SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

In this article, we describe differences in the self-perception of weight status in the United States between the two most recent National Health and Nutrition Examination Survey (NHANES) periods (1988–1994 and 1999–2004), and test the hypothesis that secular increases in adult mean BMI, adult obesity, and childhood obesity contributed to changes over time in weight perceptions. We find that the probability of self-classifying as overweight is significantly lower on average in the more recent survey, for both women and men, controlling for objective weight status and other factors. Among women, the decline in the tendency to self-classify as overweight is concentrated in the 17–35 age range, and is more pronounced among women with normal BMI than those with overweight BMI. Among men, the shift away from feeling overweight is roughly equal across age groups. Overweight men exhibit a sharper decline in feeling overweight than normal weight men. Despite the declines in feeling overweight between surveys, weight misperception did not increase significantly for men and decreased by a sizable margin among women. We interpret the findings as evidence of a generational shift in social norms related to body weight. As a result, people may be less likely to desire weight loss than previously, limiting the effectiveness of public health campaigns aimed at weight reduction. On the other hand, there may be health benefits associated with improved body image.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

A significant share of overweight and obese individuals perceive their body weight to be either “about right,” “appropriate,” or “acceptable,” or express satisfaction with their body size (1,2,3,4,5,6,7,8). Misperception of overweight status is more prevalent among individuals with low socioeconomic status, among African Americans, and among men compared with high SES individuals, white (non-Hispanic) Americans, and women, respectively (7,9). Researchers have hypothesized that departures of self-assessments from objective standards indicate that sociocultural influences, and social weight comparisons in particular, play an important role in such assessments (3,10,11,12,13). A number of recent papers find evidence in cross-sectional data that supports this hypothesis (5,14). Also consistent with the social comparison hypothesis, Johnson et al. (13) find that the self-perception of overweight declined in the United Kingdom between 1999 and 2007, a period during which rates of overweight and obesity increased significantly.

In the wake of dramatic increases in obesity rates worldwide in recent decades, obesity has become a top public health priority in the United States and many other countries. Among recent interventions aimed at obesity reduction and prevention, some have advocated for improved education in relation to weight self-perception. Mistaken perceptions have been linked to risky behaviors, whereas accurate perceptions have been associated with appropriate weight goals. Overweight individuals who do not perceive themselves as such are less likely to desire weight loss than overweight individuals with accurate perceptions (6,15,16), and may be more likely to smoke, have a poor diet, and be physically inactive (17). On the other end of the spectrum, disordered eating behavior has been associated with mistaken perceptions of overweight among normal-weight and underweight individuals, especially women (18).

Despite the extensive evidence on cross-sectional variation in weight perceptions, little is known about the evolution of weight perceptions over time within the United States.1 Because mean BMI and obesity increased in the United States between the National Health and Nutrition Examination Survey (NHANES) III (1988–1994) and NHANES 1999–2004 surveys, the social comparison framework (10,11) predicts that, compared to adults observed in the earlier survey, those observed in the more recent survey should be less likely to classify themselves as overweight, controlling for objective weight status. In this article, we describe differences in the self-perception of weight status in the United States between two survey cohorts observed 10 years apart on average, and we test the hypothesis that secular increases in adult mean BMI, adult obesity, and childhood obesity contributed to declines over time in the self-perception of overweight status and increases in the self-perception of normal weight status.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

Data from the NHANES III (1988–1994) and NHANES 1999–2004 were used in this analysis. NHANES is a continuous survey of the health and nutritional status of the US civilian, noninstitutionalized population; samples are selected through a complex, multistage probability design. Data are collected through interviews and physical examinations that include measurement of weight and height. Individuals aged ≥17 and ≤74 years completing the physical examination were included in this analysis; pregnant women were excluded.2,3

BMI are calculated based on measured weight and measured height. Objective weight status (“underweight,” “normal weight,” “overweight,” or “obese”) follows the CDC (Centers for Disease Control and Prevention) classification system, which consists of a uniform set of BMI thresholds for adults ages ≥20, and age-specific BMI thresholds for individuals ages ≤19. See Supplementary Table S1 online for the BMI thresholds for adolescents aged 17–19. In each survey, individuals aged 17–74 were asked whether they consider themselves to be either “underweight,” “about right,” or “overweight.”4 The choice to self-classify as obese was not included, and therefore obese individuals will be said to have an accurate self-perception if they consider themselves to be overweight.5 Otherwise, correct perception of weight status (also referred to simply as “self-perception” of weight status) occurs when self-reported status coincides with objective status. Misperception of weight status is defined as a difference between self-reported (subjective) status and objective status (for example, if an individual with a BMI of 28 selects the “about right” response to the self-perception question).

