Over, Under, or About Right: Misperceptions of Body Weight Among Food Stamp Participants

Authors


(sverploeg@ers.usda.gov)

Abstract

The purpose of this research was to investigate the associations between misperception of body weight and sociodemographic factors such as food stamp participation status, income, education, and race/ethnicity. National Health and Nutrition Examination Survey (NHANES) data from 1999–2004 and multivariate logistic regression are used to estimate how sociodemographic factors are associated with (i) the probability that overweight adults misperceive themselves as healthy weight; (ii) the probability that healthy-weight adults misperceive themselves as underweight; and (iii) the probability that healthy-weight adults misperceive themselves as overweight. NHANES data are representative of the US civilian noninstitutionalized population. The analysis included 4,362 men and 4,057 women. BMI derived from measured weight and height was used to classify individuals as healthy weight or overweight. These classifications were compared with self-reported categorical weight status. We find that differences across sociodemographic characteristics in the propensity to underestimate or overestimate weight status were more pronounced for women than for men. Overweight female food stamp participants were more likely to underestimate weight status than income-eligible nonparticipants. Among healthy-weight and overweight women, non-Hispanic black and Mexican-American women, and women with less education were more likely to underestimate actual weight status. We found few differences across sociodemographic characteristics for men. Misperceptions of weight are common among both overweight and healthy-weight individuals and vary across socioeconomic and demographic groups. The nutrition education component of the Food Stamp Program could increase awareness of healthy body weight among participants.

Introduction

A sizable portion of overweight Americans do not believe that they are actually overweight. In 1988–1994, 55% of overweight or obese men and 23% of overweight or obese women did not believe they were overweight (1). Such misperceptions have been noted as potential barriers to addressing weight control problems—it is unlikely that someone will address a problem if they do not consider it to be a problem. Misperceptions about body weight also provide behavioral clues about why some groups may be more prone to obesity (1,2,3,4). For example, misperceptions about overweight status (an overweight person who incorrectly believes he or she is a healthy weight) among some subgroups may simply reflect a higher prevalence of overweight among those subgroups, such that people do not perceive themselves as heavier than their peers because their peers are heavier too. Or tendencies among subgroups to perceive weight status incorrectly may reflect ideals of body weight among subgroups (5).

One subgroup of particular concern is those who participate in the Food Stamp Program. This program, which is run by the US Department of Agriculture, issues monthly benefits for grocery store purchases for households with low incomes (below 130% of Federal poverty guidelines) (6). Higher prevalence of overweight and obesity among some individuals who receive food stamps, particularly women (7), has led to suggestions that the program may need to do more to address the problem of obesity while ensuring adequate nutrition for low-income households (8). How food stamp participants perceive their body weight is important for two reasons. First, if participants are less likely to recognize that they are overweight, they may be less likely to adapt behavior to maintain a healthy body weight or reduce weight, and may become heavier than nonparticipants. Second, the Food Stamp Program includes a nutrition education component. This component of the program could also be used to inform participants about what is considered a healthy weight and potentially help correct misperceptions.

Previous studies of which subgroups are prone to misperceptions about their weight found differences across race/ethnicity, gender, and some education and income groups (2,3). Additionally, cross-tabulations show that overweight food stamp participants are more likely to perceive themselves as healthy weight than nonparticipants (1). However, these cross-tabulations do not control for other factors that are associated with food stamp participation although they are also associated with weight misperception—particularly race/ethnicity, income, and education.

