Soc. Sci. Med 63 (5), 1400–1413. [DOI] [PubMed] [Google Scholar]
  24. Schafer Markus H., Ferraro Kenneth F., 2011. Distal and variably proximal causes: education, obesity, and health. Soc. Sci. Med 73 (9), 1340–1348. [DOI] [PubMed] [Google Scholar]
  25. Simpkins Sandra D., Schaefer David R., Price Chara D., Vest Andrea E., 2013. Adolescent friendships, BMI, and physical activity: untangling selection and influence through longitudinal social network analysis. J. Res. Adolesc 23 (3), 537–549. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Sobal Jeffery, Stunkard Albert J., 1989. Socioeconomic status and obesity: a review of the literature. Psychol. Bull 105 (2), 260–275. [DOI] [PubMed] [Google Scholar]
  27. StataCorp, 2009. Stata Statistical Software: Release, vol. 11 StataCorp LP, College Station, TX. [Google Scholar]
  28. Thoits Peggy A., 1995. Stress, coping, and social support processes: where are we? What next? J. Health Soc. Behav 35 (Extra Issue), 53–79. [PubMed] [Google Scholar]
  29. Thoits Peggy A., 2011. Mechanisms linking social ties and support to physical and mental health. J. Health Soc. Behav 52 (2), 145–161. [DOI] [PubMed] [Google Scholar]
  30. Umberson Debra, Montez Jennifer K., 2010. Social relationships and health: a flashpoint for health policy. J. Health Soc. Behav 51 (S), S54–S66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Uphoff Eleonora P., Pickett Kate E., Cabieses Baltica, Small Neil, Wright John, 2013. A systematic review of the relationships between social capital and socioeconomic inequalities in health: a contribution to understanding the psychosocial pathway of health inequalities. Int. J. Equity Health 12, 54–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Wang Youfa, Beydoun May A., 2007. “The obesity epidemic in the United States—gender, age, socioeconomic, racial/ethnic, and geographic characteristics: a systematic review and meta-regression analysis. Epidemiol. Rev 29, 6–28. [DOI] [PubMed] [Google Scholar]
  33. Zhang Qi, Wang Youfa, 2004. Socioeconomic inequality of obesity in the United States: do gender, age, and ethnicity matter? Soc. Sci. Med 58 (6), 1171–1180. [DOI] [PubMed] [Google Scholar]

RESOURCES