We control for complex survey design within each survey using the appropriate strata and primary sampling unit variables. Using Stata's “svy” commands, the data are weighted correctly and the standard errors are clustered appropriately.

Analytic plan

We calculate cross-tabulations of subjective weight status against objective weight status, separately by survey period and sex. We also compute misperception rates by sex and survey period, for the entire BMI distribution, and separately for individuals with BMI ≥25 (that is, for overweight and obese individuals). For each of six age groups, we calculate differences between the two survey periods (NHANES 1999–2004 minus NHANES III) in the following variables: mean BMI, obesity rate, prior childhood obesity rate, accurate perception of overweight by overweight (not obese) individuals, and accurate perception of overweight by obese individuals. We compute associated standard errors and report significance levels of these between-survey differences.

We estimate the probability of self-classifying as “overweight” using a multivariate logit model. The base response is “not overweight,” which includes both the “about right” and “underweight” responses.6 To test for differences in responses between the two surveys (NHANES III and NHANES 1999–2004), we pool the data and create a dummy variable that takes the value 1 for observations from NHANES 1999–2004 and the value 0 for observations from NHANES III.7 This variable is included in all model specifications, and its effect is allowed to vary by age group and sex. We divide age into six discrete categories: 17–19, 20–25, 26–35, 36–45, 46–55, and 56–74.8 In an initial set of regressions we estimate the models separately for men and women. When we test for reference-group effects, we pool the data but still allow for gender-specific effects, for reasons discussed below. All models also include the following variables: measured (individual) BMI (and its square), race/ethnicity, educational attainment, household income, and marital status. The construction of these variables is discussed in a Supplementary Appendix online. Sample means by sex, pooled across surveys, are shown in Supplementary Table S2 online. All results are reported as odds ratios, with associated standard errors and significance levels. To facilitate understanding of the logit estimates, we plot the predicted probabilities of feeling overweight for specific BMI values, by age group and survey cohort, at representative values of the covariates.

We estimate a series of alternative specifications that include additional control variables. The first of these models adds measurements of body fat percentage and waist circumference, either of which may affect health risks and physical appearance, independent of BMI.9 To investigate the role of social comparison effects on weight perceptions, we estimate models that include, in turn, (i) the contemporaneous mean BMI in a reference group, (ii) the contemporaneous obesity rate in a reference group, and (iii) the prior childhood obesity rate in a reference group. The reference group is defined by sex, age group, survey period, and race/ethnicity in the case of the adult mean BMI, and adult obesity rate variables. In the case of the childhood obesity rate variable, the reference group is defined by sex, age group, and survey period (and not also race/ethnicity) due to data limitations. These rates can be constructed only for individuals aged 17–42 and the construction involves imputation in some cases; for details, see the Supplementary Appendix online. In the social comparison models, male and female data are pooled in order to reduce the risk of collinearity between reference-group variables and survey-by-age effects. In order to determine whether social comparison effects help to explain changes in self-perception across surveys, we conduct a series of Wald tests using the Stata “suest” command.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

In Table 1, the value in a given cell indicates the percentage of individuals in a given objective weight status category (the row variable) that reported being in a given subjective weight status category (the column variable). For example, in NHANES III, among women with normal BMI values, 56% considered their weight to be “about right” and 40% considered themselves to be “overweight.” For women in NHANES 1999–2004, the corresponding figures are 63 and 33%. Among women qualifying as overweight but not obese, 85% reported feeling overweight as of NHANES III, and only 79% said the same in 1999–2004. (All of the between-survey differences are significant at the 5% level or better.) For a given objective weight status and observation period, men are significantly less likely than women to self-classify as overweight and significantly more likely to classify as underweight. However, like women, men's self-perceptions also shifted significantly between survey periods. For example, among overweight (but not obese) men, the share that self-classified as overweight fell from 58% in NHANES III to 53% in NHANES 1999–2004. Among obese men the share that considered themselves overweight fell from 89 to 87%.10

