Media and Cultural Influences in African-American Girls' Eating Disorder Risk

Objective. To investigate media and cultural influences in eating disorder development in African-American adolescent females. Method. Fifty-seven participants were recruited through churches and community organizations to complete a questionnaire. Results. Mainstream sociocultural identification was associated with more eating disorder behavior in African-American females; cultural ethnic identification was not significantly associated with eating disorder behavior in African-American females, mainstream sociocultural identification, cultural ethnic identification, and body dissatisfaction significantly predicted eating disorder behavior; and cultural ethnic identification was positively correlated with mainstream sociocultural identification. This study provides support for the importance of eating disorder prevention interventions that focus specifically on African-American girls.


Introduction
e research on eating disorders (EDs) in minorities, specifically African-Americans, is scarce. Even rarer is literature that examines the cultural and media in�uences of eating pathology in this population, speci�cally in those from urban areas. ere is also little research on African-American middle school and early high school age females. Few studies focus on middle school adolescents [1,2]. Many of the studies on general EDs and media in�uences on EDs focus on college students [3,4]. Some �ndings suggest that within minority populations, eating disorder (ED) development is considered rare (e.g., [5]). However, recently such �ndings have come into question [6] and few studies have used solely minority populations in investigating ED risk [4,5].
e Diagnostic and Statistical Manual of Mental Disorders Fourth Edition-Text Revision (DSM-IV-TR) describes three types of EDs, anorexia nervosa (AN), bulimia nervosa (BN), and Eating Disorder-Not Otherwise Speci�ed (ED-NOS). In AN, individuals refuse to sustain normal body weight;, in BN, individuals binge eat and compensate for binging using inappropriate methods (e.g., self-induced vomiting, excessive exercise), in order to maintain weight, and in ED-NOS, individuals do not meet the complete criteria for an ED but display features of either AN or BN [7]. AN and BN are found to be less common in African-American youth (In their review of the literature, Anthony and Yager [8] stated that various studies), [9]. However, the aforementioned studies did not speci�cally focus on middle school and early high school age African-American females living in an urban area.
Compared with Caucasians, Binge Eating Disorder (BED) is shown to be the most common ED among African American adolescents and adults [10]. BED is the consumption of large quantities of food in a short period of time accompanied by a feeling of lack of control over eating behavior [11]. us, investigating BED behavior in addition to AN and BN behaviors is important.
Notably, a large part of ED development in America is exposure to the media [2,12]. Particularly, African-American females watch more television than White females [13]. American culture uses the media to promote a thin ideal, which may be a risk factor in the development of EDs, speci�cally for minorities [2]. Much of the research examining African-American female youth has compared them to Caucasians [13]. us, more research needs to be done on media impact speci�cally in African-American girls living in an urban area.
Nevertheless, despite the dearth of research about this speci�c population, the �eld is making progress in targeting minority females living in urban areas [14]. ere have been many gains in the �eld of EDs in examining cultural and media in�uences, as well as risk and protective factors in ED symptom and disease development [15]. However, there is still much room for improvement in examining African-American females living in urban areas and investigating the correlates of media and cultural in�uences. us, due to gaps in the aforementioned literature, there was a need to conduct research that utilized an all minority, female, middle school to early high school age sample living in an urban area. For the purposes of this study, African-American girls were the focus.

Cultural Manifestations of Eating Disorders
In assessing cultural aspects of EDs in minority populations, a review of research has shown that compared to Caucasian females, eating pathology is more common among Native American females, equally common among Hispanic females, and less common in Asian American and Black adolescents [16]. However, those minorities with a high acculturation to Western culture "the process of shiing values to the host culture from the culture of origin" [12, page 19] are possibly more at risk for developing EDs because they have internalized the thin ideal notion of beauty [12].
In a study of seven African-American and Latina adolescent females, high acculturation to Western culture contributed to eating disorder development [5]. Other research examined internalization of American values in college women [3,17]. Internalization of American values was found to be related to identi�cation with White culture and less satisfaction with body shape and size. A combination of DT, physical self-concept, and personal competence greatly predicts body dissatisfaction (BD) [3,17]. is �nding is consistent with a study of young black, Hispanic, and White women and men. Demarest and Allen [18] found that African American females (mean dissatisfaction score = 9.4, SD = 14.4) desired thinner bodies and were not more satis�ed with their bodies than Caucasian females (mean dissatisfaction score = 13.5, SD = 13.1). However, none of these aforementioned studies examined African-American middle school to early high school age students living in an urban setting.

