Development of a Culturally Relevant Body Image Instrument among Urban African Americans

Authors

  • Kim M. Pulvers,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • Rebecca E. Lee,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    2. Kansas Cancer Institute, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • Harsohena Kaur,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    2. Department of Pediatrics, University of Kansas Medical Center, Kansas City, Kansas
    3. Kansas Cancer Institute, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • Matthew S. Mayo,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    2. Kansas Cancer Institute, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • Marian L. Fitzgibbon,

    1. Department of Psychiatry and Behavioral Sciences, Northwestern University School of Medicine, Chicago, Illinois
    Search for more papers by this author
  • Shawn K. Jeffries,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    2. Kansas Cancer Institute, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • James Butler,

    1. Department of Center for Public Health Practice, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania
    Search for more papers by this author
  • Qingjiang Hou,

    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author
  • Jasjit S. Ahluwalia

    Corresponding author
    1. Department of Preventive Medicine and Public Health, University of Kansas Medical Center, Kansas City, Kansas
    2. Department of Pediatrics, University of Kansas Medical Center, Kansas City, Kansas
    3. Department of Internal Medicine, University of Kansas Medical Center, Kansas City, Kansas and
    4. Kansas Cancer Institute, University of Kansas Medical Center, Kansas City, Kansas
    Search for more papers by this author

  • The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Department of Preventive Medicine and Public Health, University of Kansas Medical Center/Mailstop 1008, Kansas City, KS 66160-7313. E-mail: jahluwal@kumc.edu

Abstract

Objective: To validate a culturally relevant body image instrument among urban African Americans through three distinct studies.

Research Methods and Procedures: In Study 1, 38 medical practitioners performed content validity tests on the instrument. In Study 2, three research staff rated the body image of 283 African-American public housing residents (75% women, mean age = 44 years), with the residents completing body image, BMI, and percentage body fat measures. In Study 3, 35 African Americans (57% men, mean age = 42) completed body image measures and evaluated their cultural relevance.

Results: In Study 1, 97% to 100% of practitioners sorted the jumbled figures into the correct ascending order. The correlation between the body image figures and the practitioners’ weight classifications of the figures was high (r = 0.91). In Study 2, observers arrived at similar ratings of body size with excellent consistency (α = 0.95). Ratings of body image were strongly correlated with participant BMI (r = 0.89 to 0.93 across observers and 0.81 for all participants) and percentage of body fat (r = 0.77 to 0.89 across observers and 0.76 for all participants). In Study 3, body image ratings with the new scale were positively correlated with other validated figural scales. The majority of participants reported that figures in the new body image scale looked most like themselves and other African Americans and were easiest to identify themselves with.

Discussion: The instrument displayed strong psychometric performance and cultural relevance, suggesting that the scale is a promising tool for examining body image and obesity among African Americans.

Introduction

Obesity is epidemic in the United States and the prevalence is currently highest among African-American women (1). Both African-American women and those from low socioeconomic backgrounds suffer disproportionately from comorbid conditions associated with obesity (2).

Factors have been identified that may contribute to the disproportionate levels of obesity among African-American women, including: high-fat, low-fiber diets (3,4); low physical activity (5,6); and the connection between body image and obesity (7). Specifically, African-American women tend to be satisfied with their bodies and perceive themselves to be attractive whether they are normal or overweight (8) and may not become dissatisfied with their body size until they are obese (9). Although greater acceptance of heavy body ideals may be protective against some eating disorders, it may be a risk factor for obesity and may limit motivation for weight loss or weight control (10). Some body image dissatisfaction and accurate perception of overweight status may be necessary to motivate healthy behaviors (11). However, African-American women are less likely to correctly perceive their overweight status compared with white women (12). Although obesity is less of a problem among African-American men than women, African-American men who are overweight are also less likely to correctly perceive their overweight status than white men (12).

Body image consists of both attitudinal and perceptual dimensions. Attitudinal investigations of body image explore the feelings and thoughts people have about their bodies, such as satisfaction with body and weight and feelings of attractiveness (7). Perceptual investigations of body image examine the accuracy of body size estimations and the accuracy of weight classification relative to medical standards (7). Silhouette drawing tests are a popular means of assessing body image attitudes and perceptions due to their minimal cost, low participant burden, and ease of administration, especially outside of laboratory settings.