Table 1. Weight perception, NHANES III (1988–1994), and NHANES 1999–2004
inline image

Figure 1 shows that, despite declines in the tendency to self-classify as overweight, the weight misperception rate—the share who incorrectly perceive their objective weight status—actually fell among women between the surveys and did not change significantly for men. Among overweight and obese subjects combined, misperception increased significantly between the surveys for women and was unchanged for men.

image

Figure 1. Misperception of weight, NHANES III (1988–1994) and NHANES 1999–2004, ages 17–74. Overweight includes both overweight and obese subjects. For ages 17–19, we use age-specific BMI thresholds for overweight. ***Difference in predicted probabilities between NHANES 1999–2004 and NHANES III (1988–1994) is significant at the 1% level or better.

Download figure to PowerPoint

Table 2 shows age-group-specific changes in weight perception and changes in aggregate weight outcomes between the surveys. Aggregate weight outcomes include adult mean BMI, the adult obesity rate, and the (prior) childhood obesity rate. Among men, those aged 20–25 experienced the largest declines in (accurate) self-perception of overweight, among both overweight (not obese) and obese subjects. This same group saw the largest increases between surveys in mean BMI, adult obesity rate, and childhood obesity rate.11 Among women, the 20–25-year-old age group exhibited either the largest or second-largest change in each column, although the decline in overweight perception among obese women aged 20–25 was not statistically significant. For the oldest age group (56–74), mean BMI and the adult obesity rate increased between the surveys (for both sexes) and yet no significant declines in overweight self-perception (among either overweight or obese subjects) were observed for either sex. In fact, overweight (not obese) men in this age group became more likely to self-classify as overweight in the later survey.

Table 2. Between-survey differences in mean BMI, obesity rates, and perceptions. NHANES 1999–2004 minus NHANES III, 1988–1994
inline image

Table 3 shows the results of three different models, each pertaining to individuals aged 17–74, analyzed separately by sex. The baseline model (“model 1″), includes BMI, BMI squared, and the demographic and socioeconomic variables. Model 2 includes the baseline variables plus body fat percentage. Model 3 includes all the previous variables, plus waist circumference. In the baseline specification for women, the estimated odds ratio on the NHANES 1999–2004 dummy is 0.74, suggesting that, all else equal, 36–45-year-old women from NHANES 1999–2004 are significantly less likely to consider themselves overweight than women of the same age from NHANES III.12 Women in each of the three youngest age groups (17–19, 20–25, and 26–35), exhibited significantly sharper declines in the tendency to feel overweight than those in the omitted (36–45) age group. In the corresponding model for men, 36–45 year olds experienced a large and significant decline in the probability of feeling overweight (odds ratio 0.52). The only significant age difference in the survey effect for men pertains to the oldest group, among whom the tendency to feel overweight was roughly unchanged between surveys in the baseline model.13

Table 3. Logit estimates of self-classifying as overweight, women and men
inline image

Model 2 indicates that, all else constant, a higher body fat percentage is associated with a greater risk of feeling overweight. For both men and women, the main survey effect becomes smaller (that is, the decline is greater) when body fat is included. This result suggests that the tendency to feel overweight would have declined by an even greater margin if body fat percentage had remained constant over time, conditional on BMI. Model 3 shows that a higher waist circumference is also associated with a higher risk of feeling overweight. Concerning the variables that models 1, 2, and 3 have in common, coefficient estimates are broadly similar, in both magnitude and significance, across all three models. One exception is that the coefficient on BMI declines when we include either body fat percentage (model 2) or both body fat percentage and waist circumference (model 3).

Table 3 also shows that the probability that an individual classifies as overweight increases with BMI, but the additional risk is smaller at higher BMI. High school graduates and college attendees are each more likely to self-classify as overweight than those that did not complete high school, and those with middle and high incomes are more likely to feel overweight than those with the lowest incomes. Married people are more likely to feel overweight than never-married people, and members of minority groups are significantly less likely than whites to consider themselves overweight. All of these results apply to both men and women.