�. Media �n�uences on Eating Disorders
�ery few studies have speci�cally examined African-American female adolescents and media in�uences, highlighting the need for more research in this population. One example is the aforementioned Botta [13] that examined television watching in 33 Black and 145 White female adolescents, �nding that Black girls participated in more disordered eating behaviors than White girls. e more adolescents viewed television shows loaded with thin images, the more likely they participated in disordered eating behaviors, the more dissatis�ed they were with their bodies, and the greater their drive for thinness (DT) [13]. e closest to a causal link between media exposure and disordered eating behavior is studies done in Fiji on ethnic Fijian girls. ese studies examined the impact introduction of television in the country (where television had not previously been) on adolescent Fijian high school girls' body image perception and development of eating disordered behaviors [6,19]. Results revealed a signi�cant increase in the salience of selfinduced purging behavior for weight loss and high Eating Attitudes Test (EAT) scores following extended television exposure in Fiji in 63 girls in 1995 and 65 girls in 1998 [6,19].

Body Dissatisfaction.
In minority females, evidence for the in�uence of thin ideal internalization on BD has been found [20,21]. DT predicts BD in minorities [3,17,18]. Furthermore, BD has been found to be the single strongest predictor of ED symptomatology [22]. Higher acculturation to Western values can cause minority adolescent females to internalize thin media �gures [12].

Drive for inness.
In both White and Black adolescent girls, fat has been found to be related to DT. In African American girls, it has been a predictor of criticisms about weight [23]. For young women without AN or BN, greater emphasis on shape and weight has been found to predict DT in females [24]. us, investigating DT in African-American girls is important in understanding how it affects eating behavior.

Summary and Hypotheses
is study aimed to examine media and cultural in�uences in ED development in African-American, female, middle school to early high school age girls living in an urban area. It was hypothesized that (a) mainstream sociocultural identi�cation will be associated with more ED behavior in African-American females; (b) cultural ethnic identi�cation will be associated with less ED behavior in African-American females; (c) mainstream sociocultural identi�cation, cultural ethnic identi�cation, and BD will signi�cantly predict ED behavior in African-American females; and (d) cultural ethnic identi�cation will be negatively correlated with mainstream sociocultural identi�cation in African-American females. e in�uence of body mass index (BMI) was also examined in relation to ED development in this study, although no speci�c hypothesis was proposed regarding this in�uence.

Participants. Participants (
) were recruited from community organizations, churches, and neighborhoods in an urban Midwest City and an urban east coast city. Of these ISRN Preventive Medicine 3 57 participants, 50 ( ) met criteria for participation in the study. All of the quali�ed participants self-identi�ed as African-American. Of the seven girls did not meet criteria for this study, four self-identi�ed as Hispanic, one self-identi�ed as Native American, one was African-American but did not have proper parent consent, and one was a nine-year-old African-American girl who also did not have proper parent consent.

Measures
6.2.1. Demographic Questionnaire. is questionnaire asked participants' current age, grade, height, weight, race/ethnicity, and parents'/guardians' highest level of education.

Eating Disorder Inventory-3. e Eating Disorder
Inventory-3 (EDI-3) [25] is a measure utilized to detect individuals at risk for EDs. ree subscales of the EDI-3 were used: DT, Bulimia, and BD. DT measures excessive worries about dieting, �xation with weight, and involvement in an intense quest of thinness. Bulimia measures the tendency toward binge eating followed by self-induced vomiting and other compensation methods. BD measures the belief that particular body parts (e.g., thighs, stomach, and hips) are too big [25]. Reliability was established on a sample of American adolescents, international adults, and American adults. Reliability for the DT and BD scales ranged from 0.90 to 0.97 across the three groups. Alpha Coefficients of the other subscales had medians of 0.74, 0.84, and 0.85. Internal consistency was found to be good. Construct validity at its highest was 0.83. e DT and Bulimia scales ask questions such as "I stuff myself with food" and "If I gain a pound, I worry that I will keep gaining" [26]. Although the EDI-3 was designed for females ages 13-53, a previous version of the EDI has been successfully used with younger adolescents [22], and the scales are written at a fourth to sixth grade reading level [26]. For this study, Cronbach's alpha is 0.77 for the DT scale, 0.72 for the Bulimia scale, and 0.83 for the BD scale.

