Journal of Health Communication, 19:795–812, 2014 Copyright © Taylor & Francis Group, LLC ISSN: 1081-0730 print/1087-0415 online DOI: 10.1080/10810730.2013.864727
How Accurate Are Americans’ Perceptions of Their Own Weight? LINDA SQUIERS, JEANETTE RENAUD, LAUREN MCCORMACK, JANICE TZENG, CARLA BANN, AND PAM WILLIAMS RTI International, Washington, District of Columbia, USA As obesity/overweight has increased in the United States (Centers for Disease Control and Prevention, 2009), studies have found that Americans’ perceptions of their own weight often are not aligned with their actual body mass index (BMI; Brener et al., 2004; Christakis, 2003; Johnson-Taylor et al., 2008). Taylor, Funk, and Craighill (2006) found that half of Americans whose BMI indicated they were overweight perceived their weight to be just about right. The purpose of this study was to examine factors that influence the accuracy of weight self-perceptions and whether accuracy influences health behaviors. Using data from the 2007 Health Information National Trends Survey, the authors compared respondents’ weight self-perceptions to their actual BMI to determine the accuracy of their weight self-perceptions. About 28% of respondents were obese, 35% were overweight, 35% were of normal weight, and 2% were underweight. About three quarters of the sample’s self-perceptions of weight were aligned with their BMI. About 10% of the sample had a BMI that indicated they were overweight, but they perceived themselves to be of normal weight; about 10% were of normal weight but perceived themselves to be overweight; and about 5% of respondents were of normal weight but thought they were underweight. Gender, race, and education were associated with the accuracy of respondents’ weight perceptions. Results suggest that asking patients about their weight self-perceptions could be useful in clinical settings and that weight perception accuracy could be used to segment audiences and tailor messages.
Over the past two decades, the prevalence of obesity and overweight has significantly increased among adults in the United States (Centers for Disease Control and Prevention, 2009). At present, 33.8% of adults in the United States are obese, more than double the 15% classified as obese 20 years ago (Flegal, Carroll, Ogden, & Curtin, 2010; Taylor, Funk, & Craighill, 2006). When overweight and obesity percentages are combined, the age-adjusted prevalence of overweight and obesity is 68.0%, 72.3% among men, and 64.1% among women (Flegal et al., 2010). It is interesting that rates of overweight and obesity significantly differ across racial and ethnic groups (Centers for Disease Control and Prevention, 2009; Flegal et al., 2010). For example, on the basis of the Behavioral Risk Factor Surveillance System surveys conducted between 2006 and 2008, the age-adjusted estimated prevalence of obesity among nonHispanic Blacks was highest (35.7%), followed by Hispanics (28.7%), and non-Hispanic Whites (23.7%) (Centers for Disease Control and Prevention, 2009). Non-Hispanic Black women had the greatest prevalence (39.2%), followed by non-Hispanic Black men (31.6%), Hispanic women (29.4%), Hispanic men (27.8%), non-Hispanic White Address correspondence to Linda Squiers, RTI International, 701 13th Street NW, Suite 750, Washington, DC 20005, USA. E-mail: [email protected]
795
796
L. Squiers et al.
men (25.4%), and non-Hispanic White women (21.8%) (CDC, 2009). Friedman (2011) stated that “recent increases in obesity likely result from the interaction of biologic, social, and cultural factors with an environment characterized by limited opportunities for physical activity and an abundance of high-caloric foods” (p. 75). Lack of access to healthy foods, particularly among neighborhoods with large minority populations, has also been cited as contributing to obesity rates as has lower rates of regular (nonoccupational) physical activity (Friedman, 2011). Cultural norms about body weight may vary among different racial and ethnic groups. Millstein and colleagues (2008) reported that, when compared with White women, African American and Hispanic women were more satisfied with their body size and thus less likely to try to lose weight. This trend is alarming because obesity is a risk factor for a variety of chronic conditions including type 2 diabetes; hypertension and coronary heart diseases; dyslipidemia; stroke; endometrial, breast, and colon cancers; liver and gall bladder disease; sleep apnea and respiratory problems; osteoarthritis; and gynecological problems such as infertility and abnormal menses (Flegal et al., 2010). Increases in chronic conditions lead to higher health care costs (Finkelstein, Fiebelkorn, & Wang, 2003), disability, and death. Obesity contributes to an estimated 112,000 preventable deaths each year (Flegal, Graubard, Williamson, & Gail, 2005). In general, Americans seems to be aware of the growing epidemic of overweight and obesity. For example, in a telephone survey of 2,250 Americans, Taylor and colleagues (2006) found that 85% of respondents agreed that Americans are more overweight than they were 5 years ago. Of these respondents, 67% said that they thought it was a major problem. However, perceptions of overweight prevalence varied by race, with 53% of African Americans reporting that Americans are very overweight compared with 35% of Whites and 33% of Hispanics. It is interesting that this Pew Research Center study also found that respondents were less likely to perceive themselves as overweight in comparison to the weight of other Americans. Only 5% of respondents thought of themselves as being very overweight (the Pew Research Center did not use the term obese in its questionnaire, so this category corresponds most closely to the definition of obese). Thus, a discrepancy seems to exist between the more objective estimates of obesity among adults in the United States and the more subjective perceptions of individual Americans (weight self-perceptions). Although 51% of Pew respondents perceived their weight as “just about right,” their body mass index (BMI), on the basis of self-reports of height and weight, indicated that they are overweight. Other studies (Brener, Eaton, Lowry, & McManus, 2004; Johnson-Taylor, Fisher, Hubbard, Starke-Reed, & Eggers, 2008) also found a misalignment between weight self-perceptions and BMI. For example, Chang and Christakis (2003) found that 27.5% of women and 29.8% of men misclassified their own weight. Furthermore, the direction of the misclassification varied by gender: Men were more likely to underestimate their weight (e.g., consider themselves to be about the right weight or underweight when their BMI indicated they were overweight), and women were more likely to overestimate their weight (e.g., consider themselves to be overweight when they were actually normal weight or underweight). In a survey of 2,032 students in Grades 9 through 12, Brener and colleagues (2004) found that of the students who were overweight according to their BMI, 20.0% perceived themselves to be about the right weight and 26.2% perceived themselves to be underweight. Johnson-Taylor and colleagues (2008) found that the prevalence of distorted weight perceptions has increased over time and was greatest among persons with lower income, men, and African Americans. Although BMI has some limitations as a measure of body weight, these limitations cannot account for all of the discrepancies.
Americans’ Perceptions of Their Own Weight
797
What can account for the number of discrepancies between objective weight measures and subjective weight self-perceptions? Are those who underestimate their weight consciously aware that they are misrepresenting their weight on surveys? If so, how can we explain those who overestimate their weight? And, does it matter? Chandler-Laney and colleagues (2009) suggested that the increasing prevalence of obesity may be skewing social perceptions of “healthy and appropriate” body weight. Providing some support for this notion, Christakis and Fowler (2010) found that weight is influenced by social networks; that is, a person’s chances of becoming obese increase if he or she has a friend, sibling, or spouse who is obese. Thus, individuals tend to evaluate their own weight status with reference to the weight distribution of their social group (Chang & Christakis, 2003; Dawson, 1988). For example, Chang and Christakis (2003) found that BMI, older age, never being married, non-White race/ethnicity, higher income or education, and female gender all increased the odds of respondents misclassifying themselves into a higher/heavier weight perception category. This article examines what factors influence the accuracy of weight self-perceptions and how discrepancies in weight self-perceptions influence weight-related behaviors. Identifying how weight self-perceptions vary among sociodemographic subgroups can help public health professionals better understand the social evolution of the obesity epidemic. In addition, discovering whether and how the accuracy of weight self-perceptions influence weight-related behaviors will help intervention developers understand the types of targeted and/or tailored messages and interventions that influence dietary and physical activity behavior change. Figure 1 displays a conceptual framework for examining factors that have been found to influence weight self-perceptions, their accuracy, and how they influence weight-related behaviors. The framework shows that demographic factors, weight of members within one’s social network, health status and actual weight/BMI, and knowledge and beliefs about obesity and healthy weight influence individuals’ weight self-perceptions (e.g., whether they think of themselves as being underweight, overweight, or being at the right weight for them). In some respects and in this context, weight self-perceptions may be similar to perceived susceptibility, a concept derived
Figure 1. Conceptual framework for understanding the potential effect of weight self-perceptions. BMI = body mass index.
798
L. Squiers et al.
from the Health Belief Model. Weight self-perceptions and their accuracy—that is, perceptions aligned with what the medical community considers to be a healthy weight, overweight or obese—affects individuals’ self-efficacy in taking care of their health, including losing weight (Martin, Dutton, & Brantly, 2004; Shin et al., 2011) and engaging in physical activity (Hagger, Chatzisarantis, & Biddle, 2002). In turn, weight perception accuracy and self-efficacy influence weight loss attempts and physical activity. Experiences with weight loss attempts and physical activity may feedback and influence perceive norms and actual body weight. This study examined what factors influence weight perceptions, the accuracy of weight perceptions, and the influence of weight perception accuracy on weight-related behaviors. We aimed to answer these research questions: 1. How did the public perceive their weight? 2. Were the public’s perceptions of their weight consistent with their BMI? What factors affect the degree to which they are consistent? 3. Did weight self-perceptions influence self-efficacy and weight-related behaviors (e.g., physical activity, weight loss attempts) that lead to the attainment and maintenance of healthy weight?
Method Data Source We used data from the 2007 Health Information National Trends Survey, a nationally representative biennial survey developed by the National Cancer Institute to better understand how health and cancer-related information is communicated to the public and how the public, in turn, uses this information. For the 2007 administration, the Health Information National Trends Survey included two modes of data collection: (a) random digit dialing (N = 4,092), using a computer-assisted telephone interview, and (b) postal mail using a pencil-and-paper questionnaire (N = 3,582). Response rates were 24% for the random digit dialing mode and 31% for the postal mail mode (Cantor et al., 2009). Study Variables BMI To assess body weight, respondents were asked how tall they are and how much they weigh without shoes. BMI, a widely used indicator of healthy or unhealthy body weight, was calculated using the following formula: (mass (1b) × 703)/(height (in))2 Weight Perception To assess perceptions of their weight, respondents were asked, “Right now do you feel you are (1) overweight, (2) slightly overweight, (3) underweight, (4) slightly underweight, and (5) just about the right weight for you?” Weight Perception Accuracy Accuracy of respondents’ weight perceptions was calculated by comparing their BMI with their responses to the weight perception item. Respondents were categorized as accurate estimators if their weight self-perceptions matched their BMI, underestimators
Americans’ Perceptions of Their Own Weight
799
if they perceived their weight to be less than their BMI, and overestimators if they perceived their weight to be more than their BMI. Specifically, those with a normal BMI (18.5–24.9) were categorized as accurate if they perceived themselves to be just right, as normal-weight overestimators if they perceived themselves to be slightly overweight or overweight, and normal-weight underestimators if they perceived themselves to be slightly underweight or underweight. Last, those whose BMI was >25 and who perceived themselves to be slightly overweight or overweight were categorized as accurate estimators, while those who perceived themselves to be just right, slightly underweight, or underweight as overweight underestimators. Classifications can be found in Table 1. Self-Efficacy for Taking Care of One’s Health Respondents answered the question, “Overall, how confident are you about your ability to take good care of your health?” using a 5-point Likert scale ranging from 1 (completely confident) to 5 (not at all confident). Knowledge of and Beliefs About Obesity The following three items were related to obesity: 1. “To what extent do you believe that obesity is caused by overeating and not exercising?” 2. “To what extent do you believe that obesity is inherited?” Participants responded on a 4-point scale ranging from 1 (a lot) to 4 (not at all). 3. “There are so many different messages about whether being overweight is harmful to one’s health it is hard to know what weight one should maintain to be healthy.” Participants responded on a 4-point scale ranging from 1 (strongly agree) to 4 (strongly disagree). Health Status We assessed general health status using the item, “In general, would you say your health is excellent, very good, good, fair, or poor?” Weight-Related Behaviors Weight-related behaviors focused on physical activity and weight loss attempts. We assessed physical activity using two items: “During the past month, did you participate in any physical activities or exercises such as running, yoga, golf, gardening, or walking for exercise?” (yes/no), and “In a typical week, how many days do you do any physical activity or exercise of at least moderate intensity, such as brisk walking, bicycling Table 1. Classification of accuracy of weight self-perceptions Respondents’ weight self-perceptions Body mass index
Slightly underweight or underweight
18.5–24.9
Normal-weight underestimator Overweight underestimator
>25
Just right Accurate Overweight underestimator
Slightly overweight or overweight Normal-weight overestimator Accurate
800
L. Squiers et al.
at a regular pace, swimming at a regular pace, and heavy gardening?” Respondents were also asked to indicate whether they had “tried to lose any weight in the past 12 months.” Demographic Characteristics Respondents reported their age, gender, race/ethnicity, education, household income, marital status, and occupational status. Race/ethnicity was recoded in the following four categories: White, Black or African American, Hispanic, and other (American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races). Data Analysis To accommodate the multistage sampling design of the Health Information National Trends Survey, analyses were conducted using Stata (StataCorp, 2009). Responses of refuse or don’t know were recoded as missing for all analyses. Where necessary, items were recoded so that higher numbers indicated stronger agreement. To assess potential differences in responses due to mode of data collection (phone vs. mail), we tested for mode effects in the predictor variables of interest. Crosstab analyses were conducted to assess bivariate relations between weight perception accuracy and demographics. Because the sample size for respondents who were underweight was small, underweight respondents were excluded from multivariate analyses. Multivariate analyses included sample mode as a covariate to control for effects due to mode. We conducted a series of logistic regression analyses to assess the relationships between weight perception accuracy and health outcomes. Logistic regression was used for categorical outcomes and linear regression was used for continuous outcomes. Household income was significantly correlated with education, marital status, and occupation status. Thus, to avoid multicollinearity, household income was used in lieu of these other potential covariates.
Results We found a significant mode effect for our primary variables of interest: weight perception and weight perception accuracy. Because mode effects were detected in the main outcomes of interest, frequencies and bivariate results are presented by mode. Sample Table 2 provides sample characteristics. About two thirds of respondents were between 18 and 49 years old, about half of the respondents were male, and about 70% were White. Age, gender, and race/ethnicity did not differ across modes. However, differences by mode were found for education and income. A greater proportion of respondents had completed some college in the mail sample (36.80%) than in the phone sample (31.85%). In addition, a greater proportion of respondents made less than $20,000 in the mail sample (21.14%) than in the phone sample (16.26%). BMI BMI did not vary significantly by mode. Of the sample, 35% had a normal BMI, about 35% had a BMI that indicated they were overweight, and about 28% were obese. Only about 2% were underweight.
Americans’ Perceptions of Their Own Weight
801
Table 2. Sample characteristics, by data collection mode Mail sample Characteristic Age (years) 18–34 35–49 50–64 65–74 75+ Gender Male Female Race/Ethnicity White Black/African American Hispanic Othera Education Less than high school High school graduate Some college College graduate Income 25: Overweight or obese (n = 4,545)
Total sample (N = 7,089)
Underweighta
13.10 (n = 242) Underestimator 58.95 (n = 1,500) Accurate 27.96 (n = 802) Overestimator
0.78 (n = 30) Underestimator 14.99 (n = 577) Underestimator 84.23 (n = 3, 938) Accurate
5.22 (n = 272)
Just right Overweightb
30.82 (n = 2,077) 63.96 (n = 4,740)
Note. Percentages are weighted, while sample size is unweighted for ease of interpretation. aCombines “slightly underweight” and “underweight” responses. Because of the relatively small number of respondents, this group was excluded from multivariate analyses. bCombines “slightly overweight” and “overweight” responses.
50–64 years of age had the highest proportion of accurate estimators (81.10%). Normal-weight and overweight men (8.68% and 14.26%) were significantly more likely than were normal-weight and overweight women (3.60 and 4.11%) to be underestimators, and normal-weight women (18.75%) were significantly more likely than were normal-weight men (4.35%) to be overestimators. Overweight Blacks/African Americans (16.48%) were significantly more likely to underestimate their weight than were overweight Whites (8.00%), while normal-weight Whites (12.71%) were more likely than normal-weight Blacks/African Americans (4.20%) to overestimate their weight. Lower levels of education were associated with underestimating weight and higher levels of education were associated with overestimating weight. Somewhat similar patterns were found in the phone sample (Table 5b). Those 65–74 years of age had the highest proportion of accurate estimators (80.19%). Overweight men (17.15%) were significantly more likely than were overweight women (4.90%) to be underestimators, and normal-weight women were significantly more likely than normal-weight men were to be overestimators (2.59% vs. 14.47%). Overweight Blacks/African Americans were significantly more likely to underestimate their weight than were overweight Whites (17.77% vs. 8.89%). Lower levels of education were associated with underestimating weight and higher levels of education were associated with overestimating weight. Logistic Regressions Weight perception accuracy was related to two of the three outcomes (see Table 6). Normal-weight underestimators were less likely to have exercised in the past month and less likely to have tried to lose weight in the past year than were accurate estimators (OR = 0.51, p < .05 and OR = 0.02, p < .001, respectively). Normal weight overestimators, on the other hand, were more likely to have exercised in the past month and more likely to have tried to lose weight in the past year than were accurate estimators (OR = 1.57, p < .01 and OR = 1.26, p < .10, respectively). Overweight underestimators were less likely to have tried to lose weight in the past year than were accurate estimators (OR = 0.39, p < .001). While our primary predictor of interest was weight perception accuracy, the weight- and health-related belief predictors revealed some interesting relationships. Those who were more confident in their ability to take care of their health were more likely to have exercised in the past month (OR = 1.53, p < .001) and exercised moderately
804
L. Squiers et al.
Table 5a. Bivariate relations among weight perception accuracy and demographics among mail sample
Predictor Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/ African American Hispanic Other*
Accurate estimators (95% CI)
Normal-weight underestimators (95% CI)
Normalweight overestimators (95% CI)
Overweight underestimators (95% CI)
65.02a (59.37, 70.27) 76.04b (72.74, 79.05) 81.10b (78.50, 83.45) 78.47b (74.54, 81.94) 64.19a (56.69, 71.05)
11.79a (8.36, 16.38) 3.89b,c (2.20, 6.76) 2.07b (1.32, 3.24) 3.84b,c (2.15, 6.76) 8.50a,c (5.01, 14.08)
12.74 (9.86, 16.30) 11.78 (9.85, 14.02) 9.71 (8.05, 11.66) 10.86 (8.04, 14.53) 12.19 (8.27, 17.60)
10.45 (7.06, 15.21) 8.30 (6.04, 11.30) 7.12a (5.38, 9.37) 6.83a (4.66, 9.90) 15.12b (10.29, 21.67)
72.71 (68.96, 76.16) 73.55 (70.33, 76.53)
8.68a (6.39, 11.69) 3.60b (2.57, 5.01)
4.35a (3.11, 6.06) 18.75b (16.38, 21.37)
14.26a (11.45, 17.63) 4.11b (3.11, 5.41)
74.18 (71.45, 76.73) 71.03 (64.53, 76.78)
5.11 (3.80, 6.83) 8.28 (5.06, 13.27)
12.71a (11.09, 14.54) 4.20b (2.37, 7.33)
8.00a (6.23, 10.22) 16.48b (11.44, 23.17)
70.24 (62.11, 77.27) 70.35 (58.54, 79.95)
7.77 (3.57, 16.08) 11.03 (5.36, 21.34)
10.78 (7.04, 16.17) 15.07a (9.87, 22.33)
11.21a, b (7.09, 17.27) 3.55a,c (1.78, 6.95)
8.13a (4.74, 13.60)
13.67a (9.04, 20.16)
Education Less than 69.55 8.65 (60.97, 76.95) (5.32, 13.78) high school High school 72.85 7.67 graduate (67.10, 77.93) (4.59, 12.52) Some college 75.30 4.93 (71.91, 78.40) (3.11, 7.75) College 72.40 4.77 graduate (69.48, 75.14) (3.13, 7.19) Income Less than 73.60 8.33 $20,000 (67.15, 79.18) (4.97, 13.66) $20,000– 68.38 11.01 $34,999 (60.52, 75.31) (6.31, 18.53)
7.48a 12.01a (5.55, 10.00) (8.87, 16.06) 12.05 7.72 (9.70, 14.87) (5.73, 10.34) 6.18b 16.66b (14.34, 19.28) (4.73, 8.03) 9.12 (6.24, 13.15) 10.08 (7.07, 14.18)
8.94 (6.23, 12.67) 10.52 (7.08, 15.37) (Continued)
Americans’ Perceptions of Their Own Weight
805
Table 5a. (Continued)
Predictor $35,000– $49,999 $50,000– $74,999 $75,000 or more
Accurate estimators (95% CI)
Normal-weight underestimators (95% CI)
Normalweight overestimators (95% CI)
Overweight underestimators (95% CI)
71.90 (65.35, 77.63) 75.21 (69.81, 79.91) 75.19 (71.37, 78.64)
6.96 (3.56, 13.17) 3.17 (1.49, 6.60) 3.59 (2.00, 6.36)
11.67 (8.56, 15.73) 10.85 (8.51, 13.74) 14.03 (11.64, 16.81)
9.47 (5.41, 16.06) 10.77 (7.08, 16.07) 7.20 (5.04, 10.19)
Note. Numbers with different letters within each weight perception accuracy by demographic category are significantly different from one another. *Includes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
more days in a typical week (OR = 1.37, p < .001), but less likely to have tried to lose weight in the past year (0.85, p < .01). The obesity knowledge and beliefs items were related to exercising in the past month and having tried to lose weight in the past year; however, the number of days of moderate exercise had no relationship with these items. Specifically, those who believed that obesity is caused by overeating and not exercising were more likely to have tried to lose weight in the past year (OR = 1.21, p < .01), and somewhat more likely to have exercised in the past month (OR = 1.16, p < .10). Those who believed that obesity is caused by genes were less likely to have exercised in the past month (OR = 0.87, p < .05). In addition, those who more strongly agreed that there are too many messages about weight and that it is difficult to know what is a healthy weight were less likely to have exercised in the past month (OR = 0.87, p < .01), but somewhat more likely to have tried to lose weight in the past year (OR = 1.08, p < .10). Health status was strongly related to all three outcomes. Those who indicated that they are healthier were more likely to have exercised in the past month (OR = 1.53, p < .001) and to have had more days of moderate exercise per week (OR = 1.37, p < .001) than those who indicated that they were less healthy. However, those who indicated they are less healthy were more likely to have tried to lose weight in the past year (OR = 0.85, p < .001).
Discussion On the basis of self-reported BMI data, we found that 35% of the sample were overweight and 28% were obese. Comparing BMI with perceived weight, we found that about three quarters of respondents accurately perceived their weight, similar to rates found by Chang and Christakis (2003). Those who were overweight/obese were most likely to have accurate weight self-perceptions; however, 14% of those who were overweight/obese felt their weight was just right or that they were underweight. These weight self-perceptions may indicate that overweight/obese underestimators should be a high-priority group for possible interventions because their weight may put them at increased risk for health problems and conditions. A profile of underestimators (i.e., those who perceive their weight to be in a lower category than their BMI indicates) shows that underestimators tend to be younger, male, African American or Hispanic, and have lower levels of educational attainment and income, suggesting that social norms may play a role in influencing weight self-perceptions.
806
L. Squiers et al.
Table 5b. Bivariate relations among weight perception accuracy and demographics among phone sample
Predictor Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/ African American Hispanic Other* Education Less than high school High school graduate Some college College graduate Income Less than $20,000 $20,000– $34,999 $35,000– $49,999
Accurate (95% CI)
Normal underestimators (95% CI)
Normal overestimators (95% CI)
Overweight underestimators (95% CI)
77.18 (72.15, 81.54) 76.28 (72.21, 79.93) 78.94 (75.33, 82.14) 80.19 (76.66, 83.30) 72.38 (67.24, 76.98)
3.48 (1.79, 6.66) 2.79 (1.69, 4.56) 2.53 (1.55, 4.12) 3.72 (2.01, 6.79) 5.41 (3.58, 8.09)
7.88 (5.50, 11.16) 9.76 (7.79, 12.16) 9.35 (7.44, 11.68) 6.40 (4.78, 8.51) 5.70 (3.67, 8.76)
11.46 (8.19, 15.82) 11.17 (8.33, 14.82) 9.19a (7.18, 11.69) 9.70 (7.16, 13.00) 16.51b (12.38, 21.68)
76.46 (72.66, 79.88) 78.04 (75.56, 80.34)
3.80 (2.75, 5.22) 2.59 (1.82, 3.67)
2.59a (1.79, 3.74) 14.47b (12.13, 17.17)
17.15a (14.16, 20.62) 4.90b (3.90, 6.15)
78.57 (76.36, 80.62 76.19 (66.96, 83.48)
3.30 (2.40, 4.52) 2.13 (0.91, 4.92)
9.24 (7.89, 10.80) 3.91 (1.68, 8.80)
8.89a (7.57, 10.41) 17.77b (11.11, 27.20)
73.68 (65.42, 80.55) 71.66 (58.48, 81.94)
1.15 (0.40, 3.24) 7.34 (2.91, 17.28)
7.09 (4.37, 11.30) 9.93 (4.61, 20.08)
18.08b (12.43, 25.56) 11.08 (5.39, 21.42)
68.02a (59.99, 75.11)
4.30 (2.23, 8.14)
5.50 (3.20, 9.28)
22.18a (16.05, 29.83)
78.84b (75.22, 82.05) 79.24 (74.76, 83.11) 77.80 (74.50, 80.79)
2.77 (1.76, 4.35) 2.84 (1.74, 4.60) 3.68 (2.27, 5.93)
8.75 (6.75, 11.29) 7.44 (5.60, 9.80) 11.12 (9.02, 13.64)
9.64b (7.05, 13.04) 10.48b (7.72, 14.08) 7.39b (5.62, 9.67)
71.85a (65.73, 77.25) 75.87 (70.99, 80.16) 77.89 (70.55, 83.83)
4.80 (3.13, 7.28) 3.86 (2.44, 6.07) 2.46 (1.05, 5.69)
7.26 (4.90, 10.64) 6.17 (3.79, 9.89) 7.82 (4.77, 12.57)
16.09a (10.99, 22.95) 14.09a (10.73, 18.29) 11.82 (7.23, 18.75) (Continued)
Americans’ Perceptions of Their Own Weight
807
Table 5b. (Continued)
Predictor $50,000– $74,999 $75,000 or more
Accurate (95% CI)
Normal underestimators (95% CI)
Normal overestimators (95% CI)
Overweight underestimators (95% CI)
78.18 (72.18, 83.19) 81.46b (77.99, 84.50)
3.90 (1.78, 8.32) 2.40 (1.17, 4.85)
10.25 (7.02, 14.73) 8.43 (6.51, 10.87)
7.67b (5.77, 10.21) 7.70b (5.77, 10.21)
Note. Numbers with different letters within each weight perception accuracy by demographic category are significantly different from one another. *Includes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
Similarly, the overall sociodemographic profile of overestimators indicates that this group is more likely to be female, classify their race as White or other, and have higher levels of income and education. These findings are consistent with those found by Chang and Christakis (2003), who analyzed data from the National Health and Nutrition Examination Survey and supported the idea that there are different normative beliefs about weight in different racial/ethnic groups and in different socioeconomic status groups. Our results also show that weight perception accuracy can help predict weight loss and physical activity behaviors. This assessment may be a surrogate measure for weight-loss motivation. Weight perception accuracy was related to having exercised in the past month and attempting to lose weight in the last year. Furthermore, consistent with previous literature, demographic variables, such as gender and income, were also related to weight loss and physical activity behaviors. More specifically, women were more likely than men were to have tried to lose weight in the past year; and those with higher income levels were more likely to have exercised in the past month and tried to lose weight in the last year than those in lower income groups. These finding suggest that different norms exist by gender, income, education and race/ethnicity and that, within different socioeconomic status groups, different communication approaches may be needed to help realign perceptions of what a healthy weight and overweight look like. Clinical Issues Results suggest that asking patients about their weight self-perceptions could be useful in clinical settings. Physicians are in an excellent position to help align patients’ weight perceptions with their actual BMI. Before sharing results of a BMI assessment, physicians should ask patients, “Right now do you feel you are (1) overweight, (2) slightly overweight, (3) underweight, (4) slightly underweight, and (5) just about the right weight for you?” Physicians would then be able to factor in patients’ perceptions of their own weight when counseling patients about their weight. For example, for patients who have a BMI of 18.5 or below and who are overestimators (e.g., indicate they perceive themselves to be the right weight, slightly overweight, or overweight), physicians may want to ask additional questions to screen for anorexia. A simple tool could be developed or be added to existing tools such as the Agency for Health Care Research and Quality’s Electronic Preventive Service Selector, by allowing physicians
808
L. Squiers et al.
Table 6. Multivariate logistic regressions assessing the effect of weight perception accuracy on weight-related behaviors
Predictor Mode Mail Phone Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/African American Hispanic Othera Household income Less than $20,000 $20,000–$34,999 $35,000–$49,999 $50,000–$74,999 $75,000 or more Self-efficacy Confidence in ability to take care of health Obesity knowledge and beliefs Obesity caused by overeating and not exercising Obesity caused by genes Too many messages makes it hard to know what is a healthy weight
Exercise in past month (n = 5,769)
Days of moderate exercise per week (n = 3,989)
Odds ratio
Odds ratio
95% CI
95% CI
Tried to lose weight in past 12 months (n = 5,770) Odds ratio
95% CI
Ref. 1.55*** 1.30–1.85
Ref. 1.13
Ref. 0.97–1.33 0.86
Ref. 0.83 0.78† 0.72* 0.56***
0.63–1.10 0.60–1.01 0.54–0.97 0.40–0.76
Ref. 0.83 0.97 0.91 0.88
0.64–1.08 0.77–1.21 0.70–1.20 0.62–1.26
Ref. 0.82*
0.70–0.97
Ref. 0.74***
Ref. 0.63–0.87 2.19*** 1.86–2.58
Ref. 0.86
0.64–1.15
Ref. 0.93
Ref. 0.62–1.39 0.96
0.62** 0.85
0.44–0.87 0.55–1.33
0.81 0.79
0.60–1.10 0.92 0.66–1.28 0.52–1.18 0.49*** 0.33–0.72
Ref.
Ref.
Ref. 1.05 1.03 0.82 0.44***
0.70–1.05 0.80–1.38 0.78–1.35 0.60–1.12 0.31–0.63
0.70–1.32
Ref.
0.90 1.28 1.45* 1.81***
0.69–1.17 0.92–1.79 1.04–2.01 1.30–2.50
0.86 0.81 0.78 0.68**
0.61–1.20 0.58–1.13 0.56–1.08 0.51–0.91
1.02 0.96 1.35† 1.61**
0.75–1.39 0.71–1.29 0.99–1.85 1.23–2.10
1.11†
0.99–1.23
1.22**
1.07–1.39 1.05
0.94–1.18
1.16†
0.98–1.37
1.04
0.86–1.25 1.21**
1.05–1.38
0.87*
0.78–0.98
0.93
0.83–1.04 1.08
0.98–1.18
0.87**
0.79–0.96
1.03
0.97–1.11 1.08†
0.99–1.17
(Continued)
Americans’ Perceptions of Their Own Weight
809
Table 6. (Continued)
Predictor Health status General health Weight accuracy Accurate estimators Normal-weight underestimators
Exercise in past month (n = 5,769)
Days of moderate exercise per week (n = 3,989)
Odds ratio
Odds ratio
95% CI
1.53*** 1.38–1.69 Ref. 0.51*
Normal-weight 1.57** overestimators Overweight 1.15 underestimators
1.37*** Ref.
95% CI
Tried to lose weight in past 12 months (n = 5,770) Odds ratio
1.21–1.55 0.85**
95% CI 0.77–0.94
Ref.
0.30–0.88
1.11
0.63–1.96 0.02*** 0.01–0.08
1.21–2.05
1.06
0.83–1.35 1.26†
0.87–1.54
1.32
0.93–1.87 0.39*** 0.28–0.56
0.98–1.64
Note. Ref = reference category. *p < .05. **p < .01. ***p < .001. †p < .10. aIncludes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
to enter a patient’s evaluation of their own weight using the questions above, as well as height and weight. The tool could calculate not only BMI, but also whether a patient is an under-, over-, or accurate-estimator of their own weight. These categorizations may help physicians tailor messages about weight. For example, for patients with a BMI in the overweight range who underestimate their weight, providers may first want to educate the patient about a healthy weight range and assess the patient’s level of knowledge about the link between overweight/obesity and poor health outcomes. Using these categorizations within a clinical setting may help identify patients have eating disorders or body weight distortions and allow providers to direct patients to appropriate support services. For overweight patients who are accurate estimators, providers could again assess level of knowledge and also assess motivation and barriers to weight-loss behaviors. Public Health Communication Issues Weight perception accuracy can be used in a variety of different applications and settings to target health promotion messages. Weight perception accuracy could be used to segment audiences for larger health promotion efforts. For some audiences, such as overweight underestimators, messages about diet and exercise for weight reduction could potentially fall on deaf ears as many inaccurately perceived their weight to be just right. Messages addressing perceptions of risk (e.g., the link between overweight/ obesity and chronic diseases and conditions) may also be useful for these individuals. Public health communication efforts might first focus on helping audiences know what a healthy weight means and is. Follow-up messages could help align self-perceptions with new knowledge about a healthy weight. While weight perception accuracy could help identify target audiences for public health communication programs, it also could be used to help tailor individualized
810
L. Squiers et al.
consumer health tools. Consumer focused e-health tools could also use an assessment of weight perception accuracy to develop individually-tailored messages designed to improve healthy behaviors. As previously described, when weight perceptions are inaccurate, tools can help align users’ perceptions and perhaps use pictures, messages, quizzes, and games to promote social norms around a healthy weight range. Limitations While the response rates for the mail and telephone surveys were relatively low, with the advent of Caller ID and a movement toward cell-only households, response rates for random digit dialing surveys have been generally falling across all types of survey administrations (Blumberg, Luke, & Cynamon, 2006; Fahimi, Link, Mokdad, Schwartz, & Levy, 2008). Some methodological studies have indicated that the threat from falling random digit dialing response rates may not be as strong for health surveys as originally hypothesized (Fahimi et al., 2008, Gentry et al., 1985; Nelson, PowellGriner, Town, & Kovar, 2003). The Health Information National Trends Survey has tried to deal with the random digit dialing response rates by offering multiple modes. While not all measures are sensitive to mode, the present study indicates that perceived weight does vary by mode. Those interviewed via phone were more likely to report their weight as just right and less likely to perceive themselves as slightly overweight than those who responded by postal mail. This may be due to social desirability concerns. It may be less threatening to indicate that one is overweight on a questionnaire filled out privately than to share that information with another person. Future research should assess the role of social desirability, as well as other potential factors, in differences in self-reported weight perceptions across different modes of data collection. No mode effects were found for BMI; however, BMI has been criticized for being too stringent and for inaccurately estimating fat among different ethnic and racial groups. Studies have found that BMI based on self-reported height and weight can differ significantly from BMI calculated from physical measurements assessed more objectively. Specifically, objective measurements of height and weight tend typically lead to fewer respondents being classified as normal weight and more respondents classified as overweight than when relying upon self- reports of height and weight (Brener et al., 2004). The present study relied on self-reported height and weight to calculate BMI, suggesting that BMI could have been underestimated. Last, the wording of the question that assessed participants’ perceptions of their own weight was not specifically tied to health. As a result, we do not know for certain that participants evaluated their weight in terms of their health. Given the apparent influence of sociocultural norms, another area that may need study is respondents’ beliefs about the accuracy or appropriateness of the medical community’s recommendations regarding weight Future Studies Despite these limitations and challenges, assessing weight perception accuracy has been shown to have important influences on weight-related behavior. Future studies should examine other factors associated with weight perception accuracy, including perceptions of the weight of individuals in social networks, health promotion behaviors, knowledge of the health consequences of being overweight or obese, and motivation to engage in weight-related behaviors. In addition, studies should delve further into discovering cultural norms around body weight, whether there is a perceived relationship between body weight and health, and what factors are considered when developing attitudes about what constitutes a right body weight. For example, some individuals may perceive that a larger body weight is protective (e.g., from being pushed around easily by others or sexual advances) or more attractive. In addition,
Americans’ Perceptions of Their Own Weight
811
studies should examine what individuals and subgroups believe about the medical communities’ recommendations about weight. Do they perceive the guidelines to be accurate, credible, and relevant to them? Examining the relation between motivation for attaining or maintaining a healthy weight and weight perception accuracy may prove useful (it was not possible to conduct this analysis in the present study). For example, behavior change theories suggest that individuals must perceive weight as a problem or a risk before being motivated to lose weight (Ajzen, 1985; Becker, 1974; Rosenstock, 1974), as is illustrated in the current conceptual framework (see Figure 1). Furthermore, it may be important to track weight perception accuracy over time to see if trends in weight perception change in accordance with changes in overweight/obesity rates. Perhaps individuals’ weight self-perceptions and the accuracy of their weight selfperceptions may improve if physicians work with patients to ensure that self-perceptions are aligned and accurate with what is known to be a healthy body weight. Future research could be conducted to help determine whether pointing out misaligned perceptions is beneficial to patients (e.g., is learning this information motivating or demoralizing?) and if, when followed by behavioral counseling or health information, it helps patients engage in behaviors that will help them achieve a healthy weight. Because results from this study suggest that some sociodemographic subgroups have different levels of weight perception accuracy, researchers should investigate what factors influence weight self-perceptions and perceived norms about weight within different cultures, subpopulations, and social networks. Results from these investigations should be used to develop and implement interventions that increase the public’s awareness of what constitutes a healthy weight (Johnson-Taylor et al., 2008).
References Ajzen, I. (1985). From intentions to action: A theory of planned behavior. In J. Kuhl & J. Beckman (Ed.), Action CONTROL: From cognitions to behaviors (pp. 11–39). New York, NY: Springer. Becker, M. H. (Ed.). (1974). The Health Belief Model and personal health behavior. Health Education Monographs, 2, 324–508. Blumberg, S. J., Luke, J. V., & Cynamon, M. L. (2006). Telephone coverage and health survey estimates: Evaluating the need for concern about wireless substitution. American Journal of Public Health, 96, 926–931. Brener, N., Eaton, D., Lowry, R., & McManus, T. (2004). The association between weight perceptions and BMI among high school students. Obesity Research, 12, 1966–1974. Cantor, D., Coa, K., Crystal-Mansour, S., Davis, T., Dipko, S., & Sigman, R. (2009). Health Information National Trends Survey (HINTS) 2007. Rockville, MD: Westat. Centers for Disease Control and Prevention. (2009). Differences in prevalence of obesity among Black, White and Hispanic Adults—United States, 2006–2008. Morbidity and Mortality Weekly Report, 58, 740–744. Chandler-Laney, P., Hunter, G., Ard, J., Roy, J., Brock, D., & Gower, B. (2009). Perceptions of others’ body size influences weight loss and regain for European American but not African American women. Health Psychology, 28, 414–418. Chang, V., & Christakis, N., (2003). Self-perceptions of weight appropriateness in the United States. American Journal of Preventive Medicine, 24, 332–339. Christakis, N., & Fowler, J. (2010). The spread of obesity in a large social network over 32 years. New England Journal of Medicine, 357, 370–379. Dawson, D. A. (1988). Ethnic differences in female overweight: Data from the 1985 National Health Interview Survey. American Journal of Public Health, 78, 1326–1329. Fahimi, M., Link, M., Mokdad, A., Schwartz, D. A., & Levy, P. (2008). Tracking chronic disease and risk behavior prevalence as survey participation declines: Statistics from
812
L. Squiers et al.
the Behavioral Risk Factor Surveillance System and other national surveys. Preventing Chronic Disease, 5(3), A80. Finkelstein, E. A., Fiebelkorn, I. C., & Wang, G. (2003). National medical spending attributable to overweight and obesity: How much, and who’s paying? Health Affairs, W3, 219–226. Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2010). Prevalence and trends in obesity among US adults, 1999–2008. JAMA, 303, 235–241. Flegal, K. M., Graubard, B. I., Williamson, D. F., & Gail, M. H. (2005). Excess Deaths Associated With Underweight, Overweight, and Obesity. JAMA, 293, 1861–1867. Gentry, E., Kalsbeek, W., Hogelin, G., Jones, J., Gaines, K., Forman, M., … Trowbridge, F. (1985). The behavioral risk factor surveillance surveys: II. Design, methods, and estimates from combined state data. American Journal of Preventive Medicine, 1, 9–14. Hagger, M., Chatzisarantis, N., & Biddle, S. (2002). The meta-analytic review of the theories of reasoned action and planned behavior in physical activity: Predictive validity and the contribution of additional variables. Journal of Sport and Exercise Psychology, 24, 3–32. Johnson-Taylor, W., Fisher, R., Hubbard, V., Starke-Reed, P., & Eggers, P. (2008). The change in weight perceptions of weight status among the overweight: Comparison of NHANES III (1988–1994) and 1999–2004 NHANES. International Journal of Behavioral Nutrition and Physical Activity, 5, 9. Martin, P, Dutton, G., & Brantly, P. (2004). Self-efficacy as a predictor of weight change in African-American women. Obesity Research, 12, 646–651. doi:10.1038/oby.2004.74 Millstein, R., Carlson, S., Fulton, J., Galuska, D., Zhang, J., Blanck, H., & Ainsworth, B. (2008). Relationships between body satisfaction and weight control practices among US adults. Medscape Journal of Medicine, 10, 119. Nelson, D. E., Powell-Griner, E., Town, M., & Kovar, M. G. (2003). A comparison of national estimates from the National Health Interview Survey and the Behavioral Risk Factor Surveillance System. American Journal of Public Health, 93, 1335–1341. Rosenstock, I. M. (1974). Historical origins of the health belief model. Health Education Monographs, 2, 328–335. Shin, H., Shin, J., Liu, P. Y., Dutton, G. R., Abood, D. A., & Elich, J. Z. (2011). Self-efficacy improves weight loss in overweight/obese postmenopausal women during a 6-month weight loss intervention. Nursing Research, 31, 822–828. StataCorp. (2009). Stata statistical software: Release 11. College Station, TX: Author. Taylor, P., Funk, C., & Craighill, P. (2006). Americans see weight problems everywhere but in the mirror. Pew Research Center. A Social Trends Report. Washington, DC: Pew Research Center.
How Accurate Are Americans’ Perceptions of Their Own Weight? LINDA SQUIERS, JEANETTE RENAUD, LAUREN MCCORMACK, JANICE TZENG, CARLA BANN, AND PAM WILLIAMS RTI International, Washington, District of Columbia, USA As obesity/overweight has increased in the United States (Centers for Disease Control and Prevention, 2009), studies have found that Americans’ perceptions of their own weight often are not aligned with their actual body mass index (BMI; Brener et al., 2004; Christakis, 2003; Johnson-Taylor et al., 2008). Taylor, Funk, and Craighill (2006) found that half of Americans whose BMI indicated they were overweight perceived their weight to be just about right. The purpose of this study was to examine factors that influence the accuracy of weight self-perceptions and whether accuracy influences health behaviors. Using data from the 2007 Health Information National Trends Survey, the authors compared respondents’ weight self-perceptions to their actual BMI to determine the accuracy of their weight self-perceptions. About 28% of respondents were obese, 35% were overweight, 35% were of normal weight, and 2% were underweight. About three quarters of the sample’s self-perceptions of weight were aligned with their BMI. About 10% of the sample had a BMI that indicated they were overweight, but they perceived themselves to be of normal weight; about 10% were of normal weight but perceived themselves to be overweight; and about 5% of respondents were of normal weight but thought they were underweight. Gender, race, and education were associated with the accuracy of respondents’ weight perceptions. Results suggest that asking patients about their weight self-perceptions could be useful in clinical settings and that weight perception accuracy could be used to segment audiences and tailor messages.
Over the past two decades, the prevalence of obesity and overweight has significantly increased among adults in the United States (Centers for Disease Control and Prevention, 2009). At present, 33.8% of adults in the United States are obese, more than double the 15% classified as obese 20 years ago (Flegal, Carroll, Ogden, & Curtin, 2010; Taylor, Funk, & Craighill, 2006). When overweight and obesity percentages are combined, the age-adjusted prevalence of overweight and obesity is 68.0%, 72.3% among men, and 64.1% among women (Flegal et al., 2010). It is interesting that rates of overweight and obesity significantly differ across racial and ethnic groups (Centers for Disease Control and Prevention, 2009; Flegal et al., 2010). For example, on the basis of the Behavioral Risk Factor Surveillance System surveys conducted between 2006 and 2008, the age-adjusted estimated prevalence of obesity among nonHispanic Blacks was highest (35.7%), followed by Hispanics (28.7%), and non-Hispanic Whites (23.7%) (Centers for Disease Control and Prevention, 2009). Non-Hispanic Black women had the greatest prevalence (39.2%), followed by non-Hispanic Black men (31.6%), Hispanic women (29.4%), Hispanic men (27.8%), non-Hispanic White Address correspondence to Linda Squiers, RTI International, 701 13th Street NW, Suite 750, Washington, DC 20005, USA. E-mail: [email protected]
795
796
L. Squiers et al.
men (25.4%), and non-Hispanic White women (21.8%) (CDC, 2009). Friedman (2011) stated that “recent increases in obesity likely result from the interaction of biologic, social, and cultural factors with an environment characterized by limited opportunities for physical activity and an abundance of high-caloric foods” (p. 75). Lack of access to healthy foods, particularly among neighborhoods with large minority populations, has also been cited as contributing to obesity rates as has lower rates of regular (nonoccupational) physical activity (Friedman, 2011). Cultural norms about body weight may vary among different racial and ethnic groups. Millstein and colleagues (2008) reported that, when compared with White women, African American and Hispanic women were more satisfied with their body size and thus less likely to try to lose weight. This trend is alarming because obesity is a risk factor for a variety of chronic conditions including type 2 diabetes; hypertension and coronary heart diseases; dyslipidemia; stroke; endometrial, breast, and colon cancers; liver and gall bladder disease; sleep apnea and respiratory problems; osteoarthritis; and gynecological problems such as infertility and abnormal menses (Flegal et al., 2010). Increases in chronic conditions lead to higher health care costs (Finkelstein, Fiebelkorn, & Wang, 2003), disability, and death. Obesity contributes to an estimated 112,000 preventable deaths each year (Flegal, Graubard, Williamson, & Gail, 2005). In general, Americans seems to be aware of the growing epidemic of overweight and obesity. For example, in a telephone survey of 2,250 Americans, Taylor and colleagues (2006) found that 85% of respondents agreed that Americans are more overweight than they were 5 years ago. Of these respondents, 67% said that they thought it was a major problem. However, perceptions of overweight prevalence varied by race, with 53% of African Americans reporting that Americans are very overweight compared with 35% of Whites and 33% of Hispanics. It is interesting that this Pew Research Center study also found that respondents were less likely to perceive themselves as overweight in comparison to the weight of other Americans. Only 5% of respondents thought of themselves as being very overweight (the Pew Research Center did not use the term obese in its questionnaire, so this category corresponds most closely to the definition of obese). Thus, a discrepancy seems to exist between the more objective estimates of obesity among adults in the United States and the more subjective perceptions of individual Americans (weight self-perceptions). Although 51% of Pew respondents perceived their weight as “just about right,” their body mass index (BMI), on the basis of self-reports of height and weight, indicated that they are overweight. Other studies (Brener, Eaton, Lowry, & McManus, 2004; Johnson-Taylor, Fisher, Hubbard, Starke-Reed, & Eggers, 2008) also found a misalignment between weight self-perceptions and BMI. For example, Chang and Christakis (2003) found that 27.5% of women and 29.8% of men misclassified their own weight. Furthermore, the direction of the misclassification varied by gender: Men were more likely to underestimate their weight (e.g., consider themselves to be about the right weight or underweight when their BMI indicated they were overweight), and women were more likely to overestimate their weight (e.g., consider themselves to be overweight when they were actually normal weight or underweight). In a survey of 2,032 students in Grades 9 through 12, Brener and colleagues (2004) found that of the students who were overweight according to their BMI, 20.0% perceived themselves to be about the right weight and 26.2% perceived themselves to be underweight. Johnson-Taylor and colleagues (2008) found that the prevalence of distorted weight perceptions has increased over time and was greatest among persons with lower income, men, and African Americans. Although BMI has some limitations as a measure of body weight, these limitations cannot account for all of the discrepancies.
Americans’ Perceptions of Their Own Weight
797
What can account for the number of discrepancies between objective weight measures and subjective weight self-perceptions? Are those who underestimate their weight consciously aware that they are misrepresenting their weight on surveys? If so, how can we explain those who overestimate their weight? And, does it matter? Chandler-Laney and colleagues (2009) suggested that the increasing prevalence of obesity may be skewing social perceptions of “healthy and appropriate” body weight. Providing some support for this notion, Christakis and Fowler (2010) found that weight is influenced by social networks; that is, a person’s chances of becoming obese increase if he or she has a friend, sibling, or spouse who is obese. Thus, individuals tend to evaluate their own weight status with reference to the weight distribution of their social group (Chang & Christakis, 2003; Dawson, 1988). For example, Chang and Christakis (2003) found that BMI, older age, never being married, non-White race/ethnicity, higher income or education, and female gender all increased the odds of respondents misclassifying themselves into a higher/heavier weight perception category. This article examines what factors influence the accuracy of weight self-perceptions and how discrepancies in weight self-perceptions influence weight-related behaviors. Identifying how weight self-perceptions vary among sociodemographic subgroups can help public health professionals better understand the social evolution of the obesity epidemic. In addition, discovering whether and how the accuracy of weight self-perceptions influence weight-related behaviors will help intervention developers understand the types of targeted and/or tailored messages and interventions that influence dietary and physical activity behavior change. Figure 1 displays a conceptual framework for examining factors that have been found to influence weight self-perceptions, their accuracy, and how they influence weight-related behaviors. The framework shows that demographic factors, weight of members within one’s social network, health status and actual weight/BMI, and knowledge and beliefs about obesity and healthy weight influence individuals’ weight self-perceptions (e.g., whether they think of themselves as being underweight, overweight, or being at the right weight for them). In some respects and in this context, weight self-perceptions may be similar to perceived susceptibility, a concept derived
Figure 1. Conceptual framework for understanding the potential effect of weight self-perceptions. BMI = body mass index.
798
L. Squiers et al.
from the Health Belief Model. Weight self-perceptions and their accuracy—that is, perceptions aligned with what the medical community considers to be a healthy weight, overweight or obese—affects individuals’ self-efficacy in taking care of their health, including losing weight (Martin, Dutton, & Brantly, 2004; Shin et al., 2011) and engaging in physical activity (Hagger, Chatzisarantis, & Biddle, 2002). In turn, weight perception accuracy and self-efficacy influence weight loss attempts and physical activity. Experiences with weight loss attempts and physical activity may feedback and influence perceive norms and actual body weight. This study examined what factors influence weight perceptions, the accuracy of weight perceptions, and the influence of weight perception accuracy on weight-related behaviors. We aimed to answer these research questions: 1. How did the public perceive their weight? 2. Were the public’s perceptions of their weight consistent with their BMI? What factors affect the degree to which they are consistent? 3. Did weight self-perceptions influence self-efficacy and weight-related behaviors (e.g., physical activity, weight loss attempts) that lead to the attainment and maintenance of healthy weight?
Method Data Source We used data from the 2007 Health Information National Trends Survey, a nationally representative biennial survey developed by the National Cancer Institute to better understand how health and cancer-related information is communicated to the public and how the public, in turn, uses this information. For the 2007 administration, the Health Information National Trends Survey included two modes of data collection: (a) random digit dialing (N = 4,092), using a computer-assisted telephone interview, and (b) postal mail using a pencil-and-paper questionnaire (N = 3,582). Response rates were 24% for the random digit dialing mode and 31% for the postal mail mode (Cantor et al., 2009). Study Variables BMI To assess body weight, respondents were asked how tall they are and how much they weigh without shoes. BMI, a widely used indicator of healthy or unhealthy body weight, was calculated using the following formula: (mass (1b) × 703)/(height (in))2 Weight Perception To assess perceptions of their weight, respondents were asked, “Right now do you feel you are (1) overweight, (2) slightly overweight, (3) underweight, (4) slightly underweight, and (5) just about the right weight for you?” Weight Perception Accuracy Accuracy of respondents’ weight perceptions was calculated by comparing their BMI with their responses to the weight perception item. Respondents were categorized as accurate estimators if their weight self-perceptions matched their BMI, underestimators
Americans’ Perceptions of Their Own Weight
799
if they perceived their weight to be less than their BMI, and overestimators if they perceived their weight to be more than their BMI. Specifically, those with a normal BMI (18.5–24.9) were categorized as accurate if they perceived themselves to be just right, as normal-weight overestimators if they perceived themselves to be slightly overweight or overweight, and normal-weight underestimators if they perceived themselves to be slightly underweight or underweight. Last, those whose BMI was >25 and who perceived themselves to be slightly overweight or overweight were categorized as accurate estimators, while those who perceived themselves to be just right, slightly underweight, or underweight as overweight underestimators. Classifications can be found in Table 1. Self-Efficacy for Taking Care of One’s Health Respondents answered the question, “Overall, how confident are you about your ability to take good care of your health?” using a 5-point Likert scale ranging from 1 (completely confident) to 5 (not at all confident). Knowledge of and Beliefs About Obesity The following three items were related to obesity: 1. “To what extent do you believe that obesity is caused by overeating and not exercising?” 2. “To what extent do you believe that obesity is inherited?” Participants responded on a 4-point scale ranging from 1 (a lot) to 4 (not at all). 3. “There are so many different messages about whether being overweight is harmful to one’s health it is hard to know what weight one should maintain to be healthy.” Participants responded on a 4-point scale ranging from 1 (strongly agree) to 4 (strongly disagree). Health Status We assessed general health status using the item, “In general, would you say your health is excellent, very good, good, fair, or poor?” Weight-Related Behaviors Weight-related behaviors focused on physical activity and weight loss attempts. We assessed physical activity using two items: “During the past month, did you participate in any physical activities or exercises such as running, yoga, golf, gardening, or walking for exercise?” (yes/no), and “In a typical week, how many days do you do any physical activity or exercise of at least moderate intensity, such as brisk walking, bicycling Table 1. Classification of accuracy of weight self-perceptions Respondents’ weight self-perceptions Body mass index
Slightly underweight or underweight
18.5–24.9
Normal-weight underestimator Overweight underestimator
>25
Just right Accurate Overweight underestimator
Slightly overweight or overweight Normal-weight overestimator Accurate
800
L. Squiers et al.
at a regular pace, swimming at a regular pace, and heavy gardening?” Respondents were also asked to indicate whether they had “tried to lose any weight in the past 12 months.” Demographic Characteristics Respondents reported their age, gender, race/ethnicity, education, household income, marital status, and occupational status. Race/ethnicity was recoded in the following four categories: White, Black or African American, Hispanic, and other (American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races). Data Analysis To accommodate the multistage sampling design of the Health Information National Trends Survey, analyses were conducted using Stata (StataCorp, 2009). Responses of refuse or don’t know were recoded as missing for all analyses. Where necessary, items were recoded so that higher numbers indicated stronger agreement. To assess potential differences in responses due to mode of data collection (phone vs. mail), we tested for mode effects in the predictor variables of interest. Crosstab analyses were conducted to assess bivariate relations between weight perception accuracy and demographics. Because the sample size for respondents who were underweight was small, underweight respondents were excluded from multivariate analyses. Multivariate analyses included sample mode as a covariate to control for effects due to mode. We conducted a series of logistic regression analyses to assess the relationships between weight perception accuracy and health outcomes. Logistic regression was used for categorical outcomes and linear regression was used for continuous outcomes. Household income was significantly correlated with education, marital status, and occupation status. Thus, to avoid multicollinearity, household income was used in lieu of these other potential covariates.
Results We found a significant mode effect for our primary variables of interest: weight perception and weight perception accuracy. Because mode effects were detected in the main outcomes of interest, frequencies and bivariate results are presented by mode. Sample Table 2 provides sample characteristics. About two thirds of respondents were between 18 and 49 years old, about half of the respondents were male, and about 70% were White. Age, gender, and race/ethnicity did not differ across modes. However, differences by mode were found for education and income. A greater proportion of respondents had completed some college in the mail sample (36.80%) than in the phone sample (31.85%). In addition, a greater proportion of respondents made less than $20,000 in the mail sample (21.14%) than in the phone sample (16.26%). BMI BMI did not vary significantly by mode. Of the sample, 35% had a normal BMI, about 35% had a BMI that indicated they were overweight, and about 28% were obese. Only about 2% were underweight.
Americans’ Perceptions of Their Own Weight
801
Table 2. Sample characteristics, by data collection mode Mail sample Characteristic Age (years) 18–34 35–49 50–64 65–74 75+ Gender Male Female Race/Ethnicity White Black/African American Hispanic Othera Education Less than high school High school graduate Some college College graduate Income 25: Overweight or obese (n = 4,545)
Total sample (N = 7,089)
Underweighta
13.10 (n = 242) Underestimator 58.95 (n = 1,500) Accurate 27.96 (n = 802) Overestimator
0.78 (n = 30) Underestimator 14.99 (n = 577) Underestimator 84.23 (n = 3, 938) Accurate
5.22 (n = 272)
Just right Overweightb
30.82 (n = 2,077) 63.96 (n = 4,740)
Note. Percentages are weighted, while sample size is unweighted for ease of interpretation. aCombines “slightly underweight” and “underweight” responses. Because of the relatively small number of respondents, this group was excluded from multivariate analyses. bCombines “slightly overweight” and “overweight” responses.
50–64 years of age had the highest proportion of accurate estimators (81.10%). Normal-weight and overweight men (8.68% and 14.26%) were significantly more likely than were normal-weight and overweight women (3.60 and 4.11%) to be underestimators, and normal-weight women (18.75%) were significantly more likely than were normal-weight men (4.35%) to be overestimators. Overweight Blacks/African Americans (16.48%) were significantly more likely to underestimate their weight than were overweight Whites (8.00%), while normal-weight Whites (12.71%) were more likely than normal-weight Blacks/African Americans (4.20%) to overestimate their weight. Lower levels of education were associated with underestimating weight and higher levels of education were associated with overestimating weight. Somewhat similar patterns were found in the phone sample (Table 5b). Those 65–74 years of age had the highest proportion of accurate estimators (80.19%). Overweight men (17.15%) were significantly more likely than were overweight women (4.90%) to be underestimators, and normal-weight women were significantly more likely than normal-weight men were to be overestimators (2.59% vs. 14.47%). Overweight Blacks/African Americans were significantly more likely to underestimate their weight than were overweight Whites (17.77% vs. 8.89%). Lower levels of education were associated with underestimating weight and higher levels of education were associated with overestimating weight. Logistic Regressions Weight perception accuracy was related to two of the three outcomes (see Table 6). Normal-weight underestimators were less likely to have exercised in the past month and less likely to have tried to lose weight in the past year than were accurate estimators (OR = 0.51, p < .05 and OR = 0.02, p < .001, respectively). Normal weight overestimators, on the other hand, were more likely to have exercised in the past month and more likely to have tried to lose weight in the past year than were accurate estimators (OR = 1.57, p < .01 and OR = 1.26, p < .10, respectively). Overweight underestimators were less likely to have tried to lose weight in the past year than were accurate estimators (OR = 0.39, p < .001). While our primary predictor of interest was weight perception accuracy, the weight- and health-related belief predictors revealed some interesting relationships. Those who were more confident in their ability to take care of their health were more likely to have exercised in the past month (OR = 1.53, p < .001) and exercised moderately
804
L. Squiers et al.
Table 5a. Bivariate relations among weight perception accuracy and demographics among mail sample
Predictor Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/ African American Hispanic Other*
Accurate estimators (95% CI)
Normal-weight underestimators (95% CI)
Normalweight overestimators (95% CI)
Overweight underestimators (95% CI)
65.02a (59.37, 70.27) 76.04b (72.74, 79.05) 81.10b (78.50, 83.45) 78.47b (74.54, 81.94) 64.19a (56.69, 71.05)
11.79a (8.36, 16.38) 3.89b,c (2.20, 6.76) 2.07b (1.32, 3.24) 3.84b,c (2.15, 6.76) 8.50a,c (5.01, 14.08)
12.74 (9.86, 16.30) 11.78 (9.85, 14.02) 9.71 (8.05, 11.66) 10.86 (8.04, 14.53) 12.19 (8.27, 17.60)
10.45 (7.06, 15.21) 8.30 (6.04, 11.30) 7.12a (5.38, 9.37) 6.83a (4.66, 9.90) 15.12b (10.29, 21.67)
72.71 (68.96, 76.16) 73.55 (70.33, 76.53)
8.68a (6.39, 11.69) 3.60b (2.57, 5.01)
4.35a (3.11, 6.06) 18.75b (16.38, 21.37)
14.26a (11.45, 17.63) 4.11b (3.11, 5.41)
74.18 (71.45, 76.73) 71.03 (64.53, 76.78)
5.11 (3.80, 6.83) 8.28 (5.06, 13.27)
12.71a (11.09, 14.54) 4.20b (2.37, 7.33)
8.00a (6.23, 10.22) 16.48b (11.44, 23.17)
70.24 (62.11, 77.27) 70.35 (58.54, 79.95)
7.77 (3.57, 16.08) 11.03 (5.36, 21.34)
10.78 (7.04, 16.17) 15.07a (9.87, 22.33)
11.21a, b (7.09, 17.27) 3.55a,c (1.78, 6.95)
8.13a (4.74, 13.60)
13.67a (9.04, 20.16)
Education Less than 69.55 8.65 (60.97, 76.95) (5.32, 13.78) high school High school 72.85 7.67 graduate (67.10, 77.93) (4.59, 12.52) Some college 75.30 4.93 (71.91, 78.40) (3.11, 7.75) College 72.40 4.77 graduate (69.48, 75.14) (3.13, 7.19) Income Less than 73.60 8.33 $20,000 (67.15, 79.18) (4.97, 13.66) $20,000– 68.38 11.01 $34,999 (60.52, 75.31) (6.31, 18.53)
7.48a 12.01a (5.55, 10.00) (8.87, 16.06) 12.05 7.72 (9.70, 14.87) (5.73, 10.34) 6.18b 16.66b (14.34, 19.28) (4.73, 8.03) 9.12 (6.24, 13.15) 10.08 (7.07, 14.18)
8.94 (6.23, 12.67) 10.52 (7.08, 15.37) (Continued)
Americans’ Perceptions of Their Own Weight
805
Table 5a. (Continued)
Predictor $35,000– $49,999 $50,000– $74,999 $75,000 or more
Accurate estimators (95% CI)
Normal-weight underestimators (95% CI)
Normalweight overestimators (95% CI)
Overweight underestimators (95% CI)
71.90 (65.35, 77.63) 75.21 (69.81, 79.91) 75.19 (71.37, 78.64)
6.96 (3.56, 13.17) 3.17 (1.49, 6.60) 3.59 (2.00, 6.36)
11.67 (8.56, 15.73) 10.85 (8.51, 13.74) 14.03 (11.64, 16.81)
9.47 (5.41, 16.06) 10.77 (7.08, 16.07) 7.20 (5.04, 10.19)
Note. Numbers with different letters within each weight perception accuracy by demographic category are significantly different from one another. *Includes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
more days in a typical week (OR = 1.37, p < .001), but less likely to have tried to lose weight in the past year (0.85, p < .01). The obesity knowledge and beliefs items were related to exercising in the past month and having tried to lose weight in the past year; however, the number of days of moderate exercise had no relationship with these items. Specifically, those who believed that obesity is caused by overeating and not exercising were more likely to have tried to lose weight in the past year (OR = 1.21, p < .01), and somewhat more likely to have exercised in the past month (OR = 1.16, p < .10). Those who believed that obesity is caused by genes were less likely to have exercised in the past month (OR = 0.87, p < .05). In addition, those who more strongly agreed that there are too many messages about weight and that it is difficult to know what is a healthy weight were less likely to have exercised in the past month (OR = 0.87, p < .01), but somewhat more likely to have tried to lose weight in the past year (OR = 1.08, p < .10). Health status was strongly related to all three outcomes. Those who indicated that they are healthier were more likely to have exercised in the past month (OR = 1.53, p < .001) and to have had more days of moderate exercise per week (OR = 1.37, p < .001) than those who indicated that they were less healthy. However, those who indicated they are less healthy were more likely to have tried to lose weight in the past year (OR = 0.85, p < .001).
Discussion On the basis of self-reported BMI data, we found that 35% of the sample were overweight and 28% were obese. Comparing BMI with perceived weight, we found that about three quarters of respondents accurately perceived their weight, similar to rates found by Chang and Christakis (2003). Those who were overweight/obese were most likely to have accurate weight self-perceptions; however, 14% of those who were overweight/obese felt their weight was just right or that they were underweight. These weight self-perceptions may indicate that overweight/obese underestimators should be a high-priority group for possible interventions because their weight may put them at increased risk for health problems and conditions. A profile of underestimators (i.e., those who perceive their weight to be in a lower category than their BMI indicates) shows that underestimators tend to be younger, male, African American or Hispanic, and have lower levels of educational attainment and income, suggesting that social norms may play a role in influencing weight self-perceptions.
806
L. Squiers et al.
Table 5b. Bivariate relations among weight perception accuracy and demographics among phone sample
Predictor Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/ African American Hispanic Other* Education Less than high school High school graduate Some college College graduate Income Less than $20,000 $20,000– $34,999 $35,000– $49,999
Accurate (95% CI)
Normal underestimators (95% CI)
Normal overestimators (95% CI)
Overweight underestimators (95% CI)
77.18 (72.15, 81.54) 76.28 (72.21, 79.93) 78.94 (75.33, 82.14) 80.19 (76.66, 83.30) 72.38 (67.24, 76.98)
3.48 (1.79, 6.66) 2.79 (1.69, 4.56) 2.53 (1.55, 4.12) 3.72 (2.01, 6.79) 5.41 (3.58, 8.09)
7.88 (5.50, 11.16) 9.76 (7.79, 12.16) 9.35 (7.44, 11.68) 6.40 (4.78, 8.51) 5.70 (3.67, 8.76)
11.46 (8.19, 15.82) 11.17 (8.33, 14.82) 9.19a (7.18, 11.69) 9.70 (7.16, 13.00) 16.51b (12.38, 21.68)
76.46 (72.66, 79.88) 78.04 (75.56, 80.34)
3.80 (2.75, 5.22) 2.59 (1.82, 3.67)
2.59a (1.79, 3.74) 14.47b (12.13, 17.17)
17.15a (14.16, 20.62) 4.90b (3.90, 6.15)
78.57 (76.36, 80.62 76.19 (66.96, 83.48)
3.30 (2.40, 4.52) 2.13 (0.91, 4.92)
9.24 (7.89, 10.80) 3.91 (1.68, 8.80)
8.89a (7.57, 10.41) 17.77b (11.11, 27.20)
73.68 (65.42, 80.55) 71.66 (58.48, 81.94)
1.15 (0.40, 3.24) 7.34 (2.91, 17.28)
7.09 (4.37, 11.30) 9.93 (4.61, 20.08)
18.08b (12.43, 25.56) 11.08 (5.39, 21.42)
68.02a (59.99, 75.11)
4.30 (2.23, 8.14)
5.50 (3.20, 9.28)
22.18a (16.05, 29.83)
78.84b (75.22, 82.05) 79.24 (74.76, 83.11) 77.80 (74.50, 80.79)
2.77 (1.76, 4.35) 2.84 (1.74, 4.60) 3.68 (2.27, 5.93)
8.75 (6.75, 11.29) 7.44 (5.60, 9.80) 11.12 (9.02, 13.64)
9.64b (7.05, 13.04) 10.48b (7.72, 14.08) 7.39b (5.62, 9.67)
71.85a (65.73, 77.25) 75.87 (70.99, 80.16) 77.89 (70.55, 83.83)
4.80 (3.13, 7.28) 3.86 (2.44, 6.07) 2.46 (1.05, 5.69)
7.26 (4.90, 10.64) 6.17 (3.79, 9.89) 7.82 (4.77, 12.57)
16.09a (10.99, 22.95) 14.09a (10.73, 18.29) 11.82 (7.23, 18.75) (Continued)
Americans’ Perceptions of Their Own Weight
807
Table 5b. (Continued)
Predictor $50,000– $74,999 $75,000 or more
Accurate (95% CI)
Normal underestimators (95% CI)
Normal overestimators (95% CI)
Overweight underestimators (95% CI)
78.18 (72.18, 83.19) 81.46b (77.99, 84.50)
3.90 (1.78, 8.32) 2.40 (1.17, 4.85)
10.25 (7.02, 14.73) 8.43 (6.51, 10.87)
7.67b (5.77, 10.21) 7.70b (5.77, 10.21)
Note. Numbers with different letters within each weight perception accuracy by demographic category are significantly different from one another. *Includes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
Similarly, the overall sociodemographic profile of overestimators indicates that this group is more likely to be female, classify their race as White or other, and have higher levels of income and education. These findings are consistent with those found by Chang and Christakis (2003), who analyzed data from the National Health and Nutrition Examination Survey and supported the idea that there are different normative beliefs about weight in different racial/ethnic groups and in different socioeconomic status groups. Our results also show that weight perception accuracy can help predict weight loss and physical activity behaviors. This assessment may be a surrogate measure for weight-loss motivation. Weight perception accuracy was related to having exercised in the past month and attempting to lose weight in the last year. Furthermore, consistent with previous literature, demographic variables, such as gender and income, were also related to weight loss and physical activity behaviors. More specifically, women were more likely than men were to have tried to lose weight in the past year; and those with higher income levels were more likely to have exercised in the past month and tried to lose weight in the last year than those in lower income groups. These finding suggest that different norms exist by gender, income, education and race/ethnicity and that, within different socioeconomic status groups, different communication approaches may be needed to help realign perceptions of what a healthy weight and overweight look like. Clinical Issues Results suggest that asking patients about their weight self-perceptions could be useful in clinical settings. Physicians are in an excellent position to help align patients’ weight perceptions with their actual BMI. Before sharing results of a BMI assessment, physicians should ask patients, “Right now do you feel you are (1) overweight, (2) slightly overweight, (3) underweight, (4) slightly underweight, and (5) just about the right weight for you?” Physicians would then be able to factor in patients’ perceptions of their own weight when counseling patients about their weight. For example, for patients who have a BMI of 18.5 or below and who are overestimators (e.g., indicate they perceive themselves to be the right weight, slightly overweight, or overweight), physicians may want to ask additional questions to screen for anorexia. A simple tool could be developed or be added to existing tools such as the Agency for Health Care Research and Quality’s Electronic Preventive Service Selector, by allowing physicians
808
L. Squiers et al.
Table 6. Multivariate logistic regressions assessing the effect of weight perception accuracy on weight-related behaviors
Predictor Mode Mail Phone Age (years) 18–34 35–49 50–64 65–74 75 and older Gender Male Female Race/ethnicity White Black/African American Hispanic Othera Household income Less than $20,000 $20,000–$34,999 $35,000–$49,999 $50,000–$74,999 $75,000 or more Self-efficacy Confidence in ability to take care of health Obesity knowledge and beliefs Obesity caused by overeating and not exercising Obesity caused by genes Too many messages makes it hard to know what is a healthy weight
Exercise in past month (n = 5,769)
Days of moderate exercise per week (n = 3,989)
Odds ratio
Odds ratio
95% CI
95% CI
Tried to lose weight in past 12 months (n = 5,770) Odds ratio
95% CI
Ref. 1.55*** 1.30–1.85
Ref. 1.13
Ref. 0.97–1.33 0.86
Ref. 0.83 0.78† 0.72* 0.56***
0.63–1.10 0.60–1.01 0.54–0.97 0.40–0.76
Ref. 0.83 0.97 0.91 0.88
0.64–1.08 0.77–1.21 0.70–1.20 0.62–1.26
Ref. 0.82*
0.70–0.97
Ref. 0.74***
Ref. 0.63–0.87 2.19*** 1.86–2.58
Ref. 0.86
0.64–1.15
Ref. 0.93
Ref. 0.62–1.39 0.96
0.62** 0.85
0.44–0.87 0.55–1.33
0.81 0.79
0.60–1.10 0.92 0.66–1.28 0.52–1.18 0.49*** 0.33–0.72
Ref.
Ref.
Ref. 1.05 1.03 0.82 0.44***
0.70–1.05 0.80–1.38 0.78–1.35 0.60–1.12 0.31–0.63
0.70–1.32
Ref.
0.90 1.28 1.45* 1.81***
0.69–1.17 0.92–1.79 1.04–2.01 1.30–2.50
0.86 0.81 0.78 0.68**
0.61–1.20 0.58–1.13 0.56–1.08 0.51–0.91
1.02 0.96 1.35† 1.61**
0.75–1.39 0.71–1.29 0.99–1.85 1.23–2.10
1.11†
0.99–1.23
1.22**
1.07–1.39 1.05
0.94–1.18
1.16†
0.98–1.37
1.04
0.86–1.25 1.21**
1.05–1.38
0.87*
0.78–0.98
0.93
0.83–1.04 1.08
0.98–1.18
0.87**
0.79–0.96
1.03
0.97–1.11 1.08†
0.99–1.17
(Continued)
Americans’ Perceptions of Their Own Weight
809
Table 6. (Continued)
Predictor Health status General health Weight accuracy Accurate estimators Normal-weight underestimators
Exercise in past month (n = 5,769)
Days of moderate exercise per week (n = 3,989)
Odds ratio
Odds ratio
95% CI
1.53*** 1.38–1.69 Ref. 0.51*
Normal-weight 1.57** overestimators Overweight 1.15 underestimators
1.37*** Ref.
95% CI
Tried to lose weight in past 12 months (n = 5,770) Odds ratio
1.21–1.55 0.85**
95% CI 0.77–0.94
Ref.
0.30–0.88
1.11
0.63–1.96 0.02*** 0.01–0.08
1.21–2.05
1.06
0.83–1.35 1.26†
0.87–1.54
1.32
0.93–1.87 0.39*** 0.28–0.56
0.98–1.64
Note. Ref = reference category. *p < .05. **p < .01. ***p < .001. †p < .10. aIncludes American Indian or Alaska Native, Asian, Native Hawaiian or Pacific Islander, and multiple races mentioned.
to enter a patient’s evaluation of their own weight using the questions above, as well as height and weight. The tool could calculate not only BMI, but also whether a patient is an under-, over-, or accurate-estimator of their own weight. These categorizations may help physicians tailor messages about weight. For example, for patients with a BMI in the overweight range who underestimate their weight, providers may first want to educate the patient about a healthy weight range and assess the patient’s level of knowledge about the link between overweight/obesity and poor health outcomes. Using these categorizations within a clinical setting may help identify patients have eating disorders or body weight distortions and allow providers to direct patients to appropriate support services. For overweight patients who are accurate estimators, providers could again assess level of knowledge and also assess motivation and barriers to weight-loss behaviors. Public Health Communication Issues Weight perception accuracy can be used in a variety of different applications and settings to target health promotion messages. Weight perception accuracy could be used to segment audiences for larger health promotion efforts. For some audiences, such as overweight underestimators, messages about diet and exercise for weight reduction could potentially fall on deaf ears as many inaccurately perceived their weight to be just right. Messages addressing perceptions of risk (e.g., the link between overweight/ obesity and chronic diseases and conditions) may also be useful for these individuals. Public health communication efforts might first focus on helping audiences know what a healthy weight means and is. Follow-up messages could help align self-perceptions with new knowledge about a healthy weight. While weight perception accuracy could help identify target audiences for public health communication programs, it also could be used to help tailor individualized
810
L. Squiers et al.
consumer health tools. Consumer focused e-health tools could also use an assessment of weight perception accuracy to develop individually-tailored messages designed to improve healthy behaviors. As previously described, when weight perceptions are inaccurate, tools can help align users’ perceptions and perhaps use pictures, messages, quizzes, and games to promote social norms around a healthy weight range. Limitations While the response rates for the mail and telephone surveys were relatively low, with the advent of Caller ID and a movement toward cell-only households, response rates for random digit dialing surveys have been generally falling across all types of survey administrations (Blumberg, Luke, & Cynamon, 2006; Fahimi, Link, Mokdad, Schwartz, & Levy, 2008). Some methodological studies have indicated that the threat from falling random digit dialing response rates may not be as strong for health surveys as originally hypothesized (Fahimi et al., 2008, Gentry et al., 1985; Nelson, PowellGriner, Town, & Kovar, 2003). The Health Information National Trends Survey has tried to deal with the random digit dialing response rates by offering multiple modes. While not all measures are sensitive to mode, the present study indicates that perceived weight does vary by mode. Those interviewed via phone were more likely to report their weight as just right and less likely to perceive themselves as slightly overweight than those who responded by postal mail. This may be due to social desirability concerns. It may be less threatening to indicate that one is overweight on a questionnaire filled out privately than to share that information with another person. Future research should assess the role of social desirability, as well as other potential factors, in differences in self-reported weight perceptions across different modes of data collection. No mode effects were found for BMI; however, BMI has been criticized for being too stringent and for inaccurately estimating fat among different ethnic and racial groups. Studies have found that BMI based on self-reported height and weight can differ significantly from BMI calculated from physical measurements assessed more objectively. Specifically, objective measurements of height and weight tend typically lead to fewer respondents being classified as normal weight and more respondents classified as overweight than when relying upon self- reports of height and weight (Brener et al., 2004). The present study relied on self-reported height and weight to calculate BMI, suggesting that BMI could have been underestimated. Last, the wording of the question that assessed participants’ perceptions of their own weight was not specifically tied to health. As a result, we do not know for certain that participants evaluated their weight in terms of their health. Given the apparent influence of sociocultural norms, another area that may need study is respondents’ beliefs about the accuracy or appropriateness of the medical community’s recommendations regarding weight Future Studies Despite these limitations and challenges, assessing weight perception accuracy has been shown to have important influences on weight-related behavior. Future studies should examine other factors associated with weight perception accuracy, including perceptions of the weight of individuals in social networks, health promotion behaviors, knowledge of the health consequences of being overweight or obese, and motivation to engage in weight-related behaviors. In addition, studies should delve further into discovering cultural norms around body weight, whether there is a perceived relationship between body weight and health, and what factors are considered when developing attitudes about what constitutes a right body weight. For example, some individuals may perceive that a larger body weight is protective (e.g., from being pushed around easily by others or sexual advances) or more attractive. In addition,
Americans’ Perceptions of Their Own Weight
811
studies should examine what individuals and subgroups believe about the medical communities’ recommendations about weight. Do they perceive the guidelines to be accurate, credible, and relevant to them? Examining the relation between motivation for attaining or maintaining a healthy weight and weight perception accuracy may prove useful (it was not possible to conduct this analysis in the present study). For example, behavior change theories suggest that individuals must perceive weight as a problem or a risk before being motivated to lose weight (Ajzen, 1985; Becker, 1974; Rosenstock, 1974), as is illustrated in the current conceptual framework (see Figure 1). Furthermore, it may be important to track weight perception accuracy over time to see if trends in weight perception change in accordance with changes in overweight/obesity rates. Perhaps individuals’ weight self-perceptions and the accuracy of their weight selfperceptions may improve if physicians work with patients to ensure that self-perceptions are aligned and accurate with what is known to be a healthy body weight. Future research could be conducted to help determine whether pointing out misaligned perceptions is beneficial to patients (e.g., is learning this information motivating or demoralizing?) and if, when followed by behavioral counseling or health information, it helps patients engage in behaviors that will help them achieve a healthy weight. Because results from this study suggest that some sociodemographic subgroups have different levels of weight perception accuracy, researchers should investigate what factors influence weight self-perceptions and perceived norms about weight within different cultures, subpopulations, and social networks. Results from these investigations should be used to develop and implement interventions that increase the public’s awareness of what constitutes a healthy weight (Johnson-Taylor et al., 2008).
References Ajzen, I. (1985). From intentions to action: A theory of planned behavior. In J. Kuhl & J. Beckman (Ed.), Action CONTROL: From cognitions to behaviors (pp. 11–39). New York, NY: Springer. Becker, M. H. (Ed.). (1974). The Health Belief Model and personal health behavior. Health Education Monographs, 2, 324–508. Blumberg, S. J., Luke, J. V., & Cynamon, M. L. (2006). Telephone coverage and health survey estimates: Evaluating the need for concern about wireless substitution. American Journal of Public Health, 96, 926–931. Brener, N., Eaton, D., Lowry, R., & McManus, T. (2004). The association between weight perceptions and BMI among high school students. Obesity Research, 12, 1966–1974. Cantor, D., Coa, K., Crystal-Mansour, S., Davis, T., Dipko, S., & Sigman, R. (2009). Health Information National Trends Survey (HINTS) 2007. Rockville, MD: Westat. Centers for Disease Control and Prevention. (2009). Differences in prevalence of obesity among Black, White and Hispanic Adults—United States, 2006–2008. Morbidity and Mortality Weekly Report, 58, 740–744. Chandler-Laney, P., Hunter, G., Ard, J., Roy, J., Brock, D., & Gower, B. (2009). Perceptions of others’ body size influences weight loss and regain for European American but not African American women. Health Psychology, 28, 414–418. Chang, V., & Christakis, N., (2003). Self-perceptions of weight appropriateness in the United States. American Journal of Preventive Medicine, 24, 332–339. Christakis, N., & Fowler, J. (2010). The spread of obesity in a large social network over 32 years. New England Journal of Medicine, 357, 370–379. Dawson, D. A. (1988). Ethnic differences in female overweight: Data from the 1985 National Health Interview Survey. American Journal of Public Health, 78, 1326–1329. Fahimi, M., Link, M., Mokdad, A., Schwartz, D. A., & Levy, P. (2008). Tracking chronic disease and risk behavior prevalence as survey participation declines: Statistics from
812
L. Squiers et al.
the Behavioral Risk Factor Surveillance System and other national surveys. Preventing Chronic Disease, 5(3), A80. Finkelstein, E. A., Fiebelkorn, I. C., & Wang, G. (2003). National medical spending attributable to overweight and obesity: How much, and who’s paying? Health Affairs, W3, 219–226. Flegal, K. M., Carroll, M. D., Ogden, C. L., & Curtin, L. R. (2010). Prevalence and trends in obesity among US adults, 1999–2008. JAMA, 303, 235–241. Flegal, K. M., Graubard, B. I., Williamson, D. F., & Gail, M. H. (2005). Excess Deaths Associated With Underweight, Overweight, and Obesity. JAMA, 293, 1861–1867. Gentry, E., Kalsbeek, W., Hogelin, G., Jones, J., Gaines, K., Forman, M., … Trowbridge, F. (1985). The behavioral risk factor surveillance surveys: II. Design, methods, and estimates from combined state data. American Journal of Preventive Medicine, 1, 9–14. Hagger, M., Chatzisarantis, N., & Biddle, S. (2002). The meta-analytic review of the theories of reasoned action and planned behavior in physical activity: Predictive validity and the contribution of additional variables. Journal of Sport and Exercise Psychology, 24, 3–32. Johnson-Taylor, W., Fisher, R., Hubbard, V., Starke-Reed, P., & Eggers, P. (2008). The change in weight perceptions of weight status among the overweight: Comparison of NHANES III (1988–1994) and 1999–2004 NHANES. International Journal of Behavioral Nutrition and Physical Activity, 5, 9. Martin, P, Dutton, G., & Brantly, P. (2004). Self-efficacy as a predictor of weight change in African-American women. Obesity Research, 12, 646–651. doi:10.1038/oby.2004.74 Millstein, R., Carlson, S., Fulton, J., Galuska, D., Zhang, J., Blanck, H., & Ainsworth, B. (2008). Relationships between body satisfaction and weight control practices among US adults. Medscape Journal of Medicine, 10, 119. Nelson, D. E., Powell-Griner, E., Town, M., & Kovar, M. G. (2003). A comparison of national estimates from the National Health Interview Survey and the Behavioral Risk Factor Surveillance System. American Journal of Public Health, 93, 1335–1341. Rosenstock, I. M. (1974). Historical origins of the health belief model. Health Education Monographs, 2, 328–335. Shin, H., Shin, J., Liu, P. Y., Dutton, G. R., Abood, D. A., & Elich, J. Z. (2011). Self-efficacy improves weight loss in overweight/obese postmenopausal women during a 6-month weight loss intervention. Nursing Research, 31, 822–828. StataCorp. (2009). Stata statistical software: Release 11. College Station, TX: Author. Taylor, P., Funk, C., & Craighill, P. (2006). Americans see weight problems everywhere but in the mirror. Pew Research Center. A Social Trends Report. Washington, DC: Pew Research Center.