Using the estimates from model 1 in Table 3, Figures 2 and 3 show the predicted probabilities of feeling overweight for men and women, respectively,14 by survey and age group. Probabilities are computed for the BMI values of 23 and 28, which represent the approximate midpoints, respectively, of the BMI ranges defined as “normal” and “overweight” by the CDC. The figures indicate the statistical significance of between-survey differences in the tendency to feel overweight.

image

Figure 2. Predicted probability of feeling overweight among women by age group; NHANES III (1988–1994) and NHANES 1999–2004; BMI = 23 and BMI = 28. Predictions based on logit regression model 1 (female) in Table 3. ***Difference in predicted probabilities between NHANES 1999–2004 and NHANES III (1988–1994) is significant at the 1% level; **Significance at the 5% level; *Significance at the 10% level.

Download figure to PowerPoint

image

Figure 3. Predicted probability of feeling overweight among men by age group; NHANES III (1988–1994) and NHANES 1999–2004; BMI = 23 and BMI = 28. Predictions based on logit regression model 1 (male) in Table 3. ***Difference in predicted probabilities between NHANES 1999–2004 and NHANES III (1988–1994) is significant at the 1% level; **Significance at the 5% level; *Significance at the 10% level.

Download figure to PowerPoint

Figure 2 confirms that, for women, the decline between surveys in the predicted probability of feeling overweight is more pronounced among those aged 35 and under than among those over 35. No significant difference is observed for women in the 46–55-year-old group. Ignoring the latter group, the percentage-point declines in feeling overweight are larger for women with BMI of 23 than for BMI of 28. For example, among 17–19-year-old women, the predicted probability of feeling overweight falls by 32% points assuming a BMI of 23, and falls by just 8% points assuming a BMI of 28. Figure 3 shows that, among men, there are no significant differences in the predicted responses between the survey periods for either the youngest group (17–19) or the oldest group (56–74). The differences are sizable, and similar, for each of the 20–25, 26–35, and 36–45-year-old groups. In contrast to women, declines among men in the predicted probability of feeling overweight are greater at a BMI of 28 than a BMI of 23.

Table 4 shows estimates from logit regression models in which male and female data are pooled in order to estimate models with reference-group effects. Model 1 includes all the variables from model 3 in Table 3, plus the estimated childhood obesity rate for the reference group (based on age, sex, and survey period). In model 1, the sample is restricted to individuals aged 17–42 and the omitted age category is 36–42. Model 2 includes the same variables as model 1, plus the contemporaneous mean BMI for the reference group (defined by age, sex, survey period, and race/ethnicity). Model 3 includes the same variables as model 1, plus the contemporaneous obesity rate for the reference group.

Table 4. Logit estimates of self-classifying as overweight, reference group effects
inline image

In model 1, the coefficient associated with the childhood obesity rate is <1 (0.54), but is not statistically significant. We compare the estimated survey effects in model 1 (for each age group) to the corresponding estimates in an otherwise identical model (not shown) that omits the childhood obesity rate. For each pair of estimates, we conduct a Wald test of the hypothesis that their values are equal. In each case, the hypothesis can be rejected at the 10% level. These results offer some evidence that exposure to higher childhood obesity among the more recent survey cohort partly accounts for their lower risk of feeling overweight.

Model 2 shows that higher mean BMI among one's reference group decreases the likelihood of self-classifying as overweight, where the coefficient estimate is statistically significant at the 1% level or better. Model 3 indicates that a higher adult obesity rate in the reference group also results in a significant decline in the likelihood of feeling overweight. Again we test for whether the survey effect (within each age group) differs significantly between models with and without either the mean BMI variable or the adult obesity rate variable. The tests indicate that the introduction of mean BMI results in a significant dampening of the survey effect for all but one age group (17–19 year olds), and the introduction of the adult obesity rate reduces the survey effect for all but two age groups (26–35 year olds and 56–74 year olds). These results suggest that social comparison effects help to explain changes in perception of overweight across surveys.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