Adolescent Version of the Questionnaire of Eating and
Weight Patterns. e adolescent version of the Questionnaire of Eating and Weight Patterns (QEWP-A) [11] is a 12question self-report measure designed to assess BED in individuals. Psychometric properties of the QEWP-A were assessed on a sample of 6th to 12th graders ranging in age from 10 to 18 years. Construct validity was adequate when compared with other measures of eating behaviors [27]. Questions asked include, "During the past 6 months, did you ever eat what most people, like your friends, would think was a really big amount of food?" [27, page 306] and "Did you ever not eat anything at all for at least 24 hours (a full day) to keep from gaining weight aer eating a really big amount of food?" [27, page 308]. For the purposes of this study, this scale was split into two separate scales, binge behaviors and compensatory behaviors. is was done in order to distinguish binge eating behaviors from compensatory methods. For this study, the Cronbach's alpha for the QEWP-A is 0.70.

Sociocultural Attitudes Towards Appearance Questionnaire-3. e Sociocultural Attitudes Towards Appearance
Questionnaire-3 (SATAQ-3) [28], which looks at media and social in�uences on appearance and body image, was originally normed with college women (found was good convergent validity, sufficient internal consistency, and replicable factor structure) but was assessed on middle school boys and girls [29]. e SATAQ-3 is a 30-question scale with two subscales, Awareness and Internalization of a thin body ideal. �irls in the study were given a modi�ed version of the SATAQ-3. ese subscales displayed sufficient internal consistency, adequate validity, and strong correlations with body image and the use of weight control methods, particularly the Internalization subscale. However, the scale assessed a nearly all White samples [29]. Nevertheless, for the current study, reliability was examined for African-American middle school to early high school age girls. For this study, the Cronbach's alpha for the SATAQ-3 is 0.89.

Multigroup Ethnic Identity Measure. e Multigroup
Ethnic Identity Measure (MEIM) [30], a 14-item questionnaire examining ethnic affirmation and belonging ethnic identity achievement, ethnic behaviors, and 6 additional questions on group orientation, was developed to measure ethnic identity in African American, Native American, Caucasian, Asian-American, and Hispanic adolescents and adults [30,31]. Cronbach's alpha found reliability to be 0.81 for the adolescents and 0.90 for college students assessed [30]. Two subscales of the MEIM have been assessed in African American middle school girls (ages from 11 to 14 years), Ethnic Identity Exploration and Ethnic Identity Commitment. Found was good reliability and internal consistency with the Ethnic Identity Commitment subscale. e subscale was also found to be replicable [32]. For this study, the Cronbach's alpha for the MEIM is 0.80.

Statistical Analysis.
Data was analyzed via simple linear regression, which determined a relationship between the independent and dependent variables of each study hypothesis. Hierarchical linear regression was also used examining stepwise regression. Linear regression was used because oen this method can be used to predict the value of one variable from the value of another [33]. Speci�cally, Pearson Product Moment Correlation Coefficient was used due to its ability to denote both the strength and the direction of the relationship between two variables [34]. A correlation matrix was constructed for the data (see Table 1).