In this article, we explored the determinants of misperceptions of weight status. We expanded upon previous work in several ways. First, we considered whether one particular subgroup of public policy interest—participants in the US Food Stamp Program—are more prone to misperceptions about weight status. Unlike previous work examining weight perception among food stamp participants, we used a multivariate framework and controlled for factors associated with weight misperception. Second, we used recent data from the National Health and Nutrition Examination Survey (NHANES) 1999–2004 to analyze the relationship between food stamp participation, other economic and demographic variables, and the accuracy of weight perception. This includes weight perception among the overweight and among those who are healthy weight, unlike previous studies that focused on only one of these two groups (2). Finally, we included an individual's measured BMI (kg/m2) as a covariate to ensure that our results do not simply reflect differences in average BMI across population subgroups. This variable is important because individuals with BMI closer to the cutoff values for each weight status category may be more likely to assess their weight status incorrectly. Studies of body weight misperception for the general population did not include this measure, although a study of severely obese women did and found BMI was not correlated with misperception of body weight (9). Our study tested this variable for the general population, for whom weight status may be more ambiguous compared with the severely obese.

Methods and Procedures

The NHANES uses a stratified, multistage probability sampling design and is representative of the US civilian noninstitutionalized population (10). We used data from the 1999–2004 surveys in this article. Weight and height for each survey respondent were measured by trained medical staff in the NHANES Mobile Examination Centers. BMI was calculated to classify each individual as either underweight, healthy-weight, or overweight (for adults, BMI < 18.5 is underweight; 18.5 ≤ BMI < 25 is healthy weight; and BMI ≥ 25 is overweight). A small percentage of adults are underweight (<2%). Because this is such a small percentage of the population and because members of this group who perceive themselves as overweight are probably much different than normal-weight individuals who perceive themselves as overweight, we excluded them from the study. Respondents were also asked whether they consider themselves “overweight, underweight, or about the right weight”. Responses to this question were compared with an individual's measured weight category (healthy weight or overweight). For overweight adults, we examined determinants of whether an individual underestimated his/her weight (reports weight status as either “about the right weight” or “underweight”). For healthy-weight adults, we examined determinants of whether an individual underestimated his/her actual weight (reports being underweight), and whether an individual overestimated his/her measured weight (reports being overweight). Misperceptions of body weight among both those who are healthy weight and overweight are important. Understanding what factors are associated with misperception among the overweight can help explain why people are overweight. Similarly, understanding misperception among those who are healthy weight but believe they are underweight or overweight can be used to understand why some have not yet become overweight (they already believe they are and may take action to lose weight) or about whether they may become overweight (they believe they are underweight and do not change behavior to remain at a healthy weight). Studying both populations also allows comparisons of whether the factors associated with underestimating weight are the same for overweight and healthy-weight individuals.

Logistic regression was used to estimate the probability of overestimating or underestimating weight status. For overweight individuals, the probability of underestimating weight status was estimated. For healthy-weight individuals, two separate logistic models—the probability of underestimating weight status and the probability of overestimating weight status—were estimated. Odds ratios (ORs) and their corresponding 95% confidence intervals for each covariate were reported for each model. Statistical analyses were conducted using SUDAAN (version 9.0.1; Research Triangle Park, NC) (11). Following the literature on modeling BMI, models were estimated separately for men and for women (12,13,14).

We examined whether weight misperception differs by Food Stamp Program participation and by household income. To receive food stamps, gross household income cannot exceed 130% of the Federal poverty guidelines. Consequently, we combined food stamp participation and income into four subgroups—food stamp participants, income-eligible nonparticipants (income below 130% of the poverty guidelines), moderate-income individuals (income between 130 and 300% of the poverty guidelines), and high-income individuals (income exceeds 300% poverty guidelines). Food stamp participation was measured by responses to the question “Are you now authorized to receive Food Stamps?”