We document a significant decline, on average, in the probability of self-classifying as overweight between the NHANES III and 1999–2004 survey periods. The decline was particularly pronounced among younger (ages 17–35), normal-weight women and among young-to-middle aged (20–45), overweight men. These findings cannot be explained by changes in individuals' own BMI, race/ethnicity, age, and socioeconomic characteristics across the survey periods. Changes in body fat percentage and waist circumference do not account for the changes in weight classification between surveys. In univariate analysis of misperception of weight status, we observe a net decline in misperception relative to CDC standards among women and no net change in misperception among men. Johnson et al. (13) observe similar trends in weight self-perception in the United Kingdom, based on two surveys conducted in 1999 and 2007, respectively. However, their analysis relies on self-reported weight and height data and involves a more restricted set of control variables.

Because “overweight” is contrasted with “about right” in the survey instrument, implying a normative judgment, the responses help locate subjective thresholds for overweight. Our findings suggest that, on average, the threshold for overweight increased between the earlier and later surveys, consistent with the social comparison hypothesis and the theory of endogenous weight norms (10). We find that members of groups (defined by age, sex, race/ethnicity, and birth cohort) with higher average BMI (or higher obesity rate) are less likely to classify as overweight than members of groups with lower mean BMI (or lower obesity rate), controlling for own BMI and other factors. Furthermore, these relationships may help to account for the between-survey decline in the tendency to feel overweight. Exposure to a higher childhood obesity rate is not associated (at conventional significance levels) with a reduced tendency to feel overweight in adulthood, and yet it may explain some of the reduced tendency to feel overweight in the more recent survey period. The latter association raises the concern that the current generation of children, who are experiencing an even higher childhood obesity rate than the cohort of children observed in NHANES III, will develop an even higher subjective threshold for overweight, resulting in increased rates of weight misperception.

We do not claim to identify causality between the rightward shift of the weight distribution and the increase in the subjective threshold for overweight. The changes in perceptions between surveys could have preceded the changes in the contemporaneous outcomes (mean BMI and adult obesity), or both developments could have occurred in response to a change in some other factor. A variety of social forces may influence self-perceptions of body weight, such as popular media, public health messages, and scholastic content. Any of these factors may have changed between the survey periods, contributing to a more relaxed standard of overweight.

The frequency of news stories related to obesity was dramatically higher during the later survey period than the earlier period (19). Whereas an increase in news coverage might have been expected to result in heightened awareness of overweight status, by emphasizing the increased prevalence of overweight, the stories may have contributed to the perception that overweight had become the norm. At the same time, the thin female body ideal portrayed in media imagery persisted across the two survey periods (20,21,22). Given that women's weight assessments became more forgiving during this time period, it would appear that media images do not dictate women's attitudes toward their weight or at least that they have become less influential in recent years.15 Concerning public health messages, the official CDC standard for overweight actually became stricter between the survey periods, such that individuals with BMI values between 25 and 27 should have been more likely to classify as overweight in the later survey.16 In the past decade numerous government programs have emerged that promote the development of a “healthy” body image and healthy eating behavior, for young women in particular.17 Such programs may have contributed to our finding that young, normal-weight females experienced some of the greatest reductions of any demographic group in the probability of self-classifying as overweight.

Whereas individuals in both the normal and overweight groups revealed a higher threshold for overweight in the later survey, this change resulted in both a lower misperception rate among normal weight women and a higher misperception rate among overweight and obese women, and among overweight and obese men aged 35 and under. An implication is that people may now be less likely to desire weight loss than previously, regardless of their objective weight status, and so may be less inclined toward dieting and physical activity. Such complacency among overweight and obese individuals may limit the effectiveness of public health campaigns aimed at weight reduction and associated improvements in health outcomes, including efforts to raise awareness of BMI thresholds for overweight and obesity. On the other hand, there may be health benefits associated with improved body image, such as higher self-esteem and, potentially, a decline in the incidence of eating disorders. More fundamentally, recent research questions the usefulness of the BMI classification system and in some cases recommends alternative criteria for overweight and obesity, such as waist circumference and body fat percentage (23,24,25,26). For example, individuals with large amounts of lean body mass can achieve BMI values that qualify as overweight or obese, despite having very little body fat. Such individuals may (rationally) disregard CDC standards for overweight based on BMI.