Hypothesis
One. e correlation between mainstream sociocultural identi�cation and DT was positive and significant ( , ). As the mainstream sociocultural identi�cation increases, DT increases, and there is a signi�cant relationship between the two, ; 3 85; . Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in DT, 2 2 2; 8 2 8; . e correlation between mainstream sociocultural identi�cation and BD was positive and signi�cant ( 3 5, ). As the mainstream sociocultural identi�cation increases, BD increases, and there is a signi�cant relationship between the two, 3 5; 2 3; . Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in BD, 2 33; 8 3 3; . e correlation between mainstream sociocultural identi�cation and BED symptomatology was positive and signi�cant ( 28 , 8). As mainstream sociocultural identi�cation increases, BED symptomatology increases, and there is a signi�cant relationship between the two, 28 ; 2 3 ; 8. Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in BED symptomatology, 2 ; 8 25; 8. In examining the splitting of the QEWP-A questionnaire into two questionnaires, binge behavior and compensatory behaviors, the correlation between mainstream sociocultural iden-ti�cation and compensatory behaviors was positive and sig-ni�cant ( 5 , ). As mainstream sociocultural identi�cation increases, compensatory behaviors increases, and there is a signi�cant relationship between the two, 5 ; 53 ; . Mainstream sociocultural iden-ti�cation explained a signi�cant proportion of variance in compensatory behaviors, 2 2 ; 8 2 522; . However, there was neither signi�cant correlation between mainstream sociocultural identi�cation and binge eating behaviors nor a signi�cant correlation between mainstream sociocultural identi�cation and Bulimia (see Table 2).
A hierarchical linear regression was also done of hypothesis one. Two steps were entered. e �rst step was BMI because the purpose was to control for this variable. e second step was mainstream sociocultural identi�cation. BMI was a signi�cant predictor of DT (  because it was deemed the most important variable in ED symptomatology. Results revealed that mainstream sociocultural identi�cation was a signi�cant predictor of DT ( , ), and there is a signi�cant relationship between the two, ; 2 ; , even aer controlling for BMI and cultural ethnic identi�cation. Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in DT, Δ 2 2 , . Results also revealed that BD was still a signi�cant predictor of DT ( , 2 ), and there is a signi�cant relationship between the two, ; 2 2 ; 2 , above and beyond the effects of BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation. BD explained a signi�cant proportion of variance in DT, Δ 2 , 2 . Mainstream sociocultural identi�cation was a signi�cant predictor of BED symptomatology ( , ). Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in BED symptomatology, Δ 2 2 , . However, mainstream sociocultural identi�cation was not a signi�cant predictor of BED symptomatology aer controlling for BMI and cultural ethnic identi�cation. BD was also not a signi�cant predictor of BED symptomatology aer controlling for BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation.
Mainstream sociocultural identi�cation was a signi�cant predictor of compensatory behaviors ( 2, ), and there is a signi�cant relationship between the two, 2 ; 2 ; , above and beyond the effects for BMI and cultural ethnic identi�cation. Mainstream sociocultural identi�cation explained a signi�cant proportion of variance in compensatory behaviors, Δ 2 2 , . Mainstream sociocultural identi�cation was not a sig-ni�cant predictor of Bulimia even aer controlling for BMI and cultural ethnic identi�cation. BD was also not a significant predictor of Bulimia even aer controlling for BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation. BD did not predict BED symptomatology aer controlling for BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation. Mainstream sociocultural identi�cation was not a signi�cant predictor of binge behavior even aer controlling for BMI and cultural ethnic identi�cation. BD was also not a signi�cant predictor of binge behavior even aer controlling for BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation. Lastly, BD did not predict compensatory behaviors above and beyond BMI, cultural ethnic identi�cation, and mainstream sociocultural identi�cation. was positive and signi�cant ( 2,2). As cultural ethnic identi�cation increases, mainstream sociocultural identi�cation increases, and there is a signi�cant relationship between the two, 2; ; 2. Cultural ethnic identi�cation explained a signi�cant proportion of variance in mainstream sociocultural identi-�cation, 2 8 ; 8 ; 2 (see Table 3).