Other covariates in the logistic models were race/ethnicity, (non-Hispanic white, non-Hispanic black, Mexican American, or other race), age, self-reported current smoking status, education level (less than high school, high school graduate, and some postsecondary schooling), and marital status (married or single). We also included each individual's BMI as a covariate. Individuals with BMI closer to the cutoff values for weight status designation may be more likely to assess their weight status incorrectly. For example, we expected that individuals with a BMI of 26 are more likely to perceive themselves as healthy weight incorrectly compared with individuals with a BMI of 30. To control for this, we included a measure of the difference between measured BMI and the lower end of the BMI cutoff point for each weight status. For healthy-weight individuals, this variable is the difference between the individual's measured BMI and the lower cutoff for healthy weight (BMI = 18.5). For overweight individuals, this variable is the difference between an individual's measured BMI and the lower cutoff for overweight status (BMI = 25). This is an important improvement over the methods used in previous studies, which did not control for measured BMI. If among overweight women, non-Hispanic black women are much heavier than non-Hispanic white women (such that they have greater average BMI, say 30 vs. 27), we expect that more non-Hispanic white women would incorrectly assume they are healthy weight simply because more of them are near the cutoff for overweight relative to black women. Without controlling for BMI, it may appear that variation in the propensity to overestimate or underestimate weight status is driven by race or other covariates when in reality, variation is really driven by differences in average BMI levels.

Because only respondents over the age of 16 were asked about their weight status and because the BMI classifications for children under the age of 20 are different from adults, our analysis only examined misperception in weight among adults aged ≥20. We excluded pregnant or breast-feeding women. After deleting observations with missing values, the final sample included 4,362 male and 4,057 female respondents. Table 1 shows final sample size for each of the three models estimated separately by gender.

Table 1.  Sample size and weight self-assessment percentages by measured weight status and by gender
inline image

Results

Table 1 shows the percentage of men and women who correctly or incorrectly assessed their body weight status. Among healthy-weight women, 32.4% reported being overweight, but only 4.0% reported being underweight. For healthy-weight men, 18.4% reported being overweight, 73.8% correctly perceived their own weight status, but 7.9% reported being underweight. Approximately one-third of overweight men (32.9%) underestimated their body weight (reported being underweight or about the right weight). Among overweight women, only 12.7% characterized themselves as underweight or about the right weight.

Age, education, food stamp participation, income, and racial/ethnic background distinguished women in their perception of body weight status. However, few of these variables helped explain variations in men's perception. Covariates that were statistically significant at the 95% probability level are discussed here. Full results for all covariates are reported in Tables 2 and 3.

Table 2.  Odds of underestimating weight status for overweight men and women
inline image
Table 3.  Odds of overestimating and underestimating weight status among healthy-weight men and women
inline image

Overweight adults who think they are healthy weight

Food stamp participation and income were associated with failure to recognize overweight status among overweight women (Figure 1). Overweight women who received food stamp benefits (the reference group) were more likely to state they were healthy weight compared with overweight women who were income eligible but did not receive food stamp benefits (OR = 0.55). Further, the ORs for moderate- and high-income overweight women were 0.41 and 0.24, respectively, suggesting that recognition of overweight status rose with income. Among overweight men, body weight misperception did not vary by food stamp participation or income (Table 2).

Figure 1.

Probability that an overweight woman perceives herself as healthy weight or underweight: Odds ratios and 95% confidence intervals by Food Stamp and Income Eligibility Status. Odds ratios and 95% confidence intervals estimated from logit models including controls for age, smoking status, education, marital status, race/ethnicity and the difference between measured BMI and the overweight cutoff of 25.

Among overweight women, non-Hispanic blacks and Mexican Americans were less likely to recognize they were overweight compared with non-Hispanic whites (ORs are 3.88 and 2.24, respectively) (Figure 2). The findings, consistent with previous studies (2,3,4,5,9), suggest that non-Hispanic black women and Mexican-American women perceived a higher BMI cutoff for “overweight,” compared with the BMI non-Hispanic white women perceived as “overweight”. Similarly, non-Hispanic black men appeared to have a heavier threshold for “overweight” status relative to non-Hispanic white men (Figure 2). Non-Hispanic black men were more than three times as likely relative to white men to fail to recognize they were overweight (OR = 3.11).

Figure 2.

Probability that an overweight man or woman perceives himself/herself as healthy weight or underweight: Odds ratios and 95% confidence intervals by race/ethnicity and gender. Odds ratios and 95% confidence intervals estimated from logit models including controls for age, smoking status, education, food stamp participation, income, and the difference between measured BMI and the overweight cutoff of 25.