Appendix

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

1  The Johnson et al.'s (13) study of the United Kingdom is the only one we are aware of which documents changes in self-perception over time using nationally representative surveys.

2  We use self-reported pregnancy status in order to maximize consistency across surveys. We also excluded those who responded “don't know” to the pregnancy question and those whose responses were missing. Some women are not asked whether they are pregnant, based on a prior question and based on their age, and these observations are treated as “not pregnant,” rather than as missing values. These exclusions eliminate a total of 1,048 observations across both surveys combined.

3  Ver Ploeg et al. (8) suggest omitting individuals who are objectively underweight (BMI < 18.5 for adults) from analysis of overweight self-classification. We find that our results are not significantly different depending on whether such individuals are included or omitted, and all results shown here include underweight individuals.

4  Beginning in 1999, respondents as young as 16 were asked this question. For consistency across surveys, we consider only individuals aged 17–74 throughout the analysis.

5  Data on weight perceptions are also available for NHANES 2005–2006. However, our regression analysis also makes use of body fat percentage data, which are not yet available for 2005–2006. For consistency of comparison, we omit the 2005–2006 data throughout. In models that do not employ body fat percentage, results are not significantly different when the 2005–2006 data are included.

6  Most “not overweight” responses refer to the response “about right.” In the sample at large, just 3% of women and 8% of men report feeling underweight.

7  According to Korn and Graubard (27), it is not necessary to reweight the data when pooling NHANES surveys, based on the assumption that the samples are independent across the periods.

8  Although they form a relatively small group, 17–19 year olds are placed in a separate category because the criteria for obesity differ between adolescents and adults.

9  Body fat percentage is measured differently in NHANES III and NHANES 1999–2004. In the latter survey, NHANES releases multiply imputed measures of body fat percentage. Regression analysis employs multiple imputation techniques, consistent with NHANES analytical guidelines.

10  The differences in self-reporting conditional on weight status cannot be explained by decreases in average BMI conditional on the objective weight category, because the conditional means either remained constant or increased between the survey periods.

11  The comparison of changes in childhood obesity rates is based on point estimates only, because standard errors are not available.

12  Given the interaction terms between the survey and the age groups, the survey effect refers to the effect for the omitted age group only. Survey effects for other age groups are obtained by multiplying the main survey effect by the coefficient on the appropriate age–survey interaction term.

13  This result is obtained by multiplying the odds ratio on the main survey effect and the odds ratio on the age-survey interaction for 56–74 year olds, both taken from the baseline model, to yield a value of 1.04. Doing the same for models 2 and 3 yields net survey effects for 56–74 year olds of 0.76 and 0.78, respectively.

14  All variables other than BMI are held at the sex-specific means of the combined sample (NHANES III and NHANES 1999–2004), reported in Supplementary Table S2 online.

15  An anonymous referee has suggested that the recent proliferation of alternative media outlets may have reduced the influence of any single medium such as women's magazines.

16  In 1998, on the recommendation of the World Health Organization, the CDC reduced the BMI thresholds for overweight, from 27.8 for men and 27.3 for women, down to 25 for both sexes. This new standard is applied retroactively when we measure rates of overweight and obesity in surveys conducted before 1998.

17  For example, the National Eating Disorders Association offers the educational program “GO GIRLS!,” which engages high school students to advocate for positive body images of youth in advertising, the media and major retailers. See http:www.nationaleatingdisorders.orgprograms-eventseducational-programs.phpgo-girls.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

We acknowledge the helpful feedback of Rusty Tchernis, Gerhard Glomm, and other participants at the Indiana University conference, “Cognitive, Behavioral, and Economic Approaches to Adolescent and Childhood Obesity,” 4–5 September 2008.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Appendix
  8. SUPPLEMENTARY MATERIAL
  9. Acknowledgments
  10. Disclosure
  11. REFERENCES
  12. Supporting Information

supporting Information

FilenameFormatSizeDescription
oby_1769_sm_1.doc31KSupporting info item
oby_1769_sm_2.doc69KSupporting info item
oby_1769_sm_3.doc32KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.