Research on body image perception requires well-validated instruments (13). Of those instruments for which psychometric information is known, little evidence exists to suggest their validity with African Americans (7). Among the psychometrically sound instruments that have been used with African Americans, concern has been raised regarding the cultural appropriateness of using instruments that resemble the morphology of whites (7).

Several body image scales have been specifically designed for and validated among African Americans (14,15,16). These scales have followed two general design schemes: silhouettes that are free of most details or detailed drawings with culture-specific hair, facial, and body features. A strength of the former approach is that the instruments are ethnically neutral, providing the potential for them to be used with a variety of cultures. A strength of the latter approach is that the figures seem more human-like.

We strove to capitalize on the strengths embodied by each of the aforementioned approaches by developing an instrument in which the figures were detailed and human-like but not specific to any one culture. To evaluate the psychometric properties and cultural relevance of the scale among African Americans, we undertook three distinct studies. In the first study, we assessed content validity by administering two validity tasks to medical practitioners in which they were asked to place the body image figures in order from smallest to largest and to classify the weight status of each figure. In the second study, we assessed interrater reliability and convergent and concurrent validity with a sample of African-American public housing residents. Independent observers rated the body size of participants, and participants rated their own body image and weight status. Participant body image ratings were compared with observer body size ratings, participant perception of weight status, BMI, and percentage body fat. In the third study, we assessed the criterion validity and cultural relevance of the instrument with a sample of African-American clinic patrons. We administered the new body image instrument along with two well-validated instruments, assessed perception of weight status, BMI, and percentage body fat, and examined their associations with each of the three body image instruments. Further, we explored the cultural relevance of each of the three instruments. Each of these studies was approved by the Human Subjects Committee of the University of Kansas Medical Center and the three studies were conducted in three separate settings.

Research Methods and Procedures

Initial Scale Development

Following the procedure used by Thompson and Gray (13), we commissioned a graphic artist to draw nine male and nine female front-view drawings of incremental sizes (see Figure 1). Nine-figure scales have been criticized for being too coarse (e.g., having too dramatic a difference between figures) and not being finite enough to adequately assess body image in a wide variety of sizes (16,17,18,19). Despite these perceived weaknesses, we modeled our instrument after Stunkard's widely recognized nine-figure scale (20) because it has been used in several recent large-scale studies (21,22) and has performed similarly to scales with >nine figures (23). Hair and facial features of the figures were designed to resemble persons of multiethnic background. Attention was given to consistency and graduation in size across the figures, especially at the larger end of the scale. Our guiding conceptual framework was that the figures should span a BMI of roughly 16 to 40 in increments of three BMI points. A panel of health professionals comprised of physicians, clinical psychologists, and epidemiologists reviewed and provided feedback to the artist on a successive series of revisions.

Figure 1.

Male and female body image instrument (master form).

Study 1: Examination of Content Validity

Evidence of content validity is generally derived from agreement by a panel of experts in judgments about the content covered by an instrument (24). Two content validity tasks were designed for medical practitioners who have regular contact with patients with a range of body sizes. We expected that there would be a high degree of correspondence among expert judges in their responses to two scale validity tasks.

Data were collected in the Family Medicine and Dietetics Departments at a university medical center. Three body image forms were created in which the drawings were presented in a different random order. Practitioners received two different forms of the adult male and adult female versions of the instrument and were asked to place the drawings in order from smallest to largest and classify each of the nine figures according to weight status. The instrument consisted of nine drawings of adult women (Adult Female Version) and nine drawings of adult men (Adult Male Version).

Weight status categories included “very underweight,” “a little underweight,” “weight is just right,” “a little overweight,” and “very overweight.” The weight status measure was modified from the Third National Health and Examination Survey (25) and has been correlated with BMI and attempted weight loss in an African-American sample (26).

Study 2: Examination of Interrater Reliability and Convergent and Concurrent Validity

The purpose of Study 2 was to evaluate interrater reliability (consistency among independent ratings with the same test), convergent validity (the degree to which a measure corresponds with variables with which it should theoretically relate), and concurrent validity (the degree to which predictions made by a test are confirmed by corresponding measures) to ensure that the present body image instrument was applicable to an African-American population.

Data were collected as part of an ongoing clinical trial of smoking cessation and dietary change in nine public housing developments in the greater Kansas City metropolitan area between October 2001 and July 2002. The public housing developments in this area serve a predominantly African-American population that meets the 2003 U.S. Department of Health and Human Service's poverty guidelines, which is an annual household income of $18, 400 or less per year for a family of four (27).

Public housing development residents were invited to attend a community health fair as part of this larger study. Health fairs were held on-site at the housing developments from 11 am to 3 pm on Saturdays. To participate in the health fair, an individual was required to be a resident of the respective housing development.

Three research staff (two female master of public health students, one male medical doctor) rated the residents’ body size using a sheet containing a set of male figures in a random order and a set of female figures in a random order. Participants were presented with the gender-appropriate version of the random-order body image instrument and asked to select the figure that mostly closely resembled their body image. Participants were also asked to classify their current weight status as either being very underweight, a little underweight, just right, a little overweight, or very overweight (25).

Trained health fair staff measured each participant's height (in inches) using a portable stadiometer with a telescopic measuring rod (Seca 225 Hite Mobile Measuring Device, North Bend, WA). Bioimpedance analysis (BIA)1 (weight, percentage body fat, and BMI) was determined using a leg-to-leg pressure contact electrode BIA system (Tanita Corporation of America, Arlington Heights, IL). Tanita uses a patented “foot-to-foot” pressure contact bioelectrical impedance analysis technique. The BIA technique is based on the fact that lean tissues have a high water and electrolyte content, providing a viable electrical pathway. Fat mass contains a lower percentage of body water and is a poor conductor of the electrical signal. By inducing a low-energy, high-frequency, electrical (50 kHz, 500 microamp) current, a measurement of the baseline resistance to the flow of electrical current can be made. This current is passed through the anterior electrode on the scale platform, and the voltage drop is then measured on the posterior electrode. The resistance measurement relates directly to the volume of the conductor which is used to determine total body water, lean body mass, and, finally, fat mass. Percentage body fat, as calculated by Tanita, is a well-researched proprietary formula combining impedance and weight measurements with height, gender, and age information (28,29,30).

Trained health fair staff explained the purpose and components of the health fair to the residents, obtained informed consent, and administered a brief survey that included demographic (e.g., age, sex, race/ethnicity, educational level, monthly income) and body image questions. Due to the possibility of low literacy among the residents (31,32,33), health fair staff read all items on the survey instrument. Height, weight, and percentage body fat were objectively measured. Participants removed their shoes for the height measurement and their socks for the BIA. A 2-pound weight adjustment was made for clothing before conducting BIA.

Anthropometric data were recorded on residents’ health fair results sheets, and data were entered by research staff on-site at the health fair and by university staff at the medical center. Participants received a free medical consultation in which they were given a personalized feedback sheet containing their physiological results and were reimbursed for their time with two movie tickets and a catered lunch.

Study 3: Examination of Criterion Validity and Cultural Relevance

The purpose of Study 3 was 2-fold: to examine criterion validity (the degree to which a scale corresponds with another valid scale that has been shown to measure the same construct) by evaluating body image ratings with the new instrument in comparison to two other existing instruments and to examine cultural relevance by assessing African Americans’ perceptions of the scale compared with the other instruments.

Data were collected at an urban health center. Research staff approached individuals they believed to be African American in the main lobby and invited them to participate in the study. Participants were eligible if they were 18 years or older, self-reported to be African American, and were English speaking. Research staff explained the purpose and components of the study, obtained informed consent, and administered the survey measure.

Participants were presented with the gender-appropriate version of the instrument (Scale A) and two additional validated body image instruments (16,20) in a counterbalanced order. The Williamson scale (16) (Scale B) and the Stunkard scale (20) (Scale C) were chosen because both are well-validated and widely recognized. Participants were asked to select the figure that most closely resembled them currently. In addition, they were asked which ethnicity best described the figures depicted in each instrument. After viewing all three instruments, participants were asked which of the three scales contained figures that looked most and least like themselves and other people of their race. Further, they were asked which set of figures was easiest to identify themselves with and what they did and did not like about each of the scales. Weight status was obtained using the same question as described in Study 2. BMI and percentage body fat were also obtained using the equipment and procedures described in Study 2.

Three respective versions of the survey prompted the interviewer to present each of the three body image scales in a different order so that each scale occupied the first, second, and third positions roughly an equal number of times. Each body image scale was placed on its own 8.5- × 11-inch white sheet with the male figures on one side and the female figures on the other side. [The Williamson scale (16) was designed to be administered by presenting each of the 18 figures on a separate card in a different shuffled order (16). The group administration method used in the present study has also been validated (34) and is a common approach in other body image studies (e.g., (13)).] Figures were arranged in consecutive order ranging from smallest to largest and numbered from largest to smallest (e.g., for the two nine-figure scales, the smallest figure was labeled with the number 9) following the method of Patt et al. (14). BMI and percentage body fat measures were taken last. Participants then received a personalized feedback sheet containing their physiological results and were compensated for their time with five movie tickets.

Results

Study 1: Examination of Content Validity

Participants

Participants were 38 medical practitioners (13 family medicine faculty members, 16 family medicine residents, and 9 registered dietitians) at a midwestern university medical center. Eighty-nine percent of the registered dietitians, 54% of the faculty members, and 53% of the residents were women.

Content Validity

Content validity was evaluated by examining how similarly practitioners responded to validity tests concerning the content of the scale. All of the practitioners sorted the female figures in the correct order, and 97% sorted the male figures in the correct order. The one practitioner who did not rank the same order for the nine male figures used the same figure twice instead of once. Practitioners next classified the figures by weight. The correlation between the body image figures and the practitioners’ weight classifications of the figures was high (r = 0.91). Practitioners’ ratings were also highly consistent (Cronbach's α = 0.99).

Study 2: Examination of Interrater Reliability and Convergent and Concurrent Validity

Participants

Participants were 283 African-American public housing residents. They were mostly women (75%), with a mean age of 44, an average BMI of 29, and an average percentage body fat of 35. Demographic characteristics are presented in Table 1.

Table 1.  Demographic characteristics of Study 2 sample (N = 283)
 PercentageN
Gender (n = 283)  
 Men25.171
 Women75.0212
Education (n = 283)  
 Less than high school education60.0169
 High school education or more40.0114
Income (n = 109)  
 Less than $400 a month39.042
 Between $400 and $799 a month39.543
 Between $800 and $1200 a month15.617
 More than $1200 a month6.47
 MeanSD
Age (n = 283)43.716.6
BMI (n = 236)29.18.0
 Men (n = 58)25.55.3
 Women (n = 178)30.38.4
Percentage body fat (n = 236)34.811.9
 Men (n = 58)22.88.7
 Women (n = 178)38.710.1

Scale Interrater Reliability

Interrater reliability was evaluated by examining consistency among independent observers in rating the body size of participants. The range in difference scores among the three observers spanned from 0 to 3, with the most common frequency being a difference of only 1 point on the 9-point scale. Global measures of central tendency revealed that observers appeared to arrive at the same body size rating (medians = 5.0, 5.0, and 5.0; means = 4.8, 4.9, and 5.1). Two methods were used to assess the degree of interrater reliability. First, a mixed linear model was developed assuming a compound symmetric correlation structure among raters (35). Mixed linear models allowed measurement of correlated responses and estimation of the correlation of these responses. They also provided unbiased parameter and correlation estimates even in the presence of missing data, provided the data were missing at random. A compound symmetric correlation structure represents the correlation of multiple measurements on the same individual, under the assumption that the correlation between any pair of raters is the same. The estimate of the compound symmetric correlation, a measure of interrater agreement, was high (r = 0.85). Second, Cronbach's α was calculated as 0.95, indicating excellent internal consistency among raters. When we incorporated the participant's own body image rating into the calculation, Cronbach's α remained unchanged.

Convergent Validity

Convergent validity was examined in several ways. First, we evaluated how strongly observer ratings of body size correlated with one another and with participants’ ratings of body image. The mean observer body size ratings were 4.8, 4.9, and 5.1, whereas the mean participant body image rating was 4.4. Spearman correlations of body size ratings among the three observers ranged from 0.78 to 0.89 (see Table 2). Correlations between observer body size ratings and participant body image ratings ranged from 0.56 to 0.61 for men and 0.75 to 0.83 for women.

Table 2.  Study 2 Spearman correlations
VariableRater 1Rater 2Rater 3Participant body imageWeight statusBMI
All      
 Rater 20.83     
 Rater 30.890.87    
 Participant body image0.780.740.82   
 Weight status0.640.650.690.68  
 BMI0.890.890.930.810.73 
 Body fat (%)0.890.770.890.760.640.88
Men      
 Rater 20.83     
 Rater 30.820.78    
 Participant body image0.560.610.58   
 Weight status0.590.680.670.51  
 BMI0.890.880.890.620.71 
 Body fat (%)0.850.820.880.580.670.92
Women      
 Rater 20.87     
 Rater 30.880.89    
 Participant body image0.800.750.83   
 Weight status0.650.630.680.69  
 BMI0.920.880.930.820.71 
 Body fat (%)0.850.810.880.780.640.91

Second, we examined the degree to which observer ratings of body size correlated with participants’ perception oftheir weight status and how well participants’ rating of their body image correlated with perception of their own weight status. Spearman correlations ranged from 0.59 to 0.68 across observers and were 0.51 for men and 0.69 for women.

Concurrent Validity

Concurrent validity was examined by evaluating how strongly observer ratings of body size correlated with participant BMI and percentage of body fat and how strongly participant ratings of body image correlated with participant BMI and percentage of body fat. Spearman correlations for BMI ranged from 0.88 to 0.93 across observers and were 0.62 for male participants and 0.82 for female participants. Spearman correlations for body fat ranged from 0.81 to 0.88 for observers and were 0.58 for male participants and 0.78 for female participants. A graphical display of participant body image ratings by anthropometric data revealed a fairly linear trend, such that BMI (Figure 2) and percentage body fat (Figure 3) generally increased as the figures became larger. The correlations among men may have been attenuated by the restricted range of very obese men (only two had BMIs > 32 and percentage body fat > 33%) and by the fact that only two men selected a body image larger than the figure 5 (data not shown).

Figure 2.

BMI by participant body image rating.

Figure 3.

Percentage body fat by participant body image rating.

Study 3: Examination of Criterion Validity and Cultural Relevance

Participants

Participants were 35 African-American adults (20 men, 15 women, mean age 42 years) who were visiting an urban health center. The majority had a high school education or more (71%), and income levels were split, with 40% reporting a household monthly income of less than $400, 31% reporting an income of greater than $1200, and the rest (26%) falling in between or unknown (3%). The mean BMI was 28.9 for men and 34.1 for women, and the mean percentage body fat was 24.0 for men and 39.0 for women.

Criterion Validity

Criterion validity was evaluated by comparing scores from the new instrument with scores from the two validated instruments. Cronbach's α among the three scales was 0.94 for all participants, 0.93 for men, and 0.96 for women, indicating a high level of consistency among the instruments. The new instrument was positively correlated with both the Williamson and Stunkard scales among men (r = 0.79 and 0.89, respectively) and women (r = 0.92 and 0.90), evidencing correlations similar to those between the Williamson and Stunkard scales (r = 0.75 for men and 0.92 for women) in this sample (see Table 3). Correlations between body image ratings and BMI, percentage body fat, and weight status were similarly high (r = 0.84 to 0.93) across the three scales among women. Among men, correlations between body image ratings and BMI and percentage body fat were high for the new instrument (r = 0.86 to 0.88) but less so for the other instruments (r = 0.62 to 0.75).

Table 3.  Study 3 Spearman correlations
VariableScale AScale BScale CBMIBody fat (%)
All     
 Scale B0.85    
 Scale C0.860.81   
 BMI0.890.790.76  
 Body fat (%)0.800.670.580.87 
 Weight status0.850.780.850.840.72
Men     
 Scale B0.79    
 Scale C0.890.75   
 BMI0.880.680.75  
 Body fat (%)0.860.620.680.86 
 Weight status0.850.650.870.840.71
Women     
 Scale B0.92    
 Scale C0.900.92   
 BMI0.900.910.93  
 Body fat (%)0.880.850.840.94 
 Weight status0.850.930.890.880.75

Perception of Ethnicity Depicted by Body Image Figures

The majority of men viewing the new scale reported that the figures depicted no particular race (45%) followed by whites (40%), compared with a majority of women who reported that the figures in the new scale depicted African Americans (46.7%) followed by no particular race (26.7%) (see Table 4). The majority of men viewing the Williamson scale reported that the figures resembled no particular race (40%) followed by a split (30% each) between whites and African Americans, whereas the majority of women reported that the Williamson figures depicted African Americans (46.7%) followed by no particular race (26.7%). The majority of men viewing the Stunkard scale reported that the figures depicted whites (45%) followed by no particular race (40%), compared with an equal proportion of women who reported that the Stunkard figures resembled African Americans and no particular race (40% each) followed by whites (13.3%).

Table 4.  Perception of ethnicity depicted by body image figures
Ethnic categoryNew scale (%)Williamson (%)Stunkard (%)
All   
 White31.420.031.4
 African-American/black22.937.122.9
 Hispanic/Latino2.95.72.9
 Asian2.90.02.9
 Bi-/multiracial2.92.90.0
 No particular race37.134.340.0
 Other0.00.00.0
Men   
 White40.030.045.0
 African-American/black5.030.010.0
 Hispanic/Latino0.00.05.0
 Asian5.00.00.0
 Bi-/multiracial5.00.00.0
 No particular race45.040.040.0
 Other0.00.00.0
Women   
 White20.06.713.3
 African-American/black46.746.740.0
 Hispanic/Latino6.713.30.0
 Asian0.00.06.7
 Bi-/multiracial0.06.70.0
 No particular race26.726.740.0
 Other0.00.00.0

Most Similar, Least Similar, and Easiest Set of Figures to Identify with

The majority of participants (65% of the men and 80% of the women) reported that the images in the new instrument were most similar to themselves and other African Americans. A smaller proportion (10% of the men and 13% of the women) reported that the images in the new instrument were least similar to themselves and other African Americans. Many of these participants expressed that the figures resembled whites and others said that the figures were too thin. A majority (60%) of women participants found the Stunkard scale to be least similar to themselves and other African Americans, and men participants were split (45% each) between the Stunkard and Williamson scales as least similar.

The majority of participants (65% of the men and 80% of the women) also reported that the images in the new instrument were easiest to identify themselves with. Participants found the figures in the Williamson scale the next easiest to identify themselves with (20% men, 13% women). The main theme among respondents who preferred the new scale was that they liked the detail of the figures and found them more human-like and realistic.

Discussion

The overall purpose of the three studies presented was to develop and validate a new culturally relevant body image instrument as the first step in promoting a better understanding of how body image perception may impact the development and maintenance of obesity among African-American men and women. In doing so, we have produced a tool that combines the strengths of other body image instruments that have been validated with African Americans (14,15,16). Evaluation of the instrument's psychometric properties provided preliminary evidence of its validity and reliability, with excellent interrater reliability and strong performance in validity tests.

In Study 1, the high degree of correspondence among medical practitioners in their responses to validity tasks indicated that “experts” agreed about the content of the body image instrument. In Study 2, independent observers using the instrument consistently arrived at similar ratings of participant body size, suggesting that judgments about another person's body size can be made reliably with this instrument. Some evidence showed that participants, particularly men, might view themselves differently than independent observers do. For example, correlations among the three observer ratings of men's body sizes were higher than the correlations between men's self-ratings and observer ratings. In addition, associations between participant body image ratings and BMI and percentage body fat were weaker for men than for women, whereas relationships between observer body size ratings and physiological markers were robust for both men and women.

We speculated that the weaker relationship between participant body image ratings and physiological markers for men in Study 2 might be due to the restricted range of larger men in our sample. However, given that the correlations between body size ratings and BMI and percentage fat were similarly favorable for men and women when rated by observers, it is also possible that men did not use as wide a range of figures in their ratings as observers did. Perhaps, observers were more accurate in judging the body sizes of men in Study 2. Historically, less societal value has been placed on body ideals for men than for women. Accordingly, it is possible that men spend less time evaluating their bodies, giving them an inherent disadvantage in body rating tasks and contributing to less robust relationships between body image ratings and related constructs. Another possible explanation is that the men focused on different dimensions of body image than the observers did. Studies in the literature suggest that body image is multidimensional, and individuals may focus on different components of body image (34). This provides a challenge for future studies.

Turning to gender differences in Study 3, associations between body image ratings and anthropometric measures with the new instrument were uniformly strong for men and women. However, relationships between these variables were less strong among men on the other two instruments. When we explored whether these numbers were unique to our sample, we found that our values were comparable with those obtained in previous studies (23,34).

Additional differences emerged in Study 3 in how men and women perceived the ethnicity of the figures depicted by the instruments. A substantial proportion of men reported that the figures in all three scales depicted no particular race or whites. The figures in the Williamson scale were the only set that a large proportion of men perceived as African American. This contrasts with the majority of women who perceived the figures in the new scale and the Willliamson scale to be African American or no particular race, and an equal proportion who viewed the figures in the Stunkard scale as either African American or no particular race.

Despite the relatively large proportion of men who perceived the ethnicity of the figures in the new scale to be white, the majority of both men and women reported that the figures in the new scale looked most like themselves and other African Americans and were easiest to identify themselves with. These reports may have been influenced by the fact that, taken together, the majority of all respondents perceived the figures in the new instrument as ethnically neutral (no particular race). Also, of all of the instruments, only the new scale was thought to depict members of every ethnic category. It is possible that the respondents related more to the detailed, human-like renderings of the new figures than to any particular ethnic features.

These studies contain two primary strengths and two notable limitations. First, Study 2 included a large sample of African Americans, which allowed us to gain a better understanding of body image among a population who is at-risk for obesity and obesity related diseases. Second, we are aware of no other studies examining body image that have also collected objective data on percentage fat. A limitation of Study 2 is the smaller number of men than women and the restricted range of large male body sizes. A second limitation lies in the homogeneity of our sample in Study 2, which calls for caution in generalizing to those of higher socioeconomic levels. Future directions for this work include a need to test the validity of the instrument with other ethnic groups.

Our goal was to integrate the designs of instruments with demonstrated validity in African-American populations into a new measure. Validity testing of the new measure suggests its usefulness with African Americans, which is important in light of the need for culturally relevant assessment tools (36,37) and culturally competent clinical applications (38). The present work also meets a need by providing a potentially culturally sensitive way to assess the body image of African-American men. More research is needed to reconcile why a majority of men in our sample reported that the figures in the new instrument looked most like themselves and other African Americans and also that these figures depicted whites. Although obesity is far less of a problem in African-American men than women, little information about body dissatisfaction among African-American men exists to draw conclusions about body image and obesity in this group.

This project is the first step in understanding how body image perceptions impact obesity-related health behaviors. Few studies have examined the relationship between body image and weight control behaviors among African-American women (39,40), and the available studies have yielded inconsistent results (14,39,41,42). Future studies linking body image perception with self-care would assist in understanding how African-American women control their weight and in distinguishing between those who engage in healthy and potentially effective weight control behaviors and those who engage in unhealthy and ineffective behaviors that can lead to aberrant eating patterns or eating disorders that can precipitate further weight gain.

Persons of low socioeconomic status in general and African-American women in particular are among those most prone to obesity and obesity-related diseases (1,2). Body image has been suggested as one of a variety of factors related to the development and maintenance of obesity. Therefore, we need to develop effective tools to measure body image as accurately and with as culturally relevant instruments as possible.

Acknowledgment

We thank the Kansas City, KS and Kansas City, MO Housing Authority, participating housing development managers and residents, health fair staff, University of Kansas Medical Center family medicine physicians and registered dietitians, Swope Health Central patrons and research staff, and Martha Montello for editorial assistance. This study was supported by a grant from the NIH (R01 CA 85930)

Footnotes

  • 1

    Nonstandard abbreviations: BIA, bioimpedance analysis.

Ancillary