Among overweight men and women, those with postsecondary education were less likely to perceive themselves as healthy weight compared with those who did not finish high school (OR = 0.63 for men and 0.53 for women, respectively). Overweight women with high school diplomas were also less likely to incorrectly believe they were healthy weight (OR = 0.56), compared with overweight women who do not have a high school diploma. For men, there was no difference in misperception of overweight status between those with a high school diploma and those without.

Body weight misperception also varied by age among overweight women, but not among overweight men. Older women were slightly more likely to misperceive their body weight status, compared with younger women—that is, the BMI threshold for what is perceived to be overweight rose with age. Failure to recognize overweight did not differ by smoking behavior or by martial status among overweight adults.

Healthy-weight adults who think they are underweight

None of the covariates except for BMI explained variations in the probability that healthy-weight men misperceived their weight as “underweight”, so we focused our attention on healthy-weight women. Here we found that racial and ethnic differences in the propensity to underestimate weight status also existed. Non-Hispanic black and Mexican-American women were more likely (ORs are 3.81 and 3.96, respectively) to perceive themselves as underweight even though they were healthy weight, compared with non-Hispanic white women (Table 3).

Age had an OR of 1.04, suggesting that older healthy-weight women were slightly more likely to state they were underweight than younger healthy-weight women. Relative to nonsmokers, healthy-weight women who smoke were more than two times as likely to believe they were underweight than nonsmokers (OR = 2.41). Women with high school diplomas were less likely to state they were underweight relative to those with less than a high school education, but there were no significant differences between those with less than a high school education and women with some postsecondary education. Food stamp participation was not associated with the likelihood that healthy-weight women underestimate weight status.

Healthy-weight adults who think they are overweight

Racial and ethnic differences were also noticeable among women who were healthy weight but believed they were overweight (Figure 3). The ORs were 0.26, 0.58, and 0.54 for non-Hispanic blacks, Mexican Americans, and other race, respectively, suggesting that non-Hispanic white women were more likely to believe they were overweight than women of other racial/ethnic groups. The OR for age was 0.97, suggesting that older women were less likely to perceive themselves as overweight than younger women. Women with at least some postsecondary schooling were more likely to perceive they were overweight relative to those with less than a high school education (1.69 times more likely), but there was no difference between high school graduates and those with less than a high school education. Food stamp participation, income, and marital status did not explain variations in misperceiving healthy weight as overweight among healthy-weight women (Table 3).

Figure 3.

Probability that a healthy-weight woman perceives herself as overweight: Odds ratios and 95% confidence intervals by race/ethnicity. Odds ratios and 95% confidence intervals estimated from logit models including controls for age, smoking status, education, food stamp participation, income, and the difference between measured BMI and the lower cutoff for health weight of 18.5.

Among healthy-weight men, non-Hispanic black men were much less likely to believe they were overweight relative to non-Hispanic white men (OR = 0.13), indicating that healthy-weight black men had a heavier threshold for overweight status than healthy-weight white men. Among healthy-weight men, those with higher BMI were more likely to perceive themselves as overweight. None of the other social and demographic variables helped explain variations in men's misperception of healthy weight as overweight.

Discussion

Previous work had shown that the determinants of adults' BMI differ by gender (12,13,14). Our study further suggests that misperception of body weight status also varied by gender among both healthy-weight and overweight adults. The findings support the approach to estimate separate models for men and women instead of a pooled model.

We found that overweight women who received food stamp benefits were less likely to recognize they were overweight than eligible nonparticipants and that these differences existed beyond differences in race/ethnicity and education. Some policy analysts are worried that food stamp benefits may actually cause obesity among participants as among some subgroups, food stamp participants have higher rates of obesity than nonparticipants (8). Studies that have attempted to link food stamp participation with weight gain have found no relationship or a very weak relationship between food stamp participation and weight for men and children, but for women, participation in the program seems to increase BMI and the probability of obesity slightly (15,16,17,18). Our results indicate that misperception about body weight was a potential barrier to achieving healthy weight among overweight women who received food stamp benefits, but not among men. It is unclear why female food stamp participants were more likely to misperceive their weight relative to nonparticipants. It could be the result of a “peer effect” if food stamp participants had social connections with other participants. It may also be that those who have accepted, ignored, or were oblivious to any stigma attached to receiving government assistance like food stamps were also less worried about stigma that may be attached to being overweight. Regardless, the higher rates of misperception among food stamp participants may be part of the reason why women who receive food stamps are more likely to be obese.

Food stamp nutrition education (FSNE) is a component of the program that includes provisions for states to provide nutrition education programs to participants, and almost all state agencies implement FSNE programs. FSNE programs are intended to help participants make healthier choices with respect to food purchases, including choices to help combat obesity. Our results indicate that FSNE programs may provide a useful service to participants by helping them recognize what body weight is considered healthy or unhealthy. Such an intervention might include information on calculating and interpreting BMI and on the health risks of overweight. The efficacy of such an intervention would need to be evaluated. If successful, increased attention to BMI and healthy weight may help food stamp participants prevent weight gain or reduce weight.

BMI has been shown to vary by social and demographic factors, including income, race and ethnicity, gender, and age (19). For example, among overweight women in the 1999–2004 NHANES, the average BMI for non-Hispanic blacks and non-Hispanic whites were 33.9 and 32.1, respectively. A strong and consistent finding of our study is that the higher an individual's BMI, the lower the probability of misperceiving overweight as healthy weight. BMI is also correlated with the probability of overestimating and underestimating weight among those who are healthy weight. By including an individual's BMI (deviation from the cutoffs) in the model, we are able to improve the explanatory power of our estimates of the propensity to underestimate or overestimate weight status by social and demographic factors. Our results confirm that even after controlling for BMI, non-Hispanic black women and Mexican-American women are more likely to underestimate their weight status compared with non-Hispanic white women. In other words, non-Hispanic black women and Mexican-American women are more likely to misperceive overweight as healthy weight and to misperceive healthy weight as underweight.

The consistency of these racial/ethnic results across healthy weight and overweight status suggest that misperception of body weight does have a social or cultural component. For example, weight misperception could be a result of a “peer” effect—women base their weight status perceptions on the weights of their peers. It could also be evident that groups attach different values to a particular weight status. White women indicate they feel more pressure to be thin, are more likely to diet, and have smaller ideal body sizes than black women (5,20,21). With smaller ideal body weight and more dieting behavior, white women may feel more pressure to be thinner and thus more likely to report being overweight than black women, who do not have the same thinner body weight ideals.

Like race and ethnicity, age is also a consistent predictor of misperception of weight among women. These results could also suggest that peer effects may be related to misperception of body weight, if women tend to form peer groups with women near the same age as them. But further research is needed.

Such research should consider how proximate peers (both in terms of similar socioeconomic status and demographics and in terms of actual social and neighborhood proximity) affect misperception of weight status, similar to how obesity has recently been linked among peers (22). For example, are overweight adults who live with or near many other overweight adults more likely to underestimate their weight? Research on how connections across and within races, socioeconomic status, or other demographic characteristics affect perception of body weight may also be constructive. For example, does weight misperception for an individual of one race with many connections to individuals of other races or ethnicities differ from those of someone who is the same race but has fewer cross-race connections? Further research could consider these questions.

Acknowledgment

All funding for the authors is from their respective employers. Ver Ploeg and Lin are employed by the Economic Research Service of the US Department of Agriculture. Chang is an assistant professor in the Department of Agricultural Economics at National Taiwan University, Taipei, Taiwan. The views expressed here are those of the authors and may not be attributed to the Economic Research Service or the US Department of Agriculture.

Disclosure

The authors declared no conflict of interest.

Ancillary