Discussion
To date, to the researcher's knowledge, no research study has examined media and cultural in�uences in African-American middle school to early high school age girls living in an urban setting. us, this study has contributed to the existing, albeit limited, literature on African-American females by showing the effects of acculturation on African-American females residing in urban settings. Similar in this study to other studies is the sample size being comparable to other studies that utilize Black girls, particularly in gathering approximately 50 participants [35,36]. e SES of participants and the utilization of middle school to high school age adolescents are also comparable to the current study [36][37][38]. e results from the current study suggest that in African-American girls living in an urban, primarily lower-middle class setting, the perceptions of American beauty standards and thin ideal may arise similarly as in upper-middle class, suburban minority populations [5,12].
Nevertheless, in this study there was neither signi�cant correlation between mainstream sociocultural iden-ti�cation and solely binge eating behaviors nor any sig-ni�cance between mainstream sociocultural identi�cation and Bulimia. e former �nding contradicts �ndings that heavy television viewing (i.e., images encouraging viewers to consume and purchase unhealthy foods) in children is related to binge eating behavior [41], while the latter �nding contradicts �ndings that mainstream sociocultural identi�cation is related to Bulimia symptoms in young girls [42]. Perhaps the preponderance of thin ideal images in the media are more in�uential in compensatory and dieting behaviors for this population, consistent with the strong effect of these images in other research [41]. Furthermore, in Bulimia, other research has shown that exposure to overweight television and magazine �gures is correlated with Bulimia [40]. us, exposure to thin images may have overridden the effect of exposure to overweight images in this sample [12].
No support was found in cultural ethnic identi�cation being associated with less ED behavior in African-American females, contradicting research that suggests that minority girls with less identi�cation with Western culture have lower risk of ED development [5,12]. Also contrary to predictions, cultural ethnic identi�cation was positively correlated with mainstream sociocultural identi�cation. is �nding is especially puzzling in light of research and theory that suggests that cultural ethnic identi�cation is a protective factor in rejecting Western thin ideal [5,43]. However, in this sample, none of these girls were from outside of Western culture. us, they likely endorsed the values of American culture. Individuals who have a high acculturation to American culture are possibly more at risk for developing ED symptomatology because they have internalized the thin ideal notion of beauty [12].

Limitations
ere are several limitations of this study that should be noted. e �rst limitation is that this sample was a volunteer sample recruited through predominately African-American groups. is recruitment method might have affected the results in the effect of cultural ethnic identi�cation in this sample. For example, the range of scores ( 2 , SD 8) may have been restricted due to the degree of cultural identi�cation manifest in girls strongly associated with a cultural group. Without a wide range in responses, the results may have been attenuated. Nevertheless, other studies examining this hard-to-reach population have recruited a nonrandom sample of urban youth from community organizations serving these speci�c youth and Black neighborhoods [44,45].
A second limitation is that a self-report measure was used, which could have affected accuracy of demographic information (e.g., height, weight, and parents' education). is method may also have lent itself to response bias. A third limitation is that BMI was not collected on all of the girls (24%), which could have in�uenced the results BMI had on ED symptomatology in this study. A fourth limitation of this study is that even girls who identi�ed as African-American and one or more other racial group were counted as African-American for the purposes of this study. Biracial or multiethnic identity could have impacted �ndings in some unknown way (20% of girls identi�ed as Biracial or multiethnic�. A ��h limitation of this study is that the age range of the girls varied. is sample represented the range of adolescence. Future studies should look at stages of adolescence as a possible factor. Finally, this sample was derived from a generally urban setting. As a result, �ndings may not be generalizable to African-American girls in other demographics.

Implications
is study has implications for working with African-American girls in preventing ED behavior, speci�cally in school-based groups. Many ED prevention mechanisms for adolescent girls are school-based groups which can reach a large number of girls [46], and several have found some effectiveness in preventing ED behavior in adolescent girls [38]. However, most of these groups do not focus speci�cally on African-American adolescent girls [38,47]. Nevertheless, one study suggests that minority and nonminority middle school girls do not differ in response to an ED prevention group [36].

Conclusion
Overall, further research is needed in this area in investigating ED prevention in African-American middle school to early high school age girls living in an urban setting. e current study shows the importance of mainstream sociocultural identi�cation in the development of ED symptomatology and how cultural ethnic identi�cation is not as important when girls are already part of a set cultural ethnic group. Essentially, this study is a small step in alerting researchers to the need for more studies